You are on page 1of 673

The Oxford Handbook of Depression and Comorbidity



Peter E.Nathan

Area Editors:

Clinical Psychology
David H.Barlow

Cognitive Neuroscience
Kevin N.Ochsner and Stephen M.Kosslyn

Cognitive Psychology
Daniel Reisberg

Counseling Psychology
Elizabeth M.Altmaier and Jo-Ida C.Hansen

Developmental Psychology
Philip DavidZelazo

Health Psychology
Howard S.Friedman

History of Psychology
David B.Baker

Methods and Measurement

Todd D.Little

Kenneth M.Adams

Organizational Psychology
Steve W.J. Kozlowski

Personality and Social Psychology

Kay Deaux and MarkSnyder

Editor in Chief peter e. nathan

The Oxford Handbook

of Depression and
Edited by
C. Steven Richards and Michael W.OHara

Oxford University Press is a department of the Universityof
Oxford. It furthers the Universitys objective of excellence in research,
scholarship, and education by publishing worldwide.

AucklandCape TownDar es SalaamHong KongKarachi
Kuala LumpurMadridMelbourneMexico CityNairobi
New DelhiShanghaiTaipeiToronto

With officesin
ArgentinaAustriaBrazilChileCzech RepublicFranceGreece
South KoreaSwitzerlandThailandTurkeyUkraineVietnam

Oxford is a registered trademark of Oxford UniversityPress

in the UK and certain other countries.

Published in the United States of Americaby

Oxford UniversityPress
198 Madison Avenue, NewYork, NY10016

Oxford University Press2014

All rights reserved. No part of this publication may be reproduced, storedin
a retrieval system, or transmitted, in any form or by any means, without theprior
permission in writing of Oxford University Press, or as expressly permitted bylaw,
by license, or under terms agreed with the appropriate reproduction rights organization.
Inquiries concerning reproduction outside the scope of the above should be senttothe
Rights Department, Oxford University Press, at the addressabove.

You must not circulate this work in any otherform

and you must impose this same condition on any acquirer.

A copy of this books Catalog-in-Publication Data is on file with the Library of Congress
ISBN 9780199797004

Printed in the United States of America
on acid-freepaper

Oxford Library of Psychology vii

About the Editors ix


Contributors xiii

Contents xvii

Chapters 1622



The Oxford Library of Psychology, a landmark series of handbooks, is published by Oxford

University Press, one of the worlds oldest and most highly respected publishers, with a
tradition of publishing significant books in psychology. The ambitious goal of the Oxford
Library of Psychology is nothing less than to span a vibrant, wide-ranging field and, in so
doing, to fill a clear marketneed.
Encompassing a comprehensive set of handbooks, organized hierarchically, the Library
incorporates volumes at different levels, each designed to meet a distinct need. At one level
are a set of handbooks designed broadly to survey the major subfields of psychology; at
another are numerous handbooks that cover important current focal research and scholarly
areas of psychology in depth and detail. Planned as a reflection of the dynamism of psychol-
ogy, the Library will grow and expand as psychology itself develops, thereby highlighting
significant new research that will affect the field. Adding to its accessibility and ease of use,
the Library will be published in print and, later on, electronically.
The Library surveys psychologys principal subfields with a set of handbooks that cap-
ture the current status and future prospects of those major subdisciplines. This initial set
includes handbooks of social and personality psychology, clinical psychology, counseling
psychology, school psychology, educational psychology, industrial and organizational psy-
chology, cognitive psychology, cognitive neuroscience, methods and measurements, history,
neuropsychology, personality assessment, developmental psychology, and more. Each hand-
book undertakes to review one of psychologys major subdisciplines with breadth, compre-
hensiveness, and exemplary scholarship. In addition to these broadly conceived volumes,
the Library also includes a large number of handbooks designed to explore in depth more
specialized areas of scholarship and research, such as stress, health and coping, anxiety and
related disorders, cognitive development, and child and adolescent assessment. In contrast
to the broad coverage of the subfield handbooks, each of these latter volumes focuses on an
especially productive, more highly focused line of scholarship and research. Whether at the
broadest or most specific level, however, all of the Library handbooks offer synthetic cover-
age that reviews and evaluates the relevant past and present research and anticipates research
in the future. Each handbook in the Library includes introductory and concluding chapters
written by its editor to provide a roadmap to the handbooks table of contents and to offer
informed anticipations of significant future developments in thatfield.
An undertaking of this scope calls for handbook editors and chapter authors who are
established scholars in the areas about which they write. Many of the nations and worlds
most productive and best-respected psychologists have agreed to edit Library handbooks or
write authoritative chapters in their areas of expertise.
For whom has the Oxford Library of Psychology been written? Because of its breadth,
depth, and accessibility, the Library serves a diverse audience, including graduate students
in psychology and their faculty mentors as well as scholars, researchers, and practitioners in
psychology and related fields. All will find in the Library the information they seek on the
subfield or focal area of psychology in which they work or are interested.
Befitting its commitment to accessibility, each handbook includes a comprehensive
index as well as extensive references to help guide research. And because the Library was
designed from its inception as an online as well as a print resource, its structure and contents

will be readily and rationally searchable online. Further, once the Library is released online,
the handbooks will be regularly and thoroughly updated.
In summary, the Oxford Library of Psychology will grow organically to provide a thor-
oughly informed perspective on the field of psychology, one that reflects both psychologys
dynamism and its increasing interdisciplinarity. Once published electronically, the Library
is also destined to become a uniquely valuable interactive tool, with extended search and
browsing capabilities. As you begin to consult this handbook, we sincerely hope you will
share our enthusiasm for Oxford University Presss more than 500-year tradition of excel-
lence, innovation, and quality, as exemplified by the Oxford Library of Psychology.

Peter E.Nathan
Oxford Library of Psychology

viii oxford library of psychology


C. Steven Richards
C. Steven Richards is a professor of psychology and director of graduate studies at Texas Tech
University. His primary research interests include depression, relapse prevention, comorbid-
ity, self-control, and health psychology. He has held fourteen administrative positions dur-
ing his career, at three different universities:Texas Tech University, Syracuse University, and
the University of MissouriColumbia. He has over sixty publications, numerous conven-
tion presentations, andworking with his collaboratorshas attracted grant funding from
state, regional, and federal agencies along with private foundations and societies.

Michael W. OHara
Michael W. OHara is a professor of psychology and director of clinical training at The
University of Iowa. His primary research interests include depression, interpersonal psycho-
therapy, womens health promotion, postpartum depression, comorbidity, health psychol-
ogy, and several issues in psychological assessment. He has held numerous administrative
positions at The University of Iowa and has over 140 publications, numerous convention
presentations, workshop presentations at many national and international sites, and
working with his collaboratorshas attracted grant funding from state, regional, and fed-
eral agencies along with private foundations and societies.


We thank the contributors for their excellent chapters. We appreciate the support of our
home universities, Texas Tech University and The University of Iowa, and the expert help
from Oxford University Press. Anumber of colleagues have provided helpful encourage-
ment on this and associated tasks, including Lee Cohen, Sheila Garos, Susan Hendrick,
Mike Perri, and Peter Nathan. We also want to thank our families and friends for helping
us, including Steves wife Carol Richards, Dawn Bush, Jill Richards, Jeffrey and Andrew
OHara, and Mikes wife, Jane Engeldinger. We greatly appreciate the help and encourage-
ment of these people. Thankyou.


Jamie L.Abaied Sara J. Bufferd

Department of Psychology Department of Psychology
University of Vermont Stony Brook University
Burlington,VT Stony Brook,NY
DeepikaAnand Melissa M. Buttner
Department of Psychiatry Department of Psychology
Indiana University School of Medicine The University ofIowa
University Hospital Iowa City,IA
Indianapolis,IN Dallas A. Callaway
Tracey L.Auster Department of Psychology
Department of Psychology Louisiana State University
Louisiana State University Baton Rouge,LA
Baton Rouge,LA Deborah M. Capaldi
DanielBE Oregon Social LearningCenter
Department of Psychology and Neuroscience Eugene,OR
University of Colorado, Boulder Nicole M. Capezza
Boulder,CO University of Massachusetts
VickieBhatia Medical School
Department of Psychology Worcester,MA
Stony Brook University Department of Psychology
Stony Brook,NY Stonehill College
Aaron J. Blashill Easton, MA
Department of Psychiatry Colleen E.Carney
Harvard MedicalSchool Department of Psychology
Massachusetts General Hospital Ryerson University
The Fenway Institute Toronto, Ontario,Canada
FenwayHealth Natalie Castriotta
Boston,MA Department of Psychology
Rhonda C.Boyd University of CaliforniaLos Angeles
Childrens Hospital of Philadelphia Los Angeles,CA
School of Medicine Alan J. Christensen
University of Pennsylvania Department of Psychology
Philadelphia,PA The University ofIowa
Jessica A. Brommelhoff Iowa City,IA
Brain, Behavior, and Aging ResearchCenter Lee AnnaClark
Veterans Affairs Greater Los Angeles Department of Psychology
HealthcareSystem University of NotreDame
University of CaliforniaLos Angeles Notre Dame,IN
Los Angeles,CA Alex S.Cohen
Jared M.Bruce Department of Psychology
Department of Psychology Louisiana State University
The University of MissouriKansasCity Baton Rouge,LA
Kansas City,MO

Michelle G.Craske Division of Geriatrics and Gerontology
Department of Psychology School of Medicine
University of CaliforniaLos Angeles Johns Hopkins University
Los Angeles,CA BaltimoreMD
Kelly C. Cukrowicz Janna R.Gordon
Department of Psychology Department of Psychiatry
Texas Tech University Harvard MedicalSchool
Lubbock,TX Massachusetts General Hospital
Karina W. Davidson Boston,MA
Department of Medicine Whitney L.Gore
College of Physicians and Surgeons Department of Psychology
Columbia University University of Kentucky
NewYork,NY Lexington, KY
JoanneDavila Stevie Grassetti
Department of Psychology Department of Psychology
Stony Brook University University of Delaware
Stony Brook,NY Newark,DE
Sona Dimidjian Lauren M. Greenberg
Department of Psychology and Neuroscience Department of Psychology
University of Colorado, Boulder Drexel University
Boulder,CO Philadelphia,PA
Gareth R.Dutton Laura M. Hancock
Division of Preventive Medicine Department of Psychology
University of Alabama at Birmingham The University of MissouriKansasCity
Birmingham,AL Department of Neurology
Nicholas R.Eaton The University of Kansas MedicalCenter
Department of Psychology Kansas City,MO
Stony Brook University Rachel Hershenberg
Stony Brook,NY Department of Psychology
Brian A. Feinstein Stony Brook University
Department of Psychology Stony Brook,NY
Stony Brook University Lauren A. Holland
Stony Brook,NY Department of Psychology
Jennifer N.Felder Florida State University
Department of Psychology and Neuroscience Tallahassee,FL
University of Colorado, Boulder Derek R.Hopko
Boulder,CO Department of Psychology
Jamie Feldman The University of TennesseeKnoxville
Department of Family Medicine and Knoxville,TN
CommunityHealth Robert H. Howland
Program in Human Sexuality University of Pittsburgh School of Medicine
University of Minnesota Western Psychiatric Institute andClinic
Minneapolis,MN Pittsburgh,PA
Michael Gawrysiak Pamela K.Keel
Department of Psychology Department of Psychology
The University of Tennessee Florida State University
Knoxville,TN Tallahassee,FL
Laura N.Gitlin Quinn D. Kellerman
Center for Innovative Care inAging Department of Psychology
Department of Community PublicHealth The University ofIowa
School of Nursing Iowa City,IA
Department of Psychiatry

xiv Contributors
Ronald C. Kessler Matthew J. Mimiaga
Department of Health CarePolicy Departments of Psychiatry
Harvard MedicalSchool and Epidemiology
Boston,MA Harvard MedicalSchool
HyounK.Kim The Fenway Institute
Oregon Social LearningCenter FenwayHealth
Eugene,OR Boston,MA
Daniel N.Klein SusanMineka
Department of Psychology Department of Psychology
Stony Brook University Northwestern University
Stony Brook,NY Evanston,IL
Robert F. Krueger Taryn G.Moss
Department of Psychology Department of Psychology
University of Minnesota Ryerson University
Minneapolis,MN Toronto, Ontario,Canada
Julia K.Langer Lisa M. Najavits
Department of Psychology Veterans Affairs Boston Healthcare
Washington University in SaintLouis System
St. Louis,MO Department of Psychiatry
Ma. AsuncinLara Boston University School of
Ramn de la Fuente National Institute of Medicine
Psychiatry Boston,MA
Mxico City,Mexico Belinda L. Needham
KarinLarsen Department of Epidemiology
Department of Obstetrics, Gynecology, and University of Michigan
WomensHealth Ann Arbor,MI
University of Minnesota Arthur M.Nezu
Minneapolis,MN Department of Psychology
Huynh-NhuLe Drexel University
Department of Psychology Philadelphia,PA
George Washington University Christine MaguthNezu
Washington,DC Department of Psychology
Abigail M. Lindemann Drexel University
Department of Psychology Philadelphia,PA
University of WisconsinMadison Michael W.OHara
Madison,WI Department of Psychology
Sharon G.Lynch The University ofIowa
Department of Neurology Iowa City,IA
The University of Kansas Erin K. Poindexter
MedicalCenter Department of Psychology
Kansas City,KS Texas Tech University
Kristian E.Markon Lubbock,TX
Department of Psychology C. Steven Richards
The University ofIowa Department of Psychology
Iowa City,IA Texas Tech University
Crystal C. McIndoo Lubbock,TX
Department of Psychology EunyoeRo
The University of TennesseeKnoxville Department of Psychology
Knoxville,TN University of NotreDame
Notre Dame,IN

Contributors xv
Michael E. Robinson JerrySuls
Department of Clinical and Health Psychology Behavioral Research Program
College of Public Health and Health Professions Division of Cancer Control & Populations
University of Florida Sciences
Gainesville,FL National Cancer Institute
Thomas L. Rodebaugh Bethesda, MD
Department of Psychology Jennifer A. Sumner
Washington University in St.Louis Department of Psychology
St. Louis,MO Northwestern University
DanielaRoditi Evanston,IL
Sleep ResearchLab Julia R. VanLiew
Department of Clinical and Health Psychology Departments of Psychology
College of Public Health and Health Professions The University ofIowa
University of Florida Iowa City,IA
Gainesville,FL Anna VanMeter
Karen D. Rudolph Department of Psychology
Department of Psychology University of North CarolinaChapelHill
University of Illinois Chapel Hill,NC
Champaign,IL David Watson
Steven A.Safren Department of Psychology
Department of Psychiatry University of NotreDame
Harvard Medical School/Massachusetts General Notre Dame,IN
Hospital Lori B. Waxenberg
The Fenway Institute Department of Clinical and Health
FenwayHealth Psychology
Boston,MA College of Public Health and Health
KristinE. Salber Professions
Department of Psychology University of Florida
Drexel University Gainesville,FL
Philadelphia,PA Mark A. Whisman
Kate M.Scott Department of Psychology and
Department of Psychological Medicine Neuroscience
Dunedin School of Medicine University of Colorado, Boulder
University ofOtago Boulder,CO
North Dunedin, New Zealand Thomas A. Widiger
VictoriaShahly Department of Psychology
Department of Health CarePolicy University of Kentucky
Harvard MedicalSchool Lexington,KY
Boston,MA Katie Witkiewitz
Sara M.Stasik Department of Psychology
Department of Psychology Washington State University, Vancouver
University of NotreDame Vancouver,WA
Notre Dame,IN Eric Youngstrom
Connie Stauffer Department of Psychology
Department of Psychology University of North CarolinaChapelHill
Washington State University, Vancouver Chapel Hill,NC
Vancouver,WA Alan M. Zaslavsky
Catherine B.Stroud Department of Health CarePolicy
Department of Psychology Harvard MedicalSchool
Williams College Boston,MA

xvi Contributors

1. Introduction 1
C. Steven Richards and Michael W. OHara
2. Diagnostic and Statistical Manual (DSM) 11
Thomas A. Widiger and Whitney L. Gore
3. From Comorbidity to Constructs:Recurring and Emergent
Issues in Modeling Comorbidity 29
Kristian E. Markon
4. Examining the Comorbidity Between Depression and the
Anxiety Disorders From the Perspective of the Quadripartite Model 46
David Watson and Sara M. Stasik
5. Depressive Disorders, Comorbidity Issues, and Assessment Strategies 66
Nicholas R. Eaton and Robert F. Krueger
6. Depression and Comorbidity with Panic Disorder 84
Natalie Castriotta and Michelle G. Craske
7. Depression and Posttraumatic Stress Disorder Comorbidity 100
Lisa M. Najavits and Nicole M. Capezza
8. Comorbidity of Social Anxiety Disorder and Depression 111
Julia K. Langer and Thomas L. Rodebaugh
9. Important Issues in Understanding Comorbidity Between Generalized
Anxiety Disorder and Major Depressive Disorder 129
Susan Mineka, Deepika Anand, and Jennifer A. Sumner
10. Depression and Alcohol Use 148
Katie Witkiewitz and Connie Stauffer
11. Eating Disorders 166
Pamela K. Keel and Lauren A. Holland
12. Comorbidity of Depression and Conduct Disorder 186
Deborah M. Capaldi and Hyoun K. Kim
13. Depression and Comorbidity:Personality Disorder 200
Daniel N. Klein, Sara J. Bufferd, Eunyoe Ro, and Lee Anna Clark
14. Sexual Dysfunction 218
Jamie Feldman and Karin Larsen
15. Schizophrenia 236
Alex S. Cohen, Dallas A. Callaway, and Tracey L. Auster

16. Suicide 254
Kelly C. Cukrowicz and Erin K. Poindexter
17. Comorbidity of Bipolar Disorder and Depression 268
Eric Youngstrom and Anna Van Meter
18. Depression and Cardiovascular Diseases 287
Jerry Suls and Karina W. Davidson
19. Cancer and Depression 302
Arthur M. Nezu, Christine Maguth Nezu, Lauren M. Greenberg,
and Kristin E. Salber
20. Pain 319
Daniela Roditi, Lori B. Waxenberg, and Michael E. Robinson
21. Obesity 335
Gareth R. Dutton and Belinda L. Needham
22. Sleep Disorders and Depression 349
Colleen E. Carney and Taryn G. Moss
23. Multiple Sclerosis 367
Laura M. Hancock, Jared M. Bruce, and Sharon G. Lynch
24. HIV/AIDS and Depression 382
Aaron J. Blashill, Janna R. Gordon, Matthew J. Mimiaga, and
Steven A. Safren
25. Depression in Chronic Kidney Disease:AContext for Comorbidity 398
Alan J. Christensen, Julia R. Van Liew, and Quinn D. Kellerman
26. Depression in Dementia Syndromes 411
Jessica A. Brommelhoff
27. Womens Health:Comorbidity of Depression and Type 2 Diabetes,
Fibromyalgia, and Rheumatoid Arthritis 427
Melissa M. Buttner and Michael W. OHara
28. Intimate Relationships 441
Catherine B. Stroud, Brian A. Feinstein, Vickie Bhatia, Rachel Hershenberg,
and Joanne Davila
29. Family Relationships, Emotional Processes, and Adolescent Depression 460
Jamie L. Abaied and Karen D. Rudolph
30. Perinatal Depression 476
Jennifer N. Felder, Abigail M. Lindemann, and Sona Dimidjian
31. Multidisciplinary Treatments and Medications for Depressive Disorders
Robert H. Howland
32. The Role of Community- and Home-Based Interventions
in Late-LifeDepression 511
Laura N. Gitlin

xviii contents
33. Treatment of Depressive Disorders and Comorbidity in Ethnic
Huynh-Nhu Le, Rhonda C. Boyd, and Ma. Asuncin Lara
34. Psychosocial Interventions for Depressed Breast Cancer Patients 546
Derek R. Hopko, Crystal C. McIndoo, Michael Gawrysiak,
and Stevie Grassetti
35. Cognitive Therapy for Comorbid Depression 584
Mark A. Whisman and Daniel BE
36. The Big Picture 599
Ronald C. Kessler, Kate M. Scott, Victoria Shahly, and Alan M. Zaslavsky
37. Epilogue 615
Michael W. OHara and C. Steven Richards

Index 623

contents xix

1 Introduction

C. Steven Richards and Michael W.OHara

In this chapter, we provide a brief introduction to our book. We discuss the following themes, which run
throughout this edited book on depressive disorders and comorbidity:assessment and diagnosis, theory
and methods, psychiatric comorbidity, health comorbidity, relationship comorbidity, intervention and
consultation, and future directions. Anumber of themes will be apparent, including the incredibly broad
scope of depressive comorbidity. Depression goes with many other problems. Another theme is that
the specifics of depressive comorbidityand the implications for theory, research, and practicevary
considerably as we consider one type of comorbidity versus another. For example, the comorbidity
of depression and generalized anxiety disorder has very different implications than the comorbidity
of depression and alcohol-use disorder, which in turn is different than the comorbidity of depression
and cancer, which again has different implications than the comorbidity of depression and severe
relationship dysfunction. Each of the chapters in the book highlight some of the themes and issues, but
the remarkable breadth and depth of depressive comorbidity becomes clearer as we consider all of
the chapters in total. We attempt to bridge some of these differences and look for common themes
in the Epilogue at the end of the book, as do some of the contributors in their individual chapters on
specific issues or types of comorbidity. In this Introduction, however, we focus more on the specific
chapters and a few of the themes that are highlighted in each one. Overarching themes, such as what
is meant by comorbidity, how might future efforts at assessment and treatment be improved, and what
future developments may be particularly helpful are discussed in many of the individual chapters. This
brief introduction serves to highlight a few of the issues and introduce the reader to the broad array of
chapters that await them in the rest of thebook.
Key Words: introduction, assessment and diagnosis, theory and methods, psychiatric comorbidity, health
comorbidity, relationship comorbidity, intervention and consultation, future directions

Depression is associated with many psychiat- than that. Rather, we argue it is stunning that
ric disorders, chronic health problems, and severe depressive disorders are comorbid with such a vast
dysfunction in close relationships. To observe that array of other psychiatric disorders, numerous
depression goes with many other problems may health problems and diseases, and many types of
appear somewhat like observing that severe distress severely dysfunctional relationships. This is part of
goes with many difficult challenges in life. This type why depressive comorbidity is so interestingand
of association and comorbidity seems almost ubiq- important. With the 37 chapters that follow in this
uitous. But we are not arguing that depression is a edited book, The Oxford Handbook of Depression and
proxy measure for severe distress. Depression and Comorbidity, some of the themes that are often dis-
depressive comorbidity are much more complicated cussed include assessment and diagnosis, theory and

methods, psychiatric comorbidity, health comor- approach. Moreover, Watson and Stasik place this
bidity, relationship comorbidity, intervention and discussion in the context of a conceptual model that
consultation, and future directions. Watson and his colleagues have developed called the
Quadripartite Model. Furthermore, as with most of
Assessment, Diagnosis, Theory, and the chapters in this book, Watson and Stasik discuss
Methods some of the implications of their analysis for clini-
In chapters 2 through 5, the authors discuss cians who are working with patients that exhibit
themes of assessment, diagnosis, theory, and meth- depressive disorders and comorbidity.
ods regarding depressive comorbidity. For example, In chapter5, Eaton and Krueger provide a thor-
in chapter2, Widiger and Gore discuss assessment ough discussion regarding assessment strategies
and diagnosis within the context of the various edi- in the context of depressive disorders, comorbid-
tions of the Diagnostic and Statistical Manual of ity, and the evidence for an underlying internal-
Mental Disorders (DSM), from the first edition in izing disorder liability that may explain some of
the 1950s (DSM-I) to the fifth, most recent revi- the high comorbidity rates between depression and
sion in 2013 (American Psychiatric Association, other psychiatric disorders. Eaton and Krueger
2013). Although most of the issues, findings, and also discuss some of the important implications
implications for depressive comorbidity that are for depressive comorbidity in the context of issues
discussed in this book do not depend greatly on such as age, ethnicity, and psychometric concerns.
which edition of the DSM is being used, Widiger This topic has received a considerable amount of
and Gore do explain and discuss how versions of empirical attention recently, and it is very much
the DSM may have some impact on our think- on the minds of researchers, scholars, and practi-
ing about depressive comorbidity. The authors tioners in the area (e.g., Eaton etal., 2013; Wright
particular emphasis, of course, is on the diagnosis etal.,2013).
of mood disorders such as depression (e.g., major
depressive disorder [MDD]), which is virtually Comorbidity Between Depression and
identical between the fourth (DSM-IV); fourth, Another Psychiatric Disorder or Diagnosis
text revision (DSM-IV-TR); and fifth (DSM-5) edi- For chapters 6 through 17, the emphasis is on
tions of the DSM, with the minor exceptionin the comorbidity between depressionparticularly
our opinionfor how bereavement is addressed MDDand another specific psychiatric disorder or
in the diagnostic decision making regarding MDD diagnosis. Thus, in chapter6, Castriotta and Craske
(American Psychiatric Association, 1994, 2000, discuss the comorbidity of depression and panic dis-
2013). Widiger and Gore also discuss some impor- order. They note that a majority of people with panic
tant issues regarding definitions, empirical support, disorder also experience comorbidity with MDD,
and theoretical models of classification. while a minoritybut still a substantial number
In chapter3, Markon discusses how theoretical, of individuals with MDD also have comorbidity
statistical, and methodological issues may influ- with panic disorder. Reflecting a theme that is pres-
ence our constructs and models of comorbidity. ent in virtually all of the chapters on psychiatric
Using depressive and anxiety disorders as examples, comorbidity, the specific comorbidity of depression
Markon compares and contrasts two models of and panic disorder usually predicts the following
comorbidity: one that focuses on a shared liability concerns:more severe symptoms; greater persistence
for the comorbid disorders and another that focuses of the disorders; additional treatment and follow-up
on correlated symptoms for the comorbid disorders. challenges that are more difficult to manage than in
Markon uses these contrasted models to illustrate cases in which the two disorders show little comor-
how our understanding of comorbidity can influ- bidity; and very serious implications for an array of
ence our understanding of psychopathology in day-to-day functioning domains including work,
general. home, and close relationships. Moreover, suicidal
In c hapter4, Watson and Stasik discuss how ana- risk appears to increase with this type of comorbid-
lyzing the magnitude of distress and the specificity ity, which is also a concern that is mentioned in
of comorbid symptoms may potentially improve some of the other comorbidity chapters. In addi-
assessment, diagnosis, and a more complete under- tion, Castriotta and Craske discuss important issues
standing of comorbidity. The authors use the regarding the development of comorbid depression
specific psychiatric diagnoses of MDD and post- and panic disorder and research on evidence-based
traumatic stress disorder (PTSD) to illustrate this treatment of this comorbid condition.

2 Introduction
In chapter 7, Najavits and Capezza discuss the Association, 1994), DSM-IV-TR (American
comorbidity of depression and PTSD. Following Psychiatric Association, 2000), and DSM-5
a traumatic event, the comorbidity of PTSD (American Psychiatric Association, 2013). Mineka
and depression, especially MDD, is quite com- and colleagues also provide a discussion of several
mon. Najavits and Capezza discuss a wide range types of comorbidity, including cross-sectional,
of relevant topics, including assessment, diagno- cumulative or lifetime, and sequential. The authors
sis, prevalence, risk factors, developmental issues, conclude their chapter by discussing some of the
evidence-based interventions, and future research implications for clinicians regarding assessment and
directions. They also focus on two areas that have treatment of comorbid MDD and GAD. Because
received intense research attention: military trau- the comorbidity of depressive and anxiety disorders
mas and interpersonal abuse. Najavits and Capezza presents additional treatment challenges for the
discuss the practical implications for clinicians and patient and the clinician, combined treatments that
some treatment recommendations. As with most include psychological and medication interventions
types of depressive comorbidity, the authors note are gaining in popularity among some research and
the likelihood that the comorbid condition will practitioner groups (e.g., Wetherell etal., 2013). In
be harder to treat effectively, more prone to relapse addition, the relatively high comorbidity between
and recurrence, and more likely to have a negative MDD and GAD has led investigators to look for
impact on daily living. many common risk factors that might account for
In c hapter8, Langer and Rodebaugh discuss the some of this high association. For instance, recent
comorbidity of depression and social anxiety disor- studies have identified rumination as a possible
der. Comorbidity, in the form of co-occurrence, is process that may link stressful life events to both
common with these two disorders. Moreover, the depression and anxiety (e.g., Michl, McLaughlin,
presence of one disorder increases the risk for devel- Shepherd, & Nolen-Hoeksema,2013).
oping the other. Langer and Rodebaugh discuss In chapter 10, Witkiewitz and Stauffer discuss
evidence for a shared vulnerability factor regarding the comorbidity of depression and alcohol-use dis-
depression and social anxiety disorder, with genetic order. The substance-related disorder discussed here
diatheses and traits such as positive and negative falls under the grouping of substance-related and
affect apparently playing a role in the high comor- addictive disorders in DSM-5 (American Psychiatric
bidity rates. Langer and Rodebaugh also discuss Association, 2013). Witkiewitz and Stauffer note
some interesting research on coping efforts, includ- large-scale surveys done by the World Health
ing counterproductive efforts to cope with perceived Organization and similar organizations usually find
social exclusion, which unfortunately may lead to that alcohol-use disorders and depression rank very
yet higher levels of depression and social anxiety. In highly in lists of all-cause disability, with the World
addition, the authors provide some recommenda- Health Organization findings suggesting that these
tions for clinicians regarding effective methods for risk factors may be as high as third and fourth
assessment, treatment, and follow-up. among the probable risk factors for causing disease
In c hapter9, Mineka, Anand, and Sumner dis- burden and disability. With a focus solely on disabil-
cuss the comorbidity of MDD and generalized ity, rather than also on disease burden and mortal-
anxiety disorder (GAD). This specific comorbidity ity, depression sometimes climbs to the number-one
topic, regarding MDD and GAD, has been one of cause of day-to-day disability, in Western countries
the more active areas for theory and research, with such as the United States and countries in western
a large number of theoretical papers and empirical Europe. The authors discuss the frequent comorbid-
studies published in the past 25years. Mineka and ity of depression and alcohol-use disorders, along
colleagues discuss the overlap of symptoms; high with recent research on symptoms; prevalence; risk
rates of comorbidity; diagnostic considerations; and and protective factors; development and course
further implications for theory, research, and prac- of the comorbid disorders; and the unfortunate
tice if GAD is continued to be classified as an anxi- prospect of more difficult assessment, treatment,
ety disorder versus being classified as some type of outcome, and follow-up for patients with these
mood or depressive disorder. The authors conclude co-occurring disorders. Witkiewitz and Stauffer
that the evidence best supports continuing to clas- also discuss a wide range of important treatment
sify GAD as an anxiety disorder, which is a conclu- considerations for working with patients who have
sion also reached by the task forces and study groups comorbid depressive and alcohol-use disorders, and
who have developed DSM-IV (American Psychiatric they provide some recommendations for clinicians.

Richards,OHara 3
The relatively high comorbidity of depressive and diagnostic thresholds; and some issues for future
alcohol-use disorders has led investigators, of course, efforts on theory, research, and practice. The authors
to explore many possible explanations for these also discuss clinical implications regarding assess-
strong associations; these studies include looking ment and treatment of these comorbid disorders.
for common liabilities and risks for the disorders, In chapter13, Klein, Bufferd, Ro, and Clark dis-
along with specific hypotheses such as alcohol-use cuss depression and comorbidity in the context of
being a coping strategy to self-medicate or cope personality disorders. The authors note that there
with unpleasant negative affect such as depression are many challenges in this area, including contro-
(e.g., Crum etal.,2013). versy about how to best diagnose personality dis-
In chapter 11, Keel and Holland discuss and orders, conceptual challenges in terms of symptom
evaluate the high comorbidity between depressive overlap with depressive disorders, research issues in
disorders such as MDD and eating disorders such terms of methodological concerns regarding the reli-
as anorexia nervosa, bulimia nervosa, and binge eat- ability and validity of personality diagnoses, clinical
ing disorder. The authors also explore how some of challenges in terms of the relatively large number
the substance-use disorders appear to play a role in of personality disorders and the resistance of these
shared liability, risk, and comorbidity. Perhaps few disorders to therapeutic change, and several other
subtopics in psychopathology research have enjoyed challenges and quandaries. Trait-like styles of dys-
as much research attention recently as the eating functional thinking, and similar symptoms in per-
disorders. Therefore, Keel and Holland are able to sonality disorders, have also been investigated in the
discuss some recent empirically supported develop- context of relapse prevention for depression (e.g.,
ments in our understanding of the high comorbid- van Rijsbergen et al., 2013; also see Richards &
ity of depression and eating disorders. For example, Perri, 2010). The authors discuss some recent and
models of shared liability and high risk or diathesis ongoing research regarding studies investigating the
for depressive disorders and eating disorders have associations between depressive symptoms, mal-
received some mixed support, although this area adaptive personality traits, psychosocial function-
is very complex, and inconsistent findings make it ing, and several possible diagnostic approaches. This
more challenging to untangle the various hypoth- is a complex area, and Klein and colleagues provide
eses and results. Keel and Holland also discuss some a helpful discussion that integrates many of the pri-
of the clinical issues regarding the comorbidity of mary issues and concerns. In addition, the authors
depressive disorders and eating disorders, and they discuss further directions for research, along with
suggest some possible clinical guidelines for prac- some of the clinical implications regarding assess-
titioners. Because the research on eating disorders ment and treatment of comorbid depressive and
is expanding so rapidly, the area is benefiting from personality disorders.
this expansion in numerous ways, such as a better In chapter14, Feldman and Larsen discuss the
understanding of diversity issues in this area (e.g., comorbidity of depression and sexual dysfunction.
Thompson-Brenner etal., 2013; also see Le, Boyd, The authors include a discussion of some infor-
& Lara, ch.33). mation about sexual dysfunctions per se in both
In chapter 12, Capaldi and Kim discuss the men and women. In addition, the authors bring
comorbidity of depression and conduct disorder. their disciplinary perspectives, with backgrounds
In addition to both of these disorders being quite in medicine and clinical psychology, into play as
important and disruptive for day-to-day living, this they discuss the topic of comorbid depression and
area may be an example of high comorbidity that is sexual dysfunction. Feldman and Larsen discuss a
not as anticipated by mental health care profession- wide range of issues regarding the interconnections
als as some of the other specific comorbidities dis- between neurobiology, psychology, social issues,
cussed in this book. Nevertheless, the comorbidity and depressive comorbidity with sexual dysfunc-
of depressive disorder and conduct disorder is quite tions. The authors also discuss the fluctuation of
high. Capaldi and Kim discuss numerous features depressive comorbidity rates from one sexual dys-
of this comorbidity, including models of shared risk function to another, along with various bidirec-
and causality, the generally poorer adjustment that tional causal models that may potentially explain
follows this comorbid condition versus either con- improvement (or deterioration) in one condition
dition presenting alone; relevant developmental and following improvement (or deterioration) in the
gender issues; certain distinctions in the findings for other. Finally, Feldman and Larsen include a sum-
subthreshold disorders versus those that reach DSM mary of clinical implications for the evaluation and

4 Introduction
treatment of comorbid depressive disorders and sometimes contradictory, and often difficult to inte-
sexual dysfunctions. grate clearly. The authors also discuss some of the
In chapter 15, Cohen, Callaway, and Auster clinical implications, such as one disorder likely
discuss the comorbidity of depression and schizo- impacting the course of the other, and the implica-
phrenia. Depressive disorders frequently co-occur tions of shared liabilities versus liabilities that appear
with schizophrenia spectrum disorders (American to be more specific to depression or bipolar disorder.
Psychiatric Association, 2013), and Cohen and col-
leagues discuss the complexand sometimes con- Comorbidity of Depressive Disorders and
tradictoryresearch literature on this topic. The Chronic Health Problems and Diseases
authors take a broad approach to integrating the In chapters18 to 27, and also to some extent in
relevant literature by focusing on clinical, behav- chapters 30 and 34, comorbidity between depres-
ioral, cognitive, phenomenological, and neurobio- sive disorders and specific chronic health problems
logical findings, which appear to be specific to the (or an array of associated chronic health problems)
comorbidity of depression and schizophrenia. In is emphasized. For example, in c hapter18 Suls and
addition, Cohen etal. discuss some of the clinical Davidson discuss the comorbidity of depression
implications and treatment considerations regard- and cardiovascular disease. One of the reasons this
ing patients who exhibit comorbid depression and topic is very important is that cardiovascular disease
schizophrenia. is often listed as the number-one cause of death in
In c hapter16, Cukrowicz and Poindexter discuss industrialized countries, and the comorbidity of
the important clinical concernsand the relevant this disease with severe depression such as MDD
researchregarding situations in which high sui- is usually associated with additional complications
cide risk, and suicide attempts or completions, are during acute treatment, maintenance intervention,
associated with MDD. This topic has also received and long-term follow-up care. Suls and Davidson
considerable attention in the media recently, in part discuss numerous issues, including risk factors, bio-
because of the high suicide rates among returning logically associated variables such as immune activ-
U.S. combat veterans from the wars in Iraq and ity, and behavioral/psycho-social variables such as
Afghanistan. Cukrowicz and Poindexter discuss a medical adherence and lifestyle change. The authors
broad array of relevant issues, including prevalence, also discuss a complex and sometimes confusing lit-
risk, and systematic treatment studies such as ran- erature regarding the impact of treatment programs
domized controlled trials. The authors also summa- for depression in patients who have cardiovascular
rize some clinical implications and recent treatment disease. Not surprisingly, the authors note that fur-
approaches that appear to have empirical support. ther large-scale randomized clinical trials would help
In chapter17, Youngstrom and Van Meter dis- clarify this situation. Suls and Davidson also discuss
cuss the comorbidity of depressive disorders and some of the clinical issues and unresolved questions
bipolar disorders. Although the diagnostic criteria that may impact practitioners decisions about how
for MDD is virtually unchanged between DSM-IV to best treat a cardiac patient with severe depression.
and DSM-5 (American Psychiatric Association, In chapter 19, A. Nezu, C. Nezu, Greenberg,
1994, 2000, 2013), the diagnostic criteria for bipo- and Salber discuss the comorbidity of depression
lar disorders have undergone some modest revisions and cancer. The authors note the recent explosion
(in our opinion), which will have some impact on in research (and research funding) on the comor-
diagnostic decisions but very little on the comor- bid condition of depression and cancer. Nezu etal.
bidity issues addressed in this chapter and book. discuss a number of issues in this area, including
Youngstrom and Van Meter discuss a wide range epidemiological research, risk factors, and the very
of relevant issues regarding comorbidity between negative impact that comorbid depression and can-
depressive and bipolar disorders, including the cer appears to predict for important outcome issues
long history of discussions about the association such as medical adherence, lifestyle enhancement,
of depression and mania, epidemiological research mood regulation, coping, morbidity, and mortal-
on this comorbidity, phenomenological variables, ity. As in several of the other chapters, the authors
developmental findings, family history, and relevant note surveys by the World Health Organization and
treatment studies. The authors note that method- other organizations that implicate depression as a
ological challenges in this area, along with concerns leading cause of disability. In addition, several treat-
about achieving satisfactory reliability and validity ment implications are evident in the authors dis-
in assessments, yield a database that is complicated, cussion of these issues.

Richards,OHara 5
In chapter20, Roditi, Waxenberg, and Robinson portfolio on comorbid depressive and sleep disor-
discuss the comorbidity of depressive disorders and ders. Carney and Moss discuss the comorbidity of
pain. The authors note that this type of comor- depression with several sleep disorders, including
bidity is very common and often quite disabling. hypersomnia, breathing-related sleep disturbances,
Roditi etal. discuss relevant theories, epidemiologi- and chronic insomnia. The authors indicate that
cal issues, patterns of symptomatology, and impli- there is the most empirical support for bidirectional
cations for assessment and treatment. The authors relationships between depression and sleep disor-
also discuss some neurobiological variables that are ders. Carney and Moss discuss the intricate clinical
important to this type of comorbidity. In addition, issues that should be unraveled for effective assess-
Roditi and colleagues summarize an array of current ment and treatment, given that a number of symp-
treatment approaches for comorbid depression and toms overlap in the two groups of disorders and that
pain and discuss the evidence base of support for approximately 90% of patients with MDD report
these approaches. some kind of sleep problem. Carney and Moss also
In chapter21, Dutton and Needham discuss the argue for an integrated, multicomponent approach
comorbidity of depression and chronic obesity. With to treatment, which includes interventions for both
obesity (along with smoking) very high on the cur- disorders.
rent lists of preventable causes of death, this topic is In chapter23, Hancock, Bruce, and Lynch dis-
clearly important. The authors discussion includes cuss the comorbidity of depressive disorders and
a consideration of the possible bidirectional nature multiple sclerosis (MS). Hancock et al. note that
of this comorbidity, where severe depression (e.g., depressive comorbidity is common in MS, with
MDD) may be a risk factor for chronic obesity and about 50% of MS patients receiving a diagnosis
vice versa. Dutton and Needham also discuss theo- of MDD during their lifetime. If the depression is
retical issues, common mediating biological and left untreated, suicidal risk climbs and a variety of
environmental factors, and the available research MS symptoms tend to worsen. The authors discuss
literature for trying to untangle various hypoth- some of the challenges for clinicians who are work-
eses. Possible biological mediators may include dys- ing with depressed MS patients, including the diffi-
regulation and the hypothalamic-pituitary-adrenal culty of distinguishing the symptoms of depression
axis. Potential environmental and psychological from the neurological symptoms and behavioral
mediators may include binge eating and a history consequences of MS. In addition, Hancock and
of abuse. Moreover, Dutton and Needham discuss colleagues provide some clinical guidelines and con-
some of the complexities and cautions for making siderations for the assessment, diagnosis, evaluation,
treatment recommendations in this area when the treatment, and follow-up of depressed MS patients.
association of depression and obesity is not thor- In chapter24, Blashill, Gordon, Mimiaga, and
oughly understood. This is also an example of an Safren discuss the comorbidity of depressive dis-
areaand there are many in this bookwhere a orders and HIV-positive status/AIDS. Blashill and
treatment package that is multicomponent (e.g., colleagues discuss a wide range of relevant issues,
psycho-social, medical, and lifestyle components), including epidemiological studies; comorbidity of
integrated, and multidisciplinary in nature and that HIV+/AIDS with depressive and substance-use
targets both depression and the comorbid disor- disorders; implications of depressive comorbidity
ders or chronic health problems may be a particu- for poor medical adherence, ineffective self-care,
larly attractive treatment option (e.g., see Daumit and unsafe sexual practices; and an array of asso-
et al., 2013, regarding obesity and mental illness; ciated biological and psycho-social variables. As
Richards, Cohen, Morrell, Watson, & Low, 2013, with many of the chapters in this section of the
regarding smoking, depression, and anxiety; Rose book on comorbidity with chronic health prob-
& Behm, 2013, regarding smoking and associated lems, Blashill and colleagues argue for multicom-
mood disorders; and Wetherell etal., 2013, regard- ponent, integrative treatments that target both
ing GAD, MDD, and older-adult status). the medical and psycho-social aspects of people
In chapter 22, Carney and Moss discuss the who are living with HIV+/AIDS. In addition, the
comorbidity of depression and sleep disorders. This authors provide a brief overview of the literature
is an important area that has benefited from a dra- on evidence-based interventions with a focus on
matic increase in research during the past 25years. improving self-care behaviors and medical adher-
Advances in technology and increases in research ence. The authors also discuss important direc-
funding have facilitated an expanded research tions for future research.

6 Introduction
In chapter 25, Christensen, Van Liew, and medical conditions regarding womens health and
Kellerman discuss the comorbidity of depression depression in part because the conditions are more
with chronic kidney disease. The authors discuss frequently comorbid with depression in women
the negative implications of this comorbidity, such than in men. Buttner and OHara consider a wide
as poorer quality of life, reduced medical adherence, range of relevant issues, including epidemiological
and higher morbidity and mortality for depressed research; specific features of the medical conditions
kidney disease patients. As with most of the depres- with particular attention to symptoms that overlap
sion and chronic health problem comorbidities, with depression; theoretical approaches to explain-
the challenges for clinicians include unraveling the ing the high comorbidity of these medical condi-
overlap in symptoms between depression and kid- tions with depression; future research directions;
ney disease. Unfortunately, there is not a large data- and clinical guidelines for assessing and treating
base of sophisticated randomized controlled trials depression in the context of type 2 diabetes, fibro-
investigating the effectiveness of integrative treat- myalgia, and rheumatoid arthritis in women. One
ments for depressed kidney disease patients; thus of the interesting features of this chapter is that
Christensen and colleagues discuss some promising the authors address three different chronic health
future directions for treatment research. The avail- problems and the comorbidity with depressive dis-
able treatment literature and clinical experience orders, rather than focusing on one health problem
have yielded some helpful clinical guidelines, how- as most of the other chapters on comorbidity with
ever, and Christensen and colleagues discuss these healthdo.
clinical recommendations for assessing and treating Chapters30 and 34 also discuss the comorbid-
depressed kidney disease patients. ity of depression with a health issue. In c hapter30,
In chapter 26, Brommelhoff discusses the Felder, Lindemann, and Dimidjian discuss the
comorbidity of depressive disorders and dementia topic of perinatal depression. The authors discus-
syndromes. Depressive disorders are commonly sion includes an evaluation of the epidemiological
associated with many of the dementia syndromes research, etiological theories, risk factors, specific
and neurocognitive disorders (American Psychiatric assessment and treatment challenges, and potential
Association, 2013). Brommelhoff discusses a num- negative consequences for women and offspring
ber of the important aspects of this comorbid- when a woman is pregnant and depressed or post-
ity, including theory, directional and bidirectional partum and depressed. This is another area where
models of causation, and neurobiological and important research has increased during the past
neuropsychological variables such as cerebrovas- 25years, and the authors discuss this research and
cular risk factors and negative emotional reactions future research directions. There are a number of
to cognitive decline. There is considerable overlap promising treatment interventions in this area, from
between symptoms of depression and symptoms evidence-based psychotherapy (e.g., Nylen et al.,
of dementia syndromes (and certain examples of 2010) to enhanced child-care services (e.g., Herba
the other neurocognitive disorders). Brommelhoff et al., 2013) for depressed pregnant and postpar-
notes that this overlap in symptoms complicates tum women. Felder etal. conclude the chapter with
the clinicians task of assessment and evaluation, of a discussion of clinical guidelines for practitioners
course, but that nevertheless the differential diagno- regarding the assessment and treatment of depres-
ses between depressive disorders and dementia syn- sion during the perinatal period inwomen.
dromes are essential. Brommelhoff discusses some In chapter 34, Hopko, McIndoo, Gawrysiak,
of the clinical implications and guidelines for prac- and Grassetti discuss psycho-social interventions
titioners working with depressed dementia patients. for depressed breast cancer patients. Thus this
The author also includes a discussion of the possible chapter overlaps with both the health chapters
contraindications for antidepressant medications and the treatment chapters in the book. Hopko
(given the possibility of adverse reactions or nega- and colleagues review an array of recent studies on
tive drug interactions) and the potential effective- epidemiology, theoriesespecially bidirectional
ness of certain psycho-social interventions. theoriesnegative impacts and consequences of
In chapter27, Buttner and OHara discuss the the comorbidity of depression and cancer, assess-
comorbidity of depression with three of the more ment and treatment approaches in the complex
common health problems regarding womens situation of comorbid depression and cancer, vari-
health:type 2 diabetes, fibromyalgia, and rheuma- ous directions for future theory and research, and
toid arthritis. The authors discuss these three chronic treatment outcome studies involving psychosocial

Richards,OHara 7
interventions for depressed breast cancer patients. relationshipsparticularly disrupted family rela-
The authors acknowledge that many of these stud- tionships. The authors discussion includes an
ies, including some of the major randomized con- emphasis on emotional processes, parentchild rela-
trolled trials, have methodological and practical tionships, and adolescent depression. This topic is
limitations, and therefore the authors point to another example of a subarea in depressive comor-
future directions for stronger research. In addition, bidity research that has enjoyed large-scale growth in
Hopko and colleagues discuss clinical guidelines for recent years. Abaied and Rudolph discuss the com-
the psycho-social treatment of depressed breast can- plex bidirectional models that appear to be best sup-
cer patients. ported by empirical research and that help explain
the associations and interactions between depres-
Comorbidity Between Depression and sion, family adversity, parentchild relationships,
Relationship Distress and Dysfunction parenting behavior, and adolescent depression. The
Depression is often associated with distressed authors also offer an integrative framework for con-
relationships, and distressed relationships are often ceptualizing emotional functioning as one of the
associated with depression. This type of depressive important mechanisms influencing how family rela-
comorbidity is not as classically considered in the tionships and adolescent depression influence one
research literature as is the depressive comorbidity another over a period of time. We expect research
with other psychiatric disorders or with chronic on the comorbidity of depression and relationship
health problems. But this type of depressive comor- dysfunctionincluding intimate and family rela-
bidity is very importantalmost everyone cares tionshipsto continue to grow and prosper in the
about close relationships. Moreover, the research foreseeable future.
literature on the comorbidity of depression and
relationship distress has grown exponentially in Treatment of Depression and
the past 30 years, just as the associated literature Comorbid Disorders
on close relationship issues, in general, has grown Chapters31 through 35 address some aspect of
very rapidly. In chapters28 and 29, the contribut- the treatment of depression and comorbid disor-
ing authors address some of the issues and research ders, with a focus on treatment issues throughout
regarding the comorbidity of depression and dys- each chapter rather than a brief discussion of clini-
functional relationships. cal guidelines as is the case in most of the previous
In chapter 28, Stroud, Feinstein, Bhatia, chapters.
Hershenberg, and Davila discuss depression in the In chapter 31, Howland discusses multidisci-
context of intimate relationships. The authors begin plinary treatments and medications for depressive
their chapter by providing an overview of several disorders and comorbidity. Howlands discussion
of the issues and noting that concurrent and lon- includes an analysis of many of the issues regarding
gitudinal research has thoroughly established an antidepressant medications and other psychotropic
association between depression and relationship drugs, which is informed in part by the authors
dysfunction. Stroud and colleagues discuss theory disciplinary perspective as a physician, psychia-
and conclude that bidirectional and transactional trist, and psychiatric researcher. Howland covers
theories are the most supported. The authors go on a range of relevant topics, including comorbidity
to discuss how the specific components and pro- with chronic depression and treatment-resistant
cesses of intimate relationships may be influenced depression, psychiatric and medical comorbidities,
by depression. Stroud and colleagues include a adjusting for possible treatment contraindications
discussion of research on adolescent intimate rela- in the cases of some medication regimens and for
tionships and how romantic and sexual experiences specific medical and psychiatric comorbidities, and
during the developmental interval of adolescence the increases in disability and mortality that can be
may relate to depression and vice versa. In addi- associated with depressive comorbidity. In addition,
tion, the authors discuss clinical implications for Howlandlike many of the contributors to this
counselors and other practitioners, with a focus volumeargues for multidisciplinary collaboration
on cognitive-behavioral couple therapy, which and integrated treatment programs that targetall or
appears to have some of the stronger evidence-based most features of the comorbidity situation.
support. In chapter32, Gitlin discusses the role of com-
In chapter 29, Abaied and Rudolph dis- munity and home-based interventions for cases of
cuss the comorbidity of depression and family late-life depression; moreover, these depression cases

8 Introduction
almost invariably involve some psychiatric, health, Of course many of the chapters in this book could
or relationship comorbidity. The authors discussion fit nicely into more than one section. For example,
includes a review of 23 well-evaluated community chapters 2 through 5 on assessment, diagnosis,

and home-based interventions that have yielded theory, and methods could also fit nicely in The
generally positive and promising results. The ben- Big Picture section below. Chapters 18 through
efits of such evidence-based interventions may be 27 (and 30 and 34) on comorbidity of depressive
enormous, given the severely negative morbidity disorders and chronic health problems also discuss
and mortality rates that may accompany comorbid some of the psychiatric comorbidity issues that
depression in older adults. Gitlin notes that there are emphasized in chapters 6 through 17. Indeed,
are also many attractive features to community and the topic of this bookdepressive disorders and
home-based interventions for depressed older adults, comorbidityhas a number of common issues and
including practical advantages to clients; economic themes that run through most of the chapters in
advantages to providers and clients; increased access thebook.
to nonpharmacologic interventions such as training In chapter35, Whisman and BE discuss cogni-
in problem-solving skills, social support increases, tive therapy for comorbid depression. One of the
and lifestyle enhancement; and increased access to important features of this chapter is that the authors
treatment for underserved populations. Gitlin also give clinical guidelines regarding specific types of
discusses some of the intervention programs that depressive comorbidity, such as the comorbidity of
she considers most outstanding and points to some depression with anxiety disorders, substance-use dis-
future directions in this area regarding intervention orders, and personality disorders. Whisman and BE
research and practice. discuss some of the complex issues and challenges
In chapter 33, Le, Boyd, and Lara discuss the that confront clinicians regarding the assessment,
treatment of depressive disorders and comorbid- treatment, and follow-up of depressed patients with
ity, particularly in the context of ethnic minority co-occurring conditions. The authors use cognitive
groups. This is a very important area, and ethnic therapy, which is one of the evidence-based thera-
minorities are often among the most underserved pies in this area, to illustrate some of their clinical
populations regarding interventions for comorbid recommendations and treatment guidelines. This
depression. (This point is also made by Gitlin in chapter also echoes a theme that is evident in many
chapter 32.) Le and colleagues cover a number of of the other chapters:Comorbid depression is more
relevant topics, including depressive comorbidity difficult to treat effectively than depression by itself.
in ethnic minority adults, treatment interventions
that include some tailoring to ethnic minority pop- The Big Picture
ulations, the challenges presented by considerable Chapters36 and 37 discuss some of the big pic-
underrecognition and undertreatment of comorbid ture issues regarding this book on depressive disor-
depression in ethnic minority groups in the United ders and comorbidity. In chapter36, Kessler, Scott,
States, and an array of relevant needs and direc- Shahly, and Zaslavsky discuss broad and integrative
tions for future research and practice. The authors issues regarding comorbid depression. Kessler and
also discuss how culturally sensitive adaptations of colleagues evaluate some of the relevant epidemio-
evidence-based treatments for depressive comorbid- logical research literature, with their perspective
ity may be successfully adapted to ethnic minor- informed by their own extensive involvement in this
ity groups. In addition, Le and colleagues discuss type of research. The authors discussion includes an
clinical guidelines for practitioners in the context analysis of research on latent liabilitiesparticu-
of treating comorbid depression in ethnic minority larly regarding the dynamics of comorbidity that
groups. involves depression and temporally primary and
In chapter 34, Hopko, McIndoo, Gawrysiak, secondary psychiatric disorders. The authors also
and Grassetti discuss psychosocial interventions for discuss the comorbidity of depression and chronic
depressed breast cancer patients. We discussed this physical problems and diseases. Kessler and col-
chapter earlier under Comorbidity of Depressive leagues suggest that future research on depressive
Disorders and Chronic Health Problems and comorbidity would be enhanced by distinguishing
Diseases. With the consistent emphasis on treat- between psychiatric and medical comorbidities spe-
ment issues throughout this chapter, however, the cific to MDD, versus those that appear to reflect a
Hopko et al. chapter also fits in this section on broader latent liability and associations with a wider
treatment of depression and comorbid disorders. range of disorders and dysfunctions.

Richards,OHara 9
In chapter 37, we discuss some of the integra- symptoms with the incidence and persistence of alcohol
tive themes that run through much of this book, dependence. JAMAPsychiatry, 70, 718726.
Daumit, G. L., Dickerson, F. B., Wang, N.-Y., Dalcin, A.,
which is usually the charge of an epilogue. As the Jerome, G. J., Anderson, C. A. M.,...Appel, L. J. (2013).
editors, we are in a good place to note themes that A behavioral weight-loss intervention in persons with seri-
seem particularly salient, important, and frequently ous mental illness. New England Journal of Medicine, 368,
mentioned throughout the book. Thus, for example, 15941602.
we note that depressive comorbidity is common and Eaton, N. R., Krueger, R. F., Markon, K. E., Keyes, K. M.,
Skodol, A. E., Wall, M.,...Grant, B. F. (2013). The structure
pervasive. We discuss some of the explanatory mod- and predictive validity of the internalizing disorders. Journal
els of causation, including shared liability, overlap- of Abnormal Psychology, 122,8692.
ping symptoms, common biological factors, specific Herba, C. M., Tremblay, R. E., Boivin, M., Liu, X., Mongeau,
models of causal direction (e.g., unidirectional ver- C., Seguin, J. R., & Cote, S. M. (2013). Maternal depres-
sus bidirectional), and intervention side effects that sive symptoms and childrens emotional problems:Can early
child care help children of depressed mothers? JAMA
may cause or exacerbate depression. We discuss the Psychiatry, 70, 830838.
remarkable increase in disability and disease burden Michl, L. C., McLaughlin, K. A., Shepherd, K., &
(morbidity and mortality) that often accompanies Nolen-Hoeksema, S. (2013). Rumination as a mechanism
comorbid depression. In part because the comorbid- linking stressful life events to symptoms of depression and
ity rates are so very high with certain comorbidities, anxiety: Longitudinal evidence in early adolescence and
adults. Journal of Abnormal Psychology, 122, 339352.
we take a look at a specific example:the comorbidity Nylen, K. J., OHara, M. W., Brock, R., Moel, J., Gorman, L.,
of depression and anxiety disorders. We discuss an & Stuart, S. (2010). Predictors of the longitudinal course
array of assessment issues, which are usually more of postpartum depression following interpersonal psycho-
complex in cases of depression co-occurring with therapy. Journal of Consulting and Clinical Psychology, 78,
other disorders. We comment on the complexities 757763.
Richards, C. S., Cohen, L. M., Morrell, H. E.R., Watson, N. L.,
and challenges that accompany intervention pro- & Low, B. E. (2013). Treating depressed and anxious smok-
grams for comorbid depression. Treating depression ers in smoking cessation programs. Journal of Consulting and
that is comorbid with other psychiatric disorders, Clinical Psychology, 81, 263273.
chronic health problems, or relationship dysfunc- Richards, C. S., & Perri, M. G. (Eds.). (2010). Relapse preven-
tion is very difficultfor both the practitioner and tion for depression. Washington, DC:American Psychological
the patient (and often for significant others too). We Rose, J. E., & Behm, F. M. (2013). Adapting smoking cessation
conclude the Epilogue with some final comments treatment according to initial response to precessation nico-
on treatment issues and the seemingly ubiquitous tine patch. American Journal of Psychiatry, 170, 860867.
nature of depressive comorbiditydepression goes Thompson-Brenner, H., Franko, D. L., Thompson, D. R.,
with many other problems. The 37 chapters in this Grilo, C. M., Boisseau, C. L., Roehrig, J. P.,...Wilson, G. T.
(2013). Race/ethnicity, education, and treatment parameters
volume, The Oxford Handbook of Depression and as moderators and predictors of outcome in binge eating
Comorbidity, do an effective job in covering the disorder. Journal of Consulting and Clinical Psychology, 81,
extensive landscape of this importanttopic. 710721.
van Rijsbergen, G. D., Bockting, C. L. H., Burger, H.,
Spinhoven, P., Koeter, M. W. J., Ruhe, H. G.,...Schene,
References A. H. (2013). Mood reactivity rather than cognitive reactiv-
American Psychiatric Association. (1994). Diagnostic and sta- ity is predictive of depressive relapse: A randomized study
tistical manual of mental disorders (4th ed.). Washington, with 5.5-year follow-up. Journal of Consulting and Clinical
DC:Author. Psychology, 81, 508517.
American Psychiatric Association. (2000). Diagnostic and statisti- Wetherell, J. L., Petkus, A. J., White, K. S., Nguyen, H., Kornblith,
cal manual of mental disorders (4th ed., text rev.). Washington, S., Andreescu, C.,...Lenze, E. J. (2013). Antidepressant
DC:Author. medication augmented with cognitive-behavioral therapy for
American Psychiatric Association. (2013). Diagnostic and sta- generalized anxiety disorder in older adults. American Journal
tistical manual of mental disorders (5th ed.). Washington, of Psychiatry, 170, 782789.
DC:Author. Wright, A. G.C., Krueger, R. F., Hobbs, M. J., Markon, K. E.,
Crum, R. M., Mojtabai, R., Lazareck, S., Bolton, J. M., Eaton, N. R., & Slade, T. (2013). The structure of psychopa-
Robinson, J., Sareen, J.,...Storr, C. L. (2013). A prospec- thology:Toward an expanded quantitative empirical model.
tive assessment of reports of drinking to self-medicate mood Journal of Abnormal Psychology, 122, 281294.

10 Introduction

2 Diagnostic and Statistical


Thomas A.Widiger and Whitney L.Gore

This chapter provides a discussion of the American Psychiatric Associations classification of mental
disorders (DSM-I through DSM-5), with a particular emphasis on mood disorders and their classification
and diagnosis. It begins with the rationale for having an official, authoritative diagnostic manual and then
traces the history of the development of the first edition through the fourth edition (DSM-IV-TR, 2000).
The authors then discuss fundamental issues concerning the fifth edition (DSM-5, 2013), including the
definition of mental disorder, the empirical support for proposed revisions, the shift toward a dimensional
model of classification, and the shift toward a neurobiologically-based classification.
Key Words: DSM-IV-TR, DSM-5, classification, diagnosis, dimensional model, neurobiologically-based
classification, mental disorders, mood disorders

Diagnostic and Statistical Manual(DSM) 2002). Whether or not a child receives medication,
Persons often feel different emotions, such as a person receives disability support, a researcher
happiness, passion, vulnerability, and anger. Any obtains a publication or receives grant funding, a
normal person will at times experience intense emo- patient obtains insurance coverage for treatment,
tions, feeling strongly about some particular event or a prisoner is released from incarceration can be
or experience, either positively or negatively, and governed heavily by what is contained within the
sometimes for a significant length of time. However, DSM (Frances & Widiger, 2012). In addition,
the mental health professions of psychiatry, clini- persons think in terms of their language, and the
cal psychology, social work, and nursing consider predominant classifications of psychopathology
the emotions or moods of some persons to repre- in English are encompassed by the DSM-IV-TR
sent mental disorders. What constitutes a mental and, in all other languages, by the tenth edition
disorder is largely under the authority and control of the World Health Organizations International
of the American Psychiatric Association (and the Classification of Diseases (ICD-10; WHO, 1992).
World Health Organization [WHO]), through Therefore these nomenclatures have a substantial
its Diagnostic and Statistical Manual of Mental impact on how clinicians, social agencies, students,
Disorders (DSM), the current version of which is the researchers, the government, and the general public
DSM-IV-TR (American Psychiatric Association, conceptualize aberrant, problematic, and maladap-
2000). As the authoritative, official nomenclature tive behavior.
for mental disorder diagnosis, the DSM-IV-TR is The DSM-IV-TR is certainly informed and
in fact an exceedingly powerful document, impact- guided by a substantial body of empirical research
ing many important social, forensic, clinical, and (Widiger et al., 1998). However, the authority
other professional decisions (Schwartz & Wiggins, and power of the DSM-IV-TR is not necessarily

matched by its scientific foundation. The disorders their studies have been completed a rhetorician writes
included within the DSM-IV-TR are well reasoned, a tragedy and a psychiatrist a classification. (Zilboorg,
scientifically researched and, for the most part, well 1941, p.450)
documented, representing what is currently under-
stood by a body of scientists, theorists, researchers, ICD-6 andDSM-I
and clinicians to be the predominant forms of psy- It was clearly untenable to have clinicians,
chopathology (Widiger, 2013). On the other hand, researchers, and other mental health profession-
mental disorders are not entities that exist indepen- als using different diagnoses to describe and study
dent of belief systems. Mental disorders are, to a the same psychopathology. Patients were unable to
substantial extent, constructions of clinicians and obtain consistent medical care and a scientific base
researchers rather than proven, evident diseases or of knowledge could not be accumulated.
illnesses (Boorsboom, 2008; Meehl, 2010; Strauss
The present condition with respect to the
& Smith, 2009). The mental disorders included
classification of mental diseases is chaotic. Some
within the DSM-IV-TR represent the opinions
states use no well-defined classification. In others the
and beliefs, perhaps even just the hypotheses, of
classifications used are similar in many respects but
a set of psychiatrists and clinical psychologists,
differ enough to prevent accurate comparisons. Some
and much of it is subject to considerable dispute,
states have adopted a uniform system, while others
even controversy. This chapter provides an over-
leave the matter entirely to the individual hospitals.
view of the DSM-IV-TR diagnostic nomenclature,
This condition of affairs discredits the science.
beginning with historical background (Widiger &
(Salmon, Copp, May, Abbot, & Cotton, 1917,
Crego, 2013), followed by a discussion of the major
issues facing the forthcoming DSM-5, with a par-
ticular emphasis on proposals concerning disorders By the time of the Second World War, the U.S.
ofmood. Navy, Army, and Veterans Administration had all
developed their own nomenclatures. However, the
Second World War was instrumental in compel-
Historical Background
ling clinicians from a number of different coun-
A primary purpose of an official diagnostic
tries to work directly with one another, bringing to
nomenclature is to provide a common language for
immediate attention the difficulties in not having
communication (Kendell, 1975; Sartorius et al.,
a common language. Therefore the World Health
1993). Dysfunctional, aberrant, and maladaptive
Organization (WHO) finally included a section
ways of feeling, thinking, behaving, and relating
devoted to mental disorders within the 6th edi-
to others are of substantial concern to many dif-
tion of the International Classification of Diseases
ferent professions, the members of which hold an
(ICD-6), published in 1948 (Kendell, 1975). The
equally diverse array of opinions regarding etiology,
U.S. Public Health Service commissioned a com-
pathology, and treatment. It is imperative that these
mittee, chaired by psychiatrist George Raines (with
persons be able to communicate meaningfully with
representatives from a variety of other professions
one another. The impetus for the development of
and public health agencies) to develop a variant
an official diagnostic nomenclature was the chaos
of the mental disorders section of ICD-6 for use
and confusion generated by its absence (Widiger,
within the United States. The United States, as a
2001). For a long time confusion reigned. Every
member of the WHO, was obliged to use ICD-6,
self-respecting alienist [the nineteenth-century term
but any particular country could make certain
for a psychiatrist], and certainly every professor, had
adjustments to maximize its acceptance and util-
his own classification (Kendell, 1975, p.87). In the
ity within its particular culture. Responsibility for
nineteenth century, the production of a new system
publishing and distributing this committees ver-
for classifying psychopathology was a standard rite
sion of ICD-6 within the United States was given to
of passage for the young aspiring psychiatrist.
the American Psychiatric Association (1952) under
To produce a well-ordered classification almost the title Diagnostic and Statistical Manual. Mental
seems to have become the unspoken ambition of Disorders (hereafter referred to as DSM-I).
every psychiatrist of industry and promise, as it is DSM-I was generally successful in obtaining
the ambition of a good tenor to strike a high C.This acceptance among different clinical institutions and
classificatory ambition was so conspicuous that the agencies within the United States, partly because
composer Berlioz was prompted to remark that after of its expanded coverage, notably the inclusion of

12 Diagnostic and Statistical Manual(DSM)

somatoform disorders, stress reactions, and person- 1966 and became effective in 1968. Acompanion
ality disorders, which had been of considerable inter- glossary in the spirit of Stengels recommendations
est during and after the war (Kendell, 1975). The was to be published conjointly, but work did not
mood disorders of DSM-I (American Psychiatric begin on the glossary until 1967 and it was not
Association, 1952) included involutional psychotic completed until 1972. This delay greatly reduced
reaction (characterized mostly by depression), [its] usefulness, and also [its] authority (Kendell,
manic-depressive reaction and psychotic depres- 1975, p. 95). In 1965, the American Psychiatric
sive reaction (within the affective reactions section), Association appointed a committee, chaired by
depressive reaction (within the psychoneurotic dis- Ernest M.Gruenberg, to revise DSM-I to be com-
orders section), and cyclothymic personality (within patible with ICD-8 and yetalso be suitable for use
the personality disorders). within the United States. The final version was
At this time, though, fundamental criticisms approved in 1967, with publication in1968.
regarding the reliability and validity of psychiat- The diagnosis of mental disorders, however,
ric diagnosis were also being raised (e.g., Zigler & was continuing to receive fundamental criticism
Phillips, 1961), with some even suggesting that the (e.g., Rosenhan, 1973). Amajor problem was the
concept of a mental disorder was largely a myth absence of empirical support for the reliability,
(e.g., Scheff, 1966; Szasz, 1960). A widely cited letalone validity, of the diagnoses (e.g., Blashfield
reliability study by Ward, Beck, Mendelson, Mock, & Draguns, 1976). Some researchers, therefore,
and Erbaugh (1962) concluded that the poor reli- took to heart the recommendations of Stengel
ability of mental disorder diagnosis was due largely (1959) to develop more specific and explicit crite-
to the inadequacies of DSM-I; specifically, its failure rion sets (Blashfield, 1984). The most influential of
to provide specific, explicit guidelines as to the diag- these initiatives was provided by a group of psychi-
nostic criteria for each respective disorder. atrists and psychologists at Washington University
The ICD-6 did not fare well internationally. The in St. Louis. They demonstrated that their relatively
mental disorders section [of ICD-6] failed to gain specific and explicit diagnostic criteria could obtain
[international] acceptance and eleven years later replicable research findings. Their criterion sets gen-
was found to be in official use only in Finland, New erated so much interest that they were published
Zealand, Peru, Thailand, and the United Kingdom separately in what became one of the most widely
(Kendell, 1975, p. 91). The WHO commissioned cited papers in psychiatry (i.e., Feighner et al.,
a review by the English psychiatrist Erwin Stengel, 1972). As expressed recently by Kendler, Munoz,
who, in 1959, reiterated the importance of estab- and Murphy (2010), the renewed interest in diag-
lishing an official nomenclature: A...serious nostic reliability in the early 1970ssubstantially
obstacle to progress in psychiatry is difficulty of influenced by the Feighner criteriaproved to be
communication. Everybody who has followed the a critical corrective and was instrumental in the
literature and listened to discussions concerning renaissance of psychiatric research witnessed in the
mental illness soon discovers that psychiatrists, even subsequent decades (p.141).
those apparently sharing the same basic orientation, The Feighner et al. (1972) criterion sets were
often do not speak the same language (Stengel, confined to just the 14 psychiatric illnesses and
1959, p.601). one secondary condition of primary interest to
Stengel (1959) attributed the failure of clinicians the Washington University researchers. Included
to accept the mental disorders section of ICD-6 therein were primary affective disorder (depression
to the presence of theoretical disputes, cynicism and mania) and secondary depression. Secondary
regarding psychiatric diagnosis (some theoreti- depression was not considered a real psychiatric
cal perspectives opposed the use of any diagnostic illness, as it occurred in the context of another
terms), and the presence of abstract, highly infer- mental disorder or a life-threatening medical illness.
ential diagnostic criteria that hindered consistent, Their approach to diagnosis was greatly expanded by
uniform applications. Robert Spitzer, a technical consultant for DSM-II
(American Psychiatric Association, 1968), into a
ICD-8 andDSM-II manual that covered a much wider variety of disor-
Work began on ICD-8 soon after Stengels 1959 ders, titled the Research Diagnostic Criteria (RDC)
report (ICD-6 had been revised to ICD-7 in 1955, (Spitzer, Endicott, & Robins, 1978), which was
but there were no revisions to the mental disorders subsequently adopted by many research programs
section). The text was approved by the WHO in around theworld.

Widiger,Gore 13
ICD-9 and DSM-III DSM-III, psychiatric diagnosis became a major
By the time the work of Feighner etal. (1972) was focus of scientific investigation. It was not dif-
published, work was nearing completion on ICD-9. ficult to find persons who wanted to be involved
The authors of ICD-9 had decided to include a sup- in the development of DSM-III-R, and everyone
plementary glossary in the spirit of Stengel (1959) who was involved (and many more who were not)
to facilitate reliable diagnosis, but it was appar- wanted to have a significant impact. Ironically,
ent that they would not include the more specific there were considerably more persons working
and explicit criterion sets developed and used in on DSM-III-R than had worked on DSM-III, yet
research settings (Kendell, 1975). Robert Spitzer its stated mission was far more conservative. Not
was appointed to chair the revision of DSM-II in a surprisingly, in the end there were many propos-
manner that would be compatible with ICD-9 but als for major revisions and even new diagnoses,
also incorporate many of the advances in diagnosis some of which proved to be highly controversial.
currently being developed. DSM-III was published Four of the diagnoses approved for inclusion by
in 1980 and was remarkably innovative, including Spitzer and the DSM-III-R central committee (i.e.,
(1)a multiaxial diagnostic system (most mental dis- late luteal phase dysphoric disorder [the name for
orders were diagnosed on Axis I, personality and which was subsequently changed to premenstrual
specific developmental disorders were diagnosed on dysphoric disorder], self-defeating personality dis-
Axis II, medical disorders on Axis III, psychosocial order, sadistic personality disorder, and paraphiliac
stressors on Axis IV, and level of functioning on rapism) were vetoed by the Board of Trustees of the
Axis V), (2)specific and explicit criterion sets for all American Psychiatric Association. Aconcern com-
but one of the disorders (i.e., schizoaffective), (3)a mon to all of them was that their inclusion might
substantially expanded text to facilitate diagnosis result in significant negative social consequences,
(e.g., age of onset, sex ratio, course, and familial to women in particular. For example, premenstrual
pattern), and (4) removal of terms (e.g., neurosis) dysphoric disorder could contribute to a consider-
that appeared to favor a particular theoretical model able amount of stigmatization with respect to nor-
(American Psychiatric Association, 1980; Spitzer, mal premenstrual dysphoria, as well as potentially
Endicott, & Robins, 1975; Spitzer, Williams, & excessive and/or unnecessary pharmacotherapy.
Skodol,1980). Arelated concern was the lack of sufficient empiri-
cal support to address or offset the concerns regard-
DSM-III-R ing potential harm and misusage. A compromise
Many of the criterion sets developed for was eventually reached in which late luteal phase
DSM-III lacked much prior research or field test- dysphoric disorder (Endicott, 2000) and the two
ing. Most were constructed by work group mem- personality disorders were included in an appendix;
bers with little guidance as to how they would paraphiliac rapism was deleted entirely.
actually perform in general clinical practice or even
research settings. As a result, a number of obvious ICD-10 andDSM-IV
errors occurred (e.g., panic disorder in DSM-III Work on DSM-III-R was supposed to have been
could not be diagnosed in the presence of major completed in 1985, but given the ever-expanding
depression). Criteria were not entirely clear, were breadth of its new additions and revisions, by the
inconsistent across categories, or were even contra- time work was completed on DSM-III-R work
dictory (American Psychiatric Association, 1987, had already begun on ICD-10. The decision of the
p. xvii). The American Psychiatric Association authors of DSM-III to develop an alternative to
therefore authorized the development of a revision ICD-9 (i.e., include specific and explicit criterion
to DSM-III to make corrections and refinements. sets) was instrumental in developing a highly inno-
Fundamental revisions were to be tabled until work vative and internationally popular manual (Kendell,
began on ICD-10. 1991; Spitzer etal., 1980). However, its innovations
However, it might have been unrealistic to also came at the cost of decreasing compatibility
expect the authors of DSM-III-R to confine their with the ICD-9 nomenclature, used throughout
efforts to refinement and clarification, given the much of the rest of the world, which is problem-
impact and acclaim of DSM-III (Blashfield, 1984; atic to the stated purpose of providing a common
Klerman, 1986). Prior to DSM-III few psychia- language of communication. In 1988 the American
trists or psychologists were particularly interested Psychiatric Association appointed the DSM-IV Task
in diagnosis and classification. Subsequent to Force chaired by Allen Frances (Frances, Widiger,

14 Diagnostic and Statistical Manual(DSM)

& Pincus, 1989). Mandates for DSM-IV included DSM-IV-TR
better coordination with ICD-10 and improved One of the innovations of DSM-III was the
documentation of empirical support. inclusion of a relatively detailed text discussion
The DSM-IV committee aspired to use a more of each disorder, including information on age of
conservative threshold for the inclusion of new onset, course, gender, and familial pattern (Spitzer
diagnoses and to have decisions be governed more et al., 1980). This text was expanded in DSM-IV
openly and explicitly by the scientific literature to include culture and ethnicity, life-span develop-
(Frances etal., 1989). Proposals for additions, dele- ment, and laboratory and physical exam findings
tions, or revisions were guided by literature reviews, (American Psychiatric Association, 1994; Frances,
which were required to use a specific meta-analysis First, & Pincus, 1995). Largely excluded from the
format that maximized the potential for informative text was information concerning etiology, pathol-
critical review, containing (for example) a method ogy, and treatment, as this material was considered
section that documented explicitly the criteria for to be too theoretically specific and more suitable for
including and excluding studies and the process by academic texts. Nevertheless, it had also become
which the literature had been reviewed (Widiger & apparent that DSM-IV was being used in some
Trull, 1993). The purpose of this structure was to settings as a textbook, and the material on age,
make it easier to discover whether the author was course, prevalence, and family history was quickly
confining his or her review only to studies that were becoming outdated as new information was being
consistent with a particular proposal and failing gathered.
to acknowledge opposing perspectives. It was not Therefore, in 1997, the American Psychiatric
unusual in the development of DSM-IV to find Association appointed the DSM-IV Text Revision
that proponents of a proposed revision attempted Work Group, chaired by Michael First (editor of
to limit their review largely to the studies that sup- the text and criterion sets for DSM-IV) and Harold
ported their proposal, neglecting to acknowledge Pincus (vice-chair for DSM-IV) to update the text
issues and findings inconsistent with their posi- material. No substantive changes in the criterion
tion (Frances & Widiger, 2012; Widiger & Trull, sets were to be considered, nor were any new addi-
1993). The literature reviews were distributed for tions, subtypes, deletions, or other changes in the
critical review, many were submitted to journals status of any diagnoses to be implemented. In addi-
for peer review, and all were published within the tion, each of the proposed revisions to the text had
three-volume DSM-IV Sourcebook (e.g., Widiger to be supported by a systematic literature review
etal.,1994). critiqued by multiple advisors. The DSM-IV Text
Testable questions that could be addressed with Revision (DSM-IV-TR) was published in 2000
existing datasets were also explored in additional (American Psychiatric Association,2000).
studies, which emphasized the aggregation of mul- The outcome, however, was not entirely consis-
tiple datasets from independent researchers, prefera- tent with the original intentions. Although it was
bly with opposing theoretical perspectives (Widiger stated in the introduction to DSM-IV-TR that no
& Trull, 1993). In addition, 12 field trials were substantive changes in the criteria sets were con-
conducted to provide reliability and validity data sidered (American Psychiatric Association, 2000,
on proposed revisions. The primary purposes of the p.xxix), substantive revisions were in fact made to
field trials were to address fundamental questions the criterion sets for tic disorders and for the para-
or concerns with regard to a particular proposal, philias involving a nonconsenting victim (First &
test alternative proposals, and compare and contrast Pincus, 2002), the latter owing to concerns of mis-
the proposals to the existing DSM-III-R (American application within forensic settings (First & Halon,
Psychiatric Association, 1987), often with respect 2008; Frances, 2010). In addition, no documenta-
to external validators. The results of the field tri- tion of the scientific support for the text revisions
als were published in the fourth volume of the was ever provided owing to the inconsistency in the
DSM-IV Sourcebook (Widiger etal., 1998). Critical quality of the effort. Some authors provided excel-
reviews of these additional projects were obtained lent documentation; however, others confined their
by sending initial drafts to advisers or consultants citations largely to their own work and still others
for a respective work group, by presenting drafts at provided grossly inadequate to no documentation
relevant conferences, and by submitting reports to at all. Rather than have inconsistent and/or inad-
peer-reviewed journals (Widiger, Frances, Pincus, equate documentation, it was decided to have none
Davis, & First, 1991; Widiger & Trull,1993). atall.

Widiger,Gore 15
Issues forDSM-5 within the individual (i.e., DSM-IV-TR does not
Work is now well under way for DSM-5, chaired include relationship disorders); the behavior or syn-
by Drs. David Kupfer and Darrel Regier, with an drome is associated with clinically significant dis-
anticipated publication date of 2013. DSM-5 is tress, disability, impairment, and/or loss of freedom;
likely to include a number of major revisions. The the behavior or syndrome is not an expectable and
proposals were posted online February 10, 2010, culturally sanctioned response to a particular event
and subsequently revised in January 2011; June (such as the death of a loved one); and the behav-
2011; and April 2012 (see The ior or syndrome is not simply a culturally deviant
content and process of DSM-5 has been controver- or denounced behavior (e.g., political, religious,
sial (Frances, 2009). Four issues discussed here are or sexual) that places the person in conflict with
(1)the definition of mental disorder, (2)the empiri- society (American Psychiatric Association, 2000).
cal support for proposed revisions, (3)shifting to a Contributing to the decision that homosexuality
dimensional model, and (4)shifting to a neurobio- was not a mental disorder was the conclusion that
logical model (Widiger & Crego,2013). the distress and impairment experienced by persons
receiving the diagnosis were due largely to formal
Definition of Mental Disorder and informal condemnation of homosexuality by a
A fundamental concern of the diagnostic manual significant proportion of society (Spitzer,1981).
is what constitutes a mental disorder. The bound- The intense controversy that surrounded the
aries of the DSM have been increasing with each inclusion of homosexuality within prior editions
edition (Kirk, 2005) and there has long been vocal of the diagnostic manual has largely abated, but
concern that much of this expansion represents the issues raised in reference to the definition of
an encroachment into normal problems of living mental disorder continue to fester. Consider, for
(Caplan, 1995; Folette & Houts, 1996; Maddux, example, major depressive disorder and bereave-
Gosselin, & Winstead, 2008). The authors of ment. Currently, clinicians are discouraged from
DSM-5 have been proposing quite a few new diag- diagnosing persons with a mood disorder if the
noses, such as paraphilic coercive disorder, hyper- depression is in response to the loss of a loved one
sexual disorder, olfactory reference syndrome, and has lasted less than two months (American
hoarding, skin picking disorder, pedohebephilia, Psychiatric Association, 2000). Two months is,
disinhibited social engagement disorder, nonsui- of course, a rather arbitrary point of demarcation
cidal self-injury, behavioral addiction, minor neuro- between normal and abnormal grieving but, most
cognitive disorder, attenuated psychosis syndrome, importantly, the exclusion criterion is clearly an
and binge eating disorder. Of particular relevance effort of the American Psychiatric Association to
to mood disorder researchers and clinicians would avoid pathologizing what is considered by most per-
be the proposals for disruptive mood dysregulation sons within society to be a normal, natural response
disorder (previously titled temper dysregulation to the loss of a loved one (Horwitz & Wakefield,
of childhood), mixed anxiety-depressive disorder, 2007). Wakefield and colleagues have argued,
and premenstrual dysphoric disorder. There is also however, that it is also arbitrary to single out the
a proposal to remove or modify the bereavement loss of a loved one with such unique importance
exclusion criterion for major depressive disorder as (e.g., Wakefield, Schmitz, First, & Horwitz, 2007).
well as to include a new diagnosis of persistent com- Persons also become understandably depressed in
plex bereavement disorder. response, for instance, to the loss of employment,
Ideally one should be able to determine what is marital dissolution, or a life-threatening physical
or is not a mental disorder based upon an explicit illness. Wakefield et al., therefore, have suggested
definition of what constitutes a mental disorder. expanding the exclusion criterion to include these
The definition of mental disorder provided in (and other) additional losses.
DSM-IV-TR (American Psychiatric Association, The DSM-5 Mood Disorders Work Group,
2000) was the result of an effort by the authors of however, did not embrace Wakefields argument.
DSM-III to develop specific and explicit criteria to On the contrary, they proposed to eliminate the
be used for deciding whether a behavior pattern bereavement exclusion criterion in part because it
(homosexuality in particular) should be classified is illogical and inconsistent to distinguish the loss of
as a mental disorder (Spitzer & Williams, 1982). a loved one from other comparable stressors, such
The primary features of the definition of mental as rape, betrayal by a spouse, or developing cancer
disorder included in DSM-IV-TR are occurrence (Kendler, 2010). A removal of the bereavement

16 Diagnostic and Statistical Manual(DSM)

exclusion criterion would be explicitly inconsistent objection. The DSM-5 Mood Disorders Work
with the DSM-IV-TR definition of mental dis- Group therefore modified the proposal somewhat.
order stating that the syndrome or pattern must They acknowledged that a normal and expected
not be merely an expectable and culturally sanc- response to a significant loss (e.g., bereavement, as
tioned response to a particular event, for example, well as financial ruin or natural disaster) will include
the death of a loved one (American Psychiatric feelings of intense sadness, rumination, insomnia,
Association, 2000, p.xxxi) as well as the proposed poor appetite, and weight loss and therefore could
definition for DSM-5, which continues, so far, to resemble an actual mood disorder. When these
state that an expectable response to the loss of a symptoms reach the level of feelings of worthless-
loved one would not be a mental disorder (Stein ness, suicidal ideation, psychomotor retardation,
et al., 2010). However, the DSM-5 definition of and severe impairment, they would be considered
mental disorder would likely be revised to be con- to be a mood disorder even if they are in response to
sistent with the Mood Disorders Work Group deci- the loss of a loved one (or in response to any other
sion if the latter committee decides to remove this loss). Proposed for an appendix to the DSM-5 (for
exclusion criterion. disorders requiring further research) would be a
Research has questioned whether there is any diagnosis of persistent complex bereavement dis-
real distinction between a depressive response to the order that involves a preoccupation with the loss
loss of a loved one and a depressive response to some of the loved one for more than 12months (for chil-
other type of loss (e.g., Kendler, Myers, & Zisook, dren only 6months).
2008; Wakefield et al., 2007; Zisook & Kendler, Wakefield (1992) developed an alternative
2007), nor does there appear to be a meaningful harmful dysfunction definition of mental disor-
difference between depression secondary to bereave- der wherein dysfunction is a failure of an internal
ment and major depressive disorder independent of mechanism to perform a naturally selected function
bereavement (Lamb, Pies, & Zisook, 2010) other (e.g., the capacity to experience feelings of guilt in
than perhaps simply the severity of the mood (e.g., a person with antisocial personality disorder) and
Wakefield etal., 2007). First (2011) and Wakefield harm is a value judgment that the design failure
(2011; Wakefield & First, 2012) argue that removal is harmful to the individual (e.g., failure to learn
of the exclusion criterion will likely contribute from mistakes results in repeated punishments,
to a considerable amount of pathologizing what arrests, loss of employment, and eventual impov-
is largely a normal human response (Horwitz & erishment). Wakefields definition has received
Wakefield, 2007). However, the fact that the depres- substantial attention and was being considered for
sion is an expectable response to the loss of a loved inclusion in DSM-5 (Rounsaville et al., 2002).
one might not negate the occurrence of a disorder However, his proposal has also received quite a bit
any more than it does in response to any other loss of critical review (e.g., Bergner, 1997; Kirmayer &
or stressor. Instead, it could be providing a partial Young, 1999; Lilienfield & Marino, 1999; Widiger
but necessary explanation for the etiology of the & Sankis, 2000) and is no longer under serious con-
disorder, just as it does for other losses (Widiger, sideration (Stein etal.,2010).
2013). It is natural and normal to develop the flu in Missing from Wakefields (1992) definition
response to contact with a virus or to die in response of mental disorder and the DSM-IV-TR is any
to a severe trauma to the head. Because the response reference to dyscontrol (unless one stretches the
is natural and expected does not imply or suggest concept of the loss of freedom included within
that the response is healthy or should not be con- the DSM-IV-TR definition of mental disorder to
sidered a disorder. It is true that in many instances include most forms of dyscontrol). Mental dis-
of a depressive reaction in response to the loss of orders are perhaps best understood as dyscon-
a loved one the depression resolves without profes- trolled impairments in psychological functioning
sional intervention (Wakefield & First, 2012), but (Kirmayer & Young, 1999; Klein, 1999; Widiger &
this is also true for many minor physical illnesses Trull, 1991). Involuntary impairment remains the
and injuries. The fact that medical treatment is not key inference (Klein, 1999, p.424). Dyscontrol is
required for the depression to remit does not imply a fundamental component within Bergners (1997)
or suggest that a disorder did not in factoccur. significant restriction and Widiger and Sankiss
Nevertheless, the DSM-5 proposal that depres- (2000) dyscontrolled maladaptivity definitions of
sion in response to the death of a loved one is a mental disorder. Dyscontrol might even provide a
mental illness received considerable criticism and basis for a fundamental distinction between mental

Widiger,Gore 17
and physical disorder, as dyscontrol is not a mean- through their research empirical support for a pro-
ingful consideration for a physical disorder. posed diagnosis, accompanied by relatively little
It is perhaps the inability or difficulty to alter familiarity or even interest in how the diagnosis
or adjust problematic feelings, thoughts, or behav- might be misapplied within routine clinical prac-
iors that suggests the presence of a mental disorder. tice, the potential impact of drug company mar-
To the extent that persons willfully, intentionally, keting, health economics, forensic misuse, and the
freely, or voluntarily engage in harmful sexual acts, many other social-clinical impacts ofaDSM.
gambling, drug usage, or child abuse, they would Frances, the chair of DSM-IV, had suggested
not be considered to have a mental disorder. Persons that the major innovation of DSM-IV will not
seek professional intervention in large part to obtain be in its having surprising new content but rather
the insights, techniques, skills, or other tools (e.g., will reside in the systematic and explicit method
medications) that help increase their ability to bet- by which DSM-IV will be constructed and docu-
ter control and manage their mood, thoughts, or mented (Frances et al., 1989, p. 375). Frances
behavior. Of course, if a mental disorder is present (2009) has suggested more recently that the authors
when there is both dyscontrol and impairment, of DSM-5 may have flipped this priority on its
then depression in response to the loss of a loved head, with emphasis being given to surprising new
one would be classified as a mental disorder to the content and inadequate attention to first conduct
extent that the person was unable to control his or systematic, thorough, and balanced reviews to
her feelings of depression and these depressive feel- ensure that the proposals have adequate justifica-
ings were resulting in significant social, personal, or tion and empirical support and that the potential
occupational impairment. costs and risks of the proposed diagnosis have also
been addressed. It is perhaps ironic that a drug com-
Empirical Support pany cannot release a medicine into clinical practice
In the absence of a clear consensus as to what without first addressing, empirically, the potential
constitutes a mental disorder, the lack of any gold costs and risks of the medicine along with its pur-
standard for any particular diagnosis, and the pres- ported benefits, yet there is no comparable control
ence of considerable dispute and debate within the over the release of new diagnoses that will likely be
field regarding the validity of respective mental treated with a variety of potentially harmful medi-
disorder diagnoses, one might anticipate that deci- cations. Concerns with respect to the process with
sions for major revisions would be supported by a which DSM-5 was being constructed were perhaps
substantial body of empirical evidence. Proposals first raised by Robert Spitzer, Chair of DSM-III and
for DSM-III-R were rejected by the American DSM-III-R, after having been denied access to the
Psychiatric Associations Board of Trustees because minutes of DSM-5 Work Group meetings (Decker,
there was inadequate research to address concerns 2010). Frances and Spitzer eventually submitted a
with respect to possible risks and costs (Endicott, joint letter to the American Psychiatric Associations
2000). Amandate given to the authors of DSM-IV Board of Trustees on July 7, 2009, expressing a
was to be more cautious and to provide better docu- variety of concerns with respect to the process with
mentation for the decisions that were made (Frances which DSM-5 was being constructed.
etal.,1989). The chair and vice-chair of DSM-5 have stated
DSM Work Group members are typically spe- that the development of DSM-5 is following the
cialists within a particular area of diagnosis and clas- procedure used for DSM-IV, including literature
sification. Such experts are necessary, but they tend reviews, data reanalyses, and field trials (Regier,
to be focused primarily on the perceived failure of Narrow, Kuhl, & Kupfer, 2009). Kendler, Kupfer,
the diagnostic manual to provide adequate cover- Narrow, Phillips, and Fawcett (2009) developed
age (Frances & Widiger, 2012). Their primary and guidelines for DSM-5 work group members indi-
at times perhaps sole concern is with the potential cating that any change to the manual should be
false negativesthe missed diagnosis or the patient accompanied by a discussion of possible unin-
who does not fit within an existing diagnosis. They tended negative effects of this proposed change, if
tend to be relatively disinterested in the false posi- it is made, and a consideration of arguments against
tives: patients who receive unnecessary diagnosis, making this change should also be included (p.2).
treatment, or stigma and consequently must bear Kendler et al. further stated that the larger and
unnecessary expense. Experts also tend to be very more significant the change, the stronger should
confident about their positions, having developed be the required level of [empirical] support (p.2).

18 Diagnostic and Statistical Manual(DSM)

Some of the DSM-5 literature reviews posted on the or stimulate research. The same spirit would seem
DSM-5 website do appear to meet the spirit of the to be present for DSM-5, as the official guidelines
Kendler et al.s guidelines (e.g., see the review for state that a new diagnosis should have support
hypersexual disorder; However, from several high priority validators and should
others perhaps not so much (e.g., see the reviews for rarely if ever be based solely on reports from a single
mixed-anxiety depressive disorder and premenstrual researcher or research team (Kendler etal., 2009,
dysphoric disorder; p. 5). However, Regier, the vice chair of DSM-5,
The letter by Frances and Spitzer was initiated acknowledged in an interview with Medscape Today
because the field trials for DSM-5 were about to that some of the proposals are not as well studied
begin before the proposals had received any critical as others and we recognize that, but we cant move
or external review. Their letter was perhaps instru- forward without some of these put into practice
mental in the delay of the field trial, allowing time (Brauser [interviewer] and Regier [interviewee],
for the proposals to be posted on a website for pub- 2011). He noted that it is difficult to get researchers
lic review (Decker, 2010). Some of the proposals to study disorders that are not included within the
have since been curtailed or significantly modified diagnostic manual and indicated that a purpose of
in response to the external critiques, but concerns the DSM-5 will be to generate the research neces-
about the process do still remain. For example, in a sary for the validation of the proposal. Thats what
step back from the field trials that were conducted the DSM isa set of scientific hypotheses that are
for DSM-IV, the DSM-5 field trials do not necessar- intended to be tested and disproved if the evidence
ily include the DSM-IV-TR criterion sets or exter- isnt found to support them (Brauer [interviewer]
nal validators, and will therefore in many instances and Regier [interviewee], 2011). This perspec-
be unable to provide information concerning a shift tive, however, may be shortsighted, failing to con-
in the prevalence, reliability, or validity of the diag- sider the clinical implications of including new,
nostic manual resulting from proposed revisions. lesser-researched diagnoses. An alternative perspec-
The threshold for the inclusion of a new diagno- tive is that the DSM is a manual used by clinicians
sis within the DSM has varied across each edition. and various social and forensic agencies to identify
For DSM-III, Spitzer and colleagues indicated that what the American Psychiatric Association has
the threshold for inclusion was fairly liberal: determined are sufficiently well-validated disorders
to warrant official recognition.
Because the DSM-III classification is intended for the
This tension between the liberal inclusive
entire profession, and because our current knowledge
approach and a more conservative restrictive
about mental disorder is so limited, the Task Force
approach is evident with respect to the proposal
has chosen to be inclusive rather than exclusive.
to include in DSM-5 mood dysregulation disor-
In practice, this means that whenever a clinical
der (previously temper dysregulation disorder with
condition can be described with clarity and relative
dysphoria), a disorder involving outbursts of anger
distinctness, it is considered for inclusion. If there is
and rage with an onset prior to age 10. The impetus
general agreement among clinicians, who would be
for the proposal is the marked increase in the rate
expected to encounter the condition, that there are a
at which children are being diagnosed with bipolar
significant number of patients who have it and that
mood disorder in the absence of adequate scientific
its identification is important in their clinical work,
support for the validity of this diagnosis in children
it is included in the classification. (Spitzer, Sheehy, &
(Blader & Carlson, 2007; Fawcett, 2010; Moreno
Endicott, 1977,p.3)
et al., 2007). As acknowledged by the DSM-5
This liberal threshold, though, led to a number Childhood and Adolescent Disorders Work Group
of controversial proposals for DSM-III-R (e.g., (2010), the work has been done predominately
premenstrual dysphoric disorder) that were, as by one research group in a select research setting,
noted earlier, vetoed by the American Psychiatric and many questions remain unanswered (p. 4).
Associations Board of Trustees. The threshold for This point is reiterated within a joint statement by
inclusion in DSM-IV was considerably more con- the DSM-5 Mood Disorders and Childhood and
servative (Frances etal., 1989), in a spirit of holding Adolescent Disorders Work Groups (2010) that
the line on proliferation (Pincus, Frances, Davis, both work groups are concerned by the fact that
First, & Widiger, 1992, p.112). Specifically, it was the work on severe mood dysregulation [in children]
indicated that DSM-IV would be informed and has been done predominately by one research group
guided by research; its purpose was not to generate in a select research setting (p.6). Nevertheless, it is

Widiger,Gore 19
felt that problematic overdiagnosis in clinical prac- treatment of one or another, each presumably with
tice could be addressed through the development of its own etiology and pathology (Feinstein, 1970).
specific and explicit diagnostic criteria that might be There are indeed many instances in which the pres-
sufficiently restrictive in its coverage. Of course, the ence of multiple diagnoses suggests the presence of
existence of the diagnosis within the manual may distinct yet comorbid psychopathologies. There is
also lend the condition a considerable degree of cred- clearly a substantial body of research on the many
ibility and perceived validity that it does not really disorders and conditions with which depression
deserve (Hyman, 2010). The authors of the proposal is comorbid, including substance use disorders,
suggest that its inclusion will generate the research schizophrenia, sexual dysfunctions, coronary heart
needed for its validation. Indeed, it can be argued disease, cancer, diabetes, and many, many more
that one of the major take-home messages from the covered within this book. The high prevalence of
controversy about the diagnosis of pediatric bipolar diagnostic comorbidity throughout the diagnostic
disorder is the fact that the research needs of a large manual was itself the primary focus of the National
population of children with severe irritability are not Comorbidity Survey (NCS), a congressionally man-
being met, particularly with respect to clinical tri- dated study designed to identify extent of mental dis-
als (DSM-5 Mood Disorders and Childhood and order comorbidity within an extensive (N=8,098)
Adolescent Disorders Work Groups, 2010, p.8). As and representative sample of the United States, aged
expressed by the DSM-5 Childhood and Adolescent 15 to 54. Kessler etal. (1994) reported on the basis
Disorders Work Group (2010), the inclusion of this of this study that the most common lifetime disor-
new diagnosis will help to jump-start research on der was major depression (17%) and that the vast
severe irritability in youth (p.9). majority of the lifetime disorders were comorbid
(i.e., 79%). They concluded that the major burden
Shifting to a Dimensional of psychiatric concentrated in a group
Model of Classification? of highly comorbid people (p.11). Areplication of
DSM-IV-TR is a categorical classification the NCS was subsequently conducted (N=9,090)
that divides mental disorders into types based on in a face-to-face survey from 2001 to 2002 using
criterion sets with defining features (American DSM-IV criterion sets (Kessler, Berglund et al.,
Psychiatric Association, 2000, p.xxxi). This categor- 2003). They reported a lifetime prevalence rate of
ical classification is consistent with the medical tra- 16.2% for major depressive disorder, with 72% of
dition in which it is believed (and often confirmed these persons having a comorbid conditionmost
in other areas of medicine) that each disorder has a often an anxiety disorder (59%), substance use dis-
unique and quite specific etiology, pathology, and order (24%), or impulse control disorder (30%).
treatment (Zachar & Kendler, 2007). The intention More recently, a third-wave NCS data collection
of the diagnostic manual is to help the clinician was confined to adolescents (N = 10,123) aged
determine which particular disorder is present, the 13 to 18. These investigators reported a lifetime
diagnosis of which would indicate the presence of prevalence rate of 14% for a mood disorder (which
a distinct pathology that would explain the occur- actually included borderline personality disorder)
rence of the symptoms and suggest a specific treat- (Merikangas etal., 2010) and a previous 12-month
ment that would ameliorate the patients suffering prevalence of 8% (Kessler et al., 2012). Again,
(Frances etal., 1995; Kendell,1975). about 40% of affected youth...reported more than
It is evident, however, that DSM-IV-TR rou- one class of lifetime disorder, with mood disorder
tinely fails in guiding a clinician to the identification being the most likely to co-occur with other classes
of one specific disorder. Despite the best efforts of (Merikangas etal., 2010, p.987).
the authors of each revision to the diagnostic man- Comorbidity, however, can in some cases suggest
ual to revise the criterion sets to increase their speci- a problem for the diagnostic manual; more specifi-
ficity, multiple diagnoses are common (Craddock & cally, the lack of distinct boundaries between two
Owen, 2005; Kendell & Jablensky, 2003; Widiger independent diagnoses and perhaps even the pres-
& Clark, 2000; Widiger & Gore, 2012). Diagnostic ence of a common shared pathology (Widiger &
comorbidity, a primary focus of this text, is perhaps Gore, 2012). For example, Krueger and Markon
the norm rather than the exception (Krueger & (2006) have written extensively about the meanings
Markon,2006). of comorbidity; they discuss whether or not effec-
The term comorbidity refers to the co-occurrence tive treatments target situational mental disorders or
of distinct disorders, impacting the course or more core, long-term processes (Krueger & Markon,

20 Diagnostic and Statistical Manual(DSM)

2006). As expressed by the vice chair of DSM-5, reliably distinct boundaries, an absence of a credible
the failure of DSM-III criteria to specifically rationale for diagnostic thresholds, an inadequate
define individuals with only one disorder served as coverage of existing clinical populations, and the
an alert that the strict neo-Kraepelinian categorical absence of specific etiologies and treatments (Kupfer
approach to mental disorder diagnoses advocated etal., 2002; Widiger & Samuel, 2005). Nowhere is
by Robins and Guze (1970), Spitzer, Endicott, & this more evident than in the classification of mood
Robins (1978), and others could have some seri- disorders. For example, depression is a section of the
ous problems (Regier, 2008, p.xxi). In an effort to diagnostic manual that does have considerable dif-
address one form of comorbidity involving anxiety ficulty identifying or defining a clear boundary with
and depression, a diagnosis of mixed-anxiety depres- normal sadness. Narrow, Rae, Robins, and Regier
sive disorder was proposed for DSM-5 (as well as (2002) proposed raising the threshold for a mood
for DSM-IV) that explicitly acknowledged the large disorder diagnosis in response to the perception that
numbers of persons who could not be distinguished the prevalence rates reported in the NCS and other
as having either an anxiety or a mood disorder comparable studies were felt to be excessively high
(Katon & Roy-Byrne, 1991; Presig, Merikangas, & and certainly much larger than could be currently
Angst, 2001), albeit this disorder would have been treated. However, even subthreshold cases of depres-
arbitrarily classified as a mood disorder in DSM-5 sion (i.e., persons with depressive symptoms below
(this proposal, however, has been withdrawn). the existing threshold for a DSM-IV-TR mental
Most (if not all) mental disorders appear to be disorder diagnosis) are responsive to pharmacologic
the result of a complex interaction of an array of interventions, seek treatment for their sadness, and
biological vulnerabilities and dispositions with a are often being treated within primary care settings
number of significant environmental/psychosocial (Judd, Schettler, & Akiskal, 2002; Magruder &
events that often exert their progressive effects over Calderone, 2000; Pincus, McQueen, & Elinson,
a developing period of time (Rutter, 2003). There 2003). Kessler, Merikangas et al. (2003) reported
may never be a single gene that is the cause of a par- that elevated risk of [negative life] outcomes among
ticular mental disorder. Each individuals mental mild cases versus noncases is consistently larger than
disorder is more likely to be the result of an array the elevated risk among moderate cases versus mild
of genetic dispositions and vulnerabilities (Frances cases (p.1121). Kessler, Merikangas etal. argued
& Widiger, 2012). The symptoms and pathologies that mild cases should be retained in the defini-
of mental disorders are also highly responsive to a tion of disorders (p. 1121) to acknowledge that
wide variety of neurobiologic, interpersonal, cogni- mental disorders (like physical disorders) vary in
tive, and other mediating and moderating variables severity (p. 1121). In fact, because persons with
that help to develop, shape, and form a particular subthreshold levels of depression are seeking and
individuals psychopathology profile. This complex are responsive to treatment, there was a proposal
etiological history and individual psychopathology to include within an appendix to DSM-IV (but
profile are unlikely to be well described by single not considered for DSM-5) a diagnosis of minor
diagnostic categories that attempt to make distinc- depressive disorder, which it is acknowledged can
tions at nonexistent discrete joints along the con- be difficult to distinguish from periods of sadness
tinuous distributions (Widiger & Samuel, 2005). that are an inherent part of everyday life (American
As expressed by Kupfer, First, and Regier (two of Psychiatric Association, 2000, p.776).
whom are the chair and vice chair of DSM-5): A common view is that most instances of sad-
ness do not constitute a mental disorder and that
Epidemiologic and clinical studies have shown
a major depressive disorder is qualitatively distinct
extremely high rates of comorbidities among the
from normal sadness (Horwitz & Wakefield, 2007;
disorders, undermining the hypothesis that the
Wakefield, 2001). However, a simple inspection of
syndromes represent distinct etiologies. Furthermore,
the diagnostic criteria for major depressive disorder
epidemiologic studies have shown a high degree of
would not lend confidence to a conceptualization of
short-term diagnostic instability for many disorders.
this condition as being something uniquely differ-
With regard to treatment, lack of treatment
ent from normal depression or sadness (Andrews
specificity is the rule rather than the exception.
etal., 2008). Persons who are just very sad may have
(Kupfer, First, & Regier, 2002, p.xvii)
most of the same attributes as those with a major
The categorical model of classification is failing depressive disorder (if not all of them) but just at
in many ways, including an absence of a provision of a lesser degree of severity (Lamb etal., 2010). The

Widiger,Gore 21
diagnostic criteria for major depressive disorder or natural discontinuity to support the selection of
include depressed mood, loss of interest or pleasure, any particular cutoff point to distinguish between
weight loss (or gain), insomnia (or hypersomnia), normal and clinical depression.
psychomotor retardation (or agitation), loss of Taxometric analyses is a series of related statistical
energy, feelings of worthlessness, and/or diminished techniques to detect whether a set of items is opti-
capacity to make decisions (American Psychiatric mally understood as describing (assessing) a dimen-
Association, 2000). Each of these diagnostic crite- sional or a categorical construct, providing thereby
ria is readily placed along a continuum of severity a direct test of which structural model is most
that would shade imperceptibly into what would valid in characterizing the set of items or variables
be considered a normal sadness or depression. (Ruscio, & Ruscio, 2004). Anumber of taxometric
DSM-IV-TR, therefore, includes specific thresholds studies have been conducted on measures of depres-
for each of them, but these required thresholds are sion. The first was provided by Ruscio and Ruscio
clearly arbitrary points of demarcation that simply (2000), analyzing items from the Beck Depression
distinguish a relatively higher level of severity from Inventory and, independently, items from the Zung
a lower level of severity (e.g., nearly every day or Self-Rating Depression Scale in a sample of 996
markedly diminished, and at least a two week male veterans who had received a diagnosis of post-
period; American Psychiatric Association, 2000, traumatic stress disorder but also had a high preva-
p. 356). The diagnosis requires five of these nine lence rate of major depressive disorder; they also
criteria, with no apparent rationale for this thresh- studied a sample of 8,045 individuals from the gen-
old other than it would appear to be severe enough eral population (60% female) who completed the
to be defensible to be titled as a major depressive items from the Depression scale of the Minnesota
episode as distinguished from a minor depressive Multiphasic Personality Inventory. The investiga-
episode, which is then distinguished from normal tors indicated that results of both studies, drawing
sadness (American Psychiatric Association,2000). on three widely used measures of depression, cor-
There is little empirical support for the belief roborated the dimensionality of depression (Ruscio
that there is a qualitative distinction between nor- & Ruscio, 2000, p.473). The taxometric findings of
mal sadness and clinical depression (Andrews etal., Ruscio and Ruscio (2000) have subsequently been
2008; Goldberg, 1996). Kessler, Zhao, Blazer, and replicated many times over (e.g., Prisciandaro &
Swartz (1997) examined the distribution of minor Roberts, 2005), with only a few scattered failures
and major symptoms of depression using data from (Widiger & Edmundson,2011).
the first NCS study. They considered the relation- As expressed by the chair and vice-chair of
ship of these symptoms with parental history of DSM-5, we have decided that one, if not the major,
mental disorder, number and duration of depres- difference between DSM-IV and DSM-V will be
sive episodes, and comorbidity with other forms of the more prominent use of dimensional measures
psychopathology. Respective relationships increased in DSM-V (Regier, Narrow, Kuhl, & Kupfer,
with increasing number of symptoms, with no clear 2009, p. 649; [the original acronym for DSM-5
distinct break. This finding was replicated in the was DSM-V]). In 1999, the DSM-5 Research
second NCS study when Kessler, Merikangas etal. Planning Conference was held under joint sponsor-
(2003) compared four levels of severity of depression ship of the American Psychiatric Association and
(as well as other disorders) with respect to a 10-year the National Institute of Mental Health (NIMH),
follow-up of 5,463 cases with respect to hospital- the purpose of which was to set research priorities
ization, work disability, and suicide attempts. No that would optimally inform future classifications.
consistent inflection point in the gradient of out- An impetus for this effort was frustration with the
come risk [could] be found (p.1121). Ustun and existing nomenclature (Kupfer etal., 2002). At this
Sartorius (1995) conducted a study of 5,000 primary conference, research planning work groups were
care patients in 14 countries and reported a simple formed to develop white papers that would set a
linear relationship between disability and number of research agenda for DSM-5. The Nomenclature
depressive symptoms. Sakashita, Slade, and Andrews Work Group, charged with addressing fundamen-
(2007) examined the relationship between the num- tal assumptions of the diagnostic system, concluded
ber of symptoms of depression and four measures of that it is important that consideration be given to
impairment using data from the Australian National advantages and disadvantages of basing part or all
Survey of Mental Health and Well-Being and found of DSM-V on dimensions rather than categories
that the relationship was again linear, with no clear (Rounsaville etal., 2002, p.12).

22 Diagnostic and Statistical Manual(DSM)

The white papers developed by the research of DSM-III removed terms (e.g., neurosis) that
planning work groups were followed by a series of appeared to refer to psychodynamic constructs
international conferences whose purpose was to in order to have the manual be atheoretical or at
further enrich the empirical data base in prepara- least be reasonably neutral with respect to alter-
tion for the eventual development of DSM-5. The native models of psychopathology (Spitzer et al.,
final conference was devoted to developing dimen- 1980). This approach, however, has not been well
sional approaches for classification across the entire received by proponents of these alternative theo-
diagnostic manual (Helzer, Kraemer et al., 2008), retical perspectives. Interpersonal and systems
including major depressive disorder (Andrews etal., theorists, who consider dysfunctional behavior
2008). It is evident that a primary goal for DSM-5 to be due to a pathology of a wider social system
is to shift the manual toward a dimensional clas- rather than simply within the individual, consider
sification (Helzer, Wittchen, Krueger, & Kraemer, the organismic diagnoses of DSM-IV-TR to be
2008). The single most important precondition fundamentally antithetical to their theoretical per-
for moving forward to improve the clinical and spective (Reiss & Emde, 2003). Psychodynamically
scientific utility of DSM-V will be the incorpora- oriented clinicians bemoan the fact that as the
tion of simple dimensional measures for assessing succeeding editions of the manual have become
syndromes within broad diagnostic categories and increasingly objective, descriptive, and atheoreti-
supraordinate dimensions that cross current diag- cal, they have inevitably minimized the inferential,
nostic boundaries (Regier etal., 2009, p.649). dynamic aspects of diagnosis on which most psy-
Nevertheless, the shifts to be taken in DSM-5 chodynamically oriented clinicians depend. They
are unlikely to be fundamental or perhaps even sig- have now constructed their own diagnostic manual,
nificant. What is being proposed for DSM-V is the Psychodynamic Diagnostic Manual (PDM Task
not to substitute dimensional scales for categorical Force, 2006). Behaviorists argue that the organis-
diagnoses, but to add a dimensional option to the mic perspective of DSM-IV-TR is inconsistent with
usual categorical diagnoses for DSM-V (Kraemer, the situational context of dysfunctional behavior
2008, p. 9). As acknowledged by Helzer, Kraemer, (Folette & Houts, 1996). Even neurobiologically
and Krueger (2006), our proposal not only preserves oriented psychiatry is dissatisfied. Although there
categorical definitions but also does not alter the pro- is a large body of research that indicates that a neu-
cess by which these definitions would be developed. robiological basis for most mental disorders, the
Those charged with developing criteria for specific DSM definitions are virtually devoid of biology
mental disorders would operate just as their prede- (Charney et al., 2002, pp. 3132). Of course the
cessors have (p. 1675). Dimensional proposals for dissatisfaction of each theoretical perspective with
DSM-5 are made only to develop supplementary the DSM-IV-TR might be consistent with the effort
dimensional approaches to the categorical defini- to avoid any theoretical bias or favoritism.
tions that would also relate back to the categorical The aspiration to be atheoretical, though, is
definitions (Helzer, Wittchen, etal., 2008, p.116). slowly but surely dissipating. Whereas DSM-I
These dimensions will serve only as ancillary descrip- favored a psychodynamic perspective (American
tions that will lack any official representation within Psychiatric Association, 1952; Spitzer etal., 1980),
a patients medical record (i.e., they will have no offi- the American Psychiatric Association and NIMH
cial alphanumerical code and may then not even be are shifting explicitly toward a neurobiologic orien-
communicated to any public health-care agency). For tation. This is evident in a number of ways, both
major depression, for instance, the proposal is limited implicitly and explicitly. For example, a reading of
to the inclusion of an optional supplementary rating the table of contents of any issue of the two lead-
of severity of depression. In sum, what is being pro- ing journals of psychiatry (i.e., American Journal of
posed for DSM-V is not to substitute dimensional Psychiatry and Archives of General Psychiatry) will evi-
scales for categorical diagnoses, but to add a dimen- dence a strong neurobiologic orientation for most
sional option to the usual categorical diagnoses for of the articles. DSM-IV included a new section of
DSM-V (Kraemer, 2008, p.9). In the end, DSM-5 the text devoted to laboratory and physical exam
will remain a categorical diagnostic system. findings (Frances etal., 1995). All of the laboratory
tests included therein were concerned with neuro-
Shifting to a NeurobiologicalModel biologic findings, with no reference to any labora-
The first editions of the DSM favored the tory test that would be of particular relevance to a
psychodynamic theoretical model. The authors cognitive, psychodynamic, or interpersonal-systems

Widiger,Gore 23
clinician (this section is likely to be modified and/ an equal playing field and would bias subsequent
or extended in DSM-5 to include genetic and physi- scientific research and discourse (Wakefield, 1998).
ologic risk factors; Kupfer & Reiger, 2011). The defi- It might be impossible to construct a diagnostic
nition of mental disorder in DSM-5 will likely refer manual that is truly theoretically neutral, but this is
to an underlying psychobiological dysfunction in not a compelling reason for abandoning the effort,
recognition that all behavior and psychology are particularly if the manual is to be used for research
dependent upon brain processes (Stein etal., 2010, attempting to determine the validity of alternative
p.5). The head of NIMH has indicated that priority theoretical perspectives.
for funding in the future will be given to studies that
formally adopt a clinical neuroscience perspective Conclusions
that contributes to an understanding of mental disor- Nobody is fully satisfied with DSM-IV-TR or
ders as developmental brain disorders (Insel, 2009, lacks valid criticisms of it. This is unlikely to change
p.132). This is being accomplished in part through with DSM-5. DSM-IV-TR is a well-reasoned docu-
the NIMH development of research domain crite- ment developed by leading experts based on their
ria (RDoC; Insel etal., 2010) with a strong focus review of a considerable body of existing research,
on biological processes, and emphasis on neural cir- but there is no way to prove conclusively the pres-
cuits (Sanislow etal., 2010, p.633). The RDoC ence and/or validity of any particular mental dis-
framework conceptualizes mental illnesses as brain order diagnosis (Meehl, 2010; Strauss & Smith,
disorders (Garvey etal., 2010, p.749). As indicated 2009). There will always be at least some degree of
by Miller (2010), over the next 2 to 3years, NIMH controversy. In addition, as an official diagnostic
will encourage researchers to shift from using DSM nomenclature, the DSM is an exceedingly powerful
criteria in their grant proposals to using the RDoC and influential social document, impacting a wide
categories (p. 1437), such as fear circuitry dis- range of social issues and personal lives. Nobody,
order. In an editorial concerning neuroscience, however, appears to suggest that all official diagnos-
clinical evidence, and the future of psychiatric classi- tic nomenclatures should be abandoned. The ben-
fication in DSM-5, Kupfer and Regier (2011, p.1), efits do appear to outweigh the costs (Salmon etal.,
the chair and vice-chair of DSM-5, embraced this 1917; Stengel, 1959). Everybody finds fault with
shifting tide, stating that this NIMH objective is this language, but it is clearly important and per-
consistent with our research planning conferences haps necessary for consistent clinical and research
and conclusions (p.2). communication to have an official diagnostic
It is unlikely that one could create a diagnos- nomenclature.
tic manual that is entirely neutral or atheoretical.
However, a professional nomenclature should prob- References
ably at least aspire to remain neutral with respect American Psychiatric Association. (1952). Diagnostic and statisti-
cal manual. Mental disorders. Washington, DC:Author.
to alternative theoretical models, particularly in the
American Psychiatric Association. (1968). Diagnostic and sta-
absence of any obvious superiority of one theoreti- tistical manual of mental disorders (2nd ed.). Washington,
cal perspective relative to another with respect to DC:Author.
etiology, pathology, and treatment (Widiger, 2013). American Psychiatric Association. (1980). Diagnostic and sta-
Mental disorders cannot currently be diagnosed tistical manual of mental disorders (3rd ed.). Washington,
solely on the basis of neurobiological mechanisms
American Psychiatric Association. (1987). Diagnostic and statisti-
or fully understood solely with respect to them cal manual of mental disorders (3rd ed., rev. ed.). Washington,
(Kendler, 2005). The DSM is used by clinicians and DC:Author.
researchers from a wide variety of theoretical per- American Psychiatric Association. (1994). Diagnostic and sta-
spectives, including (but not limited to) neurobio- tistical manual of mental disorders (4th ed.). Washington,
logical, psychodynamic, interpersonal, cognitive,
American Psychiatric Association. (2000). Diagnostic and statisti-
behavioral, humanistic, and interpersonal-systems cal manual of mental disorders. Text revision. (4th ed., rev. ed.).
theoretical perspectives. An important function of Washington, DC:Author.
the manual is to provide a common means for com- Andrews, G., Brugha, T., Thase, M., Duffy, F. F., Rucci, P., &
munication among and research concerning these Slade, T. (2008). Dimensionality and the category of major
depressive episode. In J. E. Helzer, H. C. Kraemer, R.
competing theoretical orientations in a language
F.Krueger, H-U. Wittchen, P. J.Sirovatka, & D. A.Regier
that would not favor one perspective over the other (Eds.), Dimensional approaches to diagnostic classifica-
(Frances et al., 1989). A language that purposely tion. Refining the research agenda for DSM-V (pp. 3551).
favors one particular perspective would not provide Washington, DC:American Psychiatric Association.

24 Diagnostic and Statistical Manual(DSM)

Bergner, R. M. (1997). What is psychopathology? And so what? Folette, W. C., & Houts, A. C. (1996). Models of scientific prog-
Clinical Psychology:Science and Practice, 4, 235248. ress and the role of theory in taxonomy development:Acase
Blader, J. C., & Carlson, G. A. (2007). Increased rates of bipolar study of the DSM. Journal of Consulting and Clinical
disorder diagnoses among U.S.child, adolescent, and adult Psychology, 64, 11201132.
inpatients, 19962004. Biological Psychiatry, 62, 107114. Frances, A. J. (2009, June 26). A warning sign on the road to
Blashfield, R. K. (1984). The classification of psychopathol- DSM-V:Beware of its unintended consequences. Psychiatric
ogy. Neo-Kraepelinian and quantitative approaches. Times, 26(8),14.
NewYork:Plenum. Frances, A. J. (2010). The forensic risks of DSM-V and how to
Blashfield, R. K., & Draguns, J. G. (1976). Evaluative criteria avoid them. Journal of the American Academy of Psychiatry
for psychiatric classification. Journal of Abnormal Psychology, and Law, 38,1114.
85, 140150. Frances, A. J., First, M. B., & Pincus, H. A. (1995). DSM-IV
Boorsboom, D. (2008). Psychometric perspectives on diagnostic guidebook. Washington, DC:American PsychiatricPress.
systems. Journal of Clinical Psychology, 64, 10891108. Frances, A. J., & Widiger, T. A. (2012). Psychiatric diagno-
Brauser, D. (interviewer) and Regier, D. (interviewee). sis: Lessons from the DSM-IV past and cautions for the
(Nov. 9, 2011). APA answers DSM-5 critics. Medscape DSM-5 future. Annual Review of Clinical Psychology, 8,
Today. Retrieved from 109130.
viewarticle/753255 Frances, A. J., Widiger, T. A., & Pincus, H. A. (1989). The
Caplan, P. J. (1995). They say youre crazy. How the worlds development of DSM-IV. Archives of General Psychiatry, 46,
most powerful psychiatrists decide whos normal. Reading, 373375.
MA:Addison-Wesley. Garvey, M., Heinssein, R., Pine, D. S., Quinn, K., Sanislow,
Charney, D. S., Barlow, D. H., Botteron, K., Cohen, J. D., C., & Wang, P. (2010). Research domain criteria (RDoc);
Goldman, D., Gur, R. E., .
Zaleman, S. J. (2002). Toward a new classification framework for research on men-
Neuroscience research agenda to guide development of tal disorders. American Journal of Psychiatry, 167, 748751.
a pathophysiologically based classification system. In D. Goldberg, D. A. (1996). A dimensional model for common
J. Kupfer, M. B. First, & D. A. Regier (Eds.), A research mental disorders. British Journal of Psychiatry, 168(suppl,
agenda for DSM-V (pp. 3184). Washington, DC:American 30),4449.
Psychiatric Association. Helzer, J. E., Kraemer, H. C, & Krueger, R. F. (2006). The
Craddock, N, & Owen, M. J. (2005). The beginning of the end feasibility and need for dimensional psychiatric diagnoses.
for the Kraepelinian dichotomy. British Journal of Psychiatry, Psychological Medicine, 36, 16711680.
186, 364366. Helzer, J. E., Kraemer, H. C., Krueger, R. F., Wittchen, H. U.,
Decker, H. S. (2010). A moment of crisis in the history of Sirovatka, P. J., & Regier, D. A. (Eds.). (2008). Dimensional
American psychiatry. Retrieved from http://historypsychia- approaches in diagnostic classification. Washington, DC:American Psychiatric Association.
istory-of-american-psychiatry Helzer, J. E., Wittchen, H-U., Krueger, R. F., & Kraemer, H.
DSM-5 Childhood and Adolescent Disorders Work Group. C. (2008). Dimensional options for DSM-V: the way
(2010). Justification for temper dysregulation disorder forward. In J. E. Helzer, H. C. Kraemer, R. F. Krueger,
with dysphoria. Retrieved from H. U. Wittchen, P. J. Sirovatka, & D. A. Regier (Eds.),
ProposedRevision/Pages/proposedrevision.aspx?rid=397# Dimensional approaches to diagnostic classification. Refining
DSM-5 Mood Disorders and Childhood and Adolescent the research agenda for DSM-V (pp. 115127). Washington,
Disorders Work Group. (2010). Issues pertinent to a devel- DC:American Psychiatric Association.
opmental approach to bipolar disorder. Retrieved from Horwitz, A. V, & Wakefield, J. C. (2007). The loss of sadness:How psychiatry transformed normal sorrow into depressive disorder.
vision.aspx?rid=397# NewYork:Oxford UniversityPress.
Endicott, J. (2000). History, evolution, and diagnosis of pre- Hyman, S. (2010). The diagnosis of mental disorders:the prob-
menstrual dysphoric disorder. Journal of Clinical Psychiatry, lem of reification. Annual Review of Clinical Psychology, 6,
62(suppl 24),58. 155179.
Fawcett, J. (2010). An overview of mood disorders in the Insel, T. R. (2009). Translating scientific opportunity into public
DSM-5. Current Psychiatry Reports, 12, 531538. health impact. Astrategic plan for research on mental illness.
Feighner, J. P., Robins, E., Guze, S. B., Woodruff, R. A., Archives of General Psychiatry, 66, 128133.
Winokur, G., & Munoz, R. (1972). Diagnostic criteria for Insel, T. R., Cuthbert, B., Garvey, M., Heinssen, R., Pine, D. S.,
use in psychiatric research. Archives of General Psychiatry, Quinn, K., .
Wang, P. (2010). Research domain criteria
26,5763. (RDoC):Toward a new classification framework for research on
Feinstein, A. R. (1970). The pre-therapeutic classification mental disorders. American Journal of Psychiatry, 167, 748757.
of co-morbidity in chronic disease. Chronic Disease, 23, Judd, L. L., Schettler, P. J., & Akiskal, H. S. (2002). The preva-
455468. lence, clinical relevance, and public health significance
First, M. B. (2011). DSM-5 proposals for mood disorders: of subthreshold depressions. Psychiatric Clinics of North
Acost-benefit analysis. Current Opinions in Psychiatry, 24,19. America, 25, 685698.
First M, B., & Halon R. (2008). Use of DSM paraphilia diagno- Katon, W, & Roy-Byrne P.P. (1991). Mixed anxiety and depres-
ses in sexually violent predator commitment cases. Journal of sion. Journal of Abnormal Psychology, 100, 337345.
the American Academy of Psychiatry and Law, 36, 443454. Kendell, R. E. (1975). The role of diagnosis in psychiatry.
First, M. B., & Pincus, H. A. (2002). The DSM-IV Text London:Blackwell Scientific Publications.
Revision:Rationale and potential impact on clinical practice. Kendell, R. E. (1991). Relationship between the DSM-IV and
Psychiatric Services, 53, 288292. the ICD-10. Journal of Abnormal Psychology, 100, 297301.

Widiger,Gore 25
Kendell, R. E., & Jablensky, A. (2003). Distinguishing between Krueger, R. F., & Markon, K. E. (2006). Reinterpreting comor-
the validity and utility of psychiatric diagnosis. American bidity:a model-based approach to understanding and classi-
Journal of Psychiatry, 160,412. fying psychopathology. Annual Review of Clinical Psychology,
Kendler, K. S. (2005). Toward a philosophical structure for psy- 2, 111133.
chiatry. American Journal of Psychiatry, 162, 433440. Kupfer, D. J., First, M. B. & Regier, D. A. (2002). Introduction.
Kendler, K.S. (2010). A statement from Kenneth S. Kendler, In D. J. Kupfer, M. B. First, & D. A. Regier (Eds.), A
M.D., on the proposal to eliminate the grief exclusion crite- research agenda for DSM-V (pp. xvxxiii). Washington,
rion from major depression, by Kenneth S.Kendler, M.D., DC:American Psychiatric Association.
Member, DSM-5 Mood Disorder Work Group. Retrieved Kupfer, D. J., & Regier, D. A. (2011). Neuroscience, clinical evi-
from dence, and the future of psychiatric classification in DSM-5.
Kendler, K. S., Kupfer, D., Narrow, W., Phillips, K., & American Psychiatric Association, 168,13.
Fawcett, J. (2009). Guidelines for making changes to DSM-V. Lamb, K., Pies, R., & Zisook, S. (2010). The bereavement exclu-
Unpublished manuscript, Washington, DC: American sion for the diagnosis of major depression:To be or not to be.
Psychiatric Association. Psychiatry, 7,1925.
Kendler, K. S., Munoz, R. A., & Murphy, G. (2010). The devel- Lilienfeld, S. O., & Marino, L. (1999). Essentialism revis-
opment of the Feighner criteria: A historical perspective. ited:Evolutionary theory and the concept of mental disorder.
American Journal of Psychiatry, 167, 134142. Journal of Abnormal Psychology, 108, 400411.
Kendler, K. S., Myers, M. D., & Zisook, M. S. (2008) Does Maddux, J. E., Gosselin, J. T., & Winstead, B. A. (2008).
bereavement-related major depression differ from major Conceptions of psychopathology: A social construction-
depression associated with other stressful life events? ist perspective. In J. E. Maddux & B. A. Winstead (Eds.),
American Journal of Psychiatry, 165, 14491455. Psychopathology:Foundations for a contemporary understand-
Kessler, R. C., Avenevoli, S., Costello, J., Georgiades, K., ing (2nd ed., pp. 318). New York: Routledge/Taylor &
Green, J. G., Gruber, M. J.,...Merikangas, K. R. (2012). Francis.
Prevalence, persistence, and sociodemographic correlates Magruder, K. M., & Calderone, G. E. (2000). Public health con-
of DSM-IV disorders in the National Comorbidity Survey sequences of different thresholds for the diagnosis of mental
Replication Adolescent Supplement. Archives of General disorders. Comprehensive Psychiatry, 41,1418.
Psychiatry, 69, 372380. Meehl, P. (2010). Diagnostic taxa as open concepts:
Kessler, R. C., Berglund, P., Demler, O., Jin, R., Koretz, D., Metatheoretical and statistical questions about reliability
Merikangas, K. R.,...Wang, P. S. (2003). The epidemiol- and construct validity in the grand strategy of nosological
ogy of major depressive disorder. Results from the National revision. In T. Millon, R. Krueger, and E. Simonsen (Eds.),
Comorbidity Survey Replication (NCS-R). Journal of the Contemporary directions in psychopathology. Scientific founda-
American Medical Association, 289(June 18), 30953105. tions of the DSM-V and ICD-11 (pp. 174186). NewYork,
Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., NY:GuilfordPress.
Hughes, M., Eshleman, S., .
Kendler, K. S. (1994). Merikangas, K. R., He, J., Burstein, M., Swanson, S. A.,
Lifetime and 12-month prevalence of DSM-III-R psychiatric Avenevoli, S., Cui, L.,...Swendsen, J. (2010). Lifetime prev-
disorders in the United States. Archives of General Psychiatry, alence of mental disorders in U.S.adolescents:Results from
51,819. the National Comorbidity Survey Replication-Adolescent
Kessler, R. C., Merikangas, K. R., Bergland, P., Eaton, W. W., Supplement (NCS-A). Journal of the American Academy of
Koretz, D., & Walters, E. E. (2003). Mild disorders should Child & Adolescent Psychiatry, 49, 980989.
not be eliminated from the DSM-V. Archives of General Miller, G. (2010). Beyond DSM:Seeking a brain-based classifi-
Psychiatry, 60, 11171122. cation of mental illness. Science, 327(March 19),1437.
Kessler, R. C., Zhao, S., Blazer, D. G., & Swartz, M. (1997). Moreno, C., Laje, G., Blanco, C., Jiang, H., Schmidt, A. B., &
Prevalence, correlates, and course of minor depression and Olfson, M. (2007). National trends in the outpatient diag-
major depression in the National Comorbidity Survey. nosis and treatment of bipolar disorder in youth. Archives of
Journal of Affective Disorders, 45,1930. General Psychiatry, 64, 10321039.
Kirk, S. A. (Ed.). (2005). Mental disorders in the social envi- Narrow, W. E., Rae, D. S., Robins, L. N., & Regier, D. A.
ronment: Critical perspectives. New York: Columbia (2002). Revised prevalence estimates of mental disorders in
UniversityPress. the United States: Using a clinical significance criterion to
Kirmayer, L. J., & Young, A. (1999). Culture and context in the reconcile 2 surveys estimates. Archives of General Psychiatry,
evolutionary concept of mental disorder. Journal of Abnormal 59, 115123.
Psychology, 108, 446452. PDM Task Force. (2006). Psychodynamic diagnostic manual.
Klein, D. F. (1999). Harmful dysfunction, disorder, disease, ill- Silver Spring, MD:Alliance of Psychoanalytic Organizations.
ness, and evolution. Journal of Abnormal Psychology, 108, Pincus, H. A., Frances, A., Davis, W., First, M., & Widiger, T.
421429. A. (1992). DSM-IV and new diagnostic categories:Holding
Klerman, G.L. (1986). Historical perspectives on contemporary the line on proliferation. American Journal of Psychiatry, 149,
schools of psychopathology. In T. Millon & G. L.Klerman 112117.
(Eds.), Contemporary directions in psychopathology (pp. 328). Pincus, H. A., McQueen, L. E., & Elinson, L. (2003).
NewYork:GuilfordPress. Subthreshold mental disorders: Nosological and research
Kraemer, H. C. (2008). DSM categories and dimensions in clini- recommendations. In K. A. Phillips, M. B. First, & H.
cal and research contexts. In J. E. Helzer, H. C. Kraemer, A. Pincus (Eds.), Advancing DSM. Dilemmas in psychiat-
R. F. Krueger, H. U. Wittchen, P. J. Sirovatka, & D. ric diagnosis (pp. 129144). Washington, DC: American
A. Regier (Eds.), Dimensional approaches to diagnostic clas- Psychiatric Association.
sification. Refining the research agenda for DSM-V (pp. 517). Presig, M., Merikangas, K. R., & Angst, J. (2001). Clinical sig-
Washington, DC:American Psychiatric Association. nificance and comorbidity of subthreshold depression and

26 Diagnostic and Statistical Manual(DSM)

anxiety in the community. Acta Psychiatrica Scandinavica, Spitzer, R. L. (1981). The diagnostic status of homosexuality in
104, 96103. DSM-III:Areformulation of the issues. American Journal of
Prisciandaro, J. J., & Roberts, J. E. (2005). A taxometric inves- Psychiatry, 138, 210215.
tigation of unipolar depression in the National Comorbidity Spitzer, R. L., Endicott, J., & Robins E. (1975). Clinical criteria
Survey. Journal of Abnormal Psychology, 114, 718728. for psychiatric diagnosis and DSM-III. American Journal of
Regier, D. A. (2008). Forward:dimensional approaches to psy- Psychiatry, 132, 11871192.
chiatric classification. In J. E. Helzer, H. C. Kraemer, R. Spitzer, R. L., Endicott, J., & Robins E. (1978). Research diag-
F.Krueger, H. U.Wittchen, P. J.Sirovatka, & D. A.Regier nostic criteria: Rationale and reliability. Archives of General
(Eds.), Dimensional approaches to diagnostic classification. Psychiatry, 35, 773789.
Refining the research agenda for DSM-V (pp. xviixxiii). Spitzer, R. L., Sheehy, M., & Endicott, J. (1977).
Washington, DC:American Psychiatric Association. DSM-III: Guiding principles. In V. Rakoff, H. Stancer,
Regier, D. A., Narrow, W. E., Kuhl, E. A., & Kupfer, D. J. & H. Kedward (Eds.), Psychiatric diagnosis (pp. 124).
(2009). The conceptual development of DSM-V. American NewYork:Brunner/Mazel.
Journal of Psychiatry, 166, 645655. Spitzer, R. L., & Williams, J. B.W. (1982). The definition and
Reiss, D., & Emde, R. N. (2003). Relationship disorders are diagnosis of mental disorder. In W.R. Gove (Ed.), Deviance
psychiatric disorders: Five reasons they were not included and mental illness (pp. 1532). Beverly Hills, CA:Sage.
in DSM-IV. In K. A.Phillips, M. B.First, & H. A.Pincus Spitzer, R. L., Williams, J. B. W., & Skodol, A. E. (1980).
(Eds.), Advancing DSM. Dilemmas in psychiatric diagno- DSM-III: The major achievements and an overview.
sis (pp. 191223). Washington, DC: American Psychiatric American Journal of Psychiatry, 137, 151164.
Association. Stein, D. J., Phillips, K. A., Bolton, D., Fulford, K. W. M.,
Robins, E., & Guze, S. B. (1970). Establishment of diagnostic Sadler, J. Z., & Kendler, K. S. (2010). What is a mental/
validity in psychiatric illness: Its application to schizophre- psychiatric disorder? From DSM-IV to DSM-V. Psychological
nia. American Journal of Psychiatry, 126, 107111. Medicine, 40, 17591765.
Rosenhan, D. L. (1973). On being sane in insane places. Science, Stengel, E. (1959). Classification of mental disorders. Bulletin of
179, 250258. the World Health Organization, 21, 601663.
Rounsaville, B. J., Alarcon, R. D., Andrews, G., Jackson, Strauss, M., & Smith, G. (2009). Construct validity:Advances
J. S., Kendell, R. E., Kendler, K. S., & Kirmayer, L. J. in theory and methodology. Annual Review of Clinical
(2002). Toward DSM-V: Basic nomenclature issues. In D. Psychology, 5,125.
J. Kupfer, M. B. First, & D. A. Regier (Eds.), A research Szasz, T. S. (1960). The myth of mental illness. New York:
agenda for DSM-V (pp. 130). Washington, DC:American Hoeber-Harper.
PsychiatricPress. Ustun, T. B., & Sartorius, N. (Eds.). (1995). Mental illness in
Ruscio, J., & Ruscio, A. M. (2000). Informing the continuity general health care:An international study. London:Wiley.
controversy: a taxometric analysis of depression. Journal of Wakefield, J. C. (1992). The concept of mental disorder: On
Abnormal Psychology, 109, 473487. the boundary between biological facts and social values.
Ruscio, J., & Ruscio, A. M. (2004). Clarifying boundary issues American Psychologist, 47, 373388.
in psychopathology: The role of taxometrics in a compre- Wakefield, J. C. (1998). The DSMs theory-neutral nosology
hensive program of structural research. Journal of Abnormal is scientifically progressive:Response to Follette and Houts
Psychology, 113,2438. (1996). Journal of Consulting and Clinical Psychology, 66,
Rutter, M. (2003, October). Pathways of genetic influences on 846852.
psychopathology. Zubin Award Address at the 18th Annual Wakefield, J. C. (2001). The myth of DSMs invention of new
Meeting of the Society for Research in Psychopathology, categories of disorder: Houts diagnostic discontinuity
Toronto, Ontario. thesis disconfirmed. Behaviour Research and Therapy, 39,
Sakashita, C., Slade, T., & Andrews, G. (2007). An empirical 575624.
analysis of two assumptions in the diagnosis of DSM-IV Wakefield, J. C. (2011). Should uncomplicated bereavement-
major depressive episode. Australian and New Zealand related depression be reclassified as a disorder in the DSM-5?
Journal of Psychiatry, 41,1723. Response to Kenneth S.Kendlers statement defending the
Salmon, T. W., Copp, O., May, J. V., Abbot, E. S., & Cotton, proposal to eliminate the bereavement exclusion. Journal of
H. A. (1917). Report of the committee on statistics of the Nervous and Mental Disease, 199, 203208.
American Medico-Psychological Association. American Wakefield, J. C., & First, M. B. (2012). Validity of the bereave-
Journal of Insanity, 74, 255260. ment exclusion to major depression: Does the empirical
Sanislow, C. A., Pine, D., S., Quinn, K., J., Kozak, M. J., Garvey, evidence support the proposal to eliminate the exclusion in
M.A., Heinssen, R. K.,...Cuthbert, B. N. (2010). Developing DSM-5? World Psychiatry, 11,310.
constructs for psychopathology research: Research domain Wakefield, J. C., Schmitz, M. F., First, M. B., & Horwitz, A.
criteria. Journal of Abnormal Psychology, 119, 631639. V. (2007). Should the bereavement exclusion for major
Sartorius, N., Kaelber, C. T., Cooper, J. E., Roper, M., Rae, D. depression be extended to other losses? Evidence from the
S., Gulbinat, W.,...Regier, D. A. (1993). Progress toward National Comorbidity Survey. Archives of General Psychiatry,
achieving a common language in psychiatry. Archives of 64, 433440.
General Psychiatry, 50, 115124. Ward, C. H., Beck, A. T., Mendelson, M., Mock, J. E., &
Scheff, T. J. (1966). Being mentally ill: a sociological theory. Erbaugh, J. K. (1962). The psychiatric nomenclature.
NewYork:AldinePress. Reasons for diagnostic disagreement. Archives of General
Schwartz, M. A., & Wiggins, O. P. (2002). The hegemony Psychiatry, 7, 198205.
of the DSMs. In J. Sadler (Ed.), Descriptions and prescrip- Widiger, T. A. (2001). Official classification systems. In W.
tions:Values, mental disorders, and the DSM (pp. 199209). J.Livesley (Ed.), Handbook of personality disorders (pp. 60
Baltimore, MD:Johns Hopkins UniversityPress. 83). NewYork:GuilfordPress.

Widiger,Gore 27
Widiger, T. A. (2013). Classification and diagnosis:Historical devel- Adult psychopathology and diagnosis (6th ed., pp. 332).
opment and contemporary issues. In J. Maddux & B. Winstead NewYork:Wiley.
(Eds.), Psychopathology: Foundations for a contemporary under- Widiger, T. A., & Samuel, D. B. (2005). Diagnostic categories
standing (3rd ed.). NewYork:Lawrence Erlbaum Associates. or dimensions:Aquestion for DSM-V. Journal of Abnormal
Widiger, T. A., & Clark, L. A. (2000). Toward DSM-V and the Psychology, 114, 494504.
classification of psychopathology. Psychological Bulletin, 126, Widiger, T. A., & Sankis, L. M. (2000). Adult psychopathol-
946963. ogy: Issues and controversies. Annual Review of Psychology,
Widiger, T. A., & Crego, C. (2013). Diagnosis and classifica- 51, 377404.
tion. In I. Weiner, G. Stricker, & T. A.Widiger (Eds.), Wiley Widiger, T. A., & Trull, T. J. (1991). Diagnosis and clinical
handbook of clinical psychology. NewYork:Wiley. assessment. Annual Review of Psychology, 42, 109133.
Widiger, T. A., & Edmundson, M. (2011). Diagnoses, dimen- Widiger, T. A., & Trull, T. J. (1993). The scholarly develop-
sions, and DSM-V. In D. Barlow (Ed.), The Oxford hand- ment of DSM-IV. In J. A.Costa de Silva & C. C.Nadelson
book of clinical psychology (pp. 254278). NewYork:Oxford (Eds.), International review of psychiatry (Vol. 1, pp. 5978).
UniversityPress. Washington, DC:American PsychiatricPress.
Widiger, T. A., Frances, A. J., Pincus, H. A., Davis, W. W., & World Health Organization. (1992). The ICD-10 classification
First, M. B. (1991). Toward an empirical classification for of mental and behavioural disorders. Clinical descriptions and
DSM-IV. Journal of Abnormal Psychology, 100, 280288. diagnostic guidelines. Geneva:Author.
Widiger, T. A., Frances, A. J., Pincus, H. A., First, M. B., Ross, Zachar, P., & Kendler, K. S. (2007). Psychiatric disorders:Acon-
R. R., & Davis, W. W. (Eds.) (1994). DSM-IV sourcebook ceptual taxonomy. American Journal of Psychiatry, 164,
(Vol. 1). Washington, DC:American Psychiatric Association. 557565.
Widiger, T. A., Frances, A. J., Pincus, H. A., Ross, R., First, Zigler, E., & Phillips, L. (1961). Psychiatric diagnosis:Acritique.
M. B., Davis, W. W., & Kline, M. (Eds.). (1998). DSM-IV Journal of Abnormal and Social Psychology, 63, 607618.
Sourcebook (Vol. 4). Washington, DC:American Psychiatric Zilboorg, G. (1941). A history of medical psychology. NewYork:
Association. W.W. Norton.
Widiger, T. A., & Gore, W. L. (2012). Mental disorders as Zisook, S., & Kendler, K. S. (2007). Is bereavement-related
discrete clinical conditions: Dimensional versus categori- depression different than nonbereavement-related depres-
cal classification. In M. Hersen and D. C. Beidel (Eds.), sion? Psychological Medicine, 37, 779794.

28 Diagnostic and Statistical Manual(DSM)


3 From Comorbidity to Constructs:

Recurring and Emergent Issues in
Modeling Comorbidity
Kristian E.Markon

Comorbidity models have become central to psychopathology theory and research, not only because
they have clarified our understanding of how and why disorders co-occur but also because they have
clarified our understanding of what the disorders are. This chapter reviews basic types of comorbidity
models, recurring issues in comorbidity modeling, and discuss emerging issues in the area. Using recent
epidemiological, repeated-measures data on depression and anxiety as an example, two different models
of comorbidity are compared, one in which comorbidity arises due a shared liability dimension (i.e.,
a reflective or latent variable model) and another in which comorbidity arises as an epiphenomenon
of correlated symptoms (i.e., a formative or network model). This comparison, relatively novel in the
literature, illustrates a number of issues that are encountered in comorbidity modeling, and clearly
demonstrates how questions pertaining to comorbidity can shape our understanding of psychopathology
Key Words: reflective and formative measurement, midlife, development, daily interview

Introduction the study of comorbidity will also be discussed. This

Like many extensively studied constructs, chapter hopefully will illustrate that comorbidity
comorbidity has a multitude of meanings and inter- has fundamental implications for understanding
pretations and is associated with certain recurring not only the causes and consequences of depression
theoretical and methodological issues. This is per- but also the nature of depression itself.
haps especially true when discussing comorbidity
in the context of depression, which is one of the Comorbidity Paradigms:
most common and costly forms of illness worldwide Co-occurrence Versus Covariance
(Moussavi etal., 2007). Given its associations with The study of comorbidity has an extensive his-
so many medical, sociological, and psychological tory, as it applies to depression specifically as well
variables, different forms and aspects of depressive as it applies to psychopathology and illness more
comorbidity will have differential emphasis and generally. Comorbid phenomena have undoubtedly
salience in different contexts. been studied for some time, although the term was
The purpose of this chapter is to review the vari- in established use by the middle of the twentieth
ous meanings and interpretations of comorbidity in century (although Feinstein [1970] is often credited
empirical modeling and research, so as to frame dis- for the term, comorbidity was in established use
cussion, prevent confusion, and contribute to prog- prior to that time, appearing in texts in the 1960s
ress in understanding depression and its comorbidity [e.g., Gale Research Group, 1965]). Discussion and
with various conditions. Newly emergent themes in study of psychiatric comorbidity began to accelerate

in the 1980s and 1990s, following publication of of any distinct additional clinical entity that has
the third edition of the Diagnostic and Statistical existed or that may occur during the clinical course
Manual (DSM-III), and currently remains a topic of a patient who has the index disease under study,
of widespread attention. emphasizing discussion of its functional and
Relatively early in the emerging literature on intellectual effects on clinical outcome.
comorbidity, it became clear that two paradigms Figure 3.1a illustrates the exogenous co-
for understanding comorbidity predominate occurrence paradigm. In this paradigm, two disor-
(Lilienfeld, Waldman, & Israel, 1994). The first, ders, A and B, affect some outcome. Each of the
referred to here as the exogenous co-occurrence par- disorders may influence the outcome directly, but
adigm, focuses on the co-occcurrence of two or their joint presence in an individual may also have
more conditions in individuals and the effects of special significance, such that they interact with one
this co-occurrence on other variables. The second another in influencing the outcome. A prototypic
paradigm, referred to here as the endogenous covari- exogenous co-occurrence model is essentially a mod-
ance paradigm, focuses on associations between two eration or interaction model, where an additional
or more conditions in a group of individuals, and disorder moderates the effects of another variable,
explaining the causes of this covariance. The terms typically by amplifying some effect. The disorders
exogenous and endogenous are used here as are also assumed to be independent of one another,
they are used in the structural equation modeling or approximately so, and are treated as exogenous
literature (Bollen, 1989), where they refer to sce- variables (variables whose causes are assumed to
narios where causes are, respectively, outside of the be outside the scope of the model). Ahypothetical
model (and are therefore ignored) or are inside of example might involve psychosis, cancer, and their
the model (and are therefore included). Although impact on social functioning:the two former vari-
many cases of research and theory blur the distinc- ables are etiologically distinct, with causes outside
tion between these two paradigms, they neverthe- the scope of the model. They may interact, such that
less represent major approaches to understanding the combination of the two in a single individual
comorbidity. might have effects beyond theirsum.

Comorbidity as Co-occurrence Comorbidity as Covariance

The first paradigm is arguably the older of the Central to the exogenous co-occurrence para-
two and represents the original focus of comorbid- digm is the assumption that the comorbid dis-
ity theory and research. The earliest discussions of orders are distinct and that their causes can be
comorbidity emphasized effects of co-occurring treated as outside the scope of the model. Although
conditions on outcome, a perspective that continues this assumption is often appropriate for comor-
into current theory and research. Feinstein (1970), bidity between certain psychiatric and nonpsy-
for example, defined comorbidity as the presence chiatric disorders (e.g., psychosis and cancer), its

(a) A



Figure3.1 Illustrations of exogenous co-occurrence (A)and endogenous covariance (B)models. In both figures, the rectangles indicate
disorders or outcomes of interest; circles represent liabilities underlying the disorders.

30 From Comorbidit y to Constructs

appropriateness for many other forms of comorbid- version of the first model, where the correlation
ity is unclear. In many settings, the causes of the two between disorders is perfect. Finally, comorbidity
disorders are central to the theoretical model (i.e., may be induced through direct influences of condi-
are endogenous to the model), such as when the two tions on one another. Comparisons of these sorts
disorders share etiology or influence one another. In of direct-effects models against the other models
such cases, comorbidity is generally more fruitfully often require complex designs, such as behavior
conceived of in terms of covariance between disor- genetic or longitudinal designs. They can generally
ders, the etiology of which is to be explained. be distinguished from the other models in that they
As many authors have noted (Lilienfeld et al., imply that specific correlations involving a pair of
1994), the etiology of psychopathology is usu- disorders cannot easily be predicted from other cor-
ally unclear, to the extent that it is often unclear relations being modeled (e.g., that the within-twin
how to delineate distinct disorders. In such cases, correlation between two disorders is relatively dis-
comorbidity may reflect different manifestations tinct from the cross-twin correlation between those
of the same underlying pathology or state rather two disorders).
than co-occurrence of or associations between
distinct processes. Even in cases where the dis- Syndrome Models of Comorbidity
tinction between disorders is relatively clear or Over the past decade, numerous researchers have
well-established, one may influence anotheror begun to investigate and compare these different
they may mutually influence one anotherin a explanations for comorbidity between syndromes.
way that makes the assumption of exogenous causes These investigations have revealed that many syn-
inappropriate. For example, depression and diabetes dromes traditionally assumed to be distinct may
are clearly distinct disorders, but it is reasonable to actually reflect the same underlying pathologies.
expect that diabetes might affect levels of depres- More broadly, these studies have raised important
sion, or conversely, that depression might affect questions about what is shared between syndromes,
severity of diabetes (e.g., through lack of self-care). what is unique about a particular syndrome, and
Figure 3.1b illustrates commonly encountered how to conceptualize these issues.
features of endogenous covariance models. Note
that in actual application, various features of these Liability Dimensions
models might be combined in a single account, and A number of studies, for example, have modeled
that the models as illustrated are not necessarily comorbidity between multiple disorders in terms of
identified as they are presented. Also, although only underlying liability dimensions, presumed to reflect
two disorders are included in the figures, in practice pathological processes underling the disorders. In
models often incorporate multiple disorders simul- general, these studies have outlined three broad,
taneously (i.e., multimorbidity; van den Akker, replicated liability dimensions that might account
Buntinx, & Knottnerus,1996). for patterns of comorbidity (e.g., Kotov etal., 2011;
As is illustrated in the figures, comorbidity Wolf et al., 1988). The first of these, generally
between disorders may be induced by multiple referred to as internalizing, is reflected in disorders
distinct causes being correlated with one another. characterized primarily by negative emotion, such
It may also be induced by through multiple, dis- as depression, phobias, and posttraumatic stress syn-
tinct causes having multiple effects. This latter dromes. The second, generally referred to as exter-
model can be distinguished from the former with nalizing, is reflected in disorders such as conduct
enough measurements in that it requires more disorder and other antisocial behavior disorders,
parameters to capture patterns of comorbidity in as well as substance use disorders. The third liabil-
the data. This can be seen in the relative number ity dimension reflects thought disorder, including
of paths of the two models in Figure 3.1b. The sec- schizophrenia and the psychotic disorders.
ond, multiple-effects model predicts more differen- Although these studies initially focused on
tiated patterns of covariance between specific pairs cross-sectional analyses of disorders in clinical and
of disorders, because each disorder has a unique epidemiological samples, they have expanded to
path from each liability. As illustrated in the third include more etiologically informative designs, such
diagram of Figure 3.1b, comorbidity may also be as longitudinal and family studies. The results of
induced through multiple disorders acting as differ- these studies have largely supported the conclusions
ent manifestations of the same underlying process. of the cross-sectional designs and also clarified the
This can be thought of as an even more constrained important role of the superordinate liability factors

Markon 31
in prediction. For example, patterns of genetic and Parsing Shared and Unique
environmental relationships largely parallel pheno- Components of Disorders
typic relationships along the lines of the superordi- Although superordinate liabilities such as inter-
nate liability dimensions. Internalizing disorders are nalizing and externalizing account for much of
likely to segregate together among twins, as are the the comorbidity between disorders, they do not
externalizing disorders (Kendler, Prescott, Myers, account for all of it. Subsets of disorders share
& Neale, 2003). In longitudinal studies, similarly, variance beyond the broad superordinate factors.
the superordinate liability dimensions account for Among the internalizing disorders, for example, the
a large proportion of the variance in predicting sta- phobias and panic disorder appear to share com-
bility of disorder, as well as outcomes (Fergusson, mon variance beyond general internalizing liability,
Horwo-od, & Boden, 2006; Seeley, Kosty, Farmer, suggesting the presence of a specific fear liability in
& Lewinsohn,2011). addition to general internalizing. Depression and
For example, research by Fergusson etal. (2006) generalized anxiety also share variance beyond gen-
indicates that a majority of the stability in depres- eral internalizing, suggesting the presence of a dis-
sion over time can be accounted for by general tress liability (Krueger & Markon,2006).
internalizing liability rather than disorder-specific These general patterns highlight a hierarchy of
effects. Specifically, depression tends to predict other shared and unique etiologies accounting for comor-
internalizing disorders at other times at levels com- bidity between disorders. Some of these etiologies
parable to depression at other times. In other words, are shared by a large spectrum of disorders (e.g.,
the cross-time, cross-disorder correlation is similar the internalizing disorders), some of these etiologies
in magnitude to the cross-time, within-disorder are shared by a smaller subset of disorders (e.g., the
correlation. Seeley etal. (2011) similarly found that fear disorders), and some etiologies are disorder-
superordinate internalizing liability accounted for a specific. In this regard, it is important to recognize
majority of the variance in psychosocial function- that any theory explaining comorbidity must take
ing and family psychiatric history; specific disorders into account the scope of disorders being examined.
contributed little in predicting functioning or fam- A study of the comorbidity between depression
ily psychiatric status. and generalized anxiety, for example, might miss

Int Ext

Fear Distress

SpPh SoPh Pan MD GAD ASPD Alc Drug


Int Ext

SpPh SoPh Pan MD GAD ASPD Alc Drug

Figure 3.2 Illustration of two different liability models of common forms of psychopathology.
Int, internalizing; Ext, externalizing; SpPh, specific phobia; SoPh, social phobia; MD, depression; xGAD, generalized anxiety disorder;
ASPD, antisocial personality disorder; Alc, alcohol use disorder; Drug, other drug disorder.

32 From Comorbidit y to Constructs

the contribution of general internalizing liability broad liability factors influencing many disorders,
if other internalizing disorders, such as panic and more specific factors influence subsets of disorders.
phobia, were not also included. For example, in addition to being influenced by
The hierarchical nature of liabilities underlying broad internalizing liability, the phobias and panic
comorbidity has sometimes led to differing per- disorder are likely influenced by a more specific fear
spectives on how to interpret the nature of those liability, characterized by acute, paroxysmal fear
liabilities, in terms of how general and specific com- responses. Further research is needed to clarify the
ponents of liability are distributed among disorders nature of this hierarchy, to understand how liabili-
or influence those disorders. An example of this is ties can be most accurately parsed into broader and
illustrated in Figure 3.2, which represents two dif- more specific components.
ferent accounts of the nature of factors underlying
comorbidity patterns (Krueger & Markon, 2006; Symptom Models of Comorbidity
Lahey, Van Hulle, Singh, Waldman, & Rathouz, As models of comorbidity between disorders
2011). Both accounts comprise three superordinate have become more complex, and it has become clear
liabilities: an externalizing liability, a liability pri- that there are hierarchies of shared relationships,
marily affecting depression and generalized anxiety, attention has increasingly been focused on the
and a liability affecting the phobias and panic. In symptoms underlying those disorders in providing
the top account, however, two of the liabilities, dis- explanations for comorbidity. For example, under-
tress and fear, both reflect a common internalizing standing patterns of relationships among specific
dimension (Krueger & Markon, 2006); in the bot- internalizing symptoms would help clarify what is
tom account, there is one internalizing liability, and distinctive about the distress versus fear disorders
another global liability that affects all disorders, but and what is shared betweenthem.
depression and generalized anxiety most strongly One fundamental concern underlying this inter-
(Lahey etal.,2011). est in symptom-level explanation of comorbidity is
The relationship between the distress disorders heterogeneity of criteria within a single diagnosis.
(depression and generalized anxiety) and the other Although some diagnoses are relatively homoge-
mental disorders is key to distinguishing the two neous in their symptom criteria, other diagnoses
accounts. The top account predicts that the distress are more heterogeneous. If disparate pathologies
disorders will be more strongly related to the other are reflected in a single diagnosis, it becomes chal-
internalizing disorders than to the externalizing lenging to understand which pathologies account
disorders; the bottom account predicts that the dis- for comorbidity with another diagnosisall of
tress disorders will be equally related to the other the pathologies causing the disorder, or a subset
internalizing disorders and externalizing disorders. of them? Do those pathological processes relate in
Thus far, available evidence across studies seems different ways to those in another diagnosis? More
to support the top model, as the distress disorders broadly, do spurious features of the diagnostic crite-
generally seem to be more strongly related to the ria account for comorbidity patterns? For example,
fear disorders than to the externalizing disorders do disorders with the same impairment or duration
(Beesdo-Baum et al., 2009; Kessler et al., 2011; criteriawhich are relatively arbitrarycovary
Krueger & Markon, 2006). However, relatively few more than other disorders, simply because of those
studies have formally evaluated the bottommodel. criteria, independently of more core pathological
Overall, work on syndrome comorbidity sug- processes?
gests that syndromes can be understood in terms Posttraumatic stress disorder (PTSD) illustrates
of underlying liability factors. Disorders character- well complexities that arise in explaining comorbid-
ized by negative emotionthe depression, anxiety, ity in the presence of relatively heterogeneous symp-
and fear disorderscan be understood in terms tom criteria. In recent years, various studies have
of a common internalizing factor, for example. demonstrated that PTSD can be understood in
Disorders such as antisocial personality disorder, terms of four subordinate dimensions:one reflecting
conduct disorder, and the substance use disorders dysphoria, another reflecting intrusive cognitions,
can be understood in terms of a common external- another reflecting avoidant behavior, and another
izing factor. Asimilar psychosis liability factor can reflecting hypervigilance (Yufik & Simms, 2010).
also be identified (e.g., Kotov et al., 2011; Wolf Importantly, these four dimensions relate differen-
et al., 1988). These liability factors are organized tially to other disorders in theoretically meaning-
hierarchically, moreover, so that in addition to ful ways. In this account, in order to completely

Markon 33
characterize comorbidity with PTSD, it is neces- Various researchers have also examined symp-
sary to examine PTSD at the level of symptoms, toms within each of the superordinate spectra (for
rather than disorder. For example, the relation- example, only internalizing symptoms, or only
ship between PTSD and depression appears to be externalizing symptoms). Watson and colleagues
largely mediated through the dysphoria symptoms (2007), for example, have examined the structure of
of PTSD, with no contribution from the hypervigi- internalizing psychopathology at the level of symp-
lance symptoms (Gootzeit & Markon, 2011). The toms, delineating a number of subordinate factors
relationship between panic and PTSD, in contrast, that are broadly influenced by a single internalizing
appears to be mediated both through dysphoria as dimension. Krueger and colleagues, similarly, have
well as hypervigilance (Gootzeit & Markon,2011). examined the structure of externalizing symptoms
In this case, focusing on the comorbidity between (2007), identifying, in addition to a general exter-
depression and PTSD at the level of syndromes nalizing factor, superordinate aggression and sub-
would obscure more meaningful relationships that stance use factors as well. Similar analyses have been
are mediated through liability factors manifested at conducted on the structure of psychotic symptoms
the level of symptoms. This is especially important (Dikeos etal., 2006), identifying positive symptom
because the symptom liability factors exhibit differ- (e.g., hallucinations and delusions), negative symp-
ential patterns of relationships with other disorders, tom (e.g., alogia and blunted affect), and disorgani-
differentiable in magnitude as well as quality. For zation factors.
example, the relationship between depression and Interestingly, in the work of both Watson etal.
dysphoria is not only stronger than the relation- (2007) and Markon (2010), distinct superordi-
ship between depression and hypervigilance, but nate fear and distress factors were not identified
it is strong enough to suggest that depression and (although subordinate depression, anxiety, and
the dysphoria of PTSD are identical (Gootzeit & phobia factors were). These results suggest that the
Markon, 2011). This has important implications superordinate distress and fear factors suggested in
for understanding the nature of PTSD, in terms Figure 3.2 may reflect an artifact of official diagnos-
of understanding what components of PTSD are tic conventions or assessment procedures (however,
unique relative to other internalizing disorders see also Beesdo-Baum etal., 2009, and Kessler etal.,
(Yufik & Simms, 2010). It also has important 2011, who find similar results). This result also sup-
implications for clinical assessment, in that it sug- ports the model in the top of Figure 3.2, insofar as
gests some criteria of PTSD are redundant with it suggests that at the level of symptoms, depression
and might be better characterized by other forms of is not easily separated from phobias, and that forms
internalizing psychopathology (Breslau, Chase, & of internalizing psychopathology, including depres-
Anthony, 2002; Yufik & Simms,2010). sion, are more closely related to one another than to
externalizing. Future studies are needed to replicate
Liability SymptomModels these findings, however.
Various researchers have examined the liability
factors suggested by models such as those presented Formative SymptomModels
in Figure 3.2, as they affect symptoms rather than In the models just described, comorbidity
disorders. For example, I recently analyzed data between disorders, either at the level of symptoms
from the 2000 British Psychiatric Morbidity sur- or diagnoses, is explained in terms of underlying
veys, a large-scale epidemiological survey of mental liability factors. It is possible, however, to conceive
disorder in Britain, to determine how general lia- of comorbidity in terms of direct relationships
bility factors manifest in symptoms across a broad between the symptoms, that induce comorbidity in
range of disorders (Markon, 2010). These analyses the diagnoses as a sort of epiphenomenon related to
identified 20 subordinate syndromes, similar but the way disorders are defined. These sorts of models
not identical to disorders defined in official nosolo- have been referred to recently as network models
gies, that could be described in terms of 4 superor- (Cramer, Waldorp, van der Maas, & Borsboom,
dinate liability factors. In addition to internalizing, 2010), although similar ideas have been discussed
externalizing, and thought disorder, which have in the past (e.g., Caron & Rutter, 1991; Lilienfeld,
been identified in diagnostic-level studies, this anal- Waldman, & Israel, 1994; Pfeffer & Plutchik,
ysis also identified a pathological introversion fac- 1989). They are referred to here as formative mod-
tor, reflected in problems related to social anxiety, els to maintain consistency with the psychometric
lack of assertiveness, and dependency. literature, where formative measurement models

34 From Comorbidit y to Constructs

have been explored in great detail as an alterna- define disorders. Second, relatedly, the disorders do
tive to liability models (Blalock, 1964; Edwards & not exist outside of the symptomsin some sense,
Bagozzi, 2000; Fornell & Bookstein, 1982; Howell, they are summaries of certain sets of symptoms. In
Breivik, & Wilcox, 2007a, 2007b). Liability models fact, strictly speaking, there cannot be other symp-
are generally referred to as reflective models in that toms of a given disorder; a disorder defined by other
literature, as the indicators reflect underlying liabil- criteria would technically be a different disorder.
ity; formative models are referred to as such because Under a strict formative model, two disorders (for
the indicators form or cause the constructs. example, Aand B in Figure 3.3) will exhibit comor-
Figure 3.3 illustrates a hypothetical formative bidity only because their symptoms are directly
model. The symptoms of Disorders Aand B are all associated with one another. Comorbidity is then
correlated with one another. Although the symp- an epiphenomenon, a side effect of the fact that cor-
toms are shown as having nondirectional relation- related symptoms are used to define disorders. In
ships in the figure (i.e., with double-headed arrows), this regard, formative models differ in important but
the relationships could be directional (e.g., lack of potentially subtle ways from the models illustrated in
sleep causing concentration problems). The symp- Figure 3.1b. Unlike the first three models illustrated
toms define the disorders, and in this sense cause in Figure 3.1b, comorbidity is not caused by shared
the disorders. The direction of association between liabilities, but rather because directly correlated symp-
the symptoms and disorders goes from the former toms are used to define the two disorders. Unlike the
to the latterunlike in Figure 3.2, where the direc- last model in Figure 3.1b, the disorders do not have
tion goes from liability factors to the symptoms. causal effect themselves (for example, depression per
For example, hypersomnia and guilt would not be se cannot cause something) and are therefore not
related because they are caused by depression, but able to influence one anotherthe direct relation-
because, for instance, hypersomnia might lead to ships instead are at the level of the symptoms defin-
guilt, and the two symptoms happen to define ing the disorders. Under a strict formative model,
depression. In fact, in an important sense, under a depressive disorder and generalized anxiety disorder
formative model, depression does not exist at all as are not comorbid because they share common causes,
a real entityonly its symptoms exist, with depres- or because they influence one another, but because
sion being an externally imposed construct. some of their symptoms cause one another.
Certain features of the formative model are Very little research has been conducted to evalu-
important to emphasize. First, in a strict formative ate formative models as explanations of comorbidity.
model, the symptoms are assumed to be measured Also, as some have noted, it is unclear how different
without errorthere are no influences on the symp- formative models actually are from existing liabil-
toms that are not directly observed. Most impor- ity models (Danks, Fancsali, Glymour, & Scheines,
tantly, there are no disorders causing the symptoms, 2010; Humphry & McGrane, 2010; Molenaar,
and in a strict sense, there are no symptoms at all 2010). For instance, it is possible to model direct
there are perfectly measured criteria that are used to paths between symptoms in a liability model (e.g., a
relationship between dyssomnia and concentration
problems above and beyond depression or internal-
izing), and the two models can be treated within
A2 A the same mathematical framework. Additionally,
as has been noted, formative models in their strict
A3 form assume no measurement error, which may be
theoretically untenable (Markus, 2010; McFarland
B1 & Malta, 2010) and lead to seemingly unreasonable
implications (for example, that measures of a symp-
B2 B tom using two different instruments are in fact two
different symptoms; Howell etal., 2007a, 2007b).
Finally, although formative models do provide
one type of explanation for comorbidity, this expla-
Figure 3.3 Illustration of a strict formativemodel. nation is limited in certain respectsfor example,
Rectangles indicate criteria, circles indicate disorders. Note that describing associations between internalizing symp-
the disorders are entirely defined by the criteria; the directed
paths (with single-headed arrows) in many circumstances would toms does not explain why internalizing symptoms
be fixed and not estimated. as a set are more closely associated with one another

Markon 35
than with externalizing symptoms. This would a second wave of data (MIDUS II) was collected
require some metaproperty of the symptom net- approximately 10years later, in 20042006, when
work as an explanation, which would likely resem- participants from the initial wave were recontacted.
ble a latent variable in some abstractsense. The NSDE is a substudy conducted in both waves
In general, the greater precision of symptom mea- of MIDUS, in which a subsample of participants
surement relative to syndromal assessment has permit- were interviewed daily about psychological and
ted more accurate tests of various comorbidity models. physical symptoms, stressors, and other daily life
Among liability models, symptom-level analyses have events over a period of 8days. Data on cortisol and
supported the basic conclusions of syndrome-level other variables were also collected. Here, I focus
analyses, with some possible exceptions in how more on the NSDE II, the second wave of data collec-
specific liabilities are understood (e.g., the scope of tion, in the primary sample of 1141 participants
fear liability). As interest in symptom-level analyses who were recruited through random digit dialing.
has grown, however, interest in other models, such Further information on the NSDE is available in
as the formative models, has grown as well. Although Ryff and Almeida (2010); data from it are publicly
questions have been raised about the utility of forma- available through the Inter-university Consortium
tive models, further research is needed to compare for Political and Social Research.
them to liability models and understand how to inte-
grate the two types of frameworks. Measures and Variables
Although structured clinical interviews were
Depression and Anxiety in the not administered in the NSDE, the interviews did
National Study of Daily Experiences include measures of numerous negative and positive
In order to provide examples of these models, and emotional states (Almeida & Kessler, 1998; Mroczek
illustrate issues that arise in comorbidity modeling, & Kolarz, 1998; Watson, Clark, & Tellegen, 1988),
I compared liability and formative (i.e., network) as well as measures of physical and somatic symp-
models of depression and anxiety symptoms in the toms (King, 2008; Larsen & Kasimatis, 1991). This
National Study of Daily Experiences (NSDE; Ryff allowed me to construct indicators of multiple cri-
& Almeida, 2010), which is part of the National teria of a Major Depressive Episode (MDE) and of
Survey of Midlife Development in the United States Generalized Anxiety Disorder (GAD). Specifically,
(MIDUS; Brim etal., 2011; Ryff etal., 2012). The I included indicators of poor appetite, lack of
NSDE included measures of depression and anxiety concentration, fatigue, restlessness, anxiety, nega-
symptoms over a period of several days, allowing for tive self-cognitions (e.g., guilt or worthlessness),
an examination of the causal structure of these symp- depressed mo-od, and irritability. Fatigue and poor
toms as they unfold over time. To my knowledge, appetite were assessed by two independent sets of
this is the first time liability and formative models of items asking whether participants experienced
symptom structure have been directly compared. In fatigue or poor appetite during the day, and if they
doing so, Ihope to illustrate important considerations did, the severity of poor appetite or fatigue (each
in comorbidity modeling, such as the importance of rated on a 10-point scale, resulting in an 11-point
model comparison, parsimony, and generalizability. scale for each including those who reported no
Ialso believe that these model comparisons will help symptoms). The other criteria were assessed by
illustrate how analysis of comorbidity can help clarify items asking participants how much of the time
the nature of psychopathology constructs. during the day they felt each emotional or psycho-
logical state (rated on a 5-point scale). Anxiety was
The National Study of Daily coded as the average of the nervous and afraid
Experiences:Overview ratings, and negative self-cognitions were coded as
The NSDE is a subproject of the larger MIDUS the average of the worthless and ashamed rat-
study, which is a longitudinal, nationally represen- ings. Poor concentration was coded as the reverse of
tative study of psychosocial and health variables in the attentive rating. Measures were administered
mid-adulthood (Brim etal., 2011; Ryff etal., 2012). using a computer-assisted telephone interview (for
MIDUS itself comprises two waves of data collec- further details, see Ryff & Almeida,2010).
tion, 10years apart, on a nationally representative
sample of 7108 adults in the United States, includ- Models
ing siblings and twins. The initial wave of MIDUS Two types of models were fit to the depres-
data (MIDUS I) was collected in 19951996; sion and anxiety symptoms: liability models and

36 From Comorbidit y to Constructs

Int1 Int2 Int3

anx1 dep1 irr1 anx2 dep2 irr2 anx3 dep3 irr3

ea1 ea2 ea3

ed1 ed2 ed3
ei1 ei2 ei3

Figure 3.4Liability (top) and formative (bot-

tom) models of internalizing symptoms in the
anx1 dep1 irr1 anx2 dep2 irr2 anx3 dep3 irr3 NSDE. For simplicity, only three criteria and
three time points are shown (the models actually
include eight criteria at eight time points).

formative models. The basic characteristics of the time, such that irritability per se on one day is assumed
two models are illustrated in Figure 3.4, with the to influence irritability on the nextday.
liability model illustrated in the top, and the for- In both types of models, parameters were
mative model illustrated in the bottom. The mod- assumed to be the same across days. For example,
els are illustrated using only three variables and at in the formative model, correlations between symp-
three time points, due to space considerations; in toms were assumed to be the same across days, and
the actual models, eight symptom variables were in the liability model, the loadings were assumed
modeled over eight time points. Note also that the to be the same. Similarly, the autoregressive paths
relationship between the symptoms and diagnoses were constrained to be the same across days. This
are omitted in the models, as the focus in the mod- assumption is arguably reasonable, as it is difficult to
eling was on symptoms (for example, the portion of hypothesize a reason why those relationships should
the formative model involving the diagnoses, on the change across the week of the study (as opposed to
right side, are omitted). changing across the years of a longitudinal study,
Both types of models comprised two sets of param- for example, where long-term developmental effects
eters. First, both models comprise paths representing might become relevant).
the relationships between the symptoms at each time
point, illustrated in the top portion of the figures for Results and Discussion
both modelsinternalizing liability in the liability Results of the model comparisons are shown in
model, and the correlations between the symptoms Table 3.1. The table presents the number of freely
in the formative model. In the liability model, rela- estimated model parameters, the log-likelihood of
tionships between the symptoms are explained by an the model, and three model selection statistics. Two
underlying internalizing factor, liability, or construct; of them, the Akaike Information Criterion (AIC)
in the formative model the symptoms are directly and Bayesian Information Criterion (BIC), are rela-
associated with one another. In the liability model, tive model selection statistics, in that they provide
there are autoregressive paths from internalizing at information about how the models compare to
one time point to internalizing at another time point, one another. The other, the root mean square error
reflecting the tendency for internalizing on one day of approximation (RMSEA) is an absolute model
to influence internalizing on the nextday. selection statistic, in that it provides information
The second set of paths in both models are the resid- about how each model compares to the data (note
ual paths, which represent the unique aspect of each that for all three model selection statistics, smaller
symptom that is not shared with other symptoms. For values are preferable).
example, in both models there is assumed to be some-
thing unique about irritability that is not shared with Parsimony
other symptoms, under either model. These residual The model selection statistics in Table 3.1 illus-
factors are also assumed to influence one another over trate the importance of parsimony in comorbidity

Markon 37
Table3.1 Comparison of Liability and Formative Models of Depressive and
Anxiety Symptoms in theNSDE

Liability model 34 50,688.34 101,444.69 101,614.14 .048

Formative model 52 50,982.11 102,068.22 102,327.37 .049

Note. Values represent number of model parameters (k), the log-likelihood (lnL), root mean
square error of approximation (RMSEA), Akaikes Information Criterion (AIC), and the Bayesian
Information Criterion (BIC).

theory and modeling. For example, the RMSEA is often appropriate and reflects the primary focus
values for each model suggest that both models fit of interest (such as with exogenous co-occurrence
the data approximately equally closely (.048 for the models), in many cases, the same theories might
liability model versus.049 for the formative model), apply equally to comorbidities between multiple
and equally well. Overall, based on RSMEA values disorders. In such cases, it may be more fruitful
alone, one might conclude that both models fit to conceptualize comorbidity and its explanations
equally well, or possibly that the liability model fits in terms of multimorbidity, or the co-occurrence
the data slightly more closely. or covariance of multiple disorders simultaneously
This conclusion is misleading, however, because (associations between depression, anxiety, phobia,
the formative model achieves this fit by using and panic considered simultaneously, for example).
one-and-a-half times as many parameters as the Even in cases where only two disorders are involved,
liability model. The liability model is much more specifying theories in terms of multiple indicators
parsimonious and efficient in this regard, providing (e.g., symptoms or criteria) may be more appropri-
a similar level of fit as the formative model but with ate. In extending theories formulated for two dis-
fewer parameters and less complexity. This greater orders to multiple disorders or indicators, however,
level of parsimony is reflected in the smaller, more complexity sometimes grows exponentially, and it
optimal AIC and BIC values for the liability model becomes necessary to consider more parsimonious
(AIC = 101,444.69; BIC = 101,614.14) com- accounts.
pared to the formative model (AIC=102,068.22; For example, with only two disorders, the final
BIC=102,327.37). two models of Figure 3.1b (roughly correspond-
The greater parsimony of the liability model ing to liability and formative models) are equally
is important because the statistical literature sug- complex and parsimonious. As more disorders are
gests that conclusions based on more parsimoni- included, however, the final model of Figure 3.1b
ous models are more likely to generalize (in the becomes exponentially more complex, because
sense of cross-validating; Grunwald, 2007; Pitt & distinct relationships between each pair of dis-
Myung, 2002). That is, to some extent, estimates orders are invoked. The penultimate model in
from the formative model will reflect overfitting Figure 3.1b, in contrast, increases in complexity
to a particular sample and will generalize less well less with each additional disorder, because only
and be less replicable in other samples. Although a single additional path (from the latent variable
this is also true of the liability model, because the to the indicator) is added for each additional dis-
liability model has fewer parameters and is less com- order. In this way, a theory or model that seems
plex, there is less to overfit with, and conclusions reasonably simple with two disorders can quickly
based on it are likely to be more generalizable and become conceptually and statistically complex,
replicable. perhaps unnecessarilyso.
Although considerations of parsimony are
important in any empirical modeling, they become Theory Explication in Model Comparison
particularly important in comorbidity research. Multimorbidity and symptom models illustrate
Historically, almost by definition, comorbidity another, related important consideration in comor-
theory and research has tended to focus on explana- bidity theory and research: that theories be elabo-
tions of associations between two disorders or condi- rated to include as many differentiable features as
tions at a time (between depression and anxiety, for possible. Traditional philosophies of science gener-
example, or depression and cancer). Although this ally characterize scientific theory testing in terms of

38 From Comorbidit y to Constructs

a process whereby alternate theories are elaborated Exogenous co-occurrence models of comorbid-
to include predictions or hypotheses that distinguish ity, which emphasize the effects of comorbidity
one theory from another (e.g., Popper, 1963; van rather than its causes, also benefit from more elabo-
Fraassen, 1980). These theories are then compared rated theory testing. In these cases, including mul-
in empirical designs that produce data that would tiple disorders hypothesized to exert the same effect
be explained differently by the two theories, and the provides multiple tests of the underlying theory.
theories are compared using statistical model com- Similarly, formulating models in terms of multiple
parisons. More elaborated theories facilitate this symptom-level indicators helps provide more spe-
process by providing more, and more distinguish- cific, nuancedand therefore more powerful
able, tests against alternate theories. evaluation of a theory versus alternatives, and helps
In this regard, comorbidity theories can be com- identify specific pathways that are mediating the
pared better when they are made more specific and effects of comorbidity on outcome.
elaborated in their predictions and hypotheses, in
terms of more empirical phenomena that are being Liability and Formative Models of
explained. Formulating tests of comorbidity theo- Depression and Anxiety
ries in terms of multiple disorders that are posited Overall, the results in Table 3.1 support a basic
share similar etiologies, or in terms of specific, mul- liability model for anxiety and depressive symp-
tiple indicators is more scientifically productive in a toms relative to a basic formative model. That is,
sense, because it provides more data points by which phenomena such as irritability, depressed mood,
those theories can be compared. Somewhat ironi- anxiety, fatigue, and poor appetite are best concep-
cally, carrying this reasoning to its extreme, explana- tualized not as independent but associated experi-
tions for comorbidity may be more easily identified ences, but rather as imperfect indicators of some
by abandoning research on comorbidity per se in common underlying state or liability. Although
favor of studies focusing on underlying processes or the nature of the underlying liability is unknown,
symptoms of the disorders involved. it likely involves negative emotion; according the
This is similar in logic to the endophenotype model, this emotional liability persists somewhat
paradigm (Gottesman & Shields, 1973; Kendler & from day to day, imparting continuity in emotional
Neale, 2010), which advocates focusing on underly- state acrosstime.
ing processes that are believed to be closer to the It is important to emphasize that under this
etiologies of the disorders involved (for example, liability model the symptoms do each have some
neuropsychological indicators of perseveration, or unique component that is distinct from the under-
negative emotion rather than major depressive dis- lying liability influencing all depression and anxiety
order per se). From a philosophy of science stand- symptoms. Fatigue, for example, is not completely
point, endophenotypes may be most helpful in an indicator of emotional liabilityit is also an
elucidating comorbidity not because they involve indicator of fatigue per se, independent of anything
causally primary variables, but because they involve elseand this unique component persists across
more numerous and more specific variables by time as well. It is reasonable, moreover, to hypothe-
which theories can be evaluated. For example, some size that these unique components of each indicator
have questioned the premise of endophenotype the- could be related to one another above and beyond
ory that some variables are more causally primal, or any effects of broad internalizing liability. This sug-
atomistic in the sense of being less reducible or less gests the possibility of hybrid liability-formative
complex than other variables (e.g., Flint & Munafo, models, where there are direct interindicator rela-
2007). However, to the extent that a model is for- tionships in addition to liability effects.
mulated in terms of endophenotypes, which are For example, one might hypothesize that in
more precise in terms of hypothesized causal pro- addition to being indicators of internalizing liabil-
cesses, that model then presumably is more precise ity, fatigue and attention might be directly related,
and specific in terms of how it is formulated and in that fatigue might directly influence concentra-
the predictions it makes. The greater specificity or tion and attention. This model can be specified by
precision of that model then might afford more allowing the residual components of the fatigue and
features whereby it can be empirically evaluated. attention indicators to be directly correlated (i.e., in
Endophenotype models, under this reasoning, the notation of Figure 3.4, by adding bidirectional
might not generally be more empirically valid or paths between the e1 fatigue and attention factors,
true, but might be more testable or falsifiable. between the e2 fatigue and attention factors, and

Markon 39
so forth for each day). Fitting this model results example, are likely to resemble both the liability and
in essentially the same RMSEA value as the basic formative models in some respects. Comparisons of
liability model (.048), but improves the AIC internalizing structural models, moreover, generally
(101,391.51) and BIC (101,565.94) values slightly. indicate that distinct depression and anxiety fac-
Although the estimated correlation between fatigue tors can be identified (Krueger & Markon, 2006;
and inattention is small (.120), it is significant (the Watson etal., 2007; Markon, 2010), and cannot be
95% CI ranges from.069 to.170). entirely replaced by a single internalizing liability.
Overall, these results suggest that the processes That is, a hierarchy of processes likely exists, includ-
involved in depression and anxiety symptoms likely ing a superordinate internalizing liability that affects
combine some features of the liability and formative a wide spectrum of symptoms (e.g., dysphoria,
models. That is, depression and anxiety symptoms anhedonia, panic, irritability, and traumatic cogni-
share a common internalizing liability, but also may tions), more subordinate liabilities that influence
have small direct effects on one another above and specific types of symptoms (e.g., depression), as well
beyond this common internalizing liability. In terms as symptom-level processes per se (e.g., fatigue caus-
of comorbidity, these results suggest that comorbid- ing po-or concentration).
ity between depressive and anxiety disorders likely Ultimately, attempts to explain comorbidity
results mostly from a shared internalizing liability, suggest a significantly different conceptualization
but also in part from direct relationships between of what is meant by depression or anxiety than
the symptoms that constitute the disorders. what is specified in formal nosologies (for exam-
ple, the DSM or ICD; Markon, 2010). That is, in
Conclusions explaining the causes of covariance between two
The liability and formative models present disorders, we redefine those disorders in terms of an
remarkably different accounts of the causes of alternate set of constructs. In this way, comorbidity
comorbidity: one posits that comorbidity occurs research has resulted in a fundamental shift in our
because of a shared etiology, the other posits that understanding of how different forms of psychopa-
comorbidity occurs as an epiphenomenon of defini- thology are structured and the processes involved.
tional criteria directly influencing each other in the
absence of any shared etiology. The model compari- Future Directions
sons presented here offer some support for aspects As empirical models of comorbidity continue to
of both models, but also illustrate a more general evolve and change in focus, a number of important
point:delineating causes of comorbidity helps clar- general questions are emerging. One question, for
ify and even redefine the nature of the psychological example, is how to flexibly identify causes of asso-
disorder involved. ciation between different forms of psychopathology
Models such as the liability and formative mod- without making assumptions about the nature of
els were originally formulated to explain how two those associationsfor example, that the associa-
disorders covary. In the process of explaining this tions are necessarily only mediated by latent liabili-
covariance, however, the two theories posit accounts ties, or are only mediated through reciprocal effects
that require significant reconceptualizations of what of symptoms on each other. Another question is
is meant by the disorders involved. Although the at what level psychopathology should be exam-
two models differ in the nature of this reconceptual- ined. Finally, important questions are emerging
ization, they both propose that distinct disorders do about how to model interactions between different
not necessarily exist in the way they are traditionally pathologies, without oversimplifying or drawing
understood. In the case of the basic liability model spurious conclusions.
of depressive and anxious comorbidity, for example,
depressive disorder and anxiety disorder are in some Flexible Causal Models
sense replaced by a single internalizing liability; of Comorbidity
in the case of the basic formative model, depres- One emerging issue is how to model comor-
sive disorder and anxiety disorder are replaced by bidity, or associations between different forms of
dynamically related symptoms. psychopathology, in a way that is flexible with
As evidenced by these analyses, reality is likely to regard to causal assumptions. Formative models
be more nuanced than either a basic liability model of the sort examined here, for example, allow one
or a basic formative model. The processes affect- to flexibly model reciprocal relationships between
ing actual depressive and anxiety symptoms, for observed variables in an exploratory sense.

40 From Comorbidit y to Constructs

However, these same formative models do not They work within a graphical or structural equation
accommodate unobserved liabilities in the form modeling framework similar to Spirtes etal. (2000);
of latent variables. Conversely, traditional liability although their methods are promising, further
models do not allow one to flexibly model latent research is needed to compare these approaches to
variables and residual observed variable relation- traditional SEM methods, particularly in terms of
ships simultaneously. These models either require their ability to recover generalizable models without
specifying the nature of latent variables a priori capitalizing on chance (see also Danks etal., 2010,
(for example, using confirmatory factor analysis), for an example of their methods and related discus-
or prohibit exploratory identification of resid- sion of these issues).
ual correlations between observed variables (for
example, in an exploratory factor analysis). How Levels of Analysis
to explore comorbidity structures involving mix- Another emerging issue in comorbidity model-
tures of liability and formative models is currently ing is the question of what level of analysis is most
unclear. appropriate for understanding associations between
Advances in graphical modeling offer potential different forms of psychopathology. For example,
solutions to these problems. Graphical models can should comorbidity be examined at the level of
roughly be thought of as generalizations of, or forms disorders, symptoms, or underlying liability dimen-
of, structural equation models (Borgelt & Kruse, sions? If the latter, in what way should these liabili-
2002; Spirtes, Glymour, & Scheines, 2000). Various ties be examined? To some extent, these questions
authors in the graphical modeling literature have reflect broader debates within the field about how
explored ways to automatically select different struc- to conceptualize mental disorder, but they have
tural models, analogous to automatically searching, important ramifications for understanding what
comparing, and selecting optimal confirmatory fac- causes covariance and interactions among different
tor models (Silva, Scheines, Glymour, and Spirtes, forms of disorder.
2006; Spirtes etal., 2000). For example, rather than Approaching revisions to official nosologies,
specifying different models of how depression and such as the DSM and ICD, have focused attention
anxiety symptoms are related based on theoretical on general questions about how to define mental
considerations, one could use an algorithm that disorder. Among these many questions is how to
automatically generates and compares a wide vari- identify the optimal level of analysis for describ-
ety different possible models, and selects the opti- ing and explaining psychopathology. Many have
mal model among them. Spirtes etal. (2000) have advocated for focus on transdiagnostic constructs
advocated for such structural model search algo- (e.g., Nolen-Hoeksema & Watkins, 2011), which
rithms, arguing that they are more rigorous than are constructs that are relevant to understanding
model comparisons based on theory-driven model psychopathology across multiple defined disorders.
comparisons. This transdiagnostic paradigm forms the basis for
These algorithms offer a promising way to initiatives such as the National Institute of Mental
explore different accounts of comorbidity and the Healths Research Domain Criteria (RDoC; Insel
causal associations that mediate patterns of comor- etal., 2010), which seeks to identify a set of work-
bidity. How to incorporate latent variables of the ing focal constructs for psychopathology research,
sort typically encountered in psychopathology together with a set of hypothesized processes
research (e.g., multiple correlated latent variables, involved in each construct across different lev-
each with multiple indicators) has proved chal- els of analysis (e.g., genes, cells, neural circuits, or
lenging, however. Kemp and Tenenbaum (2008) behavior).
provided a framework for identifying optimal struc- Although such efforts are framed as being transdi-
tural models, including models with features some- agnostic, the implied motivation is often not simply
what analogous to latent variables. Their approach to identify constructs that operate across different
assumes that the structure has some regularity in its disorders, but to redefine the focus of psychopa-
features, however, and is limited in how the latent thology research. Specifically, the intended change
features can be parameterizedit is more similar in focus is toward constructs that vary continuously
to cluster analysis than structural equation model- and have a basis in normative neurobehavioral pro-
ing in many respects. Silva et al. (2006) describe cesses (Hyman, 2010; Insel etal., 2010). This has
methods for delineating structural models with important implications for comorbidity research, if
latent variables that does not have these limitations. for no other reason than that it implies a shift in

Markon 41
what forms of psychopathology might be consid- those that are made on the basis of overfitting to
ered to be co-occurring or covarying. particular samples, or using overly flexible models
Various authors have expressed concern that in and designs.
the process of redefining focal psychopathology Problems with spurious or ungeneralizable con-
constructs, the level of analysis might shift, inten- clusions are not new and not unique to comorbidity
tionally or unintentionally, resulting in the neglect research. However, as models of comorbid phenom-
of important etiologies and sequelae of psychopa- ena become more complex, and as the designs of
thology. For instance, in the RDoC initiative thus studies incorporate more features, the opportunities
far, although substantial effort has been made to to make inappropriate inferences increase. Various
define psychopathology across multiple levels of authors have noted how flexibility in study designs
analysis, certain levels of analysis (e.g., genes, cells, and models of resultant data can lead to unrepli-
or circuits) have at least implicitly had greater focus cable or ungeneralizable conclusions. More flexible
relative to other levels of analysis through their models, for example, allow one to capture subtleties
absence (e.g., relationships, local environments, of complex effects, but also afford more opportuni-
or culture; see also Insel et al., 2010). Some have ties to overfit to data (Pitt & Myung, 2002). More
cautioned that attempts to identify optimal levels flexible designs with more features, similarly, afford
of analysis in psychopathology, implicitly or explic- more opportunities for multiple testing (Simmons,
itly, may lead to neglect of fundamental etiological Nelson, & Simonsohn,2011).
processesthat it is impossible to fully understand Nevertheless, developments in psychopathol-
psychopathology at any fixed level of analysis. ogy create important opportunities for the devel-
Kendler (2012a), for instance, using major depres- opment of new methods for understanding how
sive disorder and alcohol use disorder as examples, different forms of psychopathology arise, and how
reviewed the literature and concluded that, for most they might interact to affect outcomes. Work in
forms of psychopathology, focusing on any given the areas of multilevel modeling (i.e., mixed effects
level of analysis would be grossly misleading in modeling) will continue to grow, accommodating
terms of understanding etiology. In a series of arti- multiple levels of analysis simultaneously. Similarly,
cles (Kendler, 2005, 2012a, 2012b), he has noted work on complex forms of moderation, at the level
that many questions can be asked about the nature of observed variables as well as latent constructs
of psychopathology, and these questions will dictate (Edwards & Lambert, 2007; Klein & Moosbrugger,
the appropriate levels of analysis involved. 2000) will help clarify how comorbidity arises
Similar issues arise when examining comorbidity and influences outcomes. Finally, as these models
per se. Effects of comorbid disorders, for example, develop, there will be a need for methods for select-
may manifest at any number of levels, and the ques- ing models and making inferences aboutthem.
tions that are asked about the nature of those effects
will frame the appropriate level of analysis. It is sim- Clinical Implications
ilarly likely that the causes of comorbidity will span One implication of the models examined here is
multiple different levels of analysis, in different ways that individuals presenting problems often reflect
for different disorders. At a more fundamental level, broad underlying liabilities, liabilities that transcend
evolution in how psychopathology is defined at specific symptoms or even diagnostic boundaries.
different levels of analysis will shape study of what That is, an individuals depression, fear, or anxiety
comorbid conditions are involved. may reflect broader problems with regulation of
negative emotion, patterns of cognitive appraisal,
Modeling Complex, Multilevel Effects or chronic stressors, rather than depression, fear, or
As the study of comorbidity shifts toward more anxiety per se. In this regard, for many clients, it may
specific mechanisms, spanning multiple levels of be more advantageous to focus on identifying broad
analysis, it is necessary to develop methodologies for precipitating factors rather than the specific details
studying this phenomenon. These methods ideally of a presenting problem as they manifest at any given
will accommodate potential interactions between time. Many clients are at as great a risk for future
multiple variables at multiple levels of analysis (e.g., problems of the same broad liability class as they
interactions between variables at the level of neu- are the problem they initially present with. For exam-
ral tracts and social environments), involving many ple, a client who initially presents with depression is
different data structures. It is also necessary for these just as likely to present with anxiety problems later as
methods to prevent spurious conclusions, such as they are to present with depression again (Fergusson

42 From Comorbidit y to Constructs

etal., 2006). In this sense, relapse in practice often with depression more specifically, would help char-
refers not to the same symptom presentation over acterize the breadth of the clients difficulties.
and over again, but the same underlying problems Many similar issues pertaining to the specific-
over and over. Focusing solely on the specific details ity of symptoms have arisen in the DSM-5 revision
of a clients difficulties as they manifest at intake may process. For example, as noted before, many have
fail to address the underlying etiologies that increase suggested that the internalizing symptoms of PTSD
the risk for other related difficulties. are not qualitatively different from those of other
On the other hand, the models reviewed here forms of internalizing psychopathology (Breslau,
also suggest that details do matter. Specific symp- Chase, & Anthony, 2002; Yufik & Simms, 2010).
toms and situational factors can themselves perpetu- This raises the question of whether the trauma cri-
ate dynamic cycles of difficulties, whereby one crisis terion of the disorder is necessary for a diagnosis.
feeds another. Intervening at the level of specific Similar questions have arisen about the DSM-5
symptoms or situational stressors can help amelio- draft criteria for major depression. For example,
rate an individuals difficulties, either by resolving the bereavement exclusionary criterion has been
primary problems, or by increasing the clients func- relaxed, allowing a client to be diagnosed with
tioning to a point where broader problems can be depression in the presence of a loss. The literature
more easily addressed. For many clients, the devil reviewed here would likely support this relaxation,
is really in the details, so to speak, and should be a insofar as the difficulties involved in bereavement
major focus of treatment. would resemble that of depression, as well as other
This tensionbetween treatment of specific internalizing problems. Seen another way, the ques-
situation-response patterns versus broader pre- tion of whether or not bereavement is distinct from
disposing factorsis a longstanding issue in the depression as a disorder is somewhat abstract, with
history of clinical psychology and psychiatry. little practical consequence in many cases:the psy-
Recurring discussions of third wave versus older chological pain of both can be similar and might be
forms of cognitive-behavioral therapy, for example, treated similarly.
have focused in part on this issue (Hayes, 2004),
as have related discussions of antecedent ver- References
sus response-focused intervention (Hofmann & Almeida, D. M., & Kessler, R. C. (1998). Everyday stressors and
Asmundson, 2008). The models and findings dis- gender differences in daily distress. Journal of Personality and
Social Psychology, 75, 670680.
cussed here suggest that this tension is in some ways
Beesdo-Baum, K., Hofler, M., Gloster, A. T., Klotsche, J., Lieb,
misguided, as both are likely to be relevant depend- R., Beauducel, A.,...Wittchen, H.-U. (2009). The structure
ing on the particular individual and circumstances. of common mental disorders:Areplication study in a com-
Assessment plays a critical role in this regard:some munity sample of adolescents and young adults. International
clients will report difficulties of a relatively specific Journal of Methods in Psychiatric Research, 18, 204220.
Blalock, H. M. (1964). Causal inferences in nonexperimen-
nature, others difficulties that are broader in scope.
tal research. Chapel Hill, NC: University of North
Models discussed here suggest that cases where CarolinaPress.
problems are broader in nature do not necessarily Bollen, K. A. (1989). Structural equations with latent variables.
reflect a greater number of distinct problems, but NewYork, NY:Wiley.
might reflect a single underlying process, one that Borgelt, C., & Kruse, R. (2002). Graphical models:Methods for
data analysis and mining. West Sussex, UK:Wiley.
nonetheless impacts a relatively broad range of out-
Breslau, N., Chase, G., & Anthony, J. (2002). The unique-
comes. Identifying clients for whom these broader ness of the DSM definition of post-traumatic stress disor-
liabilities are critical factors, versus those for whom der: Implications for research. Psychological Medicine, 32,
more specific problems are critical, can be par- 573576.
tially addressed through appropriate assessment. It Brim, O. G., Baltes, P. B., Bumpass, L. L, Cleary, P. D.,
Featherman, D. L., Hazzard, W. R.,...Shweder, R. A.
is important in this regard to assess for a range of
(2011). National Survey of Midlife Development in the United
potential problems, to delineate the boundaries of a States (MIDUS), 19951996 (ICPSR02760-v8). Ann Arbor,
clients difficulties. For example, one client present- MI: Inter-university Consortium for Political and Social
ing with extreme guilt, irritability, and poor concen- Research.
tration may have problems limited to depression in Caron, C., & Rutter, M. (1991). Comorbidity in child psycho-
pathology:Concepts, issues and research strategies. Journal of
response to a recent stressor; another client present-
Child Psychology and Psychiatry, 32(7), 10631080.
ing with the same difficulties may have problems Cramer, A. O. J., Waldorp, L. J., van der Maas, H. L. J., &
with negative appraisals more broadly. Assessing Borsbo-om, D. (2010). Comorbidity: A network perspec-
such a clients general level of internalizing, along tive. Behavioral and Brain Sciences, 33(23), 137150.

Markon 43
Danks, D., Fancsali, S., Glymour, C., & Scheines, R. (2010). of an etiologically based nosology. Molecular Psychiatry,
Comorbid science? Behavioral and Brain Sciences, 33(23), 17(1),1121.
153155. Kendler, K. S. (2012b). The dappled nature of causes of psychiat-
Dikeos, D. G., Wickham, H., McDonald, C., Walshe, M., ric illness:Replacing the organicfunctional/hardwaresoft-
Sigmundsson, T., Bramon, E., Grech, A.,...Sham, P. (2006). ware dichotomy with empirically based pluralism. Molecular
Distribution of symptom dimensions across Kraepelinian Psychiatry, 17(4), 377388.
divisions. British Journal of Psychiatry, 189(4), 346353. Kendler, K. S., & Neale, M. C. (2010). Endophenotype:Acon-
Edwards, J. R., & Bagozzi, R. P. (2000). On the nature and ceptual analysis. Molecular Psychiatry, 15(8), 789797.
direction of relationships between constructs and measures. Kendler, K. S., Prescott, C. A., Myers, J., Neale, M. C. (2003).
Psychological Methods, 5(2), 155174. The structure of genetic and environmental risk factors for
Edwards, J. R., & Lambert, L. S. (2007). Methods for integrat- common psychiatric and substance use disorders in men and
ing moderation and mediation:Ageneral analytical frame- women. Archives of General Psychiatry, 60, 929937.
work using moderated path analysis. Psychological Methods, Kessler, R. C., Ormel, J., Petukhova, M., McLaughlin, K.
12(1),122. A., Green, J. G., Russo, L. J.,...Ustun, T. B. (2011).
Feinstein, A. R. (1970). The pre-therapeutic classification of Development of lifetime comorbidity in the World Health
co-morbidity in chronic disease. Journal of Chronic Disease, Organization World Mental Health Surveys. Archives of
23, 455468. General Psychiatry, 68, 90100.
Fergusson, D. M., Horwo-od, L. J., & Boden, J. M. (2006). King, H. A. (2008). Neuroticism and indicators of daily
Structure of internalising symptoms in early adultho-od. health:Alifespan developmental approach. Unpublished mas-
British Journal of Psychiatry, 189, 540546. ters thesis, The Pennsylvania State University, University
Flint, J., & Munafo, M. R. (2007). The endophenotype concept Park,PA.
in psychiatric genetics. Psychological Medicine, 37, 163180. Klein, A., & Mo-osbrugger, H. (2000). Maximum likeliho-od
Fornell, C., & Bo-okstein, F. L. (1982). Two structural equation estimation of latent interaction effects with the LMS
models: LISREL and PLS applied to consumer exit voice method. Psychometrika, 65, 457474.
theory. Journal of Marketing Research, 19, 440452. Kotov, R., Chang, S.-W., Fochtmann, L. J., Mojtabai, R., Carlson,
Gale Research Group. (1965). Research centers directory. Detroit, G. A., Sedler, M. J., & Bromet, E. J. (2011). Schizophrenia
MI:Gale Research. in the internalizing-externalizing Framework:Athird dimen-
Go-otzeit, J., & Markon, K. (2011). Factors of PTSD:Differential sion? Schizophrenia Bulletin, 37(6), 11681178.
specificity and external correlates. Clinical Psychology Review, Krueger, R. F., & Markon, K. E. (2006). Reinterpreting comor-
31(6), 9931003. bidity:Amodel-based approach to understanding and classi-
Gottesman, I. I., & Shields, J. (1973). Genetic theorizing and fying psychopathology. Annual Review of Clinical Psychology,
schizophrenia. British Journal of Psychiatry, 122,1530. 2(1), 111133.
Grunwald, P. D. (2007). The minimum description length prin- Krueger, R. F., Markon, K. E., Patrick, C. J., Benning, S. D., &
ciple. Cambridge, MA:MITPress. Kramer, M. (2007). Linking antisocial behavior, substance
Hayes, S. C. (2004). Acceptance and commitment therapy, rela- use, and personality:An integrative quantitative model of the
tional frame theory, and the third wave of behavioral and adult externalizing spectrum. Journal of Abnormal Psychology,
cognitive therapies. Behavior Therapy, 35, 669665. 116, 645666.
Hofmann, S. G., & Asmundson, G. J. G. (2008). Acceptance Lahey, B. B., Van Hulle, C. A., Singh, A. L., Waldman, I. D., &
and mindfulness-based therapy: New wave or old hat? Rathouz, P. J. (2011). Higher-order genetic and environmen-
Clinical Psychology Review, 28,116. tal structure of prevalent forms of child and adolescent psy-
Howell, R. D., Breivik, E., & Wilcox, J. B. (2007a). Reconsidering chopathology. Archives of General Psychiatry, 68(2), 181189.
formative measurement. Psychological Methods, 12(2), 205218. Larsen, R. J., & Kasimatis, M. (1991). Day-to-day physical
Howell, R. D., Breivik, E., & Wilcox, J. B. (2007b). Is formative symptoms: Individual differences in the occurrence, dura-
measurement really measurement? Reply to Bollen (2007) tion, and emotional concomitants of minor daily illnesses.
and Bagozzi (2007). Psychological Methods, 12(2), 238245. Journal of Personality, 59, 387423.
Humphry, S. M., & McGrane, J. A. (2010). Is there a contradic- Lilienfeld, S. O., Waldman, I. D., & Israel, A. C. (1994). A criti-
tion between the network and latent variable perspectives? cal examination of the use of the term and concept of comor-
Behavioral and Brain Sciences, 33(23), 160161. bidity in psychopathology research. Clinical Psychology:
Hyman, S. E. (2010). The diagnosis of mental disorders: The Science and Practice, 1, 71103.
problem of reification. Annual Review of Clinical Psychology, Markon, K. E. (2010). Modeling psychopathology struc-
6, 155179. ture: A symptom-level analysis of Axis I and II disorders.
Insel, T., Cuthbert, B., Garvey, M., Heinssen, R., Pine, D. S., Psychological Medicine, 40(02),273.
Quinn, K., ...Wang, P. (2010). Research Domain Criteria Markus, K. A. (2010). Questions about networks, measure-
(RDoC):Toward a new classification framework for research ment, and causation. Behavioral and Brain Sciences, 33(23),
on mental disorders. American Journal of Psychiatry, 167(7), 164165.
748751. McFarland, D. J., & Malta, L. S. (2010). Symptoms as latent
Kemp, C. & Tenenbaum, J. B. (2008). The discovery of struc- variables. Behavioral and Brain Sciences, 33(23), 165166.
tural form. Proceedings of the National Academy of Sciences, Molenaar, P. C.M. (2010). Latent variable models are network
105(31), 1068710692. models. Behavioral and Brain Sciences, 33(23),166.
Kendler, K. S. (2005). Toward a philosophical structure for psy- Moussavi, S., Chatterji, S., Verdes, E., Tandon, A., Patel, V., &
chiatry. American Journal of Psychiatry, 162(3), 433440. Ustun, B. (2007). Depression, chronic diseases, and decre-
Kendler, K. S. (2012a). Levels of explanation in psychiatric and ments in health: Results from the World Health Surveys.
substance use disorders: Implications for the development Lancet, 370(9590), 851858.

44 From Comorbidit y to Constructs

Mroczek, D. K., & Kolarz, C. M. (1998). The effect of age on Silva, R., Scheines, R., Glymour, C., & Spirtes, P. (2006).
positive and negative affect:Adevelopmental perspective on Learning the structure of linear latent variable models.
happiness. Journal of Personality and Social Psychology, 75, Journal of Machine Learning Research, 7, 191246.
13331349. Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011).
Nolen-Hoeksema, S., & Watkins, E. R. (2011). A heuristic False-positive psychology:Undisclosed flexibility in data col-
for developing transdiagnostic models of psychopathol- lection and analysis allows presenting anything as significant.
ogy: Explaining multifinality and divergent trajectories. Psychological Science, 22(11), 13591366.
Perspectives on Psychological Science, 6(6), 589609. Spirtes, P., Glymour, C. & Scheines, R. (2000). Causation, pre-
Pfeffer, C. R., & Plutchik, R. (1989). Co-occurrence of psychi- diction, and search. Cambridge, MA:MITPress.
atric disorders in child psychiatric patients and nonpatients: van den Akker, M., Buntinx, F., & Knottnerus, J. A. (1996).
A circumplex model. Comprehensive Psychiatry, 30(4), Comorbidity or multimorbidity:Whats in a name? Areview
275282. of the literature. European Journal of General Practice, 2(2),
Pitt, M. A., & Myung, I. (2002). When a good fit can be bad. 6570.
Trends in Cognitive Sciences, 6(10), 421425. van Fraassen, B. C. (1980). The scientific image. London, UK:Oxford.
Popper, K. R. (1963). Conjectures and refutations. London, Watson, D., Clark, L. A., & Tellegen, A. (1988). Development
UK:Routledge. and validation of brief measures of positive and negative
Ryff, C. D., & Almeida, D. M. (2010). National Survey of Midlife affect: The PANAS scales. Journal of Personality and Social
in the United States (MIDUS II):Daily Stress Project, 2004 Psychology, 54, 10631070.
2009 (ICPSR26841-v1). Ann Arbor, MI: Inter-university Watson, D., OHara, M. W., Simms, L. J., Kotov, R.,
Consortium for Political and Social Research. Chmielewski, M., McDade-Montez, E.,...Stuart, S. (2007).
Ryff, C., Almeida, D. M., Ayanian, J. S., Carr, D. S., Cleary, P. D., Development and validation of the Inventory of Depression
Coe, C., Davidson, R.,...Williams, D. (2012). National Survey and Anxiety Symptoms (IDAS). Psychological Assessment,
of Midlife Development in the United States (MIDUS II), 2004 19(3), 253268.
2006 (ICPSR04652-v6). Ann Arbor, MI: Inter-university Wolf, A. W., Schubert, D. S. P., Patterson, M. B., Grande,
Consortium for Political and Social Research. T. P., Brocco, K. J., & Pendleton, L. (1988). Associations
Seeley, J. R., Kosty, D. B., Farmer, R. F., & Lewinsohn, P. M. among major psychiatric diagnoses. Journal of Consulting and
(2011). The modeling of internalizing disorders on the basis Clinical Psychology, 56, 292294.
of patterns of lifetime comorbidity:Associations with psycho- Yufik, T., & Simms, L. J. (2010). A meta-analytic investigation
social functioning and psychiatric disorders among first-degree of the structure of posttraumatic stress disorder symptoms.
relatives. Journal of Abnormal Psychology, 120(2), 308321. Journal of Abnormal Psychology, 119, 764776.

Markon 45

4 Examining the Comorbidity Between

Depression and the Anxiety Disorders From
the Perspective of the QuadripartiteModel
David Watson and Sara M.Stasik

Major depression and posttraumatic stress disorder (PTSD) represent heterogeneous combinations
of symptoms. Analyses focusing on these distinctive symptom dimensions can play an important role in
explicating key diagnostic phenomena such as comorbidity. We review depression and PTSD from the
perspective of the quadripartite model, which posits that it is important to consider two quantitative
elements when analyzing the properties of symptoms:(a)the magnitude of their general distress
component and (b)their level of specificity. Within both disorders, we identified certain symptoms
insomnia and appetite disturbance in the case of depression, dysphoria within PTSDthat both
(a)exhibited poor diagnostic specificity and (b)provided little or no incremental information to their
respective diagnoses. We therefore argue that deemphasizing these weak and nonspecific indicators
and focusing primarily on more specific types of symptoms potentially can improve the diagnosis and
assessment of these disorders.
Key Words: major depression, posttraumatic stress disorder, comorbidity, general distress, symptom
specificity, dysphoria, diagnosis, assessment

Introduction DSM unipolar mood and anxiety disorder diagno-

In this chapter, we review the historical progres- ses across four large national epidemiological sam-
sion of structural schemes that attempt to model ples: the National Comorbidity Survey (Krueger,
the relation between depression and the anxiety dis- 1999; N = 8,098, except for posttraumatic stress
orders. This topic has been the focus of enormous disorder [PTSD], where N=5,877) and National
interest within clinical psychology and psychiatry Comorbidity Survey Replication (Kessler, Chiu,
since the 1980s. Much of the structural work in Demler, & Walters, 2005; N = 3,199) data in
this area has been stimulated by extensive evidence the United States; Wave 1 of the Netherlands
regarding diagnostic comorbidity. Comorbidity Mental Health Survey and Incidence Study (W.
is pervasive within the Diagnostic and Statistical A.Vollebergh, personal communication, December
Manual of Mental Disorders (DSM; American 15, 2003; N=7,076) and the Australian National
Psychiatric Association, 1980, 1987, 2000), and Survey of Mental Health and Well-Being (Slade &
it is particularly problematic within the mood and Watson, 2006; N = 10,641). Diagnoses of major
anxiety disorders (Clark, Watson, & Reynolds, depression correlated .64 with generalized anxiety
1995; Krueger & Markon, 2006; Mineka, Watson, disorder (GAD), .57 with obsessive-compulsivedis-
& Clark, 1998; Widiger & Clark,2000). order (OCD), .56 with PTSD, .55 with panic dis-
For example, Watson (2009) computed order, .50 with social phobia, .48 with agoraphobia,
weighted mean tetrachoric correlations between and .45 with simple/social phobia (see Watson,

2009, for more details). Similarly, tetrachoric cor- ranged from .72 (college students) to .85 (older
relations between dysthymic disorder and the adults), with an overall weighted mean value of.78.
anxiety disorders ranged from .41 (simple/specific The role of Positive Affect. How, then, can depres-
phobia) to .66 (GAD). More generally, most indi- sion and anxiety be distinguished? Findings from
viduals with major depression also meet criteria the mood literature establish that indicators of
for a comorbid anxiety disorder and vice versa (see the higher order Positive Affect factor consistently
Mineka etal., 1998; Watson, 2005,2009). correlate negatively with depressed mood and
symptomatology and are related more weakly to
PreviousModels measures of anxiety (Mineka et al., 1998; Watson
The Two-Factor AffectiveModel & Naragon-Gainey, 2010). One partial exception,
The Big Two dimensions of affect. Why do we however, is that low Positive Affect shows consis-
see such extensive comorbidity between the mood tent negative associations with social anxiety/social
and anxiety disorders? Watson, Clark, and Carey phobia (Kashdan, 2007; Naragon-Gainey, Watson,
(1988) originally developed an explanatory model & Markon, 2009). Low positive affectivity also is
that drew on key findings from the basic mood lit- consistently linked to negative symptoms of schizo-
erature. Extensive evidence establishes the existence phrenia/schizotypy (Watson & Naragon-Gainey,
of two dominant dimensions of emotional experi- 2010). Thus Watson and Naragon-Gainey (2010)
ence: Negative Affect (or Activation) and Positive recently concluded that the reviewed data establish
Affect (or Activation; e.g., Watson & Tellegen, that low levels of positive affect are a distinguishing
1985; Watson, Wiese, Vaidya, & Tellegen, 1999). feature of depression, social anxiety and schizophre-
Negative Affect is a general dimension of subjec- nia/schizotypy (pp.846847). They further added
tive distress and dissatisfaction. It subsumes a broad that a more limited range of evidence suggests that
range of specific negative emotional states, including indicators of positive affect are more strongly and
fear, anger, sadness, guilt, and disgust. Its emergence systematically linked to depression than to these
in analyses of affect ratings indicates that these vari- other syndromes (p.847).
ous negative emotions significantly co-occur both The two-factor model of affect. Putting these data
within and across individuals. Thus an individual together, Watson et al. (1988) proposed that low
who reports feeling anxious and fearful also is likely levels of positive affectivity were a specific feature
to report substantial levels of anger, guilt, sadness, of depression that distinguishes it from the anxiety
and so on. In parallel fashion, the general Positive disorders. Thus, in this two-factor model, negative
Affect dimension reflects important co-occurrences affectivity represents a nonspecific factor common
among positive mood states; for instance, an indi- to depression and anxiety, whereas low positive
vidual who reports feeling happy and joyful also affectivity is a specific factor that is related primarily
will report feeling interested, excited, confident, to depression.
Influence of the general Negative Affect factor. The TripartiteModel
Extrapolating from these data, Watson etal. (1988) Basic elements of the model. Clark and Watson
argued that this general Negative Affect dimension (1991) subsequently extended this two-factor model
was largely responsible for the substantial overlap/ by proposing a second specific factorphysiological
comorbidity between depression and anxiety. Put hyperarousalthat is relatively specific to anxiety.
differently, this higher order factor produces strong They therefore argued that a tripartite model more
correlations among different types of negative emo- accurately captures this domain. This model classi-
tion, including sad/depressed affect and fearful/anx- fies symptoms of anxiety and depression into three
ious affect. basic subgroups. First, many symptoms are strong
This idea has received extensive empirical sup- indicators of the general distress/negative affectiv-
port (see Watson, 2005; Watson, Clark, & Stasik, ity dimension; this nonspecific group includes both
2011). For example, Watson, OHara, and Stuart anxious and depressed mood (consistent with the
(2008) examined the association between mea- earlier two-factor model), as well as other symptoms
sures of depressed mood (e.g., I felt depressed, Ifelt (e.g., insomnia, poor concentration) that are prev-
like crying) and anxious mood (e.g., I felt anxious, alent in both the mood and anxiety disorders. In
Ifound myself worrying all the time) in eight different addition, however, each syndrome is characterized
samples (overall N = 3,549). These measures were by its own cluster of symptoms:somatic tension and
strongly related in every group; the correlations hyperarousal (e.g., shortness of breath, dizziness)

Watson,Stasik 47
are unique to anxiety, whereas anhedonia and low in that it fails to account for the heterogeneous
positive affectivity (e.g., loss of interest, feeling that nature of the anxiety disorders. This heterogene-
nothing is enjoyable) are specific to depression. ity has crucial implications for structural models.
Diagnostic and assessment implications. One Most notably, it means that a single specific factor
important implication of the tripartite model is is insufficient to capture the diversity of symptoms
that differential diagnosis and assessment can be subsumed within these disorders. Indeed, subse-
enhanced by focusing more on the specific symp- quent evidence has established that anxious arousal
tom clusters and deemphasizing nonspecific mani- is not generally characteristic of the anxiety disor-
festations of distress/Negative Affect. Watson ders but instead is more specifically related to panic
and Clark (1991) explicitly tested this possibil- disorder and PTSD (Brown, Chorpita, & Barlow,
ity by creating the Mood and Anxiety Symptom 1998; Watson,2009).
Questionnaire. The questionnaire contains two To remedy this problem, Mineka et al. (1998)
anxiety scales and two depression scales that vary proposed an integrative hierarchical model that
in their specificity. One member of each scale pair incorporated elements from both the tripartite
was designed to assess the general distress symptoms model and Barlows hierarchical organization of the
traditionally associated with that construct, whereas anxiety disorders (see Zinbarg & Barlow, 1996). In
the other was constructed to tap its unique symp- this scheme, each individual syndrome is hypoth-
tom cluster. Thus, the General Distress: Anxious esized to contain both a common and a unique
Symptoms scale includes indicators of anxious component. Consistent with earlier models, the
mood, as well as other anxiety disorder symptoms shared component represents broad individual dif-
that were expected to overlap heavily with depres- ferences in general distress/negative affectivity; it is
sion; in contrast, Anxious Arousal includes various a pervasive higher order factor that is common to
manifestations of somatic hyperarousal. Similarly, both the anxiety and mood disorders and primarily
the General Distress: Depressive Symptoms scale is responsible for the overlap/comorbidity between
contains items tapping depressed mood along them. In addition, each disorder also includes
with other nonspecific symptoms of mood disor- unique features that differentiate it from all of the
der, whereas Anhedonic Depression contains items others. Thus anxious arousal no longer is viewed as
reflecting anhedonia, disinterest, and anergia (as broadly characteristic of the anxiety disorders but
well as reverse-keyed items assessing Positive Affect). instead assumes a more limited role as a specific ele-
According to the tripartite model, the two ment in syndromes such as panic disorder.
General Distress scales should be highly intercor- Mineka et al. (1998) also incorporated an
related, whereas the two specific scales should show explicitly quantitative component into this inte-
much better discriminant validity. This prediction grative scheme. They summarized a range of evi-
has been strongly confirmed in multiple studies. dence indicating that the size of these general and
For example, Watson (2000) reported weighted specific components differs markedly across disor-
mean correlations between these Mood and Anxiety ders. Specifically, major depression and GAD both
Symptom Questionnaire scales across eight samples are distress-laden disorders that clearly contain an
(total N = 3,629). As predicted, the overall corre- enormous amount of this general factor variance; in
lation between the General Distress: Depression contrast, most of the other anxiety disorders contain
Symptoms and General Distress:Anxiety Symptoms a more modest component of nonspecific negative
scales was .69 (representing 48% shared variance), affectivity (see also Watson, 2005,2009).
whereas the corresponding coefficient between Subsequent evidence has strongly supported
Anhedonic Depression and Anxious Arousal was basic aspects of the integrative hierarchical model.
only .32 (reflecting 10% shared variance). These Kotov, Gamez, Schmidt, and Watson (2010)
findings (see also Watson, 2005) highlight the reported particularly striking meta-analytic evidence
potential benefits of focusing assessment on the for neuroticism, a personality trait that essentially
unique, specific symptoms within these disorders. reflects individual differences in negative affectivity
(Watson et al., 1999). Kotov et al. compared the
Integrative HierarchicalModel mean neuroticism scores of individuals with and
The basic propositions of the tripartite model without various emotional disorders. Neuroticism
have received extensive support (see Mineka etal., displayed large effect sizes (expressed as Cohens
1988; Watson, 2000, 2005). Unfortunately, how- d; Cohen, 1988) with every analyzed disorder; for
ever, this model suffers from a significant problem example, ds (corrected for unreliability) ranged

48 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
from 1.33 to 2.25 for major depression, GAD, multilevel hierarchical models in which groups of
PTSD, panic disorder, social phobia, andOCD. symptoms are classified at varying levels of specific-
Unfortunately, this model also has run into sig- ity (p.398).
nificant problems. Most notably, the comorbidities Second, as discussed earlier, Mineka etal. (1998)
among the unipolar mood and anxiety disorders demonstrated that the size of the general distress/
cannot be adequately captured using a single non- negative affectivity component differs markedly
specific factor. That is, the integrative hierarchi- across symptoms and disorders. For example, major
cal model predicts (a) a high level of comorbidity depression, dysthymic disorder, and GAD all are
between disorders that have strong components of heavily saturated with this general distress vari-
negative affectivity but (b)weaker overlap between ance, whereas disorders such as agoraphobia and
syndromes that are less saturated with this gen- specific phobia contain a more limited component
eral factor variance. This former proposition has of negative affectivity (see also Watson, Gamez, &
received substantial support in the literature. For Simms,2005).
example, as noted earlier, Watson (2009) reported a Thus from the perspective of the quadripar-
tetrachoric correlation of .64 between major depres- tite model, we need to consider two quantita-
sion and GADtwo disorders that are strongly sat- tive elements when analyzing the properties of
urated with general factor varianceacross the four symptoms: (a) the level of specificity and (b) the
large epidemiological samples. The latter prediction magnitude of the general distress variance. Clearly,
has proven to be more problematic, however. For these two symptom properties are not indepen-
instance, Watson (2009) obtained a correlation dent of one another. Most notably, symptoms
of .61 between social phobia and agoraphobia that are heavily saturated with general distress/
two disorders containing a lesser component of negative affectivitysuch as depressed and anxious
general distressin these same samples (see also moodwill be strongly correlated and, therefore,
Watson,2005). can be expected to show a reduced level of speci-
ficity (Watson, 2005; Watson etal., 2007; Watson
The QuadripartiteModel etal.,2008).
Basic elements of the model. These comorbidity Nevertheless, it is important to consider these
data demonstrated the need for further structural two quantitative elements separately as distinct
refinements. Watson (2009) therefore articulated a properties of symptoms. Although they are most
quadripartite model that focuses on specific symp- accurately viewed as quantitative dimensions, for
tom dimensions within the mood and anxiety the sake of convenience we can use these elements
disorders (e.g., specific symptoms within PTSD). to create an organizing 2 2 classification scheme,
This new scheme represents a synthesis of the ear- which produces four basic types of symptoms:
lier tripartite and integrative hierarchical models. It
(1) high distress symptoms with limited
adopts the explicit symptom focus of the tripartite
model butsimilar to the integrative hierarchical
(2) high distress symptoms with greater specificity
modeluses the DSM mood and anxiety disorders
(3) low distress symptoms with greater specificity
as its organizing framework. This explicit link to the
(4) low distress symptoms with limited specificity.
DSM eliminates the most notable limitation of the
tripartite model by capturing the full range of con- It is the existence of this fourth group that clearly
tent within the DSM-IV anxiety disorders. establishes the quasi-independence of the specificity
In addition, this new scheme incorporates the and distress elements of the model and that most
explicit quantitative focus of the integrative hierar- sharply differentiates this new scheme from the tri-
chical model. This quantitative focus has two inter- partite model (in which nonspecific symptoms were
related elements. First, the tripartite model simply assumed to have a strong component of general
classifies anxiety and depression symptoms as either distress):From the perspective of the quadripartite
specific (e.g., anhedonia and anxious arousal) or model, even symptoms that have a relatively weak
nonspecific (e.g., anxious and depressed mood). distress component can show little or no specificity.
In contrast, Mineka et al. (1998) adopted a more Diagnostic and assessment implications. This final
nuanced quantitative approach in the integrative group of symptoms has important implications for
hierarchical model, arguing that symptom specific- differential diagnosis and assessment, particularly
ity must be viewed in relative rather than absolute with regard to depression. As noted earlier, we can
terms...we need to move toward more complex, enhance differential diagnosis and assessment by

Watson,Stasik 49
focusing greater attention on the specific symptoms the Inventory of Depression and Anxiety Symptoms
contained in the second and third groups. In addi- (IDAS).
tion, even though the distress-laden symptoms of In creating the initial IDAS item pool, Watson
the first group can create significant problems with et al. (2007) included multiple markers to define
regard to differential diagnosis and discriminant all of the symptom dimensions that potentially
validity, they obviously are enormously important could emerge in structural analyses. To ensure that
and, in fact, lie at the very heart of the mood and sufficient markers were included for each poten-
anxiety disorders (Watson, 2005; Watson et al., tial dimension, the candidate items were rationally
2007; Watson et al., 2008). Consequently, the organized into homogeneous item composites (Hogan,
symptoms included in the first three groups all play 1983), or HICs. The original IDAS item pool con-
an essential role in the diagnosis and assessment of tained 13 HICs targeting symptoms that are broadly
the mood and anxiety disorders. relevant to depression, as well as 7 HICs assessing
But how essential are the low distress, limited various anxiety-related symptoms. Nine HICs cor-
specificity symptoms of the fourth group? We responded to the basic symptom criteria for major
address this issue in subsequent analyses, using the depression in the DSM (fourth edition [DSM-IV];
quadripartite model as an organizing framework for American Psychiatric Association, 2000):depressed
understanding the comorbidity between depression mood, loss of interest or pleasure, appetite distur-
and the anxiety disorders. We focus primarily on bance, sleep disturbance, psychomotor problems,
the symptom dimensions within major depression fatigue/anergia, feelings of worthlessness and guilt,
but then consider PTSD symptoms more briefly. cognitive problems, and suicidal ideation. The
In each case, we examine two basic questions. The four remaining HICs tapped hopelessness/pes-
first question concerns the central issue of specific- simism (this targeted the hopelessness subtype of
ity/discriminant validity: To what extent do these depression; Abramson, Metalsky, & Alloy, 1989),
symptoms display specificity to their respective the symptom features of melancholic depression,
disorders, as opposed to being more broadly and angry/irritable mood (which can be an alternative
nonspecifically related to internalizing psychopa- expression of depressed mood in children and ado-
thology (e.g., is insomnia specific to depression)? lescents), and markers of high energy and Positive
The second question involves the key issue of crite- Affect (which are specifically related to depression;
rion/incremental validity:How important are these Clark & Watson, 1991; Mineka etal.,1998).
symptoms to their respective diagnoses (e.g., does The 169 items contained in these 20 HICs
insomnia make a significant incremental contribu- were administered to large samples of college stu-
tion to diagnoses of depression)? dents, psychiatric patients, and community adults
Considered together, the results of these analyses (Watson et al., 2007, Study 2). Data from these
potentially have important diagnostic and assess- three samples were subjected to separate series of
ment implications. We argue that symptoms that principal factor analyses. These analyses yielded
both (a)display poor specificity and (b)provide little several important findings that helped to clarify
incremental diagnostic information are of question- the structure of depression symptoms. First, as
able validity and, therefore, are prime candidates for expected, these analyses revealed a very large, non-
elimination. Put differently, the problem of comor- specific factor that was defined by the core affective/
bidity can be lessened by removingor at least cognitive symptoms of both depression and anxiety.
deemphasizingpoorly functioning symptoms. It is particularly noteworthy that items assessing five
of the nine DSM-IV symptom criteria for a major
Depression Symptoms depressive episodeCriterion 1 (depressed mood),
A Measurement Model for Depression Criterion 2 (anhedonia/loss of interest), Criterion
The IDAS Depression Scales. Depression symp- 5 (psychomotor problems), Criterion 7 (worth-
toms are characterized by a strong general factor lessness/guilt), and Criterion 8 (cognitive prob-
(e.g., Beck, Steer, & Garbin, 1988; Watson et al., lems)were very highly interrelated and defined
2007), a point we return to shortly. Consequently, a single common factor (see also Watson, 2009).
these symptoms typically are summed into a single Items assessing anxious mood and worry also were
overall score. Nevertheless, Watson et al. (2007) strong markers of this general dimension. Overall,
were able to differentiate several meaningful and therefore, we again see evidence of a broad factor
replicable symptom dimensions within depression; that is strongly saturated with general distress/nega-
scales assessing these dimensions were included in tive affectivity and that lies at the very core of the

50 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
mood and anxiety disorders:It subsumes more than saturated with general distress variance (e.g., insom-
half of the DSM-IV symptom criteria for a major nia, appetite loss) to show the greatest specificity
depressive episode, as well as the central features of and to be the purest indicators of depression. From
GAD. To capture the content encompassed within the perspective of the quadripartite model, however,
this broad factor, Watson etal. (2007) created the this is not necessarily the case:Specificity represents
10-item IDAS Dysphoria scale, which contains a separate quantitative element that is distinct from
symptoms assessing depressed mood, loss of inter- the magnitude of the general distress variance.
est, worry, worthlessness, guilt, hopelessness, cogni- We examine the specificity and incremental pre-
tive disturbance, and psychomotor problems. dictive power of these six IDAS depression scales
Beyond this general factor, however, several spe- in subsequent analyses. We also report results
cific symptom dimensions consistently emerged for the IDAS Well-Being scale, which contains
in these structural analyses and were used to cre- high Positive Affect items that show consider-
ate corresponding IDAS scales. Five of these able specificity to depression (Mineka etal., 1998;
scales assess classic manifestations of depression. Watson, 2009; Watson et al., 2011; Watson &
The IDAS Suicidality scale essentially represents Naragon-Gainey,2010).
DSM-IV Criterion 9 for a major depressive episode, It also is important to establish that these find-
the Insomnia scale taps the corresponding portion ings generalize across measures and methods.
of Criterion 4 (sleep disturbance), and Appetite Consequently, in addition to reporting results based
Loss and Appetite Gain jointly define Criterion 3 on the self-report IDAS scales, we present find-
(appetite disturbance). The final scaleLassitude ings based on two parallel interview measures:the
includes content related to both fatigue/anergia Clinician Rating version of the IDAS (IDAS-CR;
(Criterion 6) and the hypersomnia portion of Watson, 2009; Watson, OHara, Chmielewski,
Criterion 4. More generally, analyses of the IDAS etal., 2008) and the Interview for Mood and Anxiety
item pool consistently showed that hypersomnia Symptoms (IMAS; Gamez, Kotov, & Watson,
symptoms actually correlate much more strongly 2010; Watson et al., 2007). Before proceeding to
with fatigue/anergia (rs ranged from .61 to .69 our main analyses, we examine the convergent and
across the three samples) than with insomnia (rs discriminant validity of these measures to establish
ranged from only .14 to .26). Thus, these results that (a) they assess the same basic constructs and
(see also Koffel & Watson, 2009) suggest that (b)depression symptoms can be meaningfully dis-
the diagnosis of depression potentially could be tinguished from one another across methods.
improved by moving hypersomnia from Criterion IDAS-CR analyses. The IDAS-CR consists of a
4 to Criterion6. series of single-item ratings representing each of the
Consequently, six IDAS scalesDysphoria, IDAS scales. Each rating is made on a 3-point scale
Insomnia, Lassitude, Suicidality, Appetite Loss, and (absent, subthreshold, present). To rate each symptom,
Appetite Gainjointly capture all of the symptom the clinicians asks a standard initial probe question,
content included in the DSM-IV diagnostic criteria as well as several standard follow up questions.
for a major depressive episode. Watson (2009) exam- Watson, OHara, Chmielewski, etal. (2008) estab-
ined the relations among these depression scales lished strong interrater reliability for these ratings in
(see also Watson etal., 2007, 2012). Supporting a both students (intraclass correlations ranged from
basic tenet of the quadripartite model, the specific .65 to .95, median=.87) and patients (range=.74
symptom scales correlated quite differently with to .99, median=.89).
Dysphoria and clearly contained different amounts We present correlations between the IDAS scales
of general distress variance (see Watson, 2009, Table and IDAS-CR ratings in 804 outpatients (Table 4.1)
4.6). That is, Dysphoria correlated strongly with and 522 University of Iowa students (Table 4.2) in
Lassitude (r=.63) and Suicidality (r=.62), more the form of a heteromethod block (Campbell &
moderately with Insomnia (r = .53) and Appetite Fiske, 1959); it should be noted that these are aug-
Loss (r=.44), and relatively weakly with Appetite mented versions of the results reported previously in
Gain (r=.27). It is noteworthy, moreover, that cli- Watson, OHara, Chmielewski, etal. (2008), using
nicians ratings showed the same differential pat- additional data from samples described in Watson
tern. Thus depression symptoms clearly vary in the et al. (2012). Looking first at convergent validity,
strength of their general distress components. Based all of the IDAS scales were significantly related
on the logic of the earlier tripartite model, one to their IDAS-CR counterparts in both samples.
would expect those symptoms that are less strongly Convergent correlations in the patient sample

Watson,Stasik 51
Table 4.1Correlations Between the IDAS Scales and IDAS-CR Ratings
(Patient Sample)
IDAS-CR Rating
IDAS Scale 1 2 3 4 5 6 7

1. Appetite Loss .71 .22 .27 .32 .21 .27 .24

2. Suicidality .22 .67 .37 .06 .25 .14 .39

3. Dysphoria .29 .42 .60 .10 .43 .25 .48

4. Appetite Gain .30 .08 .13 .60 .19 .00 .01

5. Lassitude .14 .22 .37 .16 .59 .13 .25

6. Insomnia .25 .19 .27 .02 .28 .59 .24

7. Well-Being .15 .31 .43 .02 .29 .11 .54

Note. N=804. Convergent correlations are highlighted. Correlations |.08| are significant at
p< .05. IDAS=Inventory of Depression and Anxiety Symptoms; IDAS-CR=Clinician Rating
version of the IDAS. These results are based on a combined sample that includes data from both
(a)the patient group described in Watson, OHara, Chmielewski, etal. (2008) and (b)Patient
Sample 2 from Watson etal. (2012).

ranged from .54 (Well-Being) to .71 (Appetite We turn now to discriminant validity. A clas-
Loss), with a mean coefficient of .62; parallel values sic test of discriminant validity is that each of the
in the student data ranged from .39 (Well-Being) to convergent correlations should be higher than any
.62 (Appetite Loss), with a mean coefficient of .53. of the other values in its row or column of the het-
These results are particularly impressive given that eromethod block (Campbell & Fiske, 1959). With
the IDAS-CR consists of a series of single ratings. one exception (Well-Being in the student sample),
Thus, consistent with previous research (e.g., Beck the scales all met this criterion. We further quanti-
etal., 1988; Clark & Watson, 1991; Watson etal., fied these relations by conducting significance tests
2007), these data demonstrate substantial asso- (using the Williams modification of the Hotelling
ciations between self-report and interview-based test for two correlations involving a common vari-
symptom measures. able; see Kenny, 1987), comparing these convergent

Table4.2 Correlations Between the IDAS Scales and IDAS-CR Ratings

(Student Sample)
IDAS-CR Rating
IDAS Scale 1 2 3 4 5 6 7

1. Appetite Loss .62 .25 .13 .11 .14 .07 .12

2. Dysphoria .20 .61 .46 .31 .20 .21 .28

3. Lassitude .14 .44 .55 .22 .22 .17 .24

4. Insomnia .18 .29 .33 .51 .09 .12 .15

5. Appetite Gain .12 .23 .26 .08 .51 .16 .08

6. Suicidality .12 .31 .24 .18 .12 .48 .22

7. Well-Being .12 .43 .32 .22 .10 .18 .39

Note. N=522. Convergent correlations are highlighted. Correlations |.10| are significant at p < .05.
IDAS=Inventory of Depression and Anxiety Symptoms; IDAS-CR=Clinician Rating version of the IDAS.
These results are based on a combined sample that includes data from both (a)the student group described in
Watson, OHara, Chmielewski, etal. (2008) and (b)Student Sample 3 from Watson etal. (2012).

52 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
Table 4.3 Correlations Between the IDAS and IMAS Table 4.3 presents correlations between the
Scales IDAS and IMAS scales in the form of a hetero-
IMAS Scale method block. We again see strong evidence of
convergent validity, with correlations ranging from
IDAS Scale 1 2 3 4 5 .61 (Lassitude) to .74 (Insomnia; mean r = .68).
1. Insomnia .74 .27 .40 .43 .39 With regard to discriminant validity, we again con-
ducted significance tests comparing each of the
2. Suicidality .21 .71 .20 .45 .30 convergent correlations to all of the other values
3. Appetite Loss .31 .24 .70 .32 .35
in its row or column of the heteromethod block.
The results indicated that 39 of the 40 comparisons
4. Dysphoria .37 .45 .35 .64 .57 (97.5%) were significant (p < .05, one-tailed); the
sole exception was that the convergent correlation
5. Lassitude .14 .26 .24 .44 .61
for Lassitude (r=.61) was not significantly higher
Note. N=390. Convergent correlations are highlighted. All than the discriminant correlation between IDAS
correlations are significant at p < .01. IDAS=Inventory of Dysphoria and IMAS Lassitude (r=.57).
Depression and Anxiety Symptoms; IMAS=Interview for Mood
and Anxiety Symptoms. These results are based on Patient Sample 3 Overall, these data offer strong evidence of con-
from Watson etal. (2012). vergent and discriminant validity. All of the con-
vergent correlations were significant and at least
correlations to each of the 12 discriminant coeffi- moderate in magnitude, with mean coefficients of
cients in the same row or column of the block; this .53 (IDAS-CR student analyses), .62 (IDAS-CR
yielded a total of 84 tests of discriminant validity patient analyses), and .68 (IMAS analyses). The
across these seven symptom dimensions in each scales also generally showed excellent discriminant
sample. Overall, 166 of these 168 comparisons validity; indeed, 205 of the 208 individual compari-
(98.8%) were significant (p < .05, one-tailed), sons (98.6%) were significant (p < .05, one-tailed)
which offers strong evidence of discriminant valid- across these analyses. Thus our data demonstrate
ity. The two exceptions were that the convergent that specific depression symptomssuch as lassi-
coefficient for Well-Being (r=.39) did not signifi- tude, insomnia, suicidality, and appetite losscan
cantly exceed the discriminant correlations between be distinguished from one another across methods.
(a)IDAS Well-Being and IDAS-CR Dysphoria and
(b) IDAS Well-Being and IDAS-CR Lassitude in Specificity to Depression
the studentdata. Relations with the Beck Inventories. How specific are
IMAS analyses. We report IMAS data on 390 out- these symptoms to depression? This issue can be ana-
patients (for a description of this sample, see Watson lyzed in a number of different ways and from several
et al., 2012). The IMAS assesses current (i.e., past different perspectives. For example, Watson (2009)
month) symptoms; individual items are scored on a examined how the IDAS scales correlated with the
3-point rating scale (absent, subthreshold, above thresh- Beck Depression Inventory-II (BDI-II; Beck, Steer,
old). Items were derived from the mood and anxiety & Brown, 1996) and the Beck Anxiety Inventory
disorder modules of the Composite International (BAI; Beck & Steer, 1990) in two large samples
Diagnostic Interview (Kessler & stn, 2004) and (combined N=2,783). Well-Being showed the most
were designed to cover all DSM-IV mood and anxi- impressive specificity in these data, correlating much
ety disorder symptom criteria. Based on data from more strongly with the BDI-II (mean r=.56) than
prior studies (Gamez et al., 2010; Watson et al., with the BAI (mean r = .32). Lassitude (mean
2007), the IMAS was revised to improve its symp- rs = .62 and .50, respectively), Suicidality (mean
tom coverage. The revised version includes five-item rs = .58 and .47, respectively), and Dysphoria
measures of Dysphoria and Lassitude, four-item (meanrs=.81 and .71, respectively) also displayed a
Insomnia and Suicidality scales, and a three-item reasonable level of specificity in thesedata.
measure of AppetiteLoss. In contrast, the three remaining scales
Extensively trained lay interviewers administered Insomnia, Appetite Loss, and Appetite Gain
the IMAS. All interviews were recorded; a randomly showed much poorer specificity (see Watson, 2009,
selected interviewer rescored 34 tapes. Interrater Table 4.7). Indeed, Insomnia (mean rs = .50 and
reliability consistently was excellent, with intraclass .47, respectively) and Appetite Loss (mean rs=.39
correlation coefficients ranging from .97 to .99 and .39, respectively) had virtually identical corre-
across the various scales (see Watson etal., 2012). lations with the two instruments. These results are

Watson,Stasik 53
particularly striking when one considers that these mood, pessimism, past failure, anhedonia, guilt,
IDAS scales actually share overlapping item content self-dislike, crying, agitation, indecision, worth-
with the BDI-II but not theBAI. lessness, energy, hypersomnia, and loss of appetite.
Diagnostic specificity. In the discussion that fol- Thus, on the basis of this limited evidence, it appears
lows, we concentrate on the key issue of diagnostic that some symptoms are better, purer indicators of
specificity, that is, whether these symptoms are more depression than are others.
strongly related to major depression than to anxi- To examine this issue more fully, we report six
ety disorders (for earlier examinations of this issue, analyses of diagnostic specificity that enable us to
see Watson, 2009; Watson, OHara, Chmielewski, evaluate the robustness of these patterns across
et al., 2008). Although the available evidence on methods, measures, and populations. To obtain
this topic is quite limited, there is reason to ques- current DSM-IV diagnoses, all participants were
tion the diagnostic specificity of some depression interviewed by masters level psychologists using
symptoms. For instance, Benazzi (2006) examined the SCID-IV (for interrater reliability data, see
the prevalence of depression symptoms in samples Watson, OHara, Chmielewski, et al., 2008). To
of patients diagnosed with bipolar-II (BP-II) dis- examine diagnostic specificity, we computed poly-
order and major depression; his results indicated choric correlations between depression symptom
that several symptoms (as assessed by the Structured measures and various DSM-IV mood and anxiety
Clinical Interview for DSM-IV [SCID-IV]; First, disorder diagnoses. Polychoric correlations esti-
Spitzer, Gibbon, & Williams, 1997) showed poor mate the associations between normally distributed
diagnostic specificity. More specifically, symptoms latent continuous variables that are presumed to
of weight gain, increased eating, hypersomnia, psy- underlie observed scores (Flora & Curran, 2004;
chomotor agitation, worthlessness, and poor con- Schmukle & Egloff, 2009; Watson & Tellegen,
centration were actually significantly more common 1999). They retain the relative rank order informa-
in BP-II patients than in those with major depres- tion provided by Pearson correlations (i.e., the same
sion. Such an overlap in symptom presentation scales will be relatively strongor weakpredic-
poses problems for discriminating between diagno- tors of particular diagnoses) but are unaffected by
ses and could lead to the underdiagnosis or misdiag- differences in prevalence rates, thereby facilitating
nosis of BP-II (Benazzi,2006). cross-diagnosis comparisons. Diagnoses were scored
It is noteworthy, moreover, that several of the as zero = absent, 1 = present, so that positive cor-
symptom criteria for major depression are also part relations indicate that higher scores on a scale are
of the criterion sets for other mental disorders. associated with an increased likelihood of receiving
For example, insomnia is one of the Criterion D the diagnosis.
symptoms of PTSD, and difficulty concentrat- In the first three analyses, we present poly-
ing is included in the diagnostic criteria for both choric correlations between the IDAS (N = 894;
PTSD and GAD. In a related vein, Carney, Ulmer, see Table4.4), IDAS-CR (N=604; see Table 4.5),
Edinger, Krystal, and Knauss (2009) noted that and IMAS (N = 293; see Table 4.6) depression
self-report measures of depression include symp- scales and eight DSM-IV mood and anxiety dis-
toms that are also essential features of primary order diagnoses (major depression, GAD, PTSD,
insomnia (e.g., insomnia, mood disturbance, irri- panic disorder, agoraphobia, social phobia, spe-
tability, fatigue, impaired concentration, and moti- cific phobia, OCD) in outpatient samples. The
vation/anergia). Accordingly, Carney et al. (2009) fourth analysis examines how the IDAS scales cor-
assessed the accuracy of the BDI-II in detecting relate with diagnoses of (a)major depression and
major depression in patients known to meet diag- (b)GAD in a sample of 337 postpartum women
nostic criteria for insomnia. They found that sev- (see Table 4.7). The final two analyses (see Table
eral BDI-II items failed to differentiate (a) those 4.8) consider how the IDAS scales (N=307) and
with insomnia from (b) those with insomnia and IDAS-CR ratings (N = 303) correlate with diag-
comorbid depression. More specifically, there were noses of (a)major depression and (b)any anxiety
no significant differences between groups on insom- disorder in a college student sample (for more
nia, irritability, suicidal ideation, increased appetite, details regarding these samples, see Watson etal.,
decreased concentration, fatigue, decreased libido, 2007; Watson, OHara, Chmielewski, etal., 2008;
and self-critical and punishment-related thoughts. Watson etal., 2012).
The items that did successfully demonstrate group What do these analyses tell us about the specific-
differences included those related to depressed ity of depression symptoms? In order to make sense

54 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
Table 4.4Polychoric Correlations Between the IDAS Scales and DSM-IV Diagnoses
(Patient Sample)
IDAS Scale MDD GAD PTSD Panic Agora Social Specific OCD
Phobia Phobia

Dysphoria .73 .46 .40 .43 .32 .27 .19 .24

Lassitude .50 .27 .30 .26 .21 .15 .11 .12

Suicidality .51 .24 .28 .24 .19 .16 .09 .18

Insomnia .40 .29 .39 .32 .23 .19 .21 .24

Appetite Loss .38 .18 .39 .33 .19 .16 .12 .13

Appetite Gain .11 .04 .05 .07 .01 .02 .09 .06

Well-Being .49 .21 .17 .22 .24 .15 .04 .17

Note. N=894. Correlations |.40| are highlighted. IDAS=Inventory of Depression and Anxiety Symptoms;
DSM-IV=Diagnostic and Statistical Manual of Mental Disorders (fourth edition); MDD=major depression;
GAD=generalized anxiety disorder; PTSD=posttraumatic stress disorder; Agora=agoraphobia; OCD=obsessive-compulsive
disorder. These results are based on a combined sample that includes data from both (a)the patient group described in Watson,
OHara, Chmielewski, etal. (2008) and (b)Patient Sample 3 from Watson etal. (2012).

of these complex results, we provide brief summa- of these analyses. Overall, the dysphoria measures
ries of the results for each symptom dimension. had a mean correlation of .69 with major depres-
Dysphoria. Replicating and extending the results sion; in marked contrast, their average correlations
of Watson (2009), the dysphoria measures showed with specific anxiety disorder diagnoses were only
both (a)the strongest overall association with major .44 (GAD), .40 (panic disorder), .39 (PTSD), .29
depression and (b)impressive diagnostic specificity (agoraphobia), .26 (social phobia), .24 (OCD), and
in these analyses. Specifically, they had the strongest .18 (specific phobia). Thus, consistent with the find-
correlation with major depression in five of six anal- ings of Watson (2009), we again see evidence that
yses (rs ranged from .54 to .83) and in every case symptoms containing a very strong general distress
correlated more strongly with depression than with component still can display considerable specificity.
any other disorder. To quantify these effects further, Lassitude. Indicators of lassitude also were
we computed weighted mean correlations across all strongly related to depression (rs ranged from .50 to

Table 4.5Polychoric Correlations Between IDAS-CR Ratings and DSM-IV Diagnoses

(Patient Sample)
IDAS-CR Rating MDD GAD PTSD Panic Agora Social Specific OCD
Phobia Phobia

Dysphoria .66 .44 .35 .48 .38 .30 .19 .32

Lassitude .56 .27 .30 .41 .37 .30 .16 .42

Suicidality .50 .24 .21 .28 .21 .20 .09 .26

Insomnia .29 .24 .30 .33 .24 .18 .02 .18

Appetite Loss .33 .23 .32 .32 .05 .19 .06 .10

Appetite Gain .23 .01 .02 .12 .01 .06 .08 .07

Well-Being .54 .26 .25 .39 .38 .15 .06 .24

Note. N=604. Correlations |.40| are highlighted. IDAS-CR=Clinician Rating version of the IDAS;
DSM-IV=Diagnostic and Statistical Manual of Mental Disorders (fourth edition); MDD=major depression;
GAD=generalized anxiety disorder; PTSD=posttraumatic stress disorder; Agora=agoraphobia;
OCD=obsessive-compulsive disorder. These results are based on the patient sample described in Watson, OHara,
Chmielewski, etal. (2008).

Watson,Stasik 55
Table 4.6Polychoric Correlations Between the IMAS Scales and DSM-IV Diagnoses
(Patient Sample)
IMAS Scale MDD GAD PTSD Panic Agora Social Specific OCD
Phobia Phobia

Dysphoria .54 .25 .42 .16 .04 .17 .14 .08

Lassitude .61 .28 .30 .33 .23 .16 .16 .01

Suicidality .34 .14 .22 .13 .07 .09 .20 .19

Insomnia .31 .09 .26 .21 .20 .18 .28 .15

Appetite Loss .43 .09 .48 .27 .28 .08 .20 .25

Note. N=293. Correlations |.40| are highlighted. IMAS=Interview for Mood and Anxiety Symptoms;
DSM-IV=Diagnostic and Statistical Manual of Mental Disorders (fourth edition); MDD=major depression;
GAD=generalized anxiety disorder; PTSD=posttraumatic stress disorder; Agora=agoraphobia;
OCD=obsessive-compulsive disorder. These results are based on Patient Sample 3 from Watson etal. (2012).

.61) and displayed impressive diagnostic specificity (rs ranged from.42 to.56) and also showed con-
in these analyses. Lassitude was the strongest predic- siderable specificity in these analyses, particularly in
tor of depression in the IMAS data (see Table4.6) the patient data. Overall, they had a weighted mean
and correlated more strongly with depression than correlation of.50 with major depression; in con-
with any other disorder in all six analyses. Overall, trast, their average associations with specific anxiety
lassitude symptoms had a weighted mean correla- disorder diagnoses ranged from only.05 (specific
tion of .55 with major depression, whereas their phobia) to.30 (agoraphobia).
average correlations with specific anxiety diagnoses Suicidality. Suicidality symptoms displayed
ranged from only .14 (specific phobia) to .32 (panic impressive specificity in these data; in all six anal-
disorder). These findings are consistent with the yses, they correlated more strongly with major
results of Watson (2009) and indicate that lassitude depression (rs ranged from .34 to .59) than with
is a strong and specific symptom of depression. any other disorder. Overall, these scales had a mean
Well-Being. Consistent with the tripartite and
integrative hierarchical models, the well-being mea-
Table4.8 Polychoric Correlations Between the IDAS
sures were substantially related to major depression Scales, IDAS-CR Ratings, and DSM-IV Diagnoses
(Student Sample)
Table4.7 Polychoric Correlations Between the IDAS IDAS IDAS-CR
Scales and DSM-IV Diagnoses (Postpartum Sample)
Symptom MDD Any AD MDD Any AD
Dysphoria .83 .61 .69 .51
Dysphoria .66 .56
Lassitude .56 .35 .58 .44
Lassitude .57 .46
Suicidality .59 .49 .50 .33
Suicidality .48 .29
Insomnia .39 .24 .30 .25
Insomnia .42 .51
Appetite Loss .33 .14 .19 .27
Appetite Loss .37 .12
Appetite Gain .19 .10 .21 .23
Appetite Gain .15 .28
Well-Being .56 .36 .42 .43
Well-Being .44 .43
Note. N=307 (IDAS analyses), 303 (IDAS-CR analyses).
Note. N=337. Correlations |.40| are highlighted. Correlations |.40| are highlighted. IDAS=Inventory of
IDAS=Inventory of Depression and Anxiety Symptoms; Depression and Anxiety Symptoms; IDAS-CR=Clinician Rating
DSM-IV=Diagnostic and Statistical Manual of Mental Disorders version of the IDAS; DSM-IV=Diagnostic and Statistical Manual
(fourth edition); MDD=major depression; GAD=generalized of Mental Disorders (fourth edition); MDD=major depression;
anxiety disorder. These results are based on an expanded version of AD=anxiety disorder. These results are based on the student sample
the postpartum sample described in Watson etal. (2007). described in Watson, OHara, Chmielewski, etal. (2008).

56 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
correlation of .49 with major depression, whereas .35, and .17, respectively). This, in turn, raises the
their average correlations with specific anxiety dis- question of whether these weak and nonspecific
order diagnoses ranged from only .11 (specific pho- symptoms actually are necessary, that is, whether
bia) to .25 (PTSD). These results are consistent with they make significant, incremental contributions to
the findings of Watson (2009) and indicate that sui- the diagnosis of major depression. This is the focus
cidality is a relatively strong and specific symptom of our next series of analyses.
of depression.
Insomnia. In contrast, insomnia measures were Incremental PredictivePower
more modestly related to major depression (rs Given that the more specific symptom dimen-
ranged from .29 to .42) and showed weak, inconsis- sions had the strongest bivariate associations with
tent evidence of diagnostic specificity. Overall, the major depression, one would expect that they also
insomnia scales had very similar mean correlations would emerge as particularly strong contributors
with major depression (.36), PTSD (.34), panic dis- in multivariate analyses. To test this expectation,
order (.31), and GAD (.28). Thus, consistent with we conducted a series of logistic regression analy-
the results of Watson (2009), these findings indicate ses to identify the unique, incremental ability of the
that insomnia is a relatively weak and nonspecific individual depression symptoms to predict DSM-IV
indicator of depression. diagnoses of major depression. To facilitate inter-
Appetite loss. Findings for the appetite loss scales pretation, the symptom scales were standardized to
were very similar to those for insomnia:These mea- put them on the same metric.
sures had relatively weak correlations with major Table 4.9 presents the odds ratios (ORs) from
depression (rs ranged from .19 to .43) and showed these analyses. Not surprisingly, dysphoria symp-
little specificity. Overall, symptoms of appetite loss tomswhich showed the strongest bivariate asso-
had a mean correlation of only .35 with major ciation with depressionemerged as a significant
depression; they correlated very similarly with panic predictor in all six analyses, with ORs ranging from
disorder (average r=.32) and actually had a slightly 1.83 to 8.59. Only two other types of symptoms
stronger association with PTSD (mean r = .38). made a significant incremental contribution in at
Therefore, similar to insomnia, appetite loss also least two different analyses: Lassitude symptoms
is a relatively weak and nonspecific symptom of contributed significantly in all three analyses based
depression. on interview measures (ORs ranged from 1.73 to
Appetite gain. Consistent with previous results 2.60), whereas indicators of well-being yielded sig-
(see Watson, 2009; Watson et al., 2007; Watson, nificant effects in both of the patient analyses in
OHara, Chmielewski, et al., 2008), the appetite which they were assessed (ORs = 0.61 and 0.54
gain scales had the weakest overall associations with in the IDAS and IDAS-CR data, respectively).
major depression (rs ranged from .11 to .23, mean Overall, these three symptoms dimensions showed
r=.17). Their relations with the anxiety disorders incremental predictive power in 11 of 17 analyses
were inconsistent across analyses but tended to be (64.7%). Thus three of the symptoms that showed
quite low. Overall, their weighted mean correla- substantial specificitydysphoria, lassitude, and
tions with specific anxiety diagnoses ranged from well-beingalso displayed the strongest predictive
only.09 (panic disorder) to .09 (specific phobia). power in thesedata.
Thus symptoms of appetite gain do not show strong In contrast, the three symptoms that exhibited
or consistent associations with any of these mood little diagnostic specificityinsomnia, appetite
and anxiety diagnoses. loss, and appetite gainmade significant incremen-
Summary. Taken together, these analyses yield a tal contributions in only 2 of 17 analyses (11.8%).
very clear pattern. The four symptom dimensions Indeed, insomnia was the only symptom dimension
that have the strongest associations with major that failed to contribute significantly in any analy-
depressiondysphoria (mean r = .69), lassitude sis. Overall, therefore, our results indicate that the
(mean r = .55), well-being (mean r = .50), and nonspecific symptoms of depression actually con-
suicidality (mean r=.49)also show the clearest, tribute little to its diagnosis.
most compelling evidence of specificity. Put dif-
ferently, the symptoms that display less impressive Implications and Conclusions
specificityinsomnia, appetite loss, and appetite In thinking about the implications of these find-
gaindo so, in part, because of their substantially ings, it may be helpful to put them in classic assessment
weaker associations with depression (mean rs=.36, terms. Using this framework as a guide, our data yield

Watson,Stasik 57
Table4.9 Odds Ratios from Logistic Regression Analyses Predicting DSM-IV Diagnoses of
Major Depression
Symptom Patient Postpartum Student Patient Student IMAS

Dysphoria 4.12** 2.08** 8.59** 1.98** 4.66** 1.83**

Lassitude 1.15 1.51 1.09 1.73** 2.60* 2.56**

Suicidality 1.02 1.27 1.19 1.42** 1.65 0.99

Insomnia 1.15 0.92 1.09 1.12 1.12 1.26

Appetite Loss 1.07 1.28 0.81 1.43** 1.10 1.20

Appetite Gain 0.88 0.99 0.70 1.61** 1.14 .

Well-Being 0.61** 0.77 0.76 0.54** 0.58 .

Note. N=894 (IDAS:Patient), 337 (IDAS:Postpartum), 307 (IDAS:Student), 605 (IDAS-CR:Patient), 303
(IDAS-CR:Student), 293 (IMAS). DSM-IV=Diagnostic and Statistical Manual of Mental Disorders (fourth edition);
IDAS=Inventory of Depression and Anxiety Symptoms; IDAS-CR=Clinician Rating version of the IDAS;
IMAS=Interview for Mood and Anxiety Symptoms.
* p< .05.
** p< .01

three basic conclusions. First, symptoms of insomnia the trauma; five symptoms), Criterion C (emotional
and appetite disturbance show relatively poor discrimi- numbing and avoidance of trauma-related stimuli;
nant validity (i.e., diagnostic specificity) in compari- seven symptoms), and Criterion D (manifestations
son to indicators of dysphoria, lassitude, suicidality, of increased arousal; five symptoms). Although a
and well-being. Second, these same symptoms dem- few studies have supported the three-factor DSM
onstrate unimpressive convergent/criterion validity; scheme, most structural analyses indicate that
that is, they have relatively weak bivariate associations other models fit the data significantly better (for a
with DSM-IV diagnoses of major depression. Third, meta-analytic review, see Yufik & Simms, 2010).
these symptom dimensions display little or no incre- In recent years, the evidence increasingly has con-
mental validity, contributing significantly in only 2 of verged on two different four-factor structures. The
17 logistic regression analyses. These final analyses are first model, which initially was identified by King,
particularly noteworthy, in that they suggest that these Leskin, King, and Weathers (1998), follows the
weak, nonspecific symptoms actually may be superflu- DSM-IV organization, except that the Criterion C
ous in reaching a diagnosis of major depression. This symptoms are split into separate Avoidance (two
being the case, it seems reasonable to suggest that the symptoms) and Numbing (five symptoms) factors.
diagnosis of depression might be improved by focus- The second modelwhich was proposed and
ing primarily on strong and specific indicators (such as supported by Simms, Watson, and Doebbeling
dysphoria, lassitude, and well-being) and deemphasiz- (2002)contains Intrusions and Avoidance fac-
ing relatively weak and nonspecific symptoms, such as tors that are identical to those specified in the King
insomnia and appetite disturbance. etal. (1998) structure. However, its arrangement of
The findings we have presented here are pre- the 10 remaining symptoms differs markedly from
liminary, and they clearly need to be replicated and both the DSM-IV and the King etal. (1998) mod-
extended in future research. Still, they suggest that els. Similar to them, it also includes a Hyperarousal
it will be useful to scrutinize the properties of these factor, but this now is defined by only two of the
low distress, nonspecific, and poorly functioning DSM-IV Criterion D symptoms (hypervigilance,
symptoms much more closely in the future. exaggerated startle response). The fourth factor in
the Simms et al. (2002) structureDysphoria
PTSD Symptoms consists of the five Criterion C numbing symp-
A Measurement Model forPTSD toms (loss of interest, detachment, restricted affect,
DSM-IV groups PTSD symptoms into three clus- sense of foreshortened future, inability to recall
ters:Criterion B (intrusions and reexperiencing of the trauma) and the three remaining Criterion D

58 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
symptoms (sleep disturbance, irritability, difficulty the Civilian Version of the PTSD Checklist (PCL;
concentrating). Weathers, Litz, Herman, Huska, & Keane, 1993),
In a recent meta-analytic review of the specificity which has 17 items corresponding to the DSM-IV
and correlates of these PTSD symptom dimensions, symptom criteria for PTSD. As in previous analyses,
Gootzeit and Markon (2011) concluded:Overall, current DSM-IV diagnoses were obtained using the
the Simms etal. (2002) model seems better able to SCID-IV; here, we report results for major depres-
separate specific and general variance; whereas dys- sion, PTSD, and panic disorder. The interviewers
phoria clearly represents the general factor in this were graduate students and advanced undergradu-
model, the King etal. (1998) model separates this ate research assistants who all underwent exten-
factor into two components:numbing and hyper- sive training prior to data collection. Interrater
arousal (p.1000). We therefore adopt the Simms reliability for these diagnoses was examined using
et al. (2002) model as our basic measurement 51 audiotaped interviews. The resulting kappas
framework in the following discussion. indicated excellent interrater reliability:.92 (major
depression), 1.00 (PTSD), and .83 (panic disorder).
Specificity toPTSD Table 4.10 presents polychoric correlations
Relations to depression and anxiety. How specific between these DSM-IV diagnoses and the PCL; the
are these symptoms to PTSD? Although the findings table displays results both for individual items and
of individual studies have not been entirely consis- for scales based on the Simms etal. (2002) model.
tent (see, e.g., Marshall, Schell, & Miles, 2010), the It is noteworthy that the intrusions and avoidance
preponderance of the evidence clearly establishes symptoms consistently show evidence of good diag-
that indicators of dysphoria display much poorer nostic specificity, both at the item and the scale level.
specificity than other types of PTSD symptoms The five intrusions symptoms had correlations with
(see Gootzeit & Markon, 2011; Watson, 2009). PTSD ranging from .60 to .65, with a mean value
For example, in their meta-analytic review, Gootzeit of .63; in contrast, their average correlations with
and Markon (2011) examined how the Simms etal. major depression and panic disorder were only .27
(2002) dimensions correlated with depression and and .45, respectively; similarly, the two avoidance
anxiety symptoms (see their Table 4.5). Depression symptoms had mean correlations of .60 (PTSD),
symptoms correlated much more strongly with dys- .36 (major depression), and .42 (panic disorder).
phoria (r = .74) than with hyperarousal (r = .50), The hyperarousal and dysphoria symptoms
intrusions (r = .50), and avoidance (r = .44). The generally show much poorer specificity. The
same basic pattern was observed for anxiety symp- hyperarousal symptoms had virtually identical cor-
toms, which also correlated more strongly with dys- relations with PTSD and panic disorder, both at the
phoria (r = .58) than with hyperarousal (r = .49), item and the scale level. These data are consistent
intrusions (r=.40), and avoidance (r=.33). with previous evidence establishing that indica-
As was noted by Marshall etal. (2010), simple tors of anxious arousal are substantially related to
scale-based analyses potentially suffer from one both panic disorder and PTSD (Brown etal., 1998;
important confound, namely, that there are many Watson,2009).
more indicators of dysphoria (eight) than of intru- The findings for the dysphoria symptoms are
sions (five), avoidance (two), or hyperarousal (two); more complex. Five of the items (loss of interest
consequently, dysphoria measures may show stron- in activities, feeling distant or cut off, feeling emo-
ger correlations with other variables simply because tionally numb, irritability, difficulty concentrat-
of their greater reliability. Gootzeit and Markon ing) actually show specificity to depression rather
(2011) tested this possibility by creating a compos- than to PTSD:That is, they correlate much more
ite, nine-item index of Nondysphoria symptoms. strongly with diagnoses of major depression (mean
Nevertheless, the same basic pattern emerged in r=.58) than with either PTSD (mean r=.36) or
these analyses (see their Table 4.8):Depression and panic disorder (r=.28). In contrast, the item mea-
anxiety symptoms both correlated more strongly suring Criterion C3 (memory problems) showed
with Dysphoria than with Nondysphoria. some specificity to panic disorder. Finally, the two
Diagnostic specificity. What about the diagnos- remaining symptoms (feeling the future will be cut
tic specificity of these symptoms? We examined short, trouble falling or staying asleep) displayed
this issue in a sample of 247 outpatients (for a relatively weak and nonspecific associations with all
description of this sample, see Watson etal., 2012). three disorders (rs ranged from .30 to .39). In this
Self-reported PTSD symptoms were assessed using regard, it is noteworthy that we again see evidence

Watson,Stasik 59
Table4.10 Polychoric Correlations Between the PCL logistic regression analysis in which the four Simms
Scales and Items and DSM-IV Diagnoses etal. (2002) factor scales were used to predict PTSD
Paraphrased PCL Item MDD PTSD Panic diagnoses; again, scale scores were standardized
to put them on the same metric. Only Intrusions
Intrusions .29 .71 .48 (OR=2.82, p < .01) showed significant incremen-
Repeated, disturbing memories .29 .65 .42
tal power in these analyses; Avoidance (OR=1.19),
Dysphoria (OR = 1.19), and Hyperarousal
Repeated, disturbing dreams .25 .62 .44 (OR=1.25) all failed to contribute significantly to
the prediction of PTSD diagnoses.
Reliving a stressful experience .25 .60 .41

Feeling upset when reminded .29 .65 .45 Implications and Conclusions
of event PTSD symptoms show the same general pattern
observed earlier in our examination of depression.
Had physical reaction .26 .61 .55
Most notably, it now is clear that these symptoms
when reminded
vary widely in their specificity. In particular, exten-
Avoidance .37 .63 .45 sive evidence establishes that indicators of dysphoria
display much poorer specificity than other types of
Avoided thinking/talking about .35 .62 .39
PTSD symptoms: They correlated more strongly
with depression and anxiety symptoms (Gootzeit &
Avoided activities/situations .37 .57 .45 Markon, 2011; Watson, 2009) and failed to show
any diagnostic specificity to PTSD in our outpa-
Dysphoria .68 .50 .39
tient sample (and, in fact, actually had a stronger
Trouble remembering parts .10 .29 .39 association with major depression). Hyperarousal
of event symptoms also failed to demonstrate diagnostic
specificity in our data, correlating equally with
Loss of interest in activities .69 .39 .34
panic disorder. In contrast, indicators of intrusions
Feeling distant or cut off .66 .46 .29 and avoidance exhibited stronger and more consis-
tent evidence of specificity. Finally, our outpatient
Feeling emotionally numb .57 .40 .28 analyses indicated that these nonspecific dysphoria
Feeling future will be cut short .39 .37 .30 and hyperarousal symptoms had somewhat weaker
bivariate correlations with DSM-IV diagnoses of
Trouble falling or staying asleep .36 .34 .33 PTSD and failed to provide significant incremen-
Feeling irritable/angry outbursts .42 .25 .17 tal information in a logistic regression analysis.
Although these incremental validity analyses were
Difficulty concentrating .56 .31 .30 based on a single sample, they tentatively suggest
Hyperarousal .34 .51 .49 that these nonspecific symptoms may not provide
information that is essential in reaching a diagnosis
Being superalert, watchful .24 .48 .47 ofPTSD.
In light of these findings, it seems reason-
Feeling jumpy, easily startled .39 .46 .47
able to suggest that the diagnosis of PTSD might
Note. N=247. Correlations |.45| are highlighted. PCL=PTSD be improved by focusing primarily on strong and
Checklist; DSM-IV=Diagnostic and Statistical Manual of specific indicators (i.e., intrusions and avoidance)
Mental Disorders (fourth edition); MDD=major depression;
PTSD=posttraumatic stress disorder. These results are based on and deemphasizing relatively weak and nonspecific
Patient Sample 4 from Watson etal. (2012). symptoms (i.e., dysphoria and hyperarousal). The
situation here is complicated, however, by two addi-
that indicators of insomnia (as assessed here by an tional considerations. First, in their meta-analytic
item tapping PTSD Criterion D1) show relatively examination, Gootzeit and Markon (2011) reported
weak and nonspecific associations with DSM-IV that trauma history correlated more strongly with
mood and anxiety disorders. dysphoria symptoms (r=.35) than with intrusions
(r = .28), hyperarousal (r = .20), and avoidance
Incremental PredictivePower (r=.19). Gootzeit and Markon (2011) concluded
We examined the incremental predictive power that these findings suggest that dysphoria symp-
of individual PTSD symptoms by conducting a toms are an important part of the overall construct

60 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
of PTSD, and likely provide important clinical Table 4.11 presents polychoric correlations
information (p.1000). between these four PCL variants and diagnoses of
Second, in an analysis of the National Comorbidity major depression, PTSD, and panic disorder in
Survey Replication data, Elhai, Grubaugh, Kashdan, our sample of 247 outpatients. Consistent with
and Frueh (2008) examined an alternative set of the results of Elhai et al. (2008) and Grubaugh
diagnostic criteria (developed by Spitzer, First, and etal., (2010), our data indicate that dropping the
Wakefield, 2007) that eliminates the five PTSD five overlapping dysphoria symptoms has surpris-
symptoms that directly overlap with other internal- ingly little effect: Correlations with PTSD (.69
izing disorders (C3: amnesia, C4: loss of interest, vs .70) and panic disorder (.50 vs..51) essentially
D1:sleep disturbance, D2:irritability, D3:difficulty remain unchanged, whereas the association with
concentrating). These modifications eliminated five major depression declines only modestly (from .56
of the eight indicators of the Simms et al. (2002) to .47). In contrast, eliminating additional symp-
Dysphoria factor. Consequently, one would expect toms appears to offer more substantial benefits.
that these revised diagnostic criteria would contain Eliminating the entire dysphoria cluster (PCL-9)
a somewhat weaker general distress component substantially reduces the correlation with major
and show greater diagnostic specificity to PTSD. depression (r=.34) but does not affect the overlap
Nevertheless, Elhai etal. (2008) found that the rates with panic disorder (r=.52). Overall, PCL-7 (i.e.,
of comorbidity between PTSD and other disorders focusing exclusively on the intrusions and avoidance
essentially remained unchanged regardless of the symptoms) displays the best convergent/discrimi-
diagnostic criteria that were used. Grubaugh, Long, nant pattern in these data:It has the weakest asso-
Elhai, Frueh, and Magruder (2010) subsequently ciations with both major depression (r = .30) and
replicated these results in a sample of veterans. These panic disorder (r = .49) while maintaining a very
findings have been interpreted as demonstrating that strong correlation with PTSD diagnoses (r=.71).
deemphasizingor even removingnonspecific In light of these data, we continue to believe that
PTSD symptoms actually would have little effect on the diagnosis and assessment of PTSD potentially
key diagnostic problems such as comorbidity (Elhai can be improved by focusing primarily on more
etal., 2008; Grubaugh etal., 2010; see also Gootzeit specific types of symptoms. This issue merits careful
& Markon,2011). attention in future research.
Alternatively, however, these results simply may
show that it is insufficient to eliminate only a sub- Future Research Directions
set of the dysphoria symptoms. Indeed, Table 4.10 The results we have presented here need to be
indicates that two of the retained dysphoria symp- replicated and extended in several ways. First, our
toms (C5:feeling distant or cut off, C6:feeling emo- analyses were limited to unipolar mood and anxiety
tionally numb) are strongly related to depression disorder diagnoses. However, the problem of comor-
(rs=.66 and .59, respectively). Thus it remains pos- bidity is not limited to the mood and anxiety dis-
sible that eliminating all eight markers of dysphoria orders but rather pervades the entire DSM (Mineka
will enhance the diagnosis of PTSD in a significant etal., 1998; Widiger & Clark, 2000). It therefore
way. We tested this idea by creating four different
versions of the PCL. The first version (PCL-17)
includes all 17 items, thereby modeling the DSM-IV Table4.11 Polychoric Correlations Between Alternative
diagnosis of PTSD. The second version (PCL-12) Versions of the PCL and DSM-IV Diagnoses
drops the five symptoms that directly overlap with PCL Version MDD PTSD Panic
other internalizing disorders; it therefore captures
the reduced model proposed by Spitzer etal. (2007) PCL-17 .56 .69 .50
and investigated subsequently by Elhai etal. (2008) PCL-12 .47 .70 .51
and Grubaugh et al. (2010). The third version
(PCL-9) removes all eight dysphoria items, thereby PCL-9 .34 .71 .52
modeling the effects of eliminating this entire factor
PCL-7 .30 .71 .49
from the disorder. The final version (PCL-7) further
eliminates the two hyperarousal items; it therefore Note. N=247. Correlations |.45| are highlighted. See text for
estimates the effects of restricting diagnosis and details. PCL=PTSD Checklist; DSM-IV=Diagnostic and Statistical
Manual of Mental Disorders (fourth edition); MDD=major
assessment to the two most specific symptom clus- depression; PTSD=posttraumatic stress disorder. These results are
ters (intrusions and avoidance). based on Patient Sample 4 from Watson etal. (2012).

Watson,Stasik 61
will be important to examine a broader array of Finally, another key limitation of the current anal-
symptoms and a wider range of disorders in future yses is that our data are completely cross-sectional in
research. For example, as discussed earlier, Carney nature. Long-term longitudinal studies are needed
et al. (2009) noted that many depression symp- to examine how the associations between symptoms
toms are also essential features of primary insomnia and diagnoses change over time. To take one exam-
(e.g., insomnia, mood disturbance, fatigue/anergia, ple, dissociative symptoms appear to play a particu-
impaired concentration). Consequently, it is pos- larly important role in acute reactions to trauma and
sible that some depression symptoms that show may be a risk factor for the subsequent development
good specificity vis--vis anxietysuch as dyspho- of PTSD (for a discussion of this issue, see Cardea
ria and lassitudemay exhibit more limited speci- & Carlson, 2011). More generally, it seems likely
ficity when examined in relation to sleep disorders. that many symptom-disorder relations are dynamic
Conversely, symptoms of appetite gainwhich and change significantly across acute versus chronic
displayed consistently weak associations with dis- phases of disorders.
orders in our analysesappears to show stronger,
more specific associations with seasonal forms of Clinical Implications
depression (e.g., Wehr et al., 1991). More gener- Although our findings warrant further research
ally, broader, more comprehensive approaches to and replication, they potentially have important
assessment can be highly informative in explicating implications for direct clinical intervention, par-
the basic properties of symptoms and in identifying ticularly with regard to assessment and diagnosis.
shared features that give rise to comorbidity. Our first major findingthat certain symptoms do
Second, although we examined how both not demonstrate specificity to their disorderhas
self-reported and clinician-rated symptoms were clear consequences for differential diagnosis. For
related to various disorders, our diagnostic data instance, symptoms of insomnia and eating distur-
were derived from a semi-structured interview bance relate equally to measures of depression and
(the SCID-IV) that employed skip-outs; because anxiety; thus, assessing these symptoms when work-
of this, we lacked complete symptom-level inter- ing with a new client who may have major depres-
view data and, therefore, were not able to model sion would not facilitate the clinicians ability to
the direct effects of removing different types of make a diagnosis, which is typically the goal in an
symptoms from the target diagnoses. As noted initial assessment. Removing those symptoms that
previously, the work of Elhai et al. (2008) and lack specificity and demonstrate diagnostic overlap
Grubaugh et al. (2010) highlights the impor- from the initial diagnostic assessment may signifi-
tance of modeling these symptom-based effects cantly enhance the assessors ability to make a dif-
directly. For example, Grubaugh et al. (2010) ferential diagnosis.
used the Clinician-Administered PTSD Scale Our second findingthat some symptoms lack
(Blake et al., 1995)a 30-item structured inter- the power to predict their associated DSM-IV diag-
view that corresponds to the DSM-IV criteria for nosisalso has important implications for diag-
PTSDto derive PTSD diagnoses. Because the nostic efficiency. For instance, based on our data,
Clinician-Administered PTSD Scale does not it appears that assessing dysphoria and hyperarousal
involve skip-outs, Grubaugh et al. (2010) were symptoms does not provide any incremental infor-
able to compare comorbidity rates for (a)standard mation relevant to the diagnosis of PTSD, once
PTSD diagnoses versus (b) the reduced symp- intrusions and avoidance have been considered.
tom criteria proposed by Spitzer et al. (2007). Removing symptoms that do not add to the diag-
As discussed earlier, they found that the rates of nosis can streamline the assessment process and
comorbidity between PTSD and other disorders facilitate the generation of a more efficientand
essentially remained unchanged regardless of the equally accuratediagnosis. In a busy practice,
symptom criteria that were used. Our own data time-effective assessment isvital.
(see Table 4.11) strongly suggest that a more radi- Of course, symptoms of insomnia, appetite dis-
cal pruning of the symptom criteria may be effec- turbance, and hyperarousal are distressing and may
tive in substantially lowering the comorbidity be clinically important in their own right; they
between PTSD and depression; obviously, how- therefore are appropriate targets in the treatment of
ever, it is necessary to test the effects of removing clients with major depression or PTSD, just as they
these symptoms directly before any clear conclu- would be in the many other disorders in which they
sions can bedrawn. are an obvious symptom. We are not suggesting

62 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
that these types of symptoms are unimportant or that the theoretical considerations and empirical
clinically uninteresting; rather, we are arguing that data we have reviewed in this chapter will stimulate
their assessment is not a necessary part of the initial further research into these issues and, more gener-
assessment and their removal as diagnostic criteria ally, into the nature and properties of distinctive
may, in fact, enhance efficient differential diagnosis. types of symptoms.
Author Notes:We thank Michael Chmielewski,
Conclusion Lee Anna Clark, Daniel Foti, Catherine Glenn,
Disorders such as major depression and PTSD Joshua Gootzeit, Greg Hajcak, Erin Koffel, Roman
represent heterogeneous combinations of distinct Kotov, Annmarie MacNamara, Jill Malik, Kristin
types of symptoms. We believe that analyses focus- Naragon-Gainey, Michael W. OHara, Jenny
ing on these distinctive symptom dimensions can Gringer Richards, Maria Rienzi, Eunyoe Ro, and
play an important role in explicating key diagnostic Anna Weinberg for their help in the preparation of
phenomena such as comorbidity. In this chapter, we this chapter. This research was supported by NIMH
have examined symptom dimensions within depres- Grant R01-MH068472 to David Watson and
sion and PTSD from the perspective of the quadri- NIMH Grant R01-MH083830 to Lee AnnaClark.
partite model, which posits that it is important to
consider two quantitative elements when analyzing References
the properties of symptoms: (a) the magnitude of Abramson, L. Y., Metalsky, G. L., & Alloy, L. B. (1989).
their general distress component and (b)their level Hopelessness depression:Atheory based subtype of depres-
sion. Psychological Review, 96, 358372.
of specificity.
American Psychiatric Association. (1980). Diagnostic and sta-
Our review highlighted the heuristic value of tistical manual of mental disorders (3rd ed.). Washington,
this framework. With regard to depression, the evi- DC:Author.
dence establishes that some symptoms containing a American Psychiatric Association. (1987). Diagnostic and statisti-
strong general distress component (e.g., dysphoria, cal manual of mental disorders (3rd ed., rev.). Washington,
lassitude) nevertheless show substantial evidence of
American Psychiatric Association. (2000). Diagnostic and statisti-
specificity. In contrast, symptoms of insomnia and cal manual of mental disorders (4th ed., text rev.). Washington,
appetite disturbancewhich are less saturated with DC:Author.
general distress variancedemonstrate little or no Beck, A. T., & Steer, R. A. (1990). Manual for the Revised
specificity. Moreover, these same symptoms have Beck Anxiety Inventory. San Antonio, TX: Psychological
relatively weak bivariate associations with DSM-IV
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Beck Depression
diagnoses of major depression and displayed little or Inventory manual (2nd ed.). San Antonio, TX:Psychological
no incremental validity in logistic regression analy- Corporation.
ses. In the case of PTSD, our analyses indicated that Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric
it was the symptoms with the strongest general dis- properties of the Beck Depression Inventory: Twenty-five
years of evaluation. Clinical Psychology Review, 8, 77100.
tress component (i.e., indicators of the Simms etal.
Benazzi, F. (2006). Symptoms of depression as possible markers
[2002] Dysphoria factor) that showed poor speci- of bipolar II disorder. Progress in Neuro-Psychopharmacology
ficity, relatively weak bivariate associations with the & Biological Psychiatry, 30, 471477.
diagnosis, and no incremental validity in logistic Blake, D. D., Weathers, F. W., Nagy, L. M., Kaloupek, D. G.,
regression analyses. Gusman, F. D., Charney, D., S., & Keane, T. M. (1995). The
development of a clinician-administered PTSD scale. Journal
Within both disorders, we identified symptoms
of Traumatic Stress, 8,7590.
that both (a) exhibited poor diagnostic specificity Brown, T. A., Chorpita, B. F., & Barlow, D. H. (1998).
and (b)provided little or no incremental informa- Structured relationships among dimensions of the DSM-IV
tion to their respective diagnoses. Put differently, anxiety and mood disorders and dimensions of negative
our results indicate that the nonspecific symptoms affect, positive affect, and autonomic arousal. Journal of
Abnormal Psychology, 107, 179192.
of these disorders actually contribute very little to
Campbell, D. T., & Fiske, D. W. (1959). Convergent and dis-
their diagnosis. This being the case, we have sug- criminant validation by the multitraitmultimethod matrix.
gested that the diagnosis and assessment of these Psychological Bulletin, 56, 81105.
disorders potentially can be improved by focusing Cardea, E., & Carlson, E. (2011). Acute stress disorder revis-
primarily on more specific types of symptoms. The ited. Annual Review of Clinical Psychology, 7, 245267.
Carney, C. E., Ulmer, C., Edinger, J. D., Krystal, A.D., &
available evidence regarding such key issues as diag-
Knauss, F. (2009). Assessing depression symptoms in those
nostic specificity and incremental validity remains with insomnia: An examination of the Beck Depression
rather limited, however, so it clearly will be impor- Inventory Second Edition (BDI-II). Journal of Psychiatric
tant to examine them further in the future. We hope Research, 43, 576582.

Watson,Stasik 63
Clark, L. A., & Watson, D. (1991). Tripartite model of anxi- Krueger, R. F., & Markon, K. E. (2006). Reinterpreting comor-
ety and depression: Psychometric evidence and taxonomic bidity:Amodel-based approach to understanding and classi-
implications. Journal of Abnormal Psychology, 100, 316336. fying psychopathology. Annual Review of Clinical Psychology,
Clark, L. A., Watson, D., & Reynolds, S. (1995). Diagnosis and 2, 111133.
classification of psychopathology:Challenges to the current Marshall, G. N., Schell, T. L., & Miles, J. N. V. (2010). All
system and future directions. Annual Review of Psychology, PTSD symptoms are highly associated with general dis-
46, 121153. tress: Ramifications for the dysphoria symptom cluster.
Cohen, J. (1988). Statistical power analysis for the behavioral sci- Journal of Abnormal Psychology, 119, 126135.
ences (2nd. ed.). NewYork:AcademicPress. Mineka, S., Watson, D., & Clark, L. A. (1998). Comorbidity
Elhai, J. D., Grubaugh, A. L., Kashdan, T. B., & Frueh, B. C. of anxiety and unipolar mood disorders. Annual Review of
(2008). Empirical examination of a proposed refinement to Psychology, 49, 377412.
DSM-IV posttraumatic stress disorder symptoms criteria Naragon-Gainey, K., Watson, D., & Markon, K. E. (2009).
using the National Comorbidity Survey Replication data. Differential relations of depression and social anxiety symp-
Journal of Clinical Psychiatry, 69, 597602. toms to the facets of extraversion/positive emotionality.
First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B.W. (1997). Journal of Abnormal Psychology, 118, 299310.
Structured Clinical Interview for DSM-IV Axis I Disorders Schmukle, S. C., & Egloff, B. (2009). Exploring bipolarity of
Patient Edition (SCID-I/P). NewYork:Biometrics Research, affect ratings by using polychoric correlations. Cognition and
NewYork State Psychiatric Institute. Emotion, 23, 272295.
Flora, D. B., & Curran, P. J. (2004). An empirical evaluation Simms, L. J., Watson, D., & Doebbeling, B. N. (2002).
of alternative methods of estimation for confirmatory factor Confirmatory factor analyses of posttraumatic stress symp-
analysis with ordinal data. Psychological Methods, 9, 466491. toms in deployed and non-deployed veterans of the Gulf
Gamez, W., Kotov, R., & Watson, D. (2010). The validity of War. Journal of Abnormal Psychology, 111, 637647.
self-report assessment of avoidance and distress. Anxiety, Slade, T. & Watson, D. (2006). The structure of common
Stress & Coping:An International Journal, 23,8799. DSM-IV and ICD-10 mental disorders in the Australian
Gootzeit, J., & Markon, K. (2011). Factors of PTSD:Differential general population. Psychological Medicine, 36, 15931600.
specificity and external correlates. Clinical Psychology Review, Spitzer, R. L., First, M. B., & Wakefield, J. C. (2007). Saving
31, 9931003. PTSD from itself in DSM-V. Journal of Anxiety Disorders,
Grubaugh, A. L., Long, M. E., Elhai, J. D., Frueh, B. C., & 21, 233241.
Magruder, K. M. (2010). An examination of the construct Watson, D. (2000). Mood and temperament. New York:
validity of posttraumatic stress disorder with veterans using GuilfordPress.
a revised criterion set. Behaviour Research and Therapy, 48, Watson, D. (2005). Rethinking the mood and anxiety disor-
909914. ders:Aquantitative hierarchical model for DSM-V. Journal
Hogan, R. T. (1983). A socioanalytic theory of personality. In of Abnormal Psychology, 114, 522536.
M. Page (Ed.), 1982 Nebraska Symposium on Motivation (pp. Watson, D. (2009). Differentiating the mood and anxiety dis-
5589). Lincoln:University of NebraskaPress. orders: A quadripartite model. Annual Review of Clinical
Kashdan, T. B. (2007). Social anxiety spectrum and diminished Psychology, 5, 221247.
positive experiences:Theoretical synthesis and meta-analysis. Watson, D., & Clark, L. A. (1991). The Mood and Anxiety
Clinical Psychology Review, 27, 348365. Symptom Questionnaire. Unpublished manuscript, University
Kenny, D. A. (1987). Statistics for the social and behavioral sci- of Iowa, IowaCity.
ences. Boston:Little, Brown and Company. Watson, D., Clark, L. A., & Carey, G. (1988). Positive and nega-
Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. (2005). tive affectivity and their relation to anxiety and depressive
Prevalence, severity, and comorbidity of 12-month DSM-IV disorders. Journal of Abnormal Psychology, 97, 346353.
disorders in the National Comorbidity Survey Replication. Watson, D., Clark, L. A., & Stasik, S. M. (2011). Emotions and
Archives of General Psychiatry, 62, 617627. the emotional disorders:Aquantitative hierarchical perspec-
Kessler, R. C., & stn, T. B. (2004). The World Mental Health tive. International Journal of Clinical and Health Psychology,
(WMH) Survey Initiative Version of the World Health 11, 429442.
Organization (WHO) Composite International Diagnostic Watson, D., Gamez, W., & Simms, L. J. (2005). Basic dimen-
Interview (CIDI). International Journal of Methods in sions of temperament and their relation to anxiety and
Psychiatric Research, 13, 93121. depression:Asymptom-based perspective. Journal of Research
King, D. W., Leskin, G. A., King, L. A., & Weathers, F. W. (1998). in Personality, 39,4666.
Confirmatory factor analysis of the Clinician-Administered Watson, D., & Naragon-Gainey, K. (2010). On the specificity of
PTSD Scale: Evidence for the dimensionality of posttrau- positive emotional dysfunction in psychopathology:Evidence
matic stress disorder. Psychological Assessment, 10,9096. from the mood and anxiety disorders and schizophrenia/
Koffel, E., & Watson, D. (2009). The two-factor structure of schizotypy. Clinical Psychology Review, 30, 839848.
sleep complaints and its relation to depression and anxiety. Watson, D., OHara, M. W., Chmielewski, M., McDade-Montez,
Journal of Abnormal Psychology, 118, 183194. E. A., Koffel, E., Naragon, K., & Stuart, S. (2008). Further
Kotov, R., Gamez, W., Schmidt, F., & Watson, D. (2010). validation of the IDAS: Evidence of convergent, discrimi-
Linking big personality traits to anxiety, depressive, and nant, criterion, and incremental validity. Psychological
substance use disorders: A meta-analysis. Psychological Assessment, 20, 248259.
Bulletin, 136, 768821. Watson, D., OHara, M. W., Naragon-Gainey, K. Koffel, E.,
Krueger, R. F. (1999). The structure of common disorders. Chmielewski, M., Kotov, R., Stasik, S. M., & Ruggero. C.J.
Archives of General Psychiatry, 56, 921926. (2012). Development and validation of new anxiety and

64 Examining the Comorbidit y Bet ween Depression and the Anxiet y Disorders
bipolar symptom scales for an expanded version of the IDAS Weathers, F. W., Litz, B. T., Herman, D. S., Huska, J. A.,
(the IDAS-II). Assessment 19, 399420. & Keane, T. M. (1993, October). The PTSD Checklist
Watson, D., OHara, M. W., Simms, L. J., Kotov, R., Chmielewski, (PCL):Reliability, validity, and diagnostic utility. Paper pre-
M., McDade-Montez, E.,...Stuart, S. (2007). Development sented at the annual meeting of the International Society for
and validation of the Inventory of Depression and Anxiety Traumatic Stress Studies, San Antonio,TX.
Symptoms (IDAS). Psychological Assessment, 19, 253268. Wehr, T. A., Giesen, H. A., Schulz, P. M., Anderson, J. L.,
Watson, D., OHara, M. W., & Stuart, S. (2008). Hierarchical Joseph-Vanderpool, J. R., Kelly, K.,...Rosenthal, N. E.
structures of affect and psychopathology and their implica- (1991). Contrasts between symptoms of summer depres-
tions for the classification of emotional disorders. Depression sion and winter depression. Journal of Affective Disorders, 23,
and Anxiety, 25, 282288. 173183.
Watson, D., & Tellegen, A. (1985). Toward a consensual struc- Widiger, T. A., & Clark, L. A. (2000). Toward DSM-V and the
ture of mood. Psychological Bulletin, 98, 219235. classification of psychopathology. Psychological Bulletin, 126,
Watson, D., & Tellegen, A. (1999). Issues in the dimensional 946963.
structure of affectEffects of descriptors, measurement Yufik, T., & Simms, L. J. (2010). A meta-analytic investigation
error, and response formats: Comment on Russell and of the structure of posttraumatic stress disorder symptoms.
Carroll (1999). Psychological Bulletin, 125, 601610. Journal of Abnormal Psychology, 119, 764776.
Watson, D., Wiese, D., Vaidya, J., & Tellegen, A. (1999). The Zinbarg, R. E., & Barlow, D. H. (1996). Structure of anxiety
two general activation systems of affect:Structural findings, and the anxiety disorders: A hierarchical model. Journal of
evolutionary considerations, and psychobiological evidence. Abnormal Psychology, 105, 181193.
Journal of Personality and Social Psychology, 76, 820838.

Watson,Stasik 65

5 Depressive Disorders, Comorbidity

Issues, and Assessment Strategies

Nicholas R.Eaton and Robert F.Krueger

Assessment is at the very core of clinical and research endeavors to understand and ameliorate
depressive disorders. In the current chapter, we discuss pressing issues in assessment of depressive
disorders beginning with the definitional:how these disorders are conceptualized and classified. We
highlight the DSM-IV-TR nosological organization of depressive disorders, and those disorders that are
closely related (e.g., anxiety and adjustment disorders), as well as current depressive disorder proposals
for the upcoming DSM-5. The high rates of comorbidity among the depressive and related disorders
are discussed as an assessment challenge, and we propose a unified latent structure framework to
supplement clinical assessment that involves characterizing individuals levels of underlying internalizing
disorder liability. We discuss how disorders, and the latent internalizing liability, may manifest differently
across subpopulations, such as age and ethnicity/culture. Finally, we discuss psychometric issues and
conclude with a list of critical unanswered questions in depressive disorder assessment.
Key Words: assessment, depressive disorders, depression, comorbidity, internalizing, neuroticism

Assessment of psychopathology is at the very is to capture the phenomena of interest adequately.

foundation of mental health efforts. Psychologists, No single assessment instrument or process will
psychiatrists, counselors, social workers, and other be adequate for all purposes, however, complicat-
health-care workers utilize clinical assessment to ing matters further. For instance, if ones purpose
investigate the signs and symptoms of patients is to identify individuals, at the population level,
and to develop treatment plans toward symp- experiencing depressive symptomatology in hopes
tom amelioration. Similarly, the basis of science is of improving public health (e.g., for National
measurement. As such, researchers use assessment Depression Screening Day; Baer etal., 2000), a brief
procedures to characterize the psychopathology of screening measure would likely be adequate as well
their participants, drawing inferences and testing as necessary. If ones purpose to create four highly
hypotheses about the nature of mental disorder. homogeneous groups of individuals with (1)major
Managed-care entities determine reimbursement depression, (2) dysthymia, (3) a nondepressive
schedules in large part by diagnosis, and assessment mental disorder, and (4)no psychopathology (i.e.,
is at the core of these diagnoses. Assessment is, thus, healthy controls) to test the efficacy of a new anti-
a critical component of all endeavors relating to depressant medication, lengthy and in-depth assess-
the classification, understanding, and treatment of ment facilitating this complex differential diagnosis
psychopathology. will likely be required. Other considerations in eval-
Assessment in general is a broad topic with uating assessment measures include measures psy-
myriad considerations, although the ultimate goal chometric properties of reliability and validity,

expense, information source (e.g., clinician-, self-, Let us consider an example from another disci-
or informant-report), time requirements, and so pline, nutrition, to illustrate why the assessment of
on. These factors may change in different contexts psychopathology is complicated by its very defini-
(e.g., psychometric properties being different across tion. If a nutritionist is interested in measuring, say,
populations; time to complete assessment may dif- the weight of a patient, she is presented with a single
fer between depressed individuals in a university physical entity, the patient (who weighs a particular
mental-health center versus a traumatic brain injury amount). The nutritionist has a clear definition of
clinic). the construct he or she wants to assess, weight. The
Even by restricting our discussion to assessment nutritionist also has a standard metric to character-
of depressive disorders, we are still left with a topic ize the patients weight in kilograms (or pounds).
far too broad to cover fully in any single chapter Finally, the nutritionist has a highly reliable assess-
or book. Our purpose in the current chapter is to ment instrument, ascale.
highlight several of the most salient and pressing Now, consider a clinical psychologist who is
issues in the assessment of depressive disorders. asked to assess whether a given patient, who appears
These include classification issues and nosology, sad, is depressed and, if so, what level of severity the
comorbidity among the disorders, and general depression reaches. Depression, unlike weight, is a
diversity considerations, including how assessment construct that is not directly observable. Through
measuresand perhaps the depressive-disorder years of experience gained by mental-health workers,
constructs themselvesdiffer nontrivially across there has been a broad consensus that there exists,
populations. We will not provide a full review of in some patients, a phenomenon associated with
specific measures of depressive disorders, given low mood, disinterest in activities, disturbances in
that over 280 such measures have been developed thinking and concentration, and other symptoms.
(Santor, Gregus, & Welch, 2006). Several excel- When these signs and symptoms occur together in
lent resources review the characteristics of the most particular patterns and for a given duration, clini-
widely used depressive disorder assessment instru- cians have agreed to conceptualize this constellation
ments, and interested readers are referred to those of pathology as a major depressive episode. Thus,
works for additional information (see Nezu, Nezu, the phenomenon in question is not directly observ-
Friedman, & Lee, 2009; Nezu, Ronan, Meadows, able as weight is. It is the result of clinical experience
& McClure, 2000). We begin with a discussion of and consensusa latent construct that cannot be
definitional issues in depressive disorders. measured directly.
There is no equivalent of the nutritionists scale
Classification Issues in for depression; there is no gold-standard ruler or
Depressive Disorders yardstick. Instead, clinicians have attempted to
Classification of latent constructs define the signs and symptoms that appear most
Clinical assessment is primarily intended to closely related to depression, which form the diag-
determine the presence (versus absence) of psycho- nostic criteria for major depression in official classifi-
pathology, and, if present, to measure its degree of cation systems, such as the Diagnostic and Statistical
severity. Thus, mental disorder constructs are the Manual of Mental Disorders (DSM). Beginning
focus of, and foundation for, clinical assessment. As largely with DSM-III (American Psychiatric
in many other scientific disciplines, measurement Association, 1980), there was an attempt to define
in the allied mental-health fields is complicated discrete signs and symptoms of mental disorders
by the very constructs we wish to assess, because that could be assessed reliably, with the aim of yield-
they are latent. That is, psychopathology, as it ing reliable diagnoses. This is in stark contrast to
exists in nature, is unobservable directly. We must, other possible approaches, such as one requiring
thus, operationalize it as the conjunction of mul- untestable, and likely unreliable, psychodynamic
tiple manifest (directly observable) indicators. Such clinical inferences about, for instance, the origins
conjunctions are known as latent variables because of the mood disturbance. The current version of
they are inferred from multiple direct observations. the diagnostic manual, DSM-IV-TR (American
For example, major depression is not just insomnia, Psychiatric Association, 2000), lists nine diagnostic
because insomnia (although directly observable) criteria for a major depressive episode and requires
must occur in the context of additional observable at least five of these to be present during the same
indicators (e.g., self-reported low mood) to infer the two-week period, and represent a change from pre-
existence of an episode of depression. vious functioning, for a major depressive episode to

Eaton, Krueger 67
be deemed present. It is the purpose of diagnostic impairment, preoccupation with worthlessness, sui-
clinical assessment of the major depression diag- cidal ideation, psychotic symptoms, or psychomo-
nosis to ascertain if a patients symptomatology fits tor retardation. This latter consideration is known
the definition given in DSM. However, we now see as the bereavement exclusion, which putatively
that the definition itself reflects assessment consid- distinguishes between normal grief and a major
erations, and, in this way, assessment has informed depressive episode.
classification on a fundamental level. Alternatively, As we can see, the assessment of a major depres-
classification delineates the focus of clinical assess- sive episode is quite a complex undertaking if done
ment. As such, we will first turn our attention to in full compliance with the DSM-IV-TR system.
classification issues of the depressive and related The major depressive episode, however, is not con-
disorders. sidered a mental disorder, and it cannot be used as a
diagnosis; rather, it forms the foundation for a diag-
DSM-IV-TR depressive disorders nosis of major depressive disorder. Major depressive
The DSM-IV-TR divides mental disorders first disorder requires (1)the presence of a major depres-
into Axes, with Axis Idisorders representing what sive episode, (2) no history of a manic, mixed, or
are sometimes referred to as clinical disorders and hypomanic episode, and (3) the major depressive
with Axis II representing personality disorders and episode is not better accounted for, or superimposed
mental retardation. Within Axis I, disorders are on, other disorders (i.e., schizophrenia, schizoaffec-
organized into conceptually similar groups, such tive disorder, schizophreniform disorder, delusional
as anxiety disorders, eating disorders, and sleep dis- disorder, or psychotic disorder NOS). If one such
orders. Mood disorders represents another group episode has occurred, the modifier single episode
of disorders, and DSM-IV-TR divides this category is used; if more than one major depressive episode
into depressive disorders and bipolar disorders. The has occurred, the modifier recurrent is used. Thus,
primary depressive disorders listed are major depres- the diagnosis of major depressive disorder techni-
sive disorder, dysthymic disorder, and depressive cally requires (1)assessment of the possible presence
disorder NOS (not otherwise specified). To illus- of a major depressive episode currently, (2)assess-
trate clinical assessment using the DSM-IV-TR, we ment of possible previous major depressive episodes,
will now characterize the process by which one can (3)assessment of previous manic, mixed, and hypo-
receive a diagnosis of major depressive disorder. manic episodes, and (4)assessment of the rule-out
The foundation of assessment for depression disorders.
in DSM-IV-TR is the major depressive episode. After making a diagnosis of major depressive
Assessment involves characterizing whether at least disorder, a variety of specifiers are included in
five of nine diagnostic criteria have been present DSM-IV-TR to characterize the disorder further.
during the same two-week period and represent First, clinicians must specify the severity of the major
a change from previous functioning. First, either depressive episode, with severity options of mild,
(1)depressed mood or (2)anhedonia (i.e., dimin- moderate, severe without psychotic features, and
ished interest or pleasure in activities) must be severe with psychotic features. If psychotic features
present. Other criteria include (3) weight loss or are present, they can be specified as mood-congruent
gain, or increase or decrease in appetite, (4)insom- or mood-incongruent. Other possible features to be
nia or hypersomnia, (5) psychomotor agitation or assessed are catatonic, melancholic, and atypical
retardation, (6) fatigue or loss of energy, (7) feel- features. If the major depressive episode has been
ings of worthlessness or guilt, (8)diminished abil- continually present for at least two years, a specifier
ity to think or concentrate, or indecisiveness, and of chronic can be applied. Longitudinal course of
(9) recurrent thoughts of death, suicidal ideation, major depressive disorder can be specified as being
suicide planning, or a suicide attempt. The present with or without full interepisode recovery. If onset of
symptoms must cause significant distress or impair- the major depressive episode occurred within four
ment, must not be due to the direct physiological weeks of giving birth, a specifier of with postpartum
effects of a substance or general medication condi- onset can be applied. Aseasonal pattern specifier can
tion, and must not meet criteria for a mixed episode be given when the mood disturbance fluctuates in
(i.e., simultaneous manic and major depressive epi- tandem with times of the year. Finally, if the full cri-
sodes). Finally, if the individual has lost a loved one teria for a major depressive episode are not met cur-
recently, the symptoms must last longer than two rently, the major depressive disorder can be specified
months after the death or be associated with notable as being in partial or in full remission.

68 Depressive Disorders, Comorbidit y Issues, and Assessment Strategies

The DSM-IV-TR also lists dysthymic disorder disorder. DSM-IV-TR lists a variety of possibilities
and depressive disorder NOS as depressive dis- for depressive disorder NOS, including premen-
orders. Dysthymic disorder is long-term depressed strual dysphoric disordera marked change in
mood, with some associated additional symptoms, depressive and related symptoms during most men-
that is not better conceptualized as chronic major strual cycles. Other possibilities include depressive
depressive disorder, major depressive disorder in symptoms that do not meet severity or duration
partial remission, or as a bipolar mood disorder. criteria for a major depressive episode or dysthymic
Unlike major depressive disorder, which requires disorder.
the presence of a major depressive episode for diag- The NOS diagnoses of DSM-IV-TR present a
nosis, dysthymic disorder is considered present challenge for assessment. Their inclusion in the diag-
when at least two years of depressive symptomatol- nostic system allows for mental-health-care workers
ogy have occurred. During the first two years of the to provide a diagnosis (and bill for services) when
mood disturbance, no major depressive episodes the specific disorders in the nosology fail to charac-
may have occurred; major depressive episodes may terize adequately the experience of the patient. In
have occurred prior to this two-year period, pro- this way, NOS diagnoses require ruling out the pres-
viding they were in full remission before onset of ence of multiple other possible disorders, which is a
dysthymia. time-consuming and complex process. Unfortunately,
As can be seen, the construct of dysthymic dis- given limited resources and other constraints fre-
order is related quite strongly to major depressive quently placed upon clinical practice, NOS diagnoses
episodes/disorder. Indeed, there is significant over- are often improperly used to characterize symptoms
lap not only in the general focus on depressed mood when a full assessment is not completed. In other
but also in the associated symptoms. Dysthymia words, from an assessment perspective, a diagnosis of
requires the presence of two or more of the follow- depressive disorder NOS may reflect a bona fide diag-
ing: (1) poor appetite or overeating, (2) insomnia nosis of depressive psychopathology or an inability/
or hypersomnia, (3)low energy or fatigue, (4)low unwillingness on the part of the provider to engage
self-esteem, (5) poor concentration or difficulty in careful and detailed assessment. The research lit-
making decisions, and (6)feelings of hopelessness. erature suggests that NOS disorders are among the
With the exception of low self-esteem, these diag- most commonly applied diagnoses within groups
nostic criteria overlap directly with those of a major of disorders (e.g., Clark, Livesley, & Morey, 1997;
depressive episode. Fairburn & Bohn, 2005; Lenzenweger, Loranger,
Assessment, for the purpose of differential diag- Korfine, & Neff, 1997; Pagan, Oltmanns, Whitmore,
nosis, between major depressive episodes/disorder & Turkheimer, 2005). There is likely a genuine need
and dysthymia relies on careful characterization of for NOS disorders within the current classification
the length of the mood disturbance and the breadth system, but their clinical utility is greatly depleted
of the symptoms that are present. This requires ret- when their diagnosis represents lackadaisical assess-
rospection on the part of the patient, which can be ment rather than a best fit diagnosis.
fallible (Moffitt et al., 2010). Clinicians may also
seek collateral information, such as informant-report DSM-IV-TR depressive-related
information or medical records. Finally, although disorders
DSM-IV-TR states that there must be a two-year As we have seen, the current classification sys-
period of dysthymia free of major depressive episodes tem conceptualizes depressive disorders into major
at its onset, it leaves open the possibility of later diag- depressive disorder, dysthymic disorder, and depres-
nostic comorbidity. That is, after dysthymia has been sive disorder NOS. In terms of clinical assessment
established, it is possible that patients will experience and differential diagnosis, these disorders differ
an increase in symptoms, resulting in a simultaneous largely in symptom breadth (i.e., how many diag-
major depressive episode. This comorbidity pattern, nostic criteria are present) and duration. Other,
sometimes referred to as double depression, repre- nondepressive DSM-IV-TR disorders, however, can
sents a major depressive episode superimposed upon be associated with mood disturbance and can pres-
long-standing dysthymic disorder. ent with depressive features. We will briefly discuss
Depressive disorder NOS is a broad and poten- a few of these disorders, because clinical assessment
tially heterogeneous diagnostic construct. It is and differential diagnosis frequently require their
diagnosed when depressive features are present but consideration, either as diagnoses to rule out or as
diagnostic criteria are not met for any particular diagnoses that account for observed depressivity.

Eaton, Krueger 69
One diagnosis closely related to the official out. These latter, depressive diagnoses are consid-
depressive disorders is depressive personality disorder, ered unipolar disordersthat is, they are associ-
which is included in the appendix of DSM-IV-TR ated with only a single pole of a mood dimension,
as a criterion set for further study. The essential fea- the low pole. This is in contrast to bipolar disor-
ture of depressive personality disorder is a pervasive ders, which are most typically associated with the
pattern of depressive cognitions and behaviors that elevated and low poles of mood. Finally, the mood
begins by early adulthood and that occurs in a vari- disorders section also includes other mood disor-
ety of contexts (p.788). Personality disorders such ders, including mood disorder due to a general medi-
as depressive personality disorder are conceptualized cal condition, substance-induced mood disorder, and
as enduring forms of psychopathology that mani- mood disorderNOS.
fest across multiple constructs and are lengthy (e.g., Although bipolar I disorder can technically
lifelong) in duration. Given these duration consid- occur without a history of a major depressive epi-
erations, there is clear overlap between dysthymic sode, bipolar II disorder requires the presence or
disorder and depressive personality disorderboth history of a major depressive episode. The overlap
are long-term phenomena with a core of depres- here with major depressive disorder is clear, given
sivity. In terms of differential diagnosis, depressive that both disorders require a major depressive epi-
personality disorders diagnostic criteria focus more sode. The differential diagnosis between bipolar
on cognitive, interpersonal, and intrapsychic per- II disorder and the depressive disorders requires
sonality traits than do those of dysthymic disorder. assessment for the presence, or positive history, of
For instance, the diagnostic criteria of depressive a hypomanic episode. If hypomania has ever been
personality disorder include feeling pessimistic (cog- present, a diagnosis of bipolar II disorder is appro-
nitive), being judgmental and critical toward others priate rather than a diagnosis of major depressive
(interpersonal), and having a critical, blaming, and disorder. Clinically, though, assessment for hypo-
derogatory attitude toward the self (intrapsychic). mania (or mania) is complicated by the possibility
Further, the onset of depressive personality disorder of dysphoric hypomania. Rather than the elevated,
should have occurred by early adulthood, which is expansive mood often considered a hallmark of
not necessarily the case for dysthymia. These rather (hypo)mania, DSM-IV-TR allows for (hypo)manic
fine-grained distinctions require very careful clini- episodes instead to be associated with irritable
cal assessment of onset, duration, pervasiveness, and mood. Because the diagnostic criteria for a hypo-
precise nature of the depressive symptoms the patient manic episode include several signs and symptoms
is experiencing. The importance of the distinction that could be interpreted as representative of a
between these two disorders, however, is probably major depressive episode (e.g., decreased need for
one of clinical judgment; this is highlighted by the sleep, distractibility, psychomotor agitation), careful
inclusion of depressive personality disorder in a assessment to characterize the precise nature of the
supplementary appendix rather than the main body episode is necessary.
text, and given that depressive personality disorder Althougj depressive personality disorder and the
is diagnosed formally as personality disorder NOS. bipolar disorders present perhaps the most compli-
Indeed, DSM-IV-TR notes that it remains contro- cated issues in the assessment of depressive disorders,
versial whether the distinction between depressive other DSM-IV-TR disorders and phenomena can
personality disorder and Dysthymic Disorder is use- show substantial overlap as well. We will not bela-
ful (p.788). bor this point, but it is important to touch briefly
Within the mood disorders section of on three such constructs. First, anxiety disorders
DSM-IV-TR are the bipolar disorders, which relate are commonly comorbid with depressive disorders
closely to the depressive disorders in many ways. (Krueger & Markon, 2006; Watson, 2005, 2009),
These disorders, broadly, comprise bipolar I dis- and some anxiety disorders share features with some
order, bipolar II disorder, cyclothymic disorder, and depressive disorders. For instance, generalized anxi-
bipolar disorder NOS. The disorders all require ety disorder has diagnostic criteria involving psycho-
the presence of at least one hypomanic, manic, or motor agitation, fatigue, difficulty concentrating,
mixed episode. In terms of assessment of depressive irritability, and sleep disturbance. Differential diag-
disorders, it is critical to determine any history of nosis between generalized anxiety disorder and the
these episodes, because, if such an episode has been depressive disorder requires assessment of whether
present in the patients life, a diagnosis of major the symptoms arise from worry and anxiety or are
depressive disorder or dysthymia must be ruled associated with low mood disturbance, respectively.

70 Depressive Disorders, Comorbidit y Issues, and Assessment Strategies

Adjustment disorder results from exposure to one or assessment, of depressive and related disorders is
more identifiable stressors that lead to emotional tentative at best at this point. We will discuss briefly
or behavioral symptoms; adjustment disorder with the most recent developments in depressive disorder
depressed mood is a subtype defined predomi- classification toward DSM-5, based on distributed
nantly by depressed mood, feelings of hopelessness, reports from the DSM-5 Mood Disorders Work
and tearfulness, and, thus, can present similarly to Group. This information is available at http://www.
depressive disorders. Clinical assessment should for interested readers who want the most
focus on the identification of a stressor and how current information about the revision.
closely the symptoms relate to postexposure adjust- The current DSM-5 depressive-disorders pro-
ment. Finally, bereavement after the death of a loved posal has expanded the depressive disorders nota-
one, listed among other conditions that may be a bly. Major depressive disorder (single episode and
focus of clinical attention and thus not diagnos- recurrent) and chronic depressive disorder (dys-
tic, can closely mirror the symptoms of a major thymia) remain as diagnoses. Also included are
depressive episode. Major depressive disorder is substance-induced depressive disorder, depressive disor-
usually not diagnosed for at least two months after der associated with a known general medical condi-
the loss, unless atypical grief reactions (e.g., hallu- tion, and depressive conditions not elsewhere classified,
cinatory experiences, morbid preoccupation with which have clear overlap with some DSM-IV-TR
worthlessness)occur. mood disorder constructs. Disruptive mood dys-
regulation disorder is a proposed diagnosis involving
Physical problems overlapping with temper outbursts occurring on otherwise negative
depressive disorders mood. Premenstrual dysphoric disorder concerns
As we noted earlier, mood disorder due to a gen- mood changes (e.g., depressed mood, mood swings,
eral medical condition is listed in the DSM-IV-TR irritability) that are temporally linked with the
to capture depressive symptoms (or other mood menstrual cycle. Mixed anxiety/depression is defined
symptoms) resulting from physical problems. For as meeting three or four diagnostic criteria for a
instance, metabolic issues may lead to onset of major depression and also experiencing anxious dis-
mood pathology. Other somatic issues, however, tress (e.g., irrational worry, motor tension, preoc-
may produce symptoms that closely mirror those cupation with unpleasant worries).
of the depressive disorders but are not bona fide Because these revisions are currently tentative, it
depressive disorders. For instance, individuals with is not profitable to discuss them at length in the cur-
obesity may report symptoms such as fatigue, hyper- rent report. Some of these changes are substantial
somnia, psychomotor retardation, appetite changes, though, and they have evoked notable controversy.
and feelings of worthlessness or inappropriate guilt. Among the most controversial proposed changes
Individuals with obstructive sleep apnea may show to the depressive disorders has been the removal
daytime fatigue, fragmented sleep, psychomotor of the bereavement exclusion when assessing for a
retardation, and reduced ability to think or concen- major depressive episode. This removal was justified
trate due to daytime fatigue. The crucial assessment by empirical evidence that loss of a loved one does
considerations here involve the presence or absence not differ significantly from the myriad other stress-
of low mood and/or anhedonia. ors that can precipitate a major depressive episode
(Zisook & Kendler, 2007). The controversy around
DSM-5 depressive disorders this change revolves around several issues. For one,
The diagnostic classification system is cur- it is possible that this removal could pathologize
rently undergoing revision toward the next edi- normal grief reactions after the death of a loved
tion: DSM-5. (The DSM-5 was released on May one, classifying a typical and expected human expe-
22, 2013. However, this chapter was completed rience as a mental disorder. Another concern is that
before the release date.) The DSM-5 Task Force such a change could lead to increased diagnosis of
and various Work Groups are now in the process major depressive disorder, thus inflating the preva-
of drafting this next version of the official nosology, lence rates. This would give the appearance to some
which is slated for publication in 2013. (The other that major depressive disorder is occurring with
primary nosology, the International Classification of greater frequency, although it would actually rep-
Diseases [ICD] is simultaneously undergoing revi- resent expanding the purview of the major depres-
sion as well.) As such, any discussion of what the sive disorder diagnosis such that it would include
future holds for diagnostic classification, and, thus, morecases.

Eaton, Krueger 71
Classification implications for test-retest reliability would largely reflect the varia-
assessment tion in depressed mood over a three-day period.
To summarize, the DSMseditions past, current, Within a sample of individuals with premenstrual
and apparently futurehave classified depressive dysphoric disorderconceptualized as depression
psychopathology broadly by parsing it into a variety or dysthymiathe three-day test-retest reliability
of smaller units. This classification has moved over would probably be quite low. As women progressed
the years to focus on diagnostic criteria, rather than through their menstrual cycles, the questionnaire
narrative paragraphs, that can be assessed individu- would (accurately) produce significant low mood
ally. Ideally, diagnostic criteria should be assessable scores at baseline and much improved low mood
in a reliable and valid manner, and this goal has led scores only days later. This test-retest reliability
framers of the DSM to tailor the diagnostic con- could easily be interpreted as the questionnaire
structs themselves such that they can be assessed being an unreliable measure. In actuality, this assess-
well. This highlights the close, reciprocal interrela- ment device was totally reliable. It was the diagnos-
tions among assessment and psychiatric nosology:It tic systems shortcoming that produced this putative
is not simply the case that a disorder is drafted in a failure of assessment.
vacuum free of assessment concerns, and assessment We can consider a nonhypothetical example to
instruments and procedures are then later devel- further this point. Borderline personality disorder
oped to capture this phenomenon. is defined by DSM-IV-TR with nine diagnostic
Assessment can only be as good as the constructs criteria, at least five of which must be present for
it is intended to assess. Although professionals have diagnosis. (As far as we are awareand Krueger is a
a tendency to reify diagnoses, there is little reason member of the DSM-5 Personality and Personality
to believe that the DSM-IV-TR represents a defini- Disorders Work Groupthis threshold was set
tive and precise characterization of psychopathology because five is more than half of nine, not because
(Hyman, 2010). This point is clearly demonstrated there were data to support this particular thresh-
by the revision toward DSM-5, which is a clear old.) Investigation of these criteria reveals that they
admission that the current nosology, like all others, are quite heterogeneous in terms of their content,
is lacking. If an assessment instrument yields poor including frantic efforts to avoid abandonment,
reliability (and thus poor validity as well), this is not identity disturbance, affective lability, feelings of
necessarily indicative of poor assessment. Indeed, emptiness, behavioral dysregulation, unstable and
this could reflect poor construct validity of the diag- intense relationships, paranoid ideation and dis-
nostic entity itself. sociation, suicidality, and anger problems. This is
To illustrate the point that nosological shortcom- quite a broad range of symptom expression, and
ings can masquerade as assessment failures, we will research has indicated that borderline personality
consider an example. Let us assume that premen- disorder, as currently defined, reflects the conflu-
strual dysphoric disorder exists in nature, with a key ence of multiple psychopathological problems.
feature of depressed mood temporally linked to the For instance, Sanislow and colleagues (2002) used
menstrual cycle, but it has not yet been identified confirmatory factor analysis to examine the latent
by researchers and clinicians. Let us further assume structure of borderline personality disorderthat
that a hypothetical self-report questionnaire can is, the underlying factors that accounted for how
assess all the symptoms and severity of depressed the diagnostic criteria related to one anotherand
mood, with perfect reliabilitythat is, it can assess determined that three underlying factors were pres-
depressed mood without any error whatsoever. If ent. First, a disturbed relatedness factor accounted
the classification system did not include premen- for (co)variation of diagnostic criteria assessing
strual dysphoric disorder, and instead attempted to unstable relationships, identity disturbance, feelings
capture it simply with a diagnosis of major depres- of emptiness, and stress-related paranoid ideation.
sive disorder (severity mild), this would cause myr- Second, a behavioral dysregulation factor accounted
iad assessment problems with our otherwise precise for co(variation) of criteria for impulsivity and sui-
questionnaire when administered to patients with cidal and self-mutilative behaviors. Finally, an affec-
premenstrual dysphoric disorder. Within a sample tive dysregulation factor accounted for (co)variation
of depressed individuals, the three-day test-retest of criteria for affective instability, inappropriate
reliability of the questionnaire would likely be quite anger, and avoidance of abandonment. These fac-
highsince our hypothetical questionnaire assesses tors were correlated but also unique, suggesting that
current low mood with no error, fluctuations in three key psychopathological currents form the

72 Depressive Disorders, Comorbidit y Issues, and Assessment Strategies

basis for what is defined as borderline personality the assessor must also determine whether major
disorder. Another study, investigating the latent fac- depressive disorder and generalized anxiety disorder
tors that account for a diagnosis of borderline per- (and dysthymic disorder, etc.) are present. Indeed,
sonality disorder, rather than its individual criteria, comorbidity is often the rule rather than the excep-
indicated that at least two dimensions (a tendency tion, and DSM-IV-TRs exclusion criteria (e.g.,
to experience psychological distress and a tendency ruling out major depressive disorder once a manic
to engage in externalizing behaviors) accounted for episode has occurred) often do little to minimize
the diagnosis (Eaton, Krueger, Keyes, etal., 2011; this overlap in terms of assessment.
Eaton, Krueger, & Oltmanns,2011). One would expect some disorders to co-occur
Given that multiple factors appear to underlie randomly given their prevalence rates. For instance,
the diagnostic criteria, and diagnosis itself, of bor- one study of a representative sample found that the
derline personality disorder, an assessment measure past-year prevalence rates for major depressive dis-
that captures borderline personality disorder crite- order and generalized anxiety disorder were 13.3%
ria (or diagnosis) quite well might still have poor and 2.7%, respectively. If these disorders occurred
psychometric properties. In )addition to temporal independently, one would predict, based on these
issues (e.g., scores varying due to affective instabil- prevalence rates, that four individuals out of 1,000
ity), reliability judged by internal consistency could (i.e., 0.133 * 0.027. * 1,000 = 3.591, rounding
also be quite low because of this constructs mul- up to 4) would have comorbid major depressive
tidimensionality. It is quite possible that the items disorder and generalized anxiety disorder (Eaton,
assessing affective dysregulation features would South, & Krueger, 2010). However, investigation
not move closely in tandem with items assessing of the actual level of overlap revealed a different pat-
disturbed relatedness features, yielding apparently tern:17 individuals out of 1,000 actually had this
poor internal consistency even though the assess- comorbid presentation. Thus, the level of disorder
ment device is reliably measuring a (potentially co-occurrence was more than 400% higher than
problematic) diagnostic construct. Again, the puta- that expected by chance, suggesting high levels of
tive success of the assessment device is attenuated systematic comorbidity.
not by its own failures; rather, the assessment device Depressive disorders can occur comorbidly with
can only perform as well as the diagnostic con- nearly every other form of psychopathology in an
structs it is intended to measure. Diagnoses that are unsystematic way, with the overlap arising solely out
unreliable or invalid will yield unreliable or invalid of random co-occurrence reflecting prevalence rates.
assessments, no matter how precise the assessment When one looks at the empirical data, though, clear
instrument and procedure maybe. patterns emerge. It is not the case that comorbidity
of depressive disorders is unsystematic and random.
Three Assessment Considerations Rather, the depressive disorders most commonly
We have detailed earlier how the classification of co-occur with each other and also with the anxiety
a disorder and its assessment are inextricably linked. disorders (Kessler, Chiu, Demler, & Walters, 2005;
We now turn our attention to three other major Krueger, 1999; Krueger & Markon, 2006; Mineka,
considerations that can affect assessment. First, we Watson, & Clark, 1998; Slade & Watson, 2006;
discuss comorbidity of depressive disorders. Second, Vollebergh et al., 2001; Watson, 2005, 2009).
we discuss how disorders present across various Research in this field of structural psychopathology
subpopulations, such as cultural groups. Third, we has indicated that the depressive disorders and anxi-
discuss how psychometric properties, such as how ety disorders share a common underlying structure,
items of an assessment battery function, can vary which accounts for their frequent co-occurrence.
nontrivially. This latent construct is referred to as internalizing
and can be thought of as a dimensional liability to
Comorbidity experience unipolar mood and anxiety pathology
As we discussed earlier, differential diagnosis of (e.g., generalized anxiety disorder, posttraumatic
the depressive and related disorders can be a vex- stress disorder, panic disorder, agoraphobia).
ing problem for psychological classification. This In clinical assessment, the goal is to divide varia-
picture is complicated by the fact that individual tion in the depressive disorders into two parts. In
disorders do not always occur one at a time. So, classic test-theory terms, these are the true score and
in addition to assessing whether major depressive the error, which are equivalent to the signal and the
disorder or generalized anxiety disorder is present, noise. Clinical assessment of depressive disorders

Eaton, Krueger 73
attempts to capture and maximize the signal (true of transdiagnostic treatment modalities that can
score) while decreasing the noise (error). The struc- conceivably affect multiple disorders simultane-
tural psychopathology research that yielded the ously by addressing their internalizing core (Barlow
presence of the internalizing liability, however, sug- etal., 2011). Further, it appears increasingly likely
gests that variation of the depressive disorders can that the DSM-5 organization (referred to as its
be parsed into two other parts: common variance meta-structure) will be organized in such a way
and unique variance. Common variance here would that internalizing disorders are grouped proximally,
be internalizing itselfthe variation shared by the emphasizing their close empirical connections
depressive (and anxiety) disorders. It forms the core (Andrews et al., 2009; Regier, Narrow, Kuhl, &
of all depressive pathology. Unique variance, on Kupfer,2011).
the other hand, is what makes each disorder sepa- When the model of mental disorders character-
rate from the others. It is the variation that makes ized in the DSM-IV-TR and other recent editions is
major depression different from, say, dysthymia. tested empirically, it tends to be inferior to the inter-
Unique variation of one disorder does not overlap nalizing model discussed earlier (for reviews, see
with unique variation of another disorder by defi- Eaton etal.,, 2010; Krueger & Markon, 2006). For
nition. Unique variance also includes error. This instance, if depressive disorders are grouped together
error could be error associated with the assessment and anxiety disorders are grouped togetheras they
process, or it could also represent classification error are in DSM-IV-TRin a statistical testing frame-
as well. Insofar as some of a disorders diagnostic work (e.g., confirmatory factor analysis), this yields
criteria are relatively unrelated to core construct of a poorer fit to actual observed data than the inter-
the disorder, they could also be conceptualized as nalizing model, where these disorders are modeled
error in thisway. such that they all relate to a single, underlying core.
The structural psychopathology literature leaves This is another indication that clinical assessment of
us with an important question: What should depressive disorders might benefit from including
be the focus of our assessment? Should we use a assessment of internalizing liability itself in addition
disorder-focused assessment approach that attempts to more traditional disorder-specific assessment. It
to maximize the signal of each disorder, minimize may be the case that the unique variance of each dis-
the noise present in the assessment, and allow us order is not worthwhile for a given purpose and may
to differentiate DSM-IV-TR constructs from one be associated with error; if that is the case, assess-
another in a highly precise manner? If the goal of ment of internalizing liability would be more clini-
assessment is DSM-IV-TR differential diagnosis, cally informative than disorder-specific approaches.
say, for reimbursement purposes, then the answer At this point, it is important to note that, while
may be yes. Should we, instead, focus on assess- internalizing liability appears to hold promise as a
ing what appears to be at the core of the depres- means to classify and conceptualize psychopathol-
sive disordersthe internalizing liability that is at ogy, the nature of this liability has yet to be fully
the heart of the unipolar mood and anxiety disor- characterized. Most studies of the comorbidity of
ders? For purposes other than differential diagnosis, depressive and anxiety disorders have conceptualized
the answer to this question may be affirmative in internalizing liability as a continuous dimension,
some cases. Although internalizing liability contin- ranging from very low to very high. Every individ-
ues to be investigated actively, many studies have ual occupies some position along this continuum.
supported its role as important in understanding, However, internalizing liability does not always
assessing, and conceptualizing psychopathology. emerge as a unidimensional construct. Although
It accounts for lifetime patterns of comorbidity many studies have assumed internalizing liability is
of the mood and anxiety disorders (Kessler et al., best conceptualized as a single dimension (Eaton,
2011). It arises primarily from genetic mechanisms Krueger, & Oltmanns, 2011; Fergusson, Horwood,
(Hettema, Neale, Myers, Prescott, & Kendler, & Boden, 2006; Krueger etal., 1998), others have
2006; Kendler, Prescott, Myers, & Neale, 2003; used alternative conceptualizations. For instance,
Kendler et al., 2011), giving insight into the ori- some studies have found evidence that internaliz-
gins of these disorders. It accounts for why many ing liability is actually a higher-order dimension,
disorders respond to similar psychotherapeutic which subsumes two subdimensions: (1) distress
(Nathan & Gorman, 2007) and pharmacologi- (also termed anxious-misery), which is a liability
cal (Goldberg, Simms, Gater, & Kreuger, 2011) primarily to major depressive, dysthymic, and gen-
treatments, and it has prompted the development eralized anxiety disorders, and (2) fear, which is a

74 Depressive Disorders, Comorbidit y Issues, and Assessment Strategies

liability primarily to social phobia, specific phobia, One possibility of assessing internalizing lia-
and panic disorder (Eaton et al., 2012; Krueger, bility is through questionnaires, both self- and
1999; Slade & Watson, 2006; Vollebergh et al., informant-report questionnaires, developed out
2001). Other studies have conceptualized internal- of the work of Achenbach beginning decades ago
izing liability as a set of liability classesthat is, not (e.g., Achenbach & Edelbrock, 1978, 1984). Over
as dimensionalas well (W. W.Eaton etal., 1989; the years, this work focused on such measures as the
Ferdinand, de Nijs, van Lier, & Verhulst, 2005; Child Behavior Checklist (CBCL), and it has sub-
Fergusson, Horwood, & Lynskey, 1993; Kessler sequently been expanded to include assessment of
etal., 2005; Vaidynathan, Patrick, & Iacono, 2010; psychopathology across the adult lifespan as well.
Wadsworth et al., 2001), although this is beyond The Adult Self-Report (ASR) and Adult Behavior
the scope of the present chapter. Checklist (ABCL) measures has national norms for
It is presently unclear which of these mod- ages 18 to 59years, and the Older Adult Self-Report
elsthe unitary internalizing liability versus the (OASR) and Older Adult Behavior Checklist
higher-order, bifurcated internalizing liability (OABCL), which was designed for individuals
is superior, and they have rarely been compared aged 6090+ years. These measures include assess-
directly. Comparisons thus far have produced ment of broadband internalizing scores (a combi-
equivocal results by and large (e.g., Seeley et al., nation of depressive and anxious symptomatology),
2011), likely because even when distress and fear and norms provide benchmarks for comparison of
emerge as separable aspects of a broader internaliz- a given individual with the broader population.
ing domain, they are very highly correlated. It may Together, these measures and others compose the
well be the case that the superiority of one model or Achenbach System of Empirically Based Assessment
another will depend on the goals of the clinician or (ASEBA), which is available at http://www.aseba.
researcher. What does appear to be the case, though, org.
is that assessment of internalizing liability as a single A recently designed inventory, the Inventory
dimension is likely adequate for most clinical pur- of Depression and Anxiety Symptoms (IDAS)
poses at this time, particularly given the highly cor- also appears to hold significant promise in assess-
related nature of the distress and fear subliabilities. ing internalizing-related constructs (Watson et al.,
(If one were to assess subliabilities, it would appear 2007). The IDAS assesses the symptoms of major
that the distress subliability would be most germane depressive disorder as well as anxiety disorders.
to depressive disorder assessment.) Through factor-analytic results, the authors investi-
How can an interested clinician or researcher gated the structure of these symptoms, and the final
assess internalizing liability? Because this area of IDAS self-report instrument emerged. The IDAS
structural psychopathology research is relatively (1) measures a general factor (cf. internalizing),
new in many ways, there has been relatively little (2)has two broader scales, general depression, and
effort made toward creating assessment devices dysphoria, and (3)has 10 specific symptom scales
that can characterize internalizing liability well. to investigate particular manifestations of symp-
This stands in marked contrast to externalizing tomatology. These scales are suicidality, lassitude,
liabilitythe tendency to experience antisocial insomnia, appetite loss, appetite gain, ill temper,
personality disorder, conduct disorder, and sub- well-being, panic, social anxiety, and traumatic
stance abuse/dependencewhere efforts have been intrusions. These broad and fine-grained character-
made to develop appropriate assessment measures izations of internalizing symptomatology, coupled
to capture the full range of this liability dimen- with the instruments good psychometric proper-
sion (Markon & Krueger, 2005; Krueger, Marcon, ties, make the IDAS a worthwhile measure to con-
Patrick, Benning, & Kramer, 2007). Researchers sider for assessment of the depressive and related
are only now turning their attention to means by disorders.
which internalizing can be assessed. The picture is Another possible way to assess internalizing
complicated, however, by the fact that internalizing liability is to examine related scales in other mea-
liability has been identified as a latent construct in sures. One such possible measure is the MMPI-2.
multivariate statistical analyses. Thus, while these Two scales on the MMPI-2 appear to capture
investigations have demonstrated the presence of internalizing liability to varying degrees. First, the
internalizing liability, translating these findings PSY-5 scales, originating in the work of Harkness
into practical assessment strategies is another ques- (Harkness & McNulty, 1994; Harkness, McNulty, &
tion entirely. Ben-Porath, 1995), include a Negative Emotionality

Eaton, Krueger 75
Scale, which is closely linked to internalizing liabil- To summarize, a primary assessment concern
ity. In addition, the Restructured Clinical Scales with the depressive and related disorders is comor-
demoralization scale RCd (Tellegen et al., 2003) bidity. The frequent co-occurrence of these disor-
also seems related to internalizing liability. The RCd ders complicates assessment. Assessors conducting
was created to minimize the levels of overlap among differential diagnosis must determine how best
the standard clinical scales by removing common to map the presenting signs and symptoms into a
variance, which was labeled demoralization. Thus, DSM-IV-TR diagnosisor into multiple appropri-
RCd represents a negativistic tendency to endorse ate diagnoses. This may be the best approach for
items reflecting pathology, which may be related many clinical and research applications. If the asses-
to internalizing liability, although the strength sor is not bound by the DSM-IV-TR diagnostic enti-
of the association between internalizing liability ties, though, an alternative approach is to recognize
MMPI-2 Restructured Clinical Scale demoraliza- that their comorbidity seems to reflect a coherent
tion needs to be addressed analytically by subse- underlying liability dimension. This internalizing
quent research. Finally, one measure that appears to liability dimension then manifests as particular
tap into internalizing liability/negative emotional- signs and symptoms, perhaps due to environmental
ity is the Schedule for Nonadaptive and Adaptive influences. Although no gold-standard assessment
Personality (SNAP; Clark 1993). The SNAP, a mea- of the internalizing liability dimension has yet been
sure of personality strengths and weaknesses, has a agreed upon, various measures of closely related
higher-order three-factor structure of negative emo- constructs exist. Ultimately, the end purpose of
tionality, positive emotionality, and disinhibition. the assessment should direct the assessment proce-
Negative emotionality subsumes scales of negative dure and measures selected. Acomplete assessment
temperament, manipulativeness, mistrust, aggres- of the depressive disorders, however, would likely
sion, self-harm, eccentric perceptions, and depen- comprise assessing for the presence/absence of the
dency, and it appears likely to tap into key aspects DSM-IV-TR diagnostic constructs and also char-
of internalizing liability. acterizing a given individuals level of underlying
A final way to assess internalizing liability is the internalizing liability to experience various forms of
use of measures of normal (nonpathological) per- unipolar mood and anxiety psychopathology.
sonality. Recent research has demonstrated that
internalizing liability is correlated almost perfectly Subpopulations and presentation
with trait neuroticism (Griffith etal., 2010). Trait DSM-IV-TR defines the depressive disorders gen-
neuroticism (also referred to as negative emotional- erally in a one size fits all rubric. The polythetic cri-
ity, negative affectivity, or, when scores are reflected, terion system, where X of Y diagnostic criteria must
as emotional stability) plays a prime role in both be present to render a diagnosis, allows for some
Big Five and Giant Three personality measures, flexibility in disorder presentation; however, the
and it is, thus, easily assessed via widely available core constructs enshrined in the diagnostic manual
measures. For instance, neuroticism, extraversion, are relatively set. In terms of clinical assessment, if
conscientiousness, agreeableness, and openness a given disorder fits a given patients symptomatol-
to experience constitute the higher-order factors ogy well, this is a nonissue. If, on the other hand, a
of the Five-Factor Model of personality, instanti- patients presentation of depressive symptomatology
ated in the NEO-PI-R (Costa & McCrae, 1992). does not fit well into the defined disorderstypi-
Another normal personality model that could be cally requiring a not-otherwise-specified diagnosis
used to capture this aspect of internalizing liability problems arise. This is a particularly salient concern
is the Multidimensional Personality Questionnaire when it comes to changing psychopathological man-
(MPQ; Tellegen, 2000), which has higher-order ifestations across subpopulations. We now turn our
factors of negative emotionality, positive emotional- attention to two subpopulation-related presentation
ity, and constraint. Athird, and, historically, highly issues with significant implications for assessment
influential personality model was developed by of depressive disorders: (1) culture-bound syn-
Eysenck (1967, 1969, 1970; Eysenck & Eysenck, dromes that seem to reflect some depressive symp-
1991). Comprising factors of extraversion, neu- tomatology, and (2) how DSM-IV-TR diagnostic
roticism, and psychoticism, measures assessing the construct-related entities manifest differently across
Eysenckian personality model, and particularly the different population groups.
component of trait neuroticism, can likely be used We first turn our attention to culture-bound syn-
as a reasonable proxy for internalizing liability. dromes. DSM-IV-TR lists a variety of culture-bound

76 Depressive Disorders, Comorbidit y Issues, and Assessment Strategies

syndromes in its Appendix I.These forms of psycho- globally, such a review is beyond the scope of any
pathology are recurrent, locality-specific patterns chapter, particularly when subpopulations can be
of aberrant behavior and troubling experience that defined in so many ways. For instance, broad eth-
may or may not be linked to a particular DSM-IV nic groups, such as Hispanics/Latinos may show
diagnostic category (p. 898). (This is opposed to different disorder presentations from other groups;
the DSM-IV-TR primary diagnoses, which the doc- within ethnic groups, there may be meaningful dif-
ument states have presentations that can be found ferences as well, such as between Hispanics/Latinos
worldwide.) Some of the culture-bound syndromes in the United States versus South American versus
included in DSM-IV-TR show clear links to the the Caribbean. Thus, although our discussion up to
depressive disorders, and they highlight how being this point has focused mostly on the importance of
mindful of cultural considerations is critical when big picture assessment issues, and, in particular,
assessing these disorders. on how the very nosological definition of disor-
The culture-bound syndrome of brain fag, for ders impacts assessment and comorbidity, it is just
instance, can resemble depressive disorders. When as critical to consider assessment challenges at the
confronted with educational challenges, some high locallevel.
school or university students in West Africa may An example may be instructive to highlight the
report its symptoms, which are conceptualized role of subpopulations in disorder presentation
as relating to brain fatigue. Problems with think- and the resulting assessment concerns. In a review,
ing, concentration, and memory may be present. Leong, Okazaki, and Tak (2003) addressed the
Another culture-bound syndrome, ghost sickness, topic of assessment of depression in East Asia. There
resembles the depressive ideation seen in major is some evidence that depression in East Asian cul-
depressive disorder, dysthymic disorder, depressive tures may express differently than in, say, the United
personality disorder, and bereavement. In this disor- States. In particular, there is some debate about
der, primarily seen in Native American individuals, whether the psychobiological state that manifests as
individuals are often preoccupied with the deceased major depression in the United States may manifest
and death, and they may experience bad dreams, loss in a more somatic way in some East Asian cultures.
of appetite, weakness, fear, anxiety, confusion, and As the authors detail, however, these manifestations,
feelings of futility. Nervois, a form of distress seen in and proper assessment thereof, may differ nontrivi-
Latinos, can include emotional distress, sleep prob- ally between Chinese, Korean, Japanese, and other
lems, irritability, easy tearfulness, and inability to East Asian cultures. For instance, one study of three
concentrate. In China, the culture-bound syndrome groups of patients with depressive and related dis-
shenjing shuairuo presents with several symptoms orders(1) Koreans living in Seoul, (2) Korean
relating to the depressive disorders, including con- Chinese, and (3) Chinese individuals living in
centration difficulties, sleep disturbances, memory Yanbian, Chinashowed different manifestations
problems, sexual dysfunction, irritability, and feel- of these disorders (Kim, Li, & Kim, 1999). This is
ings of mental fatigue. notable for clinical assessment purposes, because pat-
Culture-bound syndromes can be considered terns of symptom endorsement on the same instru-
to be unique forms of psychopathology that mani- ment, a translated version of the Beck Depression
fest within a given subpopulation and/or locale, Inventory (BDI; Beck, Ward, Mendelson, Mock,
as opposed to the DSM-IV-TR primary disorders, & Erbaugh, 1961; Beck, Steer, & Brown, 1996;
which are defined as broadly applicable across Han etal., 1986, in Leong, Okazaki, & Tak, 2003),
subpopulations. However, is it necessarily the case differed across groups. The Korean patients expe-
that a disorder, such as major depressive disorder, rienced more affective symptoms, the Chinese
actually presents similarly across subpopulations? patients experienced more somatic symptoms, and
The answer appears to be no. Rather, the gen- the Korean Chinese patients were in between the
eral constructs enshrined in DSM-IV-TR, such as Korean and Chinese patient groups, experiencing
depression, may manifest differently across sub- a more mixed affective and somatic presentation.
populations. This causes notable challenges for This issue illustrates how good clinical assessment of
assessment of depressive and related disorders across the depressive and related disorders involves more
cultures. than simply selecting a reliable and well-validated
There is a large literature regarding cross-cultural instrument.
psychopathology, and we will not belabor this Ethnicity is not the only determiner of sub-
discussion. Given the breadth of subpopulations population characteristics, of course. Gender may

Eaton, Krueger 77
play a role as well. A classic example of gender diagnostic criteria contribute similarly to diagno-
differences in the presentation of the depressive sis. An exception can be found in the diagnostic
disorders is the role of rumination, with studies criteria for a major depressive episode, where two
consistently showing that womens depression tends criteria, depressed mood and anhedonia, are given
to manifest with more rumination compared to special status in that one or both must be present
mens depression, which tends to be associated for a diagnosis, making the presence of one or both
with more problem-solving-focused approaches necessary but not sufficient for diagnosis. Beyond
(Nolen-Hoeksema, 1987; Nolen-Hoeksema, Wisco, this caveat, however, all nine of the diagnostic cri-
& Lyubomirsky, 2008). Another critical consider- teria contribute equally toward determination of
ation for subpopulation-related assessment issues the presence of a major depressive episode. By this
is age. Just because a given assessment approach or polythetic approach, feelings of worthlessness and a
measure has been shown to be reliable and valid in depressive suicide attempt each count as one crite-
adults, this will not necessarily be the case in chil- rion present (toward the diagnostic benchmark of
dren or older adults (Balsis & Cully, 2008; Nezu five or more), but so do insomnia and psychomotor
etal., 2009). Although a clinician might choose to agitation. For diagnostic assessment that is wholly
assess for major depressive symptoms in adults by compliant with DSM-IV-TR, each criterion must
using the BDI, she might be better served by using thus be assessed with similar rigor. For clinical pur-
the Childrens Depression Inventory (Kovacs, 1992) poses, however, given limited resources and time,
in a youth sample or the Geriatric Depression Scale it may not always be possible or even preferable to
(Yesavage etal., 1983) in a sample of older adults. take such an approach. Characterizations of worth-
These can be thought of as cohort differences in a less feelings or suicidality often do (and likely often
cross-sectional assessment sense, an interindividual should) take precedence over characterizations of
difference in manifestation based on age. However, sleep disturbances or psychomotor changes. In this
this sort of assessment concern can also be viewed way, the clinician must be sensitive to the goals of
in a longitudinal sense as an intra-individual dif- the assessment and the determination of how criti-
ference in manifestation based on age. The latter cal it is that each criterion receives the same degree
scenario would be an example of heterotypic conti- of consideration given their equal weight.
nuity (Caspi & Bem, 1990; Kagan, 1969), where the Second, the current polythetic criterion system
depressive/internalizing liability within an individ- includes no possibility of differential item function-
ual would remain relatively constant over time but ing, a characteristic of test items in item response
manifest differently as the individual ages. So, while theory commonly referred to as DIF (Lord &
assessment must be concerned with subpopulation Novick, 1968). As noted in the preceding para-
members of individuals being assessed, one must graph, DSM-IV-TR allows each criterion to count
also be concerned with a given individuals transition invariably toward one unit of psychopathology up
between subpopulation groups across the lifespan. to the diagnostic benchmark, even though one cri-
terion might tap into the construct being assessed
Psychometricissues more strongly than another. Not only might crite-
Certainly, psychometric issues are among the ria relate differently to the construct, their relations
most pressing considerations when conducting to the construct might change across individuals or
assessment of depressive disorders. As noted repeat- over time. For instance, adolescents often experience
edly earlier, the reliability (including test-retest and normal sleep duration or hypersomnia, whereas
internal consistency) of assessment devices and pro- older adults often require and get less sleep nightly,
cedures, and the validity of the inferences drawn which can reflect natural aging processes or insom-
from these assessments, are crucial to good assess- nia (e.g., Morin, 1993). In terms of how the sleep
ment. Given the widespread appreciation of these disturbance criterion of a major depressive episode
psychometric issues, we will not discuss them here relates to the construct of interest, it may be the
but, instead, will refer readers to any number of case that insomnia is indicative of a more significant
canonical texts on this issue (e.g., Crocker & Algina, level of depressive episode severity if it is present in a
2006; Lord & Novick, 1968). Instead, we will note 15-year-old boy than a 90-year-old woman. This is a
briefly two related issues that are less frequently dis- hypothetical example of DIF:The insomnia criteria
cussed, particularly in applied clinical settings. taps more strongly into, and is more indicative of,
First, the way that depressive disorders are depressivity in the younger individual than the older
defined by DSM-IV-TR suggests generally that individual. One quality of good clinical assessment

78 Depressive Disorders, Comorbidit y Issues, and Assessment Strategies

of depressive disorders is mindfulness to such DIF postpartum onset. Stressful life events precipitating
issues, investigation of other potential origins of the disorder manifestation can also suggest the possibil-
criterions presence, and careful application of clini- ity of a nondepressive disorder, such as adjustment
cal judgment to ameliorate the possibility of DIF disorder. Third, subcultural, demographic, indi-
impacting diagnoses across individuals. vidual difference, and physical health variables must
be investigated, given that they can have significant
Summary impacts on the presenting symptoms. Finally, a
We have briefly detailed some key considerations careful psychosocial history is necessary to under-
present when assessing the depressive disorders. stand the ebbs and flows of mental health over an
Primary among these are definitional issues, which individuals life and to appreciate manifestation pat-
are associated with myriad assessment difficulties. terns (e.g., seasonal affect disturbances, dysthymia
Comorbidity is rampant, both within the depressive versus a major depressive episode).
disorders and with related forms of psychopathol- Assessment of all these issues can be an oner-
ogy, such as anxiety disorders. To some extent, this ous proposition, raising the question of what the
reflects the current classification systems approach; responsible clinician can feasibly do to gather rel-
however, it also seems to reflect a meaningful under- evant information. This is particularly true given
lying liability to experience internalizing disorders. that insight in depressed individuals may be biased
In addition, key depressive disorders may manifest by negativistic thinking, and this draws into ques-
differently across subpopulations, such as ethnic, tion the accuracy of self-report data from patients.
gender, and age cohort groups. Similarly, some disor- For example, although a nondepressed person
ders with depressive features may be culture bound, might perceive and report relatively few stressful
thus occurring with frequency only in certain cul- life events, a depressed individual may be more
tural groups and locales. These diversity issues com- oriented to perceive and report them. Outside of
plicate clinical assessment of depressive disorders, insight-related issues, retrospective report can be
potentially attenuating the reliability and validity biased in general (Moffitt etal.,2010).
of a given assessment measure or protocol that per- What is a clinician to do? The answer appears to
forms well in other groups. Symptom endorsement be that clinicians should conduct detailed, multi-
may change across subpopulations, and the proper- modal, multisource assessments. Thus, the clinician
ties of the diagnostic criteria (and related assessment can assess self-reported symptoms via question-
items) themselves may change. These issues clarify naire and interview methods that were broad in
how good clinical assessment requires consideration scope and focused on multiple disorder possibili-
of a multitude of factors beyond simple psychomet- ties. Behavioral observations made by the clini-
rics properties of assessment measures. cian, such as assessment of psychomotor slowing,
attention and concentration problems, and so on,
General Assessment Implications are critical. Technology can assist with such assess-
As we have seen, assessment of depressive disor- ments as well. For example, clinicians can consider
ders is complicated by a variety of factors, such as using recording devices for in vivo monitoring, hav-
imprecise and arbitrary classification, definitional ing the patient wear an actigraphy device to assess
overlap with other constructs, differential disorder diurnal behavioral activation and nocturnal sleep
manifestation across subpopulations, and high rates disturbances, and requesting the patient complete
of comorbidity. The primary implication of these ecological momentary assessments throughout the
considerations is that there is no single, circum- day on a cell phone or tablet computer to gauge
scribed method or measure that can fully character- intraday mood lability.
ize depressive disorder presentation. Rather, a full Finally, information source is a critical con-
assessment of depressive disorders will require sev- sideration. A significant implication of the com-
eral components. First, nondepressive, but related, plexity of depressive disorder assessment is that
disorders will need to be assessed to determine patient self-report is necessary but not sufficient.
comorbidity patterns and to rule out particular Clinician-report data are also key. That said, a third,
diagnoses. Second, life events need to be examined, and often overlooked, data source is the informant
because certain occurrences can influence the way report. Partners, friends, children, parents, cowork-
these disorders are conceptualized and diagnosed. ers, and other individuals can provide crucial sup-
For instance, giving birth prior to the onset of a plementary information to help understand the
major depressive episode can be indicative of a symptomatology and history of depressive patients.

Eaton, Krueger 79
Although contacting these informants can be a time It is important to expand our assessment
consuming endeavor on the part of the clinician, armamentarium to include informant-report
it can provide an excellent and beneficial outside information more commonly. It may well be
perspective of the patients functioning. One means that particular aspects of the depressive disorders
of collecting such information without a signifi- (e.g.,psychomotor slowing) may be more amenable
cant time investment for the clinician is to send to assessment by informant-report whereas other
informant-report questionnaires by mail to infor- aspects (e.g., feelings of hopelessness, suicidal
mants or to send them home with the patient for ideation) are more amenable to assessment by
distribution to appropriate informants. self-report. Moving beyond interviews and
questionnaires will likely also be key to facilitate
Future Directions
improvements in depressive disorder assessment
We now turn our attention to several unresolved
long-term. Several approaches seem likely candidates
issues in the assessment of depressive disorders that
to improve assessment, including molecular genetics,
warrant future research.
psychophysiological responses (e.g., event-related
1. What is the role of internalizing liability in potentials), and structural and functional
assessment? Perhaps the most critical question is neuroimaging. The integration of such objective
how, and when, one should assess internalizing measures with subjective reports from patients and
liability as part of, or in place of, a standard their informants may improve assessment broadly
DSM-IV-TR diagnostic assessment. Because and also serve to integrate psychopathology research
a growing number of studies indicates that with more foundational basic science research in
internalizing liability plays a key role in the etiology, psychology and neuroscience.
maintenance, and external correlates of depressive 4. What is the role of empirically supported
disorders, it seems crucial for future studies to assessment? Within clinical psychology over the past
determine (a)how to assess this liability optimally several decades, there has been a marked focus on
and (b)when this liability should be assessed. It the identification and application of empirically
may well be the case that internalizing liability supported treatments (Nathan & Gorman,
accounts for the meaningful core of most depressive 2007). Given that intervention is underpinned
psychopathology, and that disorder-specific, unique by assessment, there has been a separate, but
variation contributes relatively little to issues of somewhat quieter, call for empirically supported
clinical importance (e.g., suicidality). assessments. Indeed, it seems that empirically
2. How should depressive disorders be classified? As supported treatments and assessments (and, likely,
we discussed earlier, nosology necessarily informs classification) will form the basis of the emerging
clinical assessment, but assessment concerns have clinical science movement. Empirically supported
historically had a major impact on nosology as assessment is not only a question of determining
well. As the field moves toward the finalization whether a measure or procedure has adequate
and publication of DSM-5, the changes to the psychometric properties. It will also involve direct
depressive disorders, and the resulting changes comparison of assessment approaches and may
in their assessment, will be an important area of well lead to the determination of gold standard
study. First, it will be necessary to determine how assessments for different symptom-presentation
the diagnostic criteria for the depressive disorders patterns, subpopulations, and so on. Although
function. Do they seem to indicate coherent there have been notable efforts in this domain
latent constructs or are they heterogeneous (Hunsley & Mash, 2008; Joiner, Walker, Pettit,
and multidimensional? Do they show varying Perez, & Cukrowicz, 2005; Klein, Dougherty, &
endorsement rates across subpopulations? Second, Olino, 2005; Nezu etal., 2000), more work needs
it will be important to determine how best to to be done to establish an empirically supported
assess these criteria. What items are necessary to assessment battery and procedure for depressive
capture the diagnostic criteria? How do these items disorders.
function, particularly in a DIFsense?
3. What are the alternative methods for assessment References
Achenbach, T. M., & Edelbrock, C. S. (1978). The classification
of depressive disorders? The majority of clinical
of child psychopathology:Areview and analysis of empirical
assessment is done via diagnostic interview and/ efforts. Psychological Bulletin, 85(6), 12751301.
or questionnaires, and most of the information Achenbach, T. M., & Edelbrock, C. S. (1984). Psychopathology
gathered reflects solely self-report information. of childhood. Annual Review of Psychology, 35, 227256.

80 Depressive Disorders, Comorbidit y Issues, and Assessment Strategies

American Psychiatric Association. (1980). Diagnostic and statisti- Eaton, W. W., McCutcheon, A., Dryman, A., & Sorenson,
cal manual of mental disorders (3rd ed.). Washington, DC: A. (1989). Latent class analysis of anxiety and depression.
Author. Sociological Methods & Research, 18(1), 104125.
American Psychiatric Association. (2000). Diagnostic and statistical Eysenck, H. (1967). The biological basis of personality. Springfield,
manual of mental disorders (4th ed.). Washington, DC: Author. IL:Thomas.
Andrews, G., Goldberg, D. P., Krueger, R. F., Carpenter, W. Eysenck, H. (1969). Personality structure and measurement.
T., Jr., Hyman, S. E., Sachdev, P., & Pine, D. S. (2009). London:Routledge.
Exploring the feasibility of a meta-structure for DSM-V and Eysenck, H. (1970). The structure of human personality. London:
ICD-11:Could it improve utility and validity? Psychological Mathuen.
Medicine, 39, 19932000. Eysenck, H., & Eysenck, S. B.G. (1991). The Eysenck Personality
Baer, L., Jacobs, D. G., Meszler-Reizes, J., Blais, M., Fava, M., Questionnaire-Revised. Sevenoaks, United Kingdom:Hodder
Kessler, R.,...OLaughlen, J. (2000). Development of a and Stoughton.
brief screening instrument. The HANDS. Psychotherapy and Fairburn, C. G., & Bohn, K. (2005). Eating disorder NOS
Psychosomatics, 69,3541. (EDNOS): An example of the troublesome not otherwise
Balsis, S., & Cully, J. A. (2008). Comparing depression diag- specified (NOS) category in DSM-IV. Behaviour Research
nostic symptoms across younger and older adults. Aging & and Therapy, 43(6), 691701.
Mental Health, 12(6), 800806. Ferdinand, R. F., de Nijs, P. F.A., van Lier, P., & Verhulst, F. C.
Barlow, D. H., Farchione, T. J., Fairholme, C. P., Ellard, K. (2005). Latent class analysis of anxiety and depressive symp-
K., Boisseau, C. L., Allen, L. B., & Ehrenreich-May, J. toms in referred adolescents. Journal of Affective Disorders,
(2011). Unified protocol for transdiagnostic treatment of 88, 299306.
emotional disorders: Therapist guide. New York: Oxford Fergusson, D. M., Horwood, L. J., & Boden, J. M. (2006).
UniversityPress. Structure of internalizing symptoms in early adulthood. The
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). BDI-II manual. British Journal of Psychiatry, 189, 540546.
San Antonio, TX:The Psychological Corporation. Fergusson, D. M., Horwood, L. J., & Lynskey, M. T. (1993).
Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, Prevalence and comorbidity of DSM-III-R diagnoses in a
J. (1961). An inventory for measuring depression. Archives of birth cohort of 15year olds. Journal of the American Academy
General Psychiatry, 4, 561571. of Child & Adolescent Psychiatry, 32(6), 11271134.
Caspi, A., & Bem, D. J. (1990). Personality continuity and Goldberg, D., Simms, L. J., Gater, R., & Krueger, R. F. (2011).
change across the life course. In L. A.Pervin (Ed.), Handbook Integration of dimensional spectra for depression and anxiety
of personality:Theory and research (pp. 549569). NewYork: into categorical diagnoses for general medical practice. In D.
The GuilfordPress. A.Regier, W. E.Narrow, E. A.Kuhl, & D. J.Kupfer (Eds.)
Clark, L. A. (1993). Schedule for Nonadaptive and Adaptive Personality The conceptual evolution of DSM-5 (pp. 1935). Arlington,
(SNAP):Manual for administration, scoring, and interpretation. VA:American Psychiatric Publishing.
Minneapolis, MN:University of Minnesota PressS. Griffith, J. W., Zinbarg, R. E., Craske, M. G., Mineka, S., Rose,
Clark, L. A., Livesley, W. J., & Morey, L. (1997). Personality dis- R. D., Waters, A. M., & Sutton, J. M. (2010). Neuroticism
order assessment:The challenge of construct validity. Journal as a common dimension in the internalizing disorders.
of Personality Disorders, 16, 438453. Psychological Medicine, 40, 11251136.
Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality Han, H. M., Yeom, T. H., Shin, Y. W., Kim, K. H., Yoon, D.
Inventory (NEO-PI-R) and NEO Five-Factor Inventory J., & Chung, K. J. (1986). A standardization study of Becks
(NEO-FFI) professional manual. Odessa, FL: Psychological Depression Inventory in Korea [in Korean]. Neuropsychiatry
Assessment Resources. and Medicine (Seoul), 25, 458502.
Crocker, L., & Algina, J. (2006). Introduction to classical and Harkness, A. R., & McNulty, J. L. (1994). The Personality
modern test theory. Pacific Grove, CA:Wadsworth. Psychopathology Five (PSY-5): Issue from the pages of a
Eaton, N. R., Keyes, K. M., Krueger, R. F., Balsis, S., Skodol, A. diagnostic manual instead of a dictionary. In S. Strack and
E., Markon, K. E., Grant, B. F., & Hasin, D. S. (2012). An M. Lorr (Eds.) Differentiating normal and abnormal personal-
invariant dimensional liability model of gender differences in ity (pp. 291315). NewYork:Springer.
mental disorder prevalence:Evidence from a national sam- Harkness, A. R., McNulty, J. L., & Ben-Porath, Y. S.
ple. Journal of Abnormal Psychology, 121(1), 282288. (1995). The personality Psychopathology Five (PSY-5):
Eaton, N. R., Krueger, R. F., Keyes, K. M., Skodol, A. E., Constructs and MMPI-2 scales. Psychological Assessment,
Markon, K. E., Grant, B. F., & Hasin, D. S. (2011). 7, 104114.
Borderline personality disorder co-morbidity:Relationship Hettema, J. M., Neale, M. C., Myers, J. M., Prescott, C. A.,
to the internalizing-externalizing structure of com- & Kendler, K. S. (2006). A population-based twin study of
mon mental disorders. Psychological Medicine, 41(5), the relationship between neuroticism and internalizing disor-
10411050. ders. American Journal of Psychiatry, 163(5), 857864.
Eaton, N. R., Krueger, R. F., & Oltmanns, T. F. (2011). The Hunsley, J., & Mash, E. J. (2008). A guide to assessments that
structure and long-term stability of the internalizing spec- work. NewYork:Oxford UniversityPress.
trum of personality and psychopathology across the lifespan. Hyman, S. E. (2010). The diagnosis of mental disorders: The
Psychology and Aging, 26(4), 987993. problem of reification. Annual Review of Clinical Psychology,
Eaton, N. R., South, S. C., & Krueger, R. F. (2010). The mean- 6, 155179.
ing of comorbidity among common mental disorders. In T. Joiner, T. E., Jr., Walker, R. L., Pettit, J. W., Perez, M., &
Millon, R. Krueger, & E. Simonsen (Eds.), Contemporary direc- Cukrowicz, K. C. (2005). Evidence-based assessment
tions in psychopathology: Scientific foundations of the DSM-V of depression in adults. Psychological Assessment, 17(3),
and ICD-11 (2nd ed., pp. 223241). NewYork:Guilford. 267277.

Eaton, Krueger 81
Kagan, J. (1969). The three faces of continuity in human common mental disorders? Evidence that lifetime prevalence
development. In D. A.Goslin (Ed.), Handbook of social- rates are doubled by prospective versus retrospective ascer-
ization theory and research (pp. 5365). Chicago: Rand tainment. Psychological Medicine, 40, 899909.
McNally. Morin, C. M. (1993). Insomnia: Psychological assessment and
Kendler, K. S., Aggen, S. H., Knudsen, G. P., Rysamb, E., management. NewYork:Guilford.
Neale, M. C., Reichborn-Kjennerud, T. (2011). The struc- Nathan, P. E., & Gorman, J. M. (2007). A guide to treatments
ture of genetic and environmental risk factors for syndro- that work (3rd ed.). NewYork:Oxford UniversityPress.
mal and subsyndromal common DSM-IV axis Iand all axis Nezu, A. M., Nezu, C. M., Friedman, J., & Lee, M. (2009).
II disorders. American Journal of Psychiatry, 168,2939. Assessment of depression. In I. H.Gotlib and C. L.Hammen
Kendler, K. S., Prescott, C. A., Myers, J., & Neale, M. C. (2003). (Eds.), Handbook of depression (2nd ed., pp. 4468).
The structure of genetic and environmental risk factors for NewYork:Guilford.
common psychiatric and substance use disorders in men and Nezu, A. M., Ronan, G. F., Meadows, E. A., & McClure, K.
women. Archives of General Psychiatry, 60, 929937. S. (2000). Practitioners guide to empirically based measures of
Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. (2005). depression. NewYork:Kluwer Academic/Plenum.
Prevalence, severity, and comorbidity of 12-month DSM-IV Nolen-Hoeksema, S. (1987). Sex differences in unipolar
disorders in the National Comorbidity Survey Replication. depression: Evidence and theory. Psychological Bulletin, 2,
Archives of General Psychiatry, 62, 617627. 259282.
Kessler, R. C., Ormel, J., Petukhova, M., McLaughlin, K. Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008).
A., Green, J. G., Russo, L. J.,...Ustun, T. B. (2011). Rethinking rumination. Perspectives on Psychological Science,
Development of lifetime comorbidity in the World Health 3(5), 400424.
Organization World Mental Health Surveys. Archives of Pagan, J. L., Oltmanns, T. F., Whitmore, M. J., & Turkheimer,
General Psychiatry, 68, 90100. E. (2005). Personality disorder not otherwise speci-
Kim, K. I., Li, D., & Kim, D.-H. (1999). Depressive symptoms fied:Searching for an empirically based diagnostic threshold.
in Koreans, Korean Chinese, and Chinese: A transcultural Journal of Personality Disorders, 19(6), 674689.
study. Transcultural Psychiatry, 36, 303316. Regier, D. A., Narrow, W. E., Kuhl, E. A., & Kupfer, D. J.
Klein, D. N., Dougherty, L. R., & Olino, T. M. (2005). Toward (Eds.). (2011). The conceptual evolution of DSM-5. Arlington,
guidelines for evidence-based assessment of depression VA:American Psychiatric Publishing.
in children and adolescents. Journal of Clinical Child and Sanislow, C. A., Grilo, C. M., Morey, L. C., Bender, D. S.,
Adolescent Psychology, 34(3), 412432. Skodol, A. E., Gunderson, J. G.,...McGlashan, T. H.
Kovacs, M. (1992). Childrens Depression Inventory manual. (2002). Confirmatory factor analysis of DSM-IV crite-
North Tonawanda, NY:Multi-Health Systems. ria for borderline personality disorder: Findings from the
Krueger, R. F. (1999). The structure of common mental disor- Collaborative Longitudinal Personality Disorders Study.
ders. Archives of General Psychiatry, 56, 921926. American Journal of Psychiatry, 159, 284290.
Krueger, R. F., Caspi, A., Moffitt, T. E., & Silva, P. A. (1998). Santor, D. A., Gregus, M., & Welch, A. (2006). Eight decades of
The structure and stability of common mental disorders measurement in depression. Measurement, 4, 135155.
(DSM-III-R):Alongitudinal-epidemiological study. Journal Seeley, J. R., Kosty, D. B., Farmer, R. F., & Lewinsohn, P. M.
of Abnormal Psychology, 108, 216227. (2011). The modeling of internalizing disorders on the
Krueger, R. F., & Markon, K. E. (2006). Reinterpreting comor- basis of patterns of lifetime comorbidity:Associations with
bidity:Amodel-based approach to understanding and classi- psychosocial functioning and psychiatric disorders among
fying psychopathology. Annual Review of Clinical Psychology, first-degree relatives. Journal of Abnormal Psychology, 120(2),
2, 111133. 308321.
Krueger, R. F., Markon, K. E., Patrick, C. J., Benning, S. D., Slade, T., & Watson, D. (2006). The structure of com-
& Kramer, M. D. (2007). Linking antisocial behavior, mon DSM-IV and ICD-10 mental disorders in the
substance use, and personality: An integrative quantita- Australian general population. Psychological Medicine, 36,
tive model of the adult externalizing spectrum. Journal of 15931600.
Abnormal Psychology, 116(4), 645666. Tellegen A. (2000). Manual of the Multidimensional Personality
Lenzenweger, M. F., Loranger, A. W., Korfine, L., & Neff, C. Questionnaire. Minneapolis, MN: University of Minnesota
(1997). Detecting personality disorders in a nonclinical pop- Press.
ulation:Application of a 2-stage procedure for case identifi- Tellegen, A., Ben-Porath, Y. S., McNulty, J. L., Arbisi, P. A.,
cation. Archives of General Psychiatry, 54, 345351. Graham, J. R., & Kaemmer, B. (2003). The MMPI-2
Leong, F. T.L., Okazaki, S., Tak, J. (2003). Assessment of depres- Restructured Clinical (RC) Scales: Development, valida-
sion and anxiety in East Asia. Psychological Assessment, 15(3), tion, and interpretation. Minneapolis, MN: University of
290305. MinnesotaPress.
Lord, F. M. & Novick, M. R. (1968). Statistical theories of mental test Vaidynathan, U., Patrick, C. J., & Iacono, W. G. (2010). Patterns
scores. Reading, MA:Addison-Welsley Publishing Company. of comorbidity among mental disorders:Aperson-centered
Markon, K. E., & Krueger, R. F. (2005). Categorical and continuous approach. Comprehensive Psychiatry, 52(5), 527535.
models of liability to externalizing disorders:Adirect compari- Vollebergh, W. A.M., Iedema, J., Bijl, R. V., de Graaf, R., Smit,
son in NESARC. Archives of General Psychiatry, 62, 13521359. F., & Ormel, J. (2001). The structure and stability of com-
Mineka, S. Watson, D., & Clark, L. A. (1998). Comorbidity mon mental disorders: The NEMESIS study. Archives of
of anxiety and unipolar mood disorders. Annual Review of General Psychiatry, 58, 597603.
Psychology, 49, 377412. Wadsworth, M. E., Hudziak, J. J., Heath, A. C., & Achenbach, T.
Moffitt, T. E., Caspi, A., Taylor, A., Kokaua, J., Milne, B. J., M. (2001). Latent class analysis of Child Behavior Checklist
Polanczyk, G., & Poulton, R. (2010). How common are anxiety/depression in children and adolescents. Journal of the

82 Depressive Disorders, Comorbidit y Issues, and Assessment Strategies

American Academy of Child & Adolescent Psychiatry, 40(1), (2007). Development and validation of the Inventory of
106114. Depression and Anxiety Symptoms (IDAS). Psychological
Watson, D. (2005). Rethinking the mood and anxiety disor- Assessment, 19(3), 253268.
ders:Aquantitative hierarchical model for DSM-V. Journal Yesavage, J. A., Brink, T. L., Rose, T. L., Lum, O., Huang, V.,
of Abnormal Psychology, 114(4), 522536. Adey, M., & Leirer, V. O. (1983). Development and valida-
Watson, D. (2009). Differentiating the mood and anxiety dis- tion of a geriatric depression screening scale:Apreliminary
orders: A quadripartite model. Annual Review of Clinical report. Journal of Psychiatric Research, 17,3749.
Psychology, 5, 221247. Zisook, S., & Kendler, K. S. (2007). Is bereavement-related
Watson, D., OHara, M. W., Simms, L. J., Kotov, R., depression different than non-bereavement-related depres-
Chmielewski, M., McDade-Montez, E. A.,...Stuart, S. sion? Psychological Medicine, 37(6), 779794.

Eaton, Krueger 83

6 Depression and Comorbidity with

Panic Disorder

Natalie Castriotta and Michelle G.Craske

Comorbidity between panic disorder and major depression is found in the majority of individuals
with panic disorder and a substantial minority of individuals with major depression. Comorbidity
between panic disorder and depression is associated with substantially more severe symptoms
of each of the disorders, greater persistence of each disorder, more frequent hospitalization and
help-seeking behavior, more severe occupational impacts, and a significantly higher rate of suicide
attempts. These two disorders share many risk factors, such as neuroticism, exposure to childhood
abuse, informational processing biases, and elevated amygdala activation in response to negative facial
expressions. Research on the temporal priority of panic disorder and major depression has most
frequently found that panic attacks and other symptoms of anxiety predate the onset of the first
major depressive episode, but the first depressive episode predates the onset of full panic disorder.
Treatment studies indicate that cognitive behavioral therapy (CBT) is the most effective treatment
for panic disorder. Other forms of treatment include medication, particularly selective serotonin
reuptake inhibitors. Comorbid depression does not appear to affect the outcome of CBT for a
principal diagnosis of panic disorder, and CBT for panic disorder has positive, yet limited, effects on
symptoms of depression.
Key Words: panic disorder, panic attacks, major depression, comorbidity, cognitive behavioral therapy

Introduction to Specific ComorbidityIssues when compared to either panic disorder or depres-

Comorbidity between panic disorder and depres- sion alone. Theoretical debate has often centered on
sion is one of the strongest psychiatric comorbidi- whether these two highly comorbid disorders rep-
ties and the most frequent form of anxiety and resent one severe condition (Gorman & Coplan,
mood disorder comorbidity in both treatment 1996; Noyes etal., 1990; Roy-Byrne etal., 1992),
samples (Clayton, 1990) and general population or two co-occurring disorders with separate etiolog-
studies (Merikangas et al., 1996). Comorbidity ical underpinnings (Dolan, Bench, Brown, Scott, &
between these two disorders has been consistently Frackowiak, 1994; Nordahl etal., 1990). Research
associated with greater symptom severity and devoted to disentangling these two theories has
impairment (Brown, Schulberg, & Shear et al., focused on temporal onset of the disorders, simi-
1996), disability (Hegel etal., 2005), more suicide larities and differences in neurobiological substrates,
attempts (Johnson, Weissman, & Klerman, 1990), heritability, and environmental stressors associated
and poorer treatment response (Brown, Schulberg, with each disorder. As the co-occurrence of these
Madonia, Shear, & Houck, 1996; Brown, two disorders leads to greater symptom severity and
Schulberg, & Shear 1996; Walker et al., 2000) persistence, impairment, and higher risk of suicidal

and harmful behaviors, it is imperative for clinicians distress or impairment in important areas of func-
to assess for the presence of comorbidity given the tioning (American Psychiatric Association,2000).
significant implications for the direction and course Major depressive disorder in DSM-IV-TR
of treatment. (American Psychiatric Association, 2000) is defined
by the presence of either depressed mood or anhe-
Definitions of the Disorders donia, or both, for two weeks or more. Anhedonia
and of Comorbidity is defined as a markedly diminished interest or plea-
Panic attacks are discrete episodes characterized sure in all or almost all activities. If one or both of
by an abrupt surge of intense fear or discomfort that these symptoms is present, other symptoms assessed
reaches a peak within minutes and during which include a significant weight loss or gain (> 5% body
time four or more (of 13) physical and cognitive weight) or an increase or decrease in appetite, signifi-
symptoms occur, such as rapid heart rate, shortness cant changes in sleeping patterns such as insomnia or
of breath, and fears of losing control or dying, as hypersomnia, psychomotor agitation or retardation,
listed in the panic attacks checklist in the Diagnostic fatigue or loss of energy, feelings of worthlessness or
and Statistical Manual of Mental Disorders (fourth inappropriate guilt, changes in the ability to concen-
edition, text revision [DSM-IV-TR]; American trate or make decisions, and recurrent thoughts of
Psychiatric Association, 2000). Panic attacks are death or suicide. Atotal of at least five of these symp-
distinguished from other forms of anxiety by their toms must be present for two weeks or longer for a
sudden onset and rapid increase in arousal, as diagnosis of major depressive disorder to begiven.
opposed to a slow and gradual building of anxious In the literature, comorbidity between panic dis-
arousal. Panic attacks that occur within panic disor- order and depression has been defined in multiple
der are often described by patients as unexpected, ways. Concurrent comorbidity is when both disor-
such that they occur out of the blue or with no ders co-occur at the same time, either presently or
apparent trigger. The diagnosis of panic disorder in the past. Lifetime comorbidity is when an indi-
is given when recurrent unexpected panic attacks vidual has been given a diagnosis of both disorders
are accompanied by at least one month of persis- at some point in his or her lifetime, such that the
tent concern about future attacks and their conse- disorders may have occurred concurrently, may
quences and/or significant maladaptive changes are have overlapped at certain points in time, or may
made to behavior as a consequence of the attacks have occurred at distinctly different time points.
(American Psychiatric Association,2000). Panic attacks that occur in the absence of a diag-
Panic disorder is often, but not always, accom- nosis of panic disorder are often thought to be a
panied by agoraphobia. In DSM-IV-TR (American nonspecific severity marker of psychopathology in
Psychiatric Association, 2000), panic disorder and general (Kircanski, Craske, Epstein, & Wittchen,
agoraphobia are linked and a diagnosis of agora- 2009). As a result of this status, it is unclear
phobia is defined as a specific subset of diagnoses whether co-occurrence of panic attacks and depres-
of panic disorder. However, in the fifth edition of sion should be classified as a comorbidity, and it
the Diagnostic and Statistical Manual of Mental is also unclear if it is really panic disorder that is
Disorders (American Psychiatric Association, 2013), comorbid with depression or if it is actually the
panic disorder and agoraphobia are two distinct dis- presence of panic attacks that explain the comorbid
orders that can occur independently of one another. relationship.
In DSM-IV-TR, agoraphobia is defined by at least
six months of marked fear or anxiety about at least Prevalence, Impact, and
two public or crowded environments, such as driv- Shared Risk Factors
ing or being outside of the home alone, at malls, at Prevalence
sporting events, or in open spaces. The individual Prevalence of comorbidity between panic disor-
fears and/or avoids these situations because escape der and depression is consistently high, although
might be difficult or help might not be available in variable, in epidemiological, clinical, and pri-
the event of incapacitation or panic-like symptoms. mary care studies, with depression being present
The agoraphobic situations are either avoided, in a majority of individuals with panic disorder
require the presence of a companion, or are endured and panic disorder being present in a substan-
with great anxiety. Additionally, the fear associated tial minority of individuals with depression. One
with these situations must be disproportionate to of the largest epidemiological studies to date, the
the actual risk of danger and must cause significant National Comorbidity Survey (NCS), surveyed

Castriot ta,Craske 85
lifetime rates of Diagnostic and Statistical Manual of current anxiety and mood disorders in a clinical
Mental Disorders (third edition, revised [DSM-III]; sample of 1,127 outpatients. They found that 33%
American Psychiatric Association, 1980) disorders of patients with a current diagnosis of panic disorder
across 8,098 respondents in the United States with with agoraphobia had a current diagnosis of major
an 82.4% response rate. In the replication of this depressive disorder and that 24% of patients with a
study, 12-month prevalence rates were 2.8% for current diagnosis of major depressive disorder had
panic disorder and 6.7% for major depressive dis- a current diagnosis of panic disorder with agora-
order (Kessler, Chiu, Demler, & Walters, 2005). phobia. It is important to note that panic disorder
In the original NCS study, lifetime prevalence without agoraphobia was not significantly related
rates were 7.2% for panic attacks, 3.4% for panic to greater comorbidity with depression, suggesting
disorder, and 16.9% for major depressive episode. that the interference from agoraphobia is a predom-
About half of participants with lifetime panic attack inant contributor to high comorbidity rates. This
(50.9%) and panic disorder (55.6%) also met life- finding makes clinical sense as more restrictions on
time criteria for depression, and about one fifth an individuals mobility and the interference caused
(21.6%) of the individuals with lifetime depres- by avoidance of necessary daily activities may
sion reported a lifetime history of panic attacks and contribute to the development of depression, or
one tenth (11.2%) met criteria for lifetime panic may increase the severity of depressive symptoms,
disorder (Kessler et al., 1998). However, lifetime through marked impacts on the individuals quality
prevalence rates should be interpreted with caution of life. Similarly, Biederman and colleagues (2005)
as they are often less reliable than 12-month preva- found that in a sample of 1,031 referred clinical
lence rates and cumulative riskrates. patients, 58% of patients with current panic dis-
The odds ratios for lifetime panic with depres- order had current comorbid major depression, and
sion were found to be significant and substantial; 12.5% of those with current major depression had
the ratio for panic attack with depression was 6.2, comorbid panic disorder. This study did not assess
and the ratio for panic disorder with depression differences between panic disorder with agorapho-
was 6.8. In other words, it is 6.2 times more likely bia and without agoraphobia on comorbidityrates.
that an individual who experiences panic attacks A review of clinical studies (Katon & Roy-Byrne,
will experience a major depressive episode than an 1991) found that approximately 25% of currently
individual who does not experience panic attacks, depressed patients have a lifetime history of panic
and an individual with panic disorder is 6.8 times disorder, and 40% to 80% of patients with current
more likely to experience a major depressive epi- panic disorder have experienced a major depressive
sode than an individual without panic disorder. episode in their lifetime.
Additionally, significant episode comorbidity, or In sum, comorbidity between panic disorder and
the co-occurrence of multiple and discrete episodes major depression is found in the majority of indi-
of major depressive disorder and panic disorder at viduals with panic disorder and a substantial minor-
different points throughout the lifetime, was found ity of individuals with major depression. Rates of
among those with a lifetime history of both disor- comorbidity between panic attacks and depression
ders. This finding indicates that the onset, persis- are even higher than between full panic disorder
tence, and recurrence of one disorder are associated and depression. These findings have been consistent
with the onset, persistence, and recurrence of the across varied treatment- and nontreatment-seeking
other (Roy-Byrne etal.,2000). samples.
Another large-scale epidemiological study of
12,668 patients in the National Institute of Mental Impact
Health Epidemiologic Catchment Area Program The impact of comorbid panic disorder and
indicated that 2.1% of the population studied major depression is substantial. Patients with both
had a lifetime diagnosis of both panic disorder disorders have more severe symptoms, greater
and major depressive episode, while 4% had only functional impairment, more anxiety and somatic
major depression and 2% had only panic disorder concerns, increased likelihood of using multiple
(Andrade, Eaton, & Chilcoat, 1994). The preva- drug treatments, and increased risk of suicide than
lence of these disorders occurring together was 11 patients with only panic disorder or major depres-
times greater than expected by chance. sion (Baldwin,1998).
Brown, Campbell, Lehman, Grisham, and According to results from the NCS study
Mancill (2001) examined the comorbidity of (Roy-Byrne etal., 2000), individuals suffering from

86 Depression and Comorbidit y with Panic Disorder

both panic attacks with comorbid lifetime major disorders, greater persistence of each disorder, more
depression and major depression with comorbid frequent hospitalization and help-seeking behavior,
lifetime panic attacks reported significantly more more severe occupational impacts, and a substan-
severe symptom ratings for the primary disorder. tially higher rate of suicide attempts.
Furthermore, among individuals with both disor-
ders (either lifetime or current), a significant lin- Shared Risk Factors
ear association was found between the number of The tripartite model (Clark & Watson, 1991) has
physiological symptoms during panic attacks and offered a great deal of insight into the relationship
the number of depressive symptoms during depres- between anxiety and depression and their shared
sive episodes. The NCS study also found that when and unique risk factors. The tripartite model pro-
looking at the effects of comorbid depression on the poses that there are symptoms shared across anxiety
course of panic disorder, comorbidity was associ- and depression as well as symptoms unique to each.
ated with greater professional help-seeking, greater Shared symptoms typically are represented by a
perceived role impairment, greater prevalence of negative affect or general distress factor. Symptoms
attempted suicide, and more recent panic attacks of anhedonia and the absence of positive affect are
than in panic disorder alone. Similarly, comorbid specific to depression whereas symptoms of physi-
panic disorder affected the course of depression by ological hyperarousal are specific to anxiety. Cox,
increasing professional help-seeking, perceived role Enns, Walker, Kjernisted, and Pidlubny (2001)
impairment, prevalence of attempted suicide, and compared panic disorder and major depression on
the number of major depressive episodes. dimensions of the tripartite model and found con-
Findings from clinical studies have been simi- tinued support for the well-researched theory, such
lar. Grunhaus, Pande, Brown, and Greden (1994) that the shared higher order risk factor of negative
compared clinical symptoms and course of illness affectivity, or neuroticism, and the unique factors
in 119 patients with major depression alone and 57 of low positive affect and anxious arousal correctly
patients with major depression and comorbid panic classified over 75% of participants. However, more
disorder. Compared to the depression-only group, recent research has found that the hyperarousal fac-
the comorbid group experienced more severe symp- tor does not sufficiently explain the heterogeneity
toms; had higher ratings on feelings of inadequacy, found among anxiety disorders and, in fact, the
somatic anxiety, and phobia; required treatment anxious arousal factor appears to be unique to panic
and hospital admission earlier; and required psychi- disorder, rather than a factor relevant to all anxiety
atric hospitalization more frequently. disorders (Mineka, Watson, & Clark,1998).
Comorbid patients have been found to have Genetic risk studies have found that psychiatric
an increased rate of suicidal ideation (Fawcett, disorders largely map onto two genetic risk factors,
1992; Johnson et al., 1990) and suicide attempts internalizing (major depression, generalized anxiety
(Johnson etal., 1990), compared with patients with disorder, phobias) and externalizing disorders (alco-
panic disorder or depression alone. Johnson et al. hol and substance abuse, antisocial personality dis-
(1990) examined epidemiological data from the order, conduct disorder; Kendler, Prescott, Myers,
Epidemiologic Catchment Area study and found & Neale, 2003). Within the internalizing disorders,
that 7% of individuals with only panic disorder and two genetic factors predispose individuals to disor-
7.9% of individuals with only major depression had ders characterized by anxious-misery (major depres-
attempted suicide. However, 19.8% of individuals sion, generalized anxiety) or fear (phobias; Kendler
with comorbid panic disorder and major depression et al., 2003). From these data, large-scale genetics
had attempted suicide in their lifetime, a signifi- studies have further investigated the shared genetic
cantly higher number than in either disorder alone. risk among internalizing disorders and found that
The odds ratio comparing suicide rate for individu- the factors affecting genetic variation between panic
als with comorbid panic disorder and depression to disorder (as well as other anxiety disorders) and
individuals with no psychiatric disorder was 14.3, major depressive disorder largely reflect the same
meaning that individuals with comorbid diagnoses factors affecting variation in neuroticism (Fanous,
are 14.3 times more likely to attempt suicide than Gardner, Prescott, Cancro, & Kendler, 2002;
those with no psychiatric disorders. Hettema, Prescott, & Kendler,2004).
In sum, comorbidity between panic disorder and In a study of 9,000 twin pairs, Hettema, Neale,
depression has been associated with the presence Myers, Prescott, and Kendler (2006) found that
of more and more severe symptoms of each of the genetic factors shared with neuroticism accounted

Castriot ta,Craske 87
for between one third and one half of the genetic risk disorder to depression (Carter, Maddock, &
across the internalizing disorders, which included Maggliozi, 1992), the panic disorder group showed
major depression, panic disorder, generalized anxi- a significant interference in color naming of supra-
ety disorder, and the phobias. However, they also liminal threat words (presented on cards), as well
found that neuroticism did not capture all the as depression words, whereas the depressed group
genetic variance underlying the internalizing dis- did not. In another study, faster eye movements
orders. A second, neuroticism-independent, com- toward threatening faces were observed in general-
mon genetic factor was identified that accounted ized anxiety disorder than a depressed group, most
for a proportion of the genetic variance for the of whom had comorbid generalized anxiety disor-
non-phobic internalizing disorders (major depres- der as well (Mogg, Millar, & Bradley, 2000). Finally,
sion, generalized anxiety disorder, and panic disor- depressed patients showed an attentional bias to sad
der) that was similar to the proportion accounted facial expressions presented for 10,000 ms, whereas
for by the neuroticism-related common genetic patients with generalized anxiety disorder did not
factor. These results show that panic disorder and show an attentional bias to sad, angry, or happy
major depression share a great deal of genetic vari- faces (Gotlib, Kraspernova, Yue, & Joormann,
ability, partially attributable to shared genetic pre- 2004). Clearly, conclusions are limited from so few
dispositions to neuroticism and partially due to a studies.
separate genetic factor that is separate from neuroti- Still, Mathews and MacLeod (2005) concluded
cism and not entirely elucidatedyet. that depression is characterized by selective atten-
Exposure to childhood physical and/or sexual tion to cues that are consistent with negative affect
abuse is also a shared risk factor between panic when presented at long durations of one second
disorder and depression. In a 21-year longitudinal or more (e.g. Gotlib et al., 2004), suggesting the
study, Goodwin, Fergusson, and Horwood (2005) involvement of strategic control processes. In con-
found that young people exposed to physical trast, only anxiety is characterized by selective atten-
abuse had odds of panic disorder that were three tion to threat cues at shorter durations of 500 ms or
times higher than for those not exposed to physi- less and under masked conditions (Mogg & Bradley,
cal abuse. Similarly, those reporting childhood 2002), suggesting that selective attention toward
sexual abuse had odds of panic disorder that were threatening cues represents a more automated
2.2 times higher. These results confirmed the find- process, not dependent on conscious awareness in
ings of many previous studies that did not have anxiety disorders. These conclusions were drawn
access to longitudinal data (e.g. Safren, Gershuny, from studies that focused on either all anxiety dis-
Marzol, Otto, & Pollack, 2002; Stein etal., 1996). orders or specific anxiety disorders, including panic
Similarly, childhood physical and sexual abuse is a disorder. Conceivably, the information-processing
significant risk factor for depression, such that ado- features support threat sensitivity in persons with
lescents and young adults who experienced child- anxiety disorders, whereas they support deeper eval-
hood abuse were three times more likely to develop uation and rumination in persons with depression.
major depression than those who were not abused Finally, both anxiety and depression are associated
(e.g., Brown, Cohen, Johnson, & Smailes,1999). with appraisal biases, to interpret ambiguous infor-
An extensive body of research has evaluated mation in a negative fashion, albeit more threat
information-processing biases in relation to anxiety laden in anxiety disorders (e.g. Eysenck, Mogg,
and depression. It is not fully clear whether these May, Richards, & Mathews, 1991) and more nega-
biases serve as risk factors for the development of tive self-evaluation in depression (e.g. Lawson &
anxiety and mood disorders or whether they develop MacLeod,1999).
as a consequence of the disorders themselves. The amygdala plays a critical role in threat assess-
Research on attentional and information processing ment, in forming associations regarding danger in
biases has focused largely on anxiety disorders and the environment, and in mediating response to
unipolar depression as a whole and less specifically threat or potential threat via descending projec-
on panic disorder; however, it is still useful to note tions to regions that mediate autonomic responses
that such biases appear to be associated with both (e.g., heart rate, blood pressure, respiration, sweat-
forms of disorders, in slightly differentways. ing, etc.), and abnormal amygdala functioning has
Only a few studies have directly compared anxiety been observed and implicated as a risk factor for or
and depression in terms of information-processing consequence of anxiety disorders (see Craske etal.,
biases. In one study directly comparing panic 2009 for review).

88 Depression and Comorbidit y with Panic Disorder

This line of research has found that both anx- depression, where age of onset of panic disorder
ious (Blair et al., 2008) and depressed (Sheline was defined as the first cluster of four panic attacks
etal., 2001) individuals show an elevated amygdala or the first month of consistent worry about panic
response when viewing negative facial expressions, attacks, 48% reported that depression began at an
suggesting that abnormalities in amygdala function earlier age than panic disorder, 30.6% reported that
could mediate symptoms of abnormal threat assess- both disorders started in the same year, and 21.5%
ment, exaggerated fear responses, and abnormali- said that panic disorder started at an earlier age than
ties in learning about the dangers of environments the first depressive episode.
(Craske et al., 2009). However, specific aspects of These results appear to be at odds with previ-
amygdala function may distinguish between anxi- ous literature that has consistently found anxiety
ety disorders and depression. Some findings sug- to predate depression in individuals with lifetime
gest that anxiety disorders may be preferentially comorbidity (e.g., Hagnell & Grasbeck, 1990;
characterized by right-lateralized amygdala function Lydiard, 1991). However, when anxiety disorders
(Fredrikson & Furmark, 2003) whereas depression are analyzed in the aggregate, the NCS data find
is characterized by left-sided amygdala function, the same results, as early-onset phobias are typically
which may reflect lateralized aspects of normal the temporally primary disorders in this group. The
amygdala function in humans (Wright etal., 2001). NCS study also found that at least half of respon-
Further, elevated resting metabolism within the dents with comorbid panic disorder and depression
amygdala has been consistently found in depression have at least one other prior anxiety disorder, which
but not in anxiety disorders (Drevets, 2003). These suggests that panic disorder and depression may
studies were completed using a mix of different be part of a larger anxious-depression syndrome in
anxiety disorders, including panic disorder; how- which another anxiety disorder is temporally pri-
ever, more studies are needed that directly examine mary. Different definitions of panic onsetalso lead
shared effects of panic disorder and depression. to variable results. If the onset of panic is defined as
In sum, panic disorder and major depression the occurrence of the first panic attack, then panic
share many risk factors, most predominantly the occurring before depression is slightly more com-
personality trait of neuroticism, or the vulnerabil- mon, whereas when the onset of panic is defined as
ity to experiencing negative affect, which has been the age of onset of panic disorder, then depression
found consistently in both phenotypic and genetic occurring before panic disorder is more common.
research. Other shared risk factors include expo- In their clinical sample, Stein, Tancer, and Uhde
sure to childhood abuse, informational-processing (1990) found similar results as the NCS study,
biases, and elevated amygdala activation in response although they did not examine anxiety disorders
to negative facial expressions. as an aggregate. That is, among those with cur-
rent comorbid panic disorder and depression, 63%
Issues of Temporal Priority, reported a depressive episode before the onset of
Development, and Cause andEffect panic disorder, while 38% experienced their first
Temporal Priority depressive episode secondary to the development of
Research on the temporal priority or the tim- panic disorder. This study also did not separate the
ing of onset of panic disorder and depression first occurrence of panic attack from the develop-
indicates that the age of onset of each disorder is ment of panic disorder, leaving it unclear if the rate
variable. In the NCS study (Kessler et al., 1998), of depression preceding panic symptoms may be
retrospective age-at-onset reports were compared to smaller if the onset of panic symptoms was defined
assess the temporal priority of first onsets of panic as first occurrences of panic attacks.
attacks, panic disorder, and depression. Among the One line of research has examined the presence
302 respondents who reported at least one lifetime of panic attacks, without a full diagnosis of panic
panic attack and one lifetime episode of depression, disorder, as a specific risk factor for the subsequent
31.1% stated that their first episode of depression development of depressive episodes. Goodwin,
occurred prior to their first panic attack, 25.5% Fergusson, and Horwoods (2004) 21-year birth
stated that their first episode of depression and first cohort study conducted in Christchurch, New
panic attack occurred in the same year, and 43.4% Zealand, found that panic attacks in the preceding
said their first panic attack occurred at an earlier age three years increased the risk for developing a cur-
than their first depressive episode. Among respon- rent major depressive episode among young adults,
dents with comorbid lifetime panic disorder and after controlling for early behavioral risk factors for

Castriot ta,Craske 89
psychopathology such as past history of depression, causes of comorbidity differ depending on which
childhood abuse, and personality characteristics. disorder is temporally primary. When depression
In the prospective Early Developmental Stages occurs prior to panic attacks, if the depression has
of Psychopathology study, covering up to age 30, remitted before the development of panic attacks or
Goodwin, Lieb, et al. (2004) found that primary disorder, the presence of a past depressive episode
panic attacks were associated with increased risk for does not affect the course or severity of the panic.
incident major depression with an odds ratio of 2.8. This is consistent with findings from clinical stud-
Finally, in a 10-year prospective longitudinal exami- ies that past depressive episodes that have remitted
nation of the same data set by Beesdo and colleagues do not affect the severity of current panic attacks or
(2007), the occurrence of panic attacks in individu- panic disorder (Lesser, Rubin, Pecknold, & Rifkin,
als with social anxiety disorder (i.e., cued panic in 1988; Maddock et al., 1993). This result, taken
social situations) increased the risk for subsequent with another finding that the number of depressive
depression. This study did not evaluate unexpected symptoms increases the risk for subsequent panic
or uncued panic attacks separate from social anxiety attacks, suggests that temporally secondary panic
disorder. may serve as a severity marker for depression rather
than a comorbid condition.
Development When panic is temporally primary, it is unclear
Research on the development of comorbid whether panic attacks represent a causal factor in
panic disorder and depression has often focused and of itself or if it represents a proxy for another
on whether patients with both disorders actually underlying causal factor. This discrepancy is based
have two separate and co-occurring disorders or on the finding that respondents with a history of
one disorder that marks a more severe condition. panic attacks have the same elevated risk of later
Evidence for the theory that comorbidity represents depression regardless of whether their panic is active
one severe disorder comes primarily from studies of or in remission, which can be interpreted multiple
biological similarities of patients with panic disor- ways. First, it could be that panic causes depres-
der and depression. Such studies have found shared sion through the consequences of panic that persist
disturbances of the hypothalamic-pituitary-adrenal after the remission of panic attacks, such as avoid-
axis function, serotonergic neurotransmission, thy- ance, secondary agoraphobia, and secondary sub-
roid releasing hormone, and growth hormone (see stance abuse. It is also possible that panic attacks
Stein & Uhde, 1990, for review). However, more are a marker of some cluster of risk factors that is
recent neuroimaging studies, whose measures have associated with both panic disorder and depression.
stronger sensitivity and specificity than the previ- Regardless of the interpretation, the results point
ously used biological measures, indicate that panic to the fact that individuals with a history of panic
disorder and depression have substantially different attacks are at increased risk for the development of
neurobiological substrates (i.e., Dolan etal., 1994; both future panic disorder and depression.
Nordahl etal.,1990). In sum, research on the temporal priority of
Corroborating evidence for the theory that these panic disorder and major depression is mixed and
disorders represent two distinct phenomena comes depends largely on the definitions of onset used.
from family studies that have sought to determine Most research has found that panic attacks and
whether panic disorder and depression aggregate other symptoms of anxiety, including other anxi-
separately or together in families. These studies have ety disorders, typically predate the onset of the first
found that comorbid panic disorder and depres- major depressive episode but that the first depressive
sion does not aggregate separately from pure panic episode predates the onset of full panic disorder in
disorder and pure depression (Maier, Minges, & more than 50% of cases. Developmental research
Lichtermann, 1995; Weissman et al., 1993), indi- shows that despite high rates of co-occurrence,
cating the specific and independent transmission of panic disorder and depression appear to be two dis-
each disorder and the separation of panic disorder tinct disorders as they aggregate separately among
from depression as a distinct disorder. families of individuals with both disorders and are
associated with different neurobiological substrates.
Cause andEffect
Examination of the underlying causes of comor- Future Research Directions
bidities in the NCS study (Kessler et al., 1998) While past research has consistently shown a very
yielded mixed results, raising questions of whether high rate of comorbidity between panic disorder and

90 Depression and Comorbidit y with Panic Disorder

depression, as well as the many impacts that their relied primarily on self-report measures to examine
co-occurrence have on the symptoms and course constructs such as the experience of positive and
of each disorder, there are several lines of research negative affect, personality factors, and anxiety sen-
that would be beneficial in further elucidating the sitivity. Future research should use more behavioral
relationship between these disorders. Research on and neurobiological paradigms for testing the cur-
the temporal priority of each of the disorders has rent models as well as other dimensional constructs,
been mixed and has varied based on the use of dif- such as sensitivity to threat and reward, that extend
ferent definitions of disorder onset. The question of beyond self-report.
whether it is the occurrence of panic attacks, panic
disorder, or depression that develops primarily Assessment and Intervention
and which serves as a causal factor for the devel- Strategies
opment of secondary diagnoses remains unclear. Assessment
It is also common for other anxiety disorders to When assessing panic disorder in a patient with
co-occur with panic disorder and depression, and current or past major depression or vice versa, a
many develop temporally earlier than either panic detailed history will aid in defining the relationship
disorder or depression, pointing to the possibility between these two disorders and may yield insight
that other forms of anxiety may serve as causal fac- into how the experience of one affects the pres-
tors as well. Longitudinal research would aid in the ence and course of the other. As discussed in the
understanding of the course from panic attacks to Temporal Priority and Impact sections of this chap-
panic disorder and depression as well as how they ter, the order of development is variable, but each
each affect the development, course, and remission disorder appears to increase the likelihood of onset
of eachother. of an episode of the other and increases the severity
Further research on the neurobiological and bio- and persistence of the symptoms. Therefore, when
logical processes associated with panic attacks, panic assessing one disorder, it is important to be aware of
disorder, and depression would be greatly beneficial indicators of past and present episodes of theother.
to understanding the common and unique fac- An in-depth interview is the first step in estab-
tors associated with the development and course lishing diagnostic features and the profile of symp-
of panic disorder and depression and how they tomatic and behavioral responses. Several semi- and
affect the development and course of each other. fully structured interviews exist. The Anxiety
Such research would also give further support for Disorders Interview ScheduleFourth Edition
the current view that these two disorders represent (DiNardo, Brown, & Barlow, 1994) assesses anxiety
two diagnostically distinct entities or would support disorders primarily as well as mood disorders and
previous theories that the co-occurrence of these somatoform disorders.
two highly comorbid disorders represent one more Several standardized self-report inventories pro-
severe disorder. This research should also focus on vide useful information for treatment planning,
potential biological and neurobiological relation- as well as being sensitive markers of therapeu-
ships between panic attacks and panic disorder tic change. The Anxiety Sensitivity Index (Reiss,
and depression, which may explain whether panic Peterson, Gursky, & McNally, 1986) has received
attacks serve as a risk factor or a severity marker for wide acceptance as a trait measure of threatening
these disorders. Additionally, more work is needed beliefs about bodily sensations. It has good psycho-
on the question of how much of the shared vari- metric properties and tends to discriminate panic
ance with depression is due to the presence of panic disorder/agoraphobia from other types of anxiety
attacks versus panic disorder. disorders (e.g., Taylor, Koch, & McNally, 1992).
Finally, more research investigating dimen- More specific information about which particular
sional constructs of panic disorder and depres- bodily sensations are feared the most, and what spe-
sion is needed to be consistent with the National cific misappraisals occur most often, can be obtained
Institute of Mental Healths new Research Domain from the Body Sensations and Agoraphobia
Criteria initiative, which calls for new ways of clas- Cognitions Questionnaires (Chambless, Caputo,
sifying psychopathology based on the dimensions Bright, & Gallagher, 1984). The Mobility Inventory
of observable behavior and neurobiological mea- (Chambless, Caputo, Gracely, Jasin, & Williams,
sures. Much of the research examining the tripar- 1985) lists agoraphobic situations that are rated in
tite model and subsequent models of common and terms of degree of avoidance when alone and when
specific risk factors for anxiety and depression have accompanied.

Castriot ta,Craske 91
The Albany Panic and Phobia Questionnaire increasing accuracy in self-observation. A panic
(Rapee, Craske, & Barlow, 1995) assesses fear and attack record is to be completed as soon after each
avoidance of activities that produce feared bodily panic attack as possible and provides a description
sensations, as well as more typical agoraphobia and of cues, maximal distress, symptoms, thoughts, and
social situations. Factor analyses confirmed three dis- behaviors.
tinct factors, which have been labeled Agoraphobia, Breathing retraining. Breathing retraining
Social Phobia, and Interoceptive Fears. The Anxiety became a central component early on in the devel-
Control Questionnaire assesses perceived lack of opment of panic-control treatments because many
control over anxiety-related events and occurrences, panic patients describe hyperventilatory symptoms
such as internal emotional reactions or externally as being very similar to their panic attack symp-
threatening cues (Rapee, Craske, Brown, & Barlow, toms. In early conceptualizations, panic attacks
1996). This scale is designed to assess locus of control were related to stress-induced, respiratory changes
but in a more specific and targeted manner relevant that either provoke fear because they are perceived
to anxiety and anxiety disorders when compared to as threatening or augment fear already elicited by
more general locus of control scales. other phobic stimuli (Clark, Salkovskis, & Chalkley,
1985). Several studies illustrated a positive effect
Cognitive Behavioral Therapy of breathing retraining (e.g. Kraft & Hoogduin,
for Panic Disorder 1984). However, several studies suggest that the
Cognitive behavioral therapy (CBT), involv- addition of breathing retraining does not improve
ing most or all of the components listed in detail upon in vivo exposure alone (e.g., Schmidt et al.,
below, yields panic-free rates in the range of 70% 2000). Given the recent recognition that tolerance
to 80% and high end-state rates (i.e., within nor- of fear and anxiety may be a more critical learning
mative ranges of functioning) in the range of 50% experience than the elimination of fear (see Eifert
to 70% for panic disorder with minimal agorapho- & Forsyth, 2005), breathing retraining has been
bia (e.g., Barlow, Craske, Cerny, & Klosko, 1989; de-emphasized since it could become a method of
Clark et al., 1994). Two meta-analyses reported avoidance of physical symptoms or a safety behavior
very large effect sizes of 1.55 and 0.90 for CBT for and thereby antitherapeutic.
panic disorder (Hoffman & Smits, 2008; Westen On the other hand, advances have been made
& Morrison, 2001). Also, results generally main- using capnometry-assisted respiratory training
tain over follow-up intervals for as long as two years (CART), which targets respiratory dysregulation,
(Craske, Brown, & Barlow,1991). in particular reduced levels of carbon dioxide in
The components of CBT are as follows: the blood that can result from hyperventilation
Education. The treatment begins with education (Meuret, Wilhelm, Ritz, & Roth, 2008). CART is
about the nature of panic disorder, the causes of a four-week training program that uses immediate
panic and anxiety, and the way in which panic and feedback of end-tidal PCO2, or the partial pres-
anxiety are perpetuated by feedback loops among sure of carbon dioxide present after exhalation, to
physical, cognitive, and behavioral response systems. teach patients how to raise their subnormal levels of
In addition, specific descriptions of the psychophys- PCO2 (hyperventilation) and thereby gain control
iology of the fightflight response are provided, as over dysfunctional respiratory patterns and associ-
well as an explanation of the adaptive value of the ated panic symptoms (e.g., shortness of breath, diz-
various physiological changes that occur during ziness). CART substantially differs from traditional
panic and anxiety. The purpose of this education is breathing retraining because it focuses directly on
to correct the common myths and misconceptions the proposed mediator of the relationship between
about panic symptoms (i.e., beliefs about going breathing dysregulation and panic, PCO2 (see
crazy, dying, or losing control). The Education sec- Meuret, Wilhelm, Ritz, & Roth, 2003, for review).
tion also distinguishes the state of anxiety from the Cognitive restructuring. Cognitive restructuring
emotion of fear/panic, both conceptually and in begins with discussion of the role of thoughts in
terms of its three response modes (subjective, physi- generating emotions to provide a treatment ratio-
ological, and behavioral). This distinction is central nale. Next, thoughts are recognized as hypotheses
to the model of panic disorder and to the remainder rather than fact and therefore open to questioning
of the treatment. and challenge. Detailed self-monitoring of emotions
Self-monitoring. Self-monitoring is introduced and associated cognitions are instituted to identify
as a way of enhancing objective self-awareness and specific beliefs, appraisals, and assumptions that are

92 Depression and Comorbidit y with Panic Disorder

categorized into types of typical errors that occur Interoceptive exposure. In interoceptive exposure,
during heightened emotion, such as overestima- the goal is to deliberately induce feared physical sen-
tions of risk of negative events or catastrophizing of sations a sufficient number of times and for long
meaning of events. The process of categorization, or enough each time so that misappraisals about the
labeling of thoughts, is consistent with a personal sensations are disconfirmed and conditioned anxi-
scientist model and facilitates an objective perspec- ety responding extinguishes. Interoceptive exposure
tive by which the validity of the thoughts can be is now a standard component of CBT for panic
evaluated. Thus in labeling the type of cognitive dis- disorder (e.g., Barlow, Gorman, Shear, & Woods,
tortion, the patient is encouraged to use an empiri- 2000; Craske, Lang, Aikins, & Mystkowski, 2005),
cal approach to examine the validity of his or her although different groups give different emphases to
thoughts by considering all of the available evidence. interoceptive exposure, with some emphasizing it is
Therapists use Socratic questioning to help patients a means for extinguishing fear responding (Barlow
make guided discoveries and question their anxious & Craske, 2006) and others as a vehicle for discon-
thoughts. Next, alternative hypotheses are generated firming misappraisals (Clark,1996).
that are more evidence-based. Importantly, cognitive In terms of implementation, a standard list of
restructuring is not intended as a direct means of exercises, such as hyperventilating and spinning, are
minimizing fear, anxiety, or unpleasant symptoms. used to establish a hierarchy of interoceptive expo-
Instead, cognitive restructuring is intended to cor- sures. Using a graduated approach, exposure begins
rect distorted thinking; eventually fear and anxiety with the less distressing physical exercises and con-
are expected to subside, but their diminution is not tinues with the more distressing exercises. It is essen-
the first goal of cognitive therapy. tial that the patient endure the sensations beyond
Exposure. Exposure is a critical phase of treat- the point at which they are first noticed, for at least
ment and, once begun, is a major focus of treat- 30 seconds to a minute, because early termination
ment sessions as well as homework, since limited of the task may eliminate the opportunity to learn
exposure practice is of small benefit and may even that the sensations are not harmful and that the
be detrimental. The exposure is designed to discon- anxiety can be tolerated. The coping skills of cogni-
firm misappraisals and extinguish conditioned emo- tive restructuring and slow breathing are used after
tional responses to external situations and contexts each exercise, followed by a discussion of what was
through in vivo exposure, as well as to bodily sensa- learned during the exercise about bodily sensations,
tions, through interoceptive exposure. fear, and avoidance. These interoceptive exercises
In vivo exposure. In vivo exposure refers to repeated are practiced outside of the therapy session to con-
and systematic, real-life exposure, in this case to ago- solidate the process of learning. Exposure extends to
raphobic situations. Most often, in vivo exposure is naturalistic activities that inherently induce somatic
conducted in a graduated manner, proceeding from sensations (e.g., caffeine consumption, exercise).
the least to the most anxiety-provoking situations
on an avoidance hierarchy. However, there is some Medication
evidence to suggest that intensive or ungraduated Pharmacologic treatment of depression com-
exposure may be effective (e.g., Feigenbaum,1988). plicated by panic disorder, or panic disorder com-
In vivo exposure typically includes the removal plicated by depression, is similar to that for either
of safety signals and safety behaviors. Examples of disorder in its uncomplicated form. Selective sero-
safety signals include other people, water, money tonin reuptake inhibitors (SSRIs) have become the
(to call for help), empty or full medication bottles, most common first-line treatment for depression
exit signs, and familiar landmarks when traveling. with concurrent panic disorder, due to their docu-
Safety behaviors similarly provide a sense of safety mented efficacy in both conditions, a favorable side
and include seeking reassurance or checking for effect profile, and fewer risks of physical complica-
exits. Reliance on safety signals and safety behaviors tions than previously more common monoamine
attenuate distress in the short term but maintain oxidase inhibitors and tricyclic antidepressants.
excessive anxiety in the long term. However, recent Many positive placebo-controlled, randomized
results contradict the previous findings that safety trials support the efficacy of six different SSRI
signals are deleterious for the course of treatment drugsfluoxetine, fluvoxamine, sertraline, parox-
(Rachman, Shafran, Radomsky, & Zysk, 2011) etine, citalopram, and escitalopram (Roy-Byrne &
and have found that they do not affect treatment Cowley, 2002). Placebo-controlled trials (Bradwejn
outcome. et al., 2005) also support the efficacy for an

Castriot ta,Craske 93
extended-release form of venlafaxine in panic dis- to engage with the materials presented in treatment,
order. Therapeutic response in panic disorder is a particularly on how comorbidity may affect their
class effect, which is common to all the SSRIs, with capacities to learn the cognitive strategies involved
no evidence of differential efficacy within the class. in CBT. As impaired concentration and cognitive
Although relevant differences exist in side-effect functioning is a frequent symptom of depression,
profiles, drug interactions, and half-life, differences future research should investigate how comorbidity
in cost due to availability of the generic forms of may impede learning alternative models of cogni-
these substances (fluoxetine, paroxetine, sertraline, tive processing and cognitive interventions for panic
and citalopram are currently available in the United disorder.
States) are probably much more important. Research on response rates to CBT has shown
The older class of tricyclic antidepressants, that comorbid depression has no effect on the
although associated with more side effects (Bakker, response to treatment at posttreatment or follow-up
van Balkhom, & Spinhoven, 2002) includes drugs and in both referred and primary care settings (Allen
that are both less expensive and similarly effective etal., 2010; McLean etal., 1998; Roy-Byrne etal.,
than newer classes of antidepressants, with many 2005). Specifically, a recent large-scale, multisite
studies indicating efficacy for imipramine, desip- clinical trial examining long-term treatment strate-
ramine, clomipramine, nortriptyline, and ami- gies for panic disorder among 454 patients (Allen
triptyline, (Roy-Byrne & Cowley, 2002) and six etal., 2010) found that CBT was equally efficacious
older pre-DSM-III studies (Roy-Byrne & Cowley, in the reduction of panic disorder for individuals
2002) showing efficacy of monoamine oxidase with and without comorbid depression.
inhibitors in the phobic anxiety of individuals with It seems contradictory that comorbid depression
panic-like symptoms. These compounds, especially would have a significant impact on the severity and
monoamine oxidase inhibitors, can be useful in persistence of panic disorder but would not affect
treatment-refractory patients. the outcomes of panic disorder treatment. This may
be a product of limitations to the current treat-
Effect of Comorbid Depression on the ment literature. For example, studies have recruited
Treatment of Panic Disorder patients for the treatment of panic disorder and
As discussed in the Impact section of this chapter, have often excluded patients who are very extremely
comorbid major depression slows the natural rate of depressed or suicidal. Thus the majority of patients
recovery for panic disorder (Bruce etal., 2005) and are mildly to moderately depressed. The results may
has been associated with increased severity of panic differ with very extremely depressed patients. Many
disorder and agoraphobia and greater impairment of these studies also exclude bipolar disorder and
(e.g., Allen et al., 2010; Grunhaus et al., 1994). therefore exclude an entire group of individuals who
However, research on how comorbid depression experience major depressive episodes.
affects the course and outcome of panic disorder
treatment has yielded mixed results. Studies focused Effect of CBT for Panic Disorder
on CBT for all anxiety disorders and treatment par- on Comorbid Depression
ticipation have found that comorbidity with depres- Another reason why depression does not affect
sion is associated with increased rates of refusal to the treatment of panic disorder is that CBT for
enter treatment (Issakidis & Andrews, 2004); panic disorder is also effective in treating symptoms
however, once patients have entered treatment, the of depression. Baseline levels of depression improve
comorbidity has no effect on rates of attrition (Allen through treatment and thereby confer less of a nega-
etal., 2010; Brown, Antony, & Barlow,1995). tive impact on the individual as treatment goes on
Preliminary research investigating the effects (LaBerge etal., 1993; Tsao, Mystkowski, Zucker, &
of comorbidity on engagement with treatment Craske, 2002), even when treatment is conducted
has found that comorbid depression has no effect in a primary care setting (Roy-Byrne etal., 2005).
on compliance with CBT homework (McLean, Allen and colleagues (2010) found that symptoms
Woody, Taylor, & Koch, 1998) or compliance with (although not diagnoses) of depressive disorders not
CBT treatment as a whole (Murphy, Michelson, addressed in the treatment also showed improve-
Marchione, Marchione, & Testa, 1998), though it ment from pre- to posttreatment. There are several
does increase levels of distress associated with treat- explanations as to why CBT for panic is effective at
ment (Murphy et al., 1998). There is a need for treating depression symptoms. One is that depres-
more research on the ability of depressed patients sion sometimes develops secondary to panic attacks

94 Depression and Comorbidit y with Panic Disorder

and other anxiety disorders that are often comorbid depression is still preliminary, as most treatment
with panic. Thus when panic and anxiety symp- studies have focused primarily on effective treat-
toms are treated and reduced, secondary depression ments for either panic disorder or depression, often
may subside as well. Another possibility is that the to the point of excluding comorbid disorders. Thus
strategies involved in CBT for panic overlap heav- there is currently no common set of guidelines for
ily with the strategies used in effective treatments treating comorbidity, yet a suggested strategy can be
for depression such as cognitive therapy and behav- gleaned from the reviewed literature. This strategy
ioral activation (Dimidjian et al., 2006; Jacobson, starts by determining whether the panic disorder or
Martell, & Dimidjian, 2001). Last, there may be depression is most severe.
a down-regulation of processes that are shared by If panic disorder is the principal, or most severely
panic disorder and depression that contribute to distressing or disabling, diagnosis, the research
their initial shared risk and comorbidity, such as points to CBT focused primarily on the panic disor-
changes in levels of neuroticism and perceived con- der. As reviewed in the previous section, comorbid
trol (Craske etal.,2007). depression does not appear to affect the outcome
However, despite the positive effects of panic dis- of treatment for panic disorder. Clinicians often
order treatment on depression, rates of diagnoses of tailor treatments to accommodate impairments
depression do not decline significantly (Allen etal., related to depression. Afew potential modifications
2010; Brown, Antony, & Barlow, 1995). Individuals include the addition of pleasant events scheduling,
who do not experience much improvement in panic the expansion of cognitive restructuring work to
disorder symptoms and diagnoses after CBT treat- include depressive thoughts, and making sure that
ment can go on to develop new episodes of major the patient does not ignore or downplay his or her
depression over the period between posttreatment own accomplishments in exposure therapy.
and follow-up (Brown, Antony, & Barlow, 1995). However, significant modifications should not
The reason for this is unclear but may be related be made to the standard CBT for panic disorder as
to the severity of their panic disorder diagnosis, the it may impede elements of the treatment that con-
strength of shared risk factors between panic disor- tribute to its effectiveness. Craske and colleagues
der and depression, and/or disappointment in not (2007) compared the effects of a higher dose of CBT
seeing improvement after treatment. for panic disorder versus CBT for panic disorder
In sum, CBT is the most heavily researched with additional CBT for comorbid disorders among
and evidenced treatment for panic disorder and patients with a principal diagnosis of panic disorder
has shown to be effective in many treatment and with agoraphobia and other comorbid disorders.
research contexts. Treatment involves initial psy- They found that patients in the CBT for panic dis-
choeducation about anxiety disorders and how they order only group saw greater improvement in both
affect the physiology of the body, self-monitoring their panic disorder and their comorbid conditions
of anxious thoughts and cognitive misappraisals, at posttreatment and follow-up. The research has
breathing retraining, cognitive restructuring of dis- also shown mild improvements in depressive symp-
torted thoughts and appraisals, and interoceptive toms with CBT for panic disorder.
and in vivo exposure to activities and environments If major depressive disorder is the principal
on the individuals fear hierarchy. Other forms of diagnosis, the treatment guidelines are unclear.
treatment include medication, particularly SSRIs. Studies examining the treatment for panic disor-
Comorbid depression has mixed effects on the der have typically excluded patients with severe
course of CBT for panic disorder, although most depression and/or have recruited patients with
recent studies indicate very little effect on patient principal anxiety disorders, and there have been
response to treatment. Furthermore, CBT for panic no recent studies examining the effects of treat-
disorder has positive, yet limited, effects on symp- ment for depression on comorbid panic disorder.
toms of depression. Finally, individuals who receive However, it can be assumed that all of the studies
CBT for panic disorder and do not experience of CBT and behavioral therapies for depression
improvement may be at increased risk for future have included individuals with comorbid panic
depressive episodes. disorder and have found these therapies to be
effective overall. Thus it may be presumed that if
Clinical Guidelines for Practitioners there is a principal diagnosis of depression, treat-
Research investigating the optimal treat- ing the depression would also lead to a reduction
ment guidelines for comorbid panic disorder and in panic disorder symptoms.

Castriot ta,Craske 95
Summary and Conclusions References
In summary, comorbidity between panic disor- Allen, L. B., White, K. S., Barlow, D. H., Shear, M. K., Gorman,
der and major depression is found in the majority J. M., & Woods, S. W. (2010). Cognitive behavioral ther-
of individuals with panic disorder and a substan- apy (CBT) for panic disorder: Relationship of anxiety and
depression comorbidity with treatment Outcome. Journal of
tial minority of individuals with major depression. Psychopathology and Behavioral Assessment, 32, 185192.
Rates of comorbidity between panic attacks and American Psychiatric Association. (1980). Diagnostic and statisti-
depression are even higher than between full panic cal manual of mental disorders, (3rd ed.). Washington, DC:
disorder and depression. These findings have been Author.
consistent throughout epidemiological, clinical, American Psychiatric Association. (2000). Diagnostic and statisti-
cal manual of mental disorders, (4th ed., text rev.). Washington,
and primary care studies. DC:Author.
Comorbidity between panic disorder and depres- American Psychiatric Association. (2013). Diagnostic and sta-
sion has been associated with substantially more tistical manual of mental disorders (5th ed.). Washington,
severe impacts than either disorder alone, specifi- DC:Author.
cally the presence of more and more severe symp- Andrade, L., Eaton, W. W., & Chilcoat, H. (1994). Lifetime
comorbidity of panic attacks and major depression in a
toms of each of the disorders, greater persistence of population-based study:Symptom profiles. British Journal of
each disorder, more frequent hospitalization and Psychiatry, 165, 363369.
help-seeking behavior, more severe occupational Bakker, A., van Balkom, A. J., & Spinhoven, P. (2002). SSRIs
impacts, and a significantly higher rate of suicide vs TCAs in the treatment of panic disorder:Ameta-analysis.
attempts. Acta Psychiatrica Scandinavica 2002, 106, 163167.
Baldwin, D. S. (1998). Depression and panic: Comorbidity.
Not surprisingly, given the high rates of co- European Psychiatry, 13 (suppl. 2),6570.
occurrence, panic disorder and major depression Barlow, D. H., & Craske, M. G. (2006). Mastery of your anxiety
share many risk factors, most predominantly the and panic: Patient workbook. (4th ed.). New York: Oxford
personality trait of neuroticism, or the vulnerabil- UniversityPress.
ity to experiencing negative affect, which has been Barlow, D. H., Craske, M. G., Cerny, J. A., & Klosko, J. S.
(1989). Behavioral treatment of panic disorder. Behavior
found consistently in both phenotypic and genetic Therapy, 20, 261282.
research. Other shared risk factors include expo- Barlow, D. H., Gorman, J. M., Shear, M. K. & Woods, S. W. (2000).
sure to childhood abuse, informational-processing Cognitive-behavioral therapy, imipramine, or their combination
biases, and elevated amygdala activation in response for panic disorder:Arandomized controlled trial. JAMA:Journal
to negative facial expressions. of the American Medical Association, 283, 25292536.
Beesdo, K., Bittner, A., Pine, D. S., Stein, M. B., Hofler, M.,
Research on the development of comorbid- Lieb, R., & Wittchen, H. U. (2007). Incidence of social anx-
ity between these two disorders has found that iety disorder and the consistent risk for secondary depression
the temporal priority of panic disorder and major in the first three decades of life. Archives of General Psychiatry,
depression is mixed and depends largely on the 64, 903912.
definitions of onset used. Most research has found Biederman, J., Petty, C., Faraone, S. V., Hirsch-Becker, D.
R., Henin, A., Pollack, M. H., Rosenbaum, J. F. (2005).
that panic attacks and other symptoms of anxiety, Patterns of comorbidity in panic disorder and major depres-
including other anxiety disorders, typically predate sion: Findings from a non-referred sample. Depression and
the onset of the first major depressive episode but Anxiety, 21,5560.
that the first depressive episode predates the onset Blair, K., Shaywitz, J., Smith, B. W., Rhodes, R., Geraci, M., Jones,
of full panic disorder in more than 50% of cases. M.,...Pine, D.S. (2008). Response to emotional expressions
in generalized SOP and GAD: Evidence for separate disor-
Developmental research shows that despite high ders. American Journal of Psychiatry, 165, 11931202.
rates of co-occurrence, panic disorder and depres- Bradwejn, J., Ahokas, A., Stein, D. J., Salinas, E., Emilien, G., &
sion appear to be two distinct disorders as they Whitaker, T. (2005). Venlafaxine extended-release capsules in
aggregate separately among families of individuals panic disorder:Flexible-dose, double-blind placebo-controlled
with both disorders and are associated with different study. British Journal of Psychiatry, 187, 352359.
Brown, T. A., Antony, M. M., & Barlow, D. H. (1995).
neurobiological substrates. Diagnostic comorbidity in panic disorder: Effect on treat-
Finally, the most researched and evidenced-based ment outcome and course of comorbid diagnoses following
treatment for panic disorder is CBT. Research has treatment. Journal of Consulting and Clinical Psychology, 63,
found mixed results on the effect of comorbid 408418.
depression on the course and outcome of CBT for Brown, C., Schulberg, H. C., Madonia, M. J., Shear, M. K., &
Houck, P. R. (1996). Treatment outcomes for primary care
panic disorder, though most recent research has patients with major depression and lifetime anxiety disor-
shown that depression has little effect on treatment ders. American Journal of Psychiatry, 153, 12931300.
response. Additionally, CBT for panic disorder has Brown, C., Schulberg, H. C., & Shear, M. K., (1996).
small but positive effects on comorbid depression. Phenomenology and severity of major depression and

96 Depression and Comorbidit y with Panic Disorder

comorbid lifetime anxiety disorders in primary medical care Dimidjian, S., Hollon, S. D., Dobson, K. S., Schmaling, K. B.,
practice. Anxiety, 2, 210218. Kohlenberg, R. J., Addis, M. E.,...Jacobson, N. S. (2006).
Brown, J., Cohen, P., Johnson, J. G., & Smailes, E. M. (1999). Randomized trial of behavioral activation, cognitive therapy,
Childhood abuse and neglect:Specificity of effects on ado- and anti-depressant medication in acute treatment of adults
lescent and young adult depression and suicidality. Journal with major depression. Journal of Consulting and Clinical
of the American Academy of Child Adolescent Psychiatry, 38, Psychology, 74, 658670.
14901496. DiNardo, P., Brown, T. A., & Barlow, D. H. (1994).
Brown, T. A., Campbell, L. A., Lehman, C. L., Grisham, J. R., Anxiety Disorders Interview Schedule Fourth Edition.
& Mancill, R. B. (2001). Current and lifetime comorbidity NewYork:Oxford UniversityPress.
of the DSM-IV anxiety and mood disorders in a large clinical Dolan, R. J., Bench, C. J., Brown, R. G., Scott, L. C., &
sample. Journal of Abnormal Psychology, 110, 585599. Frackowiak, R. (1994). Neuropsychological dysfunction in
Bruce, S. E., Yonkers, K. A., Otto, M. W., Eisen, J. L., Weisberg, depression:The relationship to regional cerebral blood flow.
R. B., Pagano, M., Shea, M. T., & Keller, M. B. (2005). Psychological Medicine, 24, 849857.
Influence of psychiatric comorbidity on recovery and recur- Drevets, W. C. (2003). Neuroimaging abnormalities in the
rence in generalized anxiety disorder, social phobia, panic amygdala in mood disorders. Annals of the NewYork Academy
disorder: A 12-year prospective study. American Journal of of Science, 985, 420444.
Psychiatry, 162, 11791187. Eifert, G. H., & Forsyth, J. P. (2005). Acceptance and commit-
Carter, C. S., Maddock, R. J., & Magliozzi, J. (1992). ment therapy for anxiety disorders: A practitioners treatment
Patterns of abnormal processing of emotional information guide to using mindfulness, acceptance, and value-based behav-
in panic disorder and major depression. Psychopathology, ior change strategies. Oakland, CA:New Harbinger.
25,6570. Eysenck, M. W., Mogg, K., May, J., Richards, A., & Mathews,
Chambless, D. L., Caputo, G., Bright, P., & Gallagher, R. A. (1991). Bias in interpretation of ambiguous sentences
(1984). Assessment of fear in agoraphobics: The Body related to threat in anxiety. Journal of Abnormal Psychology,
Sensations Questionnaire and the Agoraphobic Cognitions 100, 144150.
Questionnaire. Journal of Consulting and Clinical Psychology, Fanous, A., Gardner, C. O., Prescott, C. A., Cancro, R., &
52, 10901097. Kendler, K. S. (2002). Neuroticism, major depression
Chambless, D. L., Caputo, G., Gracely, S., Jasin, E., & Williams, and gender: A population-based twin study. Psychological
C. (1985). The Mobility Inventory for Agoraphobia. Medicine, 4, 719728.
Behaviour Research & Therapy, 23,3544. Fawcett, J. (1992). Suicide risk factors in depressive disorders
Clark, D. A. (1996). Panic disorder: From theory to therapy. and in panic disorder. Journal of Clinical Psychiatry, 53
In:P. M.Salkovskis (Ed.), From frontiers of cognitive therapy (suppl. 3),913.
(pp. 318344). NewYork, GuilfordPress. Feigenbaum, W. (1988). Long-term efficacy of ungraded ver-
Clark, D., Salkovskis, P., & Chalkley, A. (1985). Respiratory sus graded massed exposure in agoraphobics. In: I. Hand
control as a treatment for panic attacks. Journal of Behavior & H. Wittchen (Eds.), Panic and phobias: Treatments
Therapy and Experimental Psychiatry, 16,2330. and variables affecting course and outcome (pp. 149158).
Clark, D. M., Salkovskis, P. M., Hackmann, A., Middleton, Berlin:Springer-Verlag.
H., Anastasiades, P., & Gelder, M. (1994). A comparison of Fredrikson, M., & Furmark, T. (2003). Amygdaloid regional
cognitive therapy, applied relaxation and imipramine in the cerebral blood flow and subjective fear during symptom
treatment of panic disorder. British Journal of Psychiatry, 164, provocation in anxiety disorders. Annals of the New York
759769. Academy of Science, 985, 341347.
Clark, L. A., & Watson, D. (1991). Tripartite model of anxi- Goodwin, R. D., Fergusson, D. M., & Horwood, L. J. (2004).
ety and depression: Psychometric evidence and taxonomic Panic attacks and the risk of depression among young adults
implications. Journal of Abnormal Psychology, 100, 316336. in the community. Psychotherapy and Psychosomatics, 73,
Clayton, P. J. (1990). The comorbidity factor:Establishing the 158165.
primary diagnosis in patients with mixed symptoms of anxi- Goodwin, R. D., Fergusson, D. M., & Horwood, L. J. (2005).
ety and depression. Journal of Clinical Psychiatry, 51,3539. Childhood abuse and familial violence and the risk of panic
Cox, B. J., Enns, M. W., Walker, J. R., Kjernisted, K., & attacks and panic disorder in young adulthood. Psychological
Pidlubny, S. R. (2001). Psychological vulnerabilities in Medicine, 35, 881890.
patients with major depression vs. panic disorder. Behaviour Goodwin, R. D., Lieb, R., Hoefler, M., Pfister, H., Bittner, A.,
Research and Therapy, 39, 567573. Beesdo, K., & Wittchen, H. U. (2004). Panic attack as a
Craske, M. G., Brown, T. A., & Barlow, D. H. (1991). Behavioral risk factor for severe psychopathology. American Journal of
treatment of panic disorder:Atwo-year follow-up. Behavior Psychiatry, 161, 22072214.
Therapy, 22, 289304. Gorman, J. M., & Coplan, J. D. (1996). Comorbidity of depres-
Craske, M. G., Farchione, T., Allen, L., Barrios, V., Stoyanova, sion and panic disorder. Journal of Clinical Psychiatry, 157
M., & Rose, R. (2007). Cognitive behavioral therapy for (suppl. 10),3441.
panic disorder and comorbidity: More of the same or less Gotlib, I. H., Krasnoperova, E., Yue, D. N., & Joormann J.
of more? Behaviour Research and Therapy, 45, 10951109. (2004). Attentional biases for negative interpersonal stimuli
Craske, M. G., Lang, A. J., Aikins, D., & Mystkowski, J. L. in clinical depression. Journal of Abnormal Psychology, 113,
(2005). Cognitive behavioral therapy for nocturnal panic. 127135.
Behavior Therapy, 36,4354. Grunhaus, L., Pande, A. C., Brown, M. B., & Greden, J. F.
Craske, M. G., Rauch, S. L., Ursano, R., Prenoveau, J., Pine, D. (1994). Clinical characteristics of patients with concurrent
S., & Zingbarg, R. E. (2009). What is an anxiety disorder? major depressive disorder and panic disorder. American
Depression and Anxiety, 26, 10661085. Journal of Psychiatry, 151, 541546.

Castriot ta,Craske 97
Hagnell, O., & Grasbeck, A. (1990). Comorbidity of anxi- agoraphobia:Frequency, severity, and response to treatment.
ety and depression in the Lundby 25 year prospective Archives of General Psychiatry, 45, 437443.
study:The pattern of subsequent episodes. In:J. D.Maser Lydiard, R. B. (1991). Coexisting depression and anxiety:Special
& C. R. Cloninger (Eds.), Comorbidity of mood and anxi- diagnostic and treatment issues. Journal of Clinical Psychiatry,
ety disorders (pp. 139152). Washington, DC: American 51 (suppl. 6),4854.
PsychiatricPress. Maddock, R. J., Carter, C. S., Blacker, K. H., Beitman, B. D.,
Hegel, M. T., Unutzer, J., Tang, L., Arean, P. A., Katon, W., Krishnan, K. R. R., Jefferson, J. W.,...Liebowiz, M. R.
Noel, P. H.,...Lin, E. H. (2005). Impact of comorbid panic (1993). Relationship of past depressive episodes to symptom
and posttraumatic stress disorder on outcomes of collabora- severity and treatment response in panic disorder and agora-
tive care for late-life depression in primary care. American phobia. Journal of Clinical Psychiatry, 54,8895.
Journal of Geriatric Psychiatry, 13,4858. Maier, W., Minges, J., & Lichtermann, D. (1995). The familial
Hettema, J. M., Neale, M. C., Myers, J. M., Prescott, C. A., relationship between panic disorder and unipolar depression.
& Kendler, K. S. (2006). A population based twin study of Journal of Psychiatric Research, 29, 375388.
the relationship between neuroticism and internalizing disor- Mathews, A., & MacLeod, C. (2005). Cognitive vulnerability
ders. American Journal of Psychiatry, 163(5), 857864. to emotional disorders. Annual Review of Clinical Psychology,
Hettema, J. M., Prescott, C. A., & Kendler, K. S. (2004). Genetic 1, 167195.
and environmental sources of covariation between general- McLean, P. D., Woody, S., Taylor, S., & Koch, W. J. (1998).
ized anxiety disorder and neuroticism. American Journal of Comorbid panic disorder and major depression:Implications
Psychiatry, 161, 15811587. for cognitive-behavioral therapy. Journal of Consulting and
Hoffman, S. G., & Smits, J. A. (2008). Cognitive behavioral Clinical Psychology, 66, 240247.
therapy for adult anxiety disorders: A meta-analysis of Merikangas, K. R., Angst, J., Eaton, W. W., Canino, G.,
randomized placebo controlled trials. Journal of Clinical Rubio-Stipec, M., Wacker, H., .
Kupfer, D. J. (1996).
Psychiatry, 69, 621632. Comorbidity and boundaries of affective disorders with anxi-
Issakidis, C., & Andrews, G. (2004). Pretreatment attrition and ety disorders and substance misuse:Results of an international
dropout in an outpatient clinic for anxiety disorders. Acta task force. British Journal of Psychiatry, 168 (suppl. 30),5867.
Psychiatrica Scandinavica, 109, 426433. Meuret, A. E., Wilhelm, F. H., Ritz, T., & Roth, W. T. (2003).
Jacobson, N. S., Martell, C. R., & Dimidjian, S. (2001). Breathing training for treating panic disorder:Useful inter-
Behavioral activation treatment for depression:Returning to vention or impediment? Behavior Modification, 27, 731754.
contextual roots. Clinical Psychology Science and Practice, 8, Meuret, A. E., Wilhelm, F. H., Ritz, T., & Roth, W. T. (2008).
255270. Feedback of end-tidal pCO2 as a therapeutic approach for
Johnson, J., Weissman, M. M., & Klerman, G. L. (1990). Panic panic disorder. Journal of Psychiatric Research, 42, 560568.
disorder, comorbidity, and suicide attempts. Archives of Mineka, S., Watson, D., & Clark, L. A. (1998). Comorbidity
General Psychiatry, 47, 805808. of anxiety and unipolar mood disorders. Annual Review of
Katon, W., & Roy-Byrne, P. P. (1991). Mixed anxiety and depres- Psychology, 49, 377412.
sion. Journal of Abnormal Psychology, 100, 337345. Mogg, K., & Bradley, B. P. (2002). Selective orienting of atten-
Kendler, K. S., Prescott, C. A., Myers, J., & Neale, M. C. (2003). tion to masked threat faces in social anxiety. Behaviour
The structure of genetic and environmental risk factors for Research and Therapy, 40, 14031414.
common psychiatric and substance use disorders in men and Mogg, K., Millar, N., & Bradley, B. P. (2000). Biases in eye
women. Archives of General Psychiatry, 60, 929937. movements to threatening facial expressions in GAD and
Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. depressive disorder. Journal of Abnormal Psychology, 109,
(2005). Lifetime prevalence and age-of-onset distributions 695704.
of DSM-IV disorders in the National Comorbidity Survey Murphy, M. T., Michelson, L. K., Marchione, K., Marchione,
Replication. Archives of General Psychiatry, 62, 593602. N., & Testa, S. (1998). The role of self-directed in vivo expo-
Kessler, R. C., Stang, P. E., Wittchen, H. U., Ustun, T. B., sure in combination with cognitive therapy, relaxation train-
Roy-Byrne, P. P., & Walters, E. E. (1998). Lifetime panic- ing, or therapist-assisted exposure in the treatment of panic
depression comorbidity in the National Comorbidity Survey. disorder with agoraphobia. Behaviour Research and Therapy,
Archives of General Psychiatry, 55, 801808. 12, 117138.
Kircanski, K., Craske, M. G., Epstein, A. M., & Wittchen, H. Nordahl, T. E., Semple, W. E., Gross, M., Mellman, T. A., Stein,
U. (2009). Subtypes of panic attacks:Acritical review of the M. B., Goyer, P.,...Cohen, R. M. (1990). Cerebral glu-
empirical literature, Depression and Anxiety, 26, 878887. cose metabolic differences in patients with panic disorder.
Kraft, A. R., & Hoogduin, C. A. (1984). The hyperventilation Neuropsychopharmacology, 3, 3261272.
syndrome: A pilot study of the effectiveness of treatment. Noyes, R., Reich, J., Christiansen, J., Suelzer, M., Pfohl,
British Journal of Psychiatry, 145, 538542. B., & Coryell, W. A. (1990). Outcome of panic disor-
Laberge, B., Gauthier, J. G., Cote, G., Plamondon, J., & der: Relationship to diagnostic subtypes and comorbidity.
Cormier, H. J. (1993). Cognitive-behavioral therapy of panic Archives of General Psychiatry, 47, 809818.
disorder with secondary major depression: a preliminary Rachman, S., Shafran, R., Radomsky, A. S., & Zysk, E. (2011).
investigation. Journal of Consulting and Clinical Psychiatry, Reducing contamination by exposure plus safety behaviour.
61, 10281037. Journal of Behavior Therapy and Experimental Psychiatry, 42,
Lawson, C., & MacLeod, C. (1999). Depression and the inter- 397404.
pretation of ambiguity. Behaviour Research and Therapy, 37, Rapee, R. M., Craske, M. G., & Barlow, D. H. (1995).
463474. Assessment instrument for panic disorder that includes fear
Lesser, I. M., Rubin, R. T., Pecknold, J. C., & Rifkin, A. of sensation-producing activities: The Albany Panic and
(1988). Secondary depression in panic disorder and Phobia Questionnaire. Anxiety, 1, 114122.

98 Depression and Comorbidit y with Panic Disorder

Rapee, R. M., Craske, M. G., Brown, T. A., & Barlow, resolves with antidepressant treatment: An fMRI study.
D. H. (1996). Measurement of perceived control over Biological Psychiatry, 50, 651658.
anxiety-related events. Behavior Therapy, 27, 279293. Stein, M. B., Tancer, M. E., & Uhde, T. W. (1990). Major
Reiss, S., Peterson, R., Gursky, D., & McNally, R. (1986). depression in patients with panic disorder:Factors associated
Anxiety sensitivity, anxiety frequency, and the prediction of with course and recurrence. Journal of Affective Disorders, 19,
fearfulness. Behaviour Research and Therapy, 24,18. 287296.
Roy-Byrne, P. P., & Cowley, D. (2002). Pharmacologic treat- Stein, M. B, & Uhde, T. W. (1990). Panic disorder and major
ments for panic disorder, generalized anxiety disorder, spe- depression: a tale of two syndromes. Psychiatric Clinics of
cific phobia and social anxiety disorders. In: P. E. Nathan North America; 11, 441461.
& J. Gorman J. (Eds.), A guide to treatments that work (pp. Stein, M. B., Walker, J. R., Anderson, G., Hazen, A. L., Ross, C.
337365). NewYork:Oxford University Press,2002. A., Eldridge, G. & Forde, D. R. (1996). Childhood physical
Roy-Byrne, P., Craske, M. G., Stein, M. B., Sullivan, G., and sexual abuse in patients with anxiety disorders and a com-
Bystritsky, A., Katon, W.,...Sherbourne, C. D. (2005). A munity sample. American Journal of Psychiatry, 153, 275277.
randomized effectiveness trial of cognitive-behavioral ther- Taylor, S., Koch, W. J., & McNally, R. J. (1992). How does anxi-
apy and medication for primary care panic disorder. Archives ety sensitivity vary across the anxiety disorders? Journal of
of General Psychiatry, 62, 290298. Anxiety Disorders, 6, 249259.
Roy-Byrne, P. P., Stang, P., Wittchen, H. U., Ustun, B., Walters, Tsao, J. C.I., Mystkowski, J. L., Zucker, B. G., & Craske, M.
E., & Kessler, R. C. (2000). Lifetime panic depression comor- G. (2002). Effects of cognitive-behavior therapy for panic
bidity in the National Comorbidity Survey:Association with disorder on comorbid conditions:replication and extension.
symptoms, impairment, course, and help-seeking. British Behavior Therapy, 33, 493509.
Journal of Psychiatry, 176, 229235. Walker, E. A., Katon, W. J., Russo, J., Von Korff, M., Lin, E.,
Roy-Byrne, P., Vitaliano, P., Cowley, D., Luciano, G., Zheng, Y., Simon, G.,...Unutzer, J. (2000). Predictors of outcome in
& Dunner, D. L. (1992). Coping in panic and major depres- a primary care depression trial. Journal of General Internal
sive disorder:Relative effects of symptom severity and diag- Medicine, 15, 859867.
nostic comorbidity. Journal of Nervous and Mental Disease, Weissman, M. M., Wickramaratne, P., Adams, P. B., Lish, J.
180, 179183. D., Horwath, E., Charney, D.,...Frosch, E. (1993). The
Safren, S. A., Gershuny, B. S., Marzol, P., Otto, M. W., & relationship between panic disorder and major depres-
Pollack, M. H. (2002). History of childhood abuse in panic sion: Anew family study. Archives of General Psychiatry, 50,
disorder, social phobia, and generalized anxiety disorder. 767780.
Journal of Nervous and Mental Disease, 190, 453456. Westen, D., & Morrison, K. (2001). A multidimensional
Schmidt, N. B., Woolaway-Bickel, K., Trakowski, J., meta-analysis of treatments for depression, panic, and gener-
Santiago, H., Storey, J., Koselka, M., & Cook, J. (2000). alized anxiety disorder:An empirical examination of the sta-
Dismantling cognitive-behavioral treatment for panic disor- tus of empirically supported therapies. Journal of Consulting
der:Questioning the utility of breathing retraining. Journal and Clinical Psychology, 69(6), 875899.
of Consulting and Clinical Psychology, 68, 417424. Wright, C. I., Fischer, H., Whalen, P. J., McInerney, S. C., Shin,
Sheline, Y. I., Barch, D. M., Donnelly, J. M., Ollinger, J. M., L. M., Rauch, S. L. (2001). Differential prefrontal cortex
Snyder, A. Z., & Mintun, M. A. (2001). Increased amygdala and amygdala habituation to repeatedly presented emotional
response to masked emotional faces in depressed subjects stimuli. Neuroreport, 12, 379383.

Castriot ta,Craske 99

7 Depression and Posttraumatic Stress

Disorder Comorbidity

Lisa M.Najavits and Nicole M.Capezza

Depression and posttraumatic stress disorder (PTSD) are highly comorbid diagnoses following a
traumatic event. In this chapter, we explore a range of topics related to comorbid depression and PTSD,
including impact, prevalence, shared risk factors, temporal priority, key research areas, intervention
strategies, and future research directions. Given the overlap in symptoms and shared risk factors, some
researchers have suggested that the comorbidity between depression and PTSD following a traumatic
event may be better understood as a single general mood disorder rather than two separate disorders.
We examine evidence supporting both possibilities. We briefly review the two research areas that have
received the most attention, namely comorbidity related to military traumas and interpersonal abuse.
Practical implications, assessments, interventions, and treatment recommendations are also discussed.
Key Words: posttraumatic stress disorder, PTSD, trauma, depression, comorbidity

Introduction Definitions of the Disorders and of

Introduction to the Specific Comorbidity in thisArea
Comorbidity andIssues PTSD. PTSD is an anxiety disorder that can
Experiencing a traumatic event is often life-altering develop after exposure to a traumatic event.
and in many cases results in serious mental health Traumatic events include experienced events, wit-
problems. Depression and posttraumatic stress dis- nessed events, or events that occur to a close other
order (PTSD) are two common mental health prob- and may involve the threat of death or physi-
lems that co-occur following a traumatic event and cal, sexual, or psychological harm to oneself or
result in serious consequences (e.g., suicidal thoughts another (e.g., rape, combat, natural disasters,
and actions). In this chapter, we explore a range of life-threatening illness, serious car accident, wit-
topics related to comorbid depression and PTSD, nessing a violent death, etc.; American Psychiatric
including prevalence, shared risk factors, temporal Association, 2000). Approximately 15% to 25% of
priority, intervention strategies, and future research individuals that experience a trauma develop PTSD
directions. Throughout the chapter we use the over- (Breslau et al., 1998; Kessler, Sonnega, Bromet,
arching term depression when referring to depres- Hughes, & Nelson, 1995). Symptoms of PTSD
sive mood in general, symptoms of depression, or a include reexperiencing the traumatic event through
history of depression and the term major depres- flashbacks or nightmares, avoidance or arousal to
sive disorder (MDD) when referring to the specific stimuli associated with the event, numbing of feel-
Diagnostic and Statistical Manual of Mental Disorders ings after the event, and hypervigilance (American
(fourth edition [DSM-IV-TR]; American Psychiatric Psychiatric Association, 2000). Formal diagnostic
Association, 2000) diagnosis. criteria per the DSM-IV-TR (American Psychiatric

Association,2000) require that the symptoms last to depressed patients without PTSD, depressed
more than one month and cause significant impair- patients with PTSD have greater psychiatric
ment in social, occupational, or other important symptom severity, higher levels of depression
areas of a personslife. and hostility, higher rates of being discharged
Prevalence. Depression and PTSD are highly against medical advice, higher suicidal behaviors,
comorbid (Kessler et al., 1995). In the United and more medical problems (Calhoun, Wiley,
States, the prevalence of MDD among adults is Dennis, & Beckham, 2009; Cougle, Resnick, &
6.7% annually and 16.6% in a lifetime, and for Kilpatrick, 2009; Holtzheimer, Russo, Zatzick,
PTSD it is 3.5% annually and 6.8% in a life- Bundy, & Roy-Byrne, 2005; Momartin, Silove,
time (Kessler et al., 2005; Kessler, Chiu, Demler, Manicavasagar, & Steel, 2004; Oquendo et al.,
Merikangas, & Walters, 2005). Twenty-six per- 2005; Shalev etal., 1998). Additionally, findings
cent of individuals with PTSD are also diagnosed from a large-scale epidemiological survey found
with current MDD (Maes, Mylle, Delmeire, & that individuals with comorbid current depression
Altamura, 2000). Indeed, within eight months of and PTSD were five times more likely to exhibit
a traumatic event, 23% of exposed adults develop functional impairment compared to those with
MDD, with or without PTSD, often within days of PTSD only (Mollica et al., 1999). These effects
the event (North etal.,1999). have been found across a range of samples, includ-
Preexisting MDD increases risk for both exposure ing clinical (Holtzheimer etal., 2005) and com-
to a traumatic event and to developing PTSD once munity samples (Shalev etal., 1998) and among
an individual is exposed to such an event (Breslau, victims of a range of traumatic events, including
Davis, Peterson, & Schultz, 1997). Additionally, the motor vehicle accidents (Blanchard, Buckley,
occurrence of first-time MDD is higher among indi- Hickling, & Taylor, 1998; Koren, Arnon, &
viduals who develop PTSD following a trauma com- Klein, 1999), war-related trauma (Momartin
pared to those that do not develop PTSD following a etal., 2004; Skodol etal., 1996), natural disasters
trauma (Breslau, Davis, Peterson, & Schultz, 2000). (Tural, Onder, & Aker, 2012), and interpersonal
Thus the relationship between depression and PTSD violence (Lipsky, Field, Caetano, & Larkin,2005).
is multifaceted; depression appears to increase risk Biological differences. In addition to the clinical
for PTSD and viceversa. findings noted above, several studies have found bio-
In the National Comorbidity Survey, among logical differences between individuals with comor-
men and women with PTSD, lifetime prevalence bid depression and PTSD and those with only one
of comorbid depression was approximately 48% disorder. For example, one study using twin data
(Kessler et al., 1995). Additionally, in a national from the Vietnam Era Twin Registry explored the
sample of adolescents ages 12 to 17, among boys genetic and environmental components of cur-
and girls with PTSD, the six-month prevalence of rent depression and PTSD. They found that the,
comorbid depression was 62% (Kilpatrick et al., best-fitting model for the MDPTSD association
2003). Similar high rates of comorbidity have been included a substantial genetic correlation (r=.77;
found in both nationally representative samples 95% confidence interval [CI] .501.00) and a
as well as community samples (Breslau, Davis, modest individual-specific environmental correla-
Andreski, & Peterson, 1991; Kessler etal., 1995). tion (r=.34; 95% CI .19.48; Koenen etal., 2008,
High comorbidity has also been found among a p. 109). Other biological findings showing dif-
variety of different samples, including military ferences between comorbid PTSD and depression
veteran primary care patients (36% comorbidity; and either PTSD or depression alone include varia-
Campbell et al., 2007), urban health care seeking tions in sleep and facial electromyographic activity
women (58% comorbidity; Gill, Page, Sharps, & (Woodward, Friedman, & Bliwise, 1996), cortisol
Campbell, 2008), and Oklahoma City bombing level (Sher, 2005), and cerebrospinal fluid (Sher
survivors (55% comorbidity; North etal.,1999). et al., 2005). Taken together these studies suggest
Impact. A number of studies have compared that there is likely a biological component to the
individuals with comorbid PTSD and depres- relationship between PTSD and depression.
sion to those with only one of these disorders. Risk factors. Given the high comorbidity between
Individuals with both current depression and depression and PTSD, several shared risk fac-
PTSD suffer more negative mental and physical tors have been identified such as a history of prior
health consequences than those with either dis- depression, traumatic event severity, childhood and
order alone (Sher, 2005). For example, compared sexual abuse, and being female (Breslau etal., 1998;

Najavits, Cape zz a 101

Carlson & Rosser-Hogan, 1991; Kendler, Gardner, proposed: (a) Depression is a reaction to PTSD,
& Prescott, 2002; Roberts, Ocaka, Browne, Oyok, (b) PTSD is a reaction to depression, (c) the two
& Sondorp, 2008). In addition, Kilpatrick and col- disorders represent a single general traumatic stress
leagues (2003) found that among a national sample factor, and (d)both disorders are separate and dis-
of adolescents, comorbid current PTSD and depres- tinct reactions to a traumaticevent.
sion was more prevalent among females (compared Moreover, the classic conceptualization of
to males), those with a history of familial drug use Meyer (1986) is an excellent overall frame-
problems (compared to no history), those who had work:(a)Disorder X may cause disorder Y; (b)dis-
witnessed violence (compared to those who had not order Y causes disorder X; (c) both X and Y are
witnessed violence), and those who had experienced caused by some other factor; (d)each disorder arises
sexual and physical assault (compared to no assault). independently, with no relation between them;
Shared symptom patterns. In addition to a num- (e)each disorder may impact the course of the other
ber of shared risk factors, depression and PTSD (improving or worsening), even if not causedbyit.
also share a number of related symptoms. Asubset These explanations highlight a key ques-
of PTSD symptoms have been found to be closely tion:What does a diagnosis of comorbid PTSD and
related to MDD symptoms, specifically numbing depression actually mean? We next review the evi-
and dysphoria (Gros, Simms, & Acierno, 2010; dence supporting some of these explanations.
Simms, Watson, & Doebbeling, 2002). Factor There is not a clear temporal order to the comor-
analyses found that these PTSD symptoms (numb- bidity of depression and PTSD. In fact, the presence
ing and dysphoria) load more strongly with MDD of one disorder increases the likelihood of onset of
symptoms than with other specific PTSD symp- the other. Specifically, individuals with preexisting
toms, such as intrusion, avoidance, and arousal MDD have twice the risk for exposure to traumatic
(Elhai, Contractor, Palmieri, Forbes, & Richardson, events and three times the risk of being diagnosed
2010; Gros et al., 2010). Among a large sample with PTSD following a traumatic event compared
of veterans, participants reported the most severe to those without a history of MDD (Breslau etal.,
current PTSD symptoms when comorbid MDD 1997; Breslau et al., 2000; Bromet, Sonnega, &
and PTSD was present; however, the PTSD only Kessler, 1998). Similarly, being diagnosed with
and the major depression only groups consistently PTSD increases the risk for developing MDD
reported similar scores on all PTSD symptom mea- (including first onset of depression) by nearly three
sures (Gros, Price, Magruder, & Frueh, 2012). These times compared to those exposed to a trauma that
findings suggest that diagnostic tests for comorbid- do not develop PTSD (Breslau et al., 2000). For
ity may not accurately distinguish between the two example, among survivors of the Oklahoma City
disorders. bombing, 55% of individuals with current PTSD
The similarity in symptoms and risk factors asso- developed MDD whereas fewer than 9% of trauma
ciated with comorbid PTSD and MDD following survivors without PTSD had MDD (North etal.,
a traumatic event has led to questions related to 1999). In sum, such findings indicate that PTSD is
whether these disorders should be considered a sin- a risk factor for MDD and vice versaa history of
gle mood disorder or if they are truly two distinct MDD increases the risk forPTSD.
disorders. According to Sher (2005), It is possible A number of studies support the idea that comor-
that some or all individuals diagnosed with comor- bidity of PTSD and depression following a trau-
bid PTSD and depression suffer from a separate matic event represents a single general traumatic
psychobiological condition that can be termed stress factor (sometimes referred to as posttraumatic
post-traumatic mood disorder. This condition is a mood disorder; see Sher, 2005). One study found
result of a trauma, has features of both PTSD and that among accident survivors from intensive care
depression, and is more severe than PTSD alone, or units (mostly motor vehicle accidents), the majority
depression alone (p.208). We address this and other of survivors diagnosed with PTSD and depression
topics related to understanding the development and did not differ from PTSD-only survivors in terms
causal relationship between these two disordersnext. of variables that differentiated them from a group
without PTSD. Specifically, at 3 and 12months the
Issues of Development, Temporal Priority, same combination of variables was able to differenti-
and Cause andEffect ate the PTSD-only group from the no-PTSD group,
Several possible explanations for the comor- and the comorbid-PTSD and depression group
bidity of PTSD and depression have been from the no-PTSD group. Variables included event

102 Depression and Post traumatic Stress Disorder Comorbidit y

characteristics (intensive care unit admission, event in the current study, we identified unique correlates
severity), individual characteristics (prior psychiat- for each condition, which support the premise that
ric and trauma history), cognitive appraisals (anxi- depression and PTSD are independent responses to
ety about the potential impact of the injury), and trauma (Chiu etal., 2011, p.207).
acute responses (reexperiencing, arousal, and depres- In sum, there exists a complex relationship
sion). Different predictors did emerge between the between depression and PTSD. All four explana-
depression-only group and the comorbid group, tions for this comorbidity have received research
suggesting that depression may exist as a separate support. More research is needed to continue to
disorder in some cases. The authors conclude: explore this intricate relationship and to determine
whether diagnoses of depression and PTSD follow-
It is clear that the bulk of psychopathology in the
ing a traumatic event are separate constructs or one
aftermath of trauma is best conceptualized as a general
traumatic stress would seem that the
PTSD and depression symptoms that constitute this
Types ofTrauma
factor are part of a shared vulnerability and thus have
The majority of research on comorbidity of
the same predictive variables. PTSD and comorbid
PTSD and depression generally falls into two cate-
PTSD/depression are effectively one and the same
gories of trauma types:(a)veterans/military-related
thing. The data suggest that depressive symptoms
trauma and (b)interpersonal violence/sexual assault.
are often integral to PTSD and that to separate
We briefly review these two areas of research.
depression out as a distinct disorder when it occurs
with PTSD is a somewhat arbitrary distinction.
(ODonnell, Creamer, & Pattison, 2004, p.1395)
Research on military veterans has consistently
These findings reinforce conclusions from the demonstrated a high prevalence of comorbid PTSD
earlier studies we reviewed (e.g., Breslau etal., 1997; and depression. Moreover, this comorbidity makes
Breslau et al., 2000; ODonnell et al., 2004) that veterans vulnerable to other significant problems
comorbid PTSD and depression in the aftermath such as suicidal ideation and behavior (Lemaire &
of trauma may best be conceptualized as a single Graham, 2010; Oquendo etal., 2005). Long-term
traumatic stress construct with shared risk factors impact of traumatic events on comorbid PTSD and
and symptoms. depression has also been found, such as in a study
The final explanation for the comorbidity of Korean War veterans conducted 50 years after
between PTSD and depression is that the disorders the war. Findings indicated that a notable minority
are separate constructs. Supporting this position, (17%) of veterans had comorbid current PTSD and
several studies indicate variations among different depression and that this comorbidity was associated
outcomes related to PTSD or depression only and with impaired life satisfaction, reduced quality of
comorbid PTSD and depression. For example, a life, and greater symptom severity (Ikin, Creamer,
study of victims of motor-vehicle accidents found Sim, & McKenzie,2010).
that PTSD and MDD were correlated, but inde- Some research has examined gender differ-
pendent, diagnoses following the trauma (Blanchard ences among veterans. For example, a recent study
etal.,1998). (Maguen, Cohen, Cohen, et al., 2012) examined
A recent study (Chiu etal., 2011) examined risk health care utilization among male and female
factors associated with comorbid depression and Operation Enduring Freedom/Operation Iraqi
PTSD in an attempt to elucidate whether depres- Freedom veterans with current PTSD. Findings
sion and PTSD represent separate constructs or a indicate that males and females with comorbid
single general stress reaction to a traumatic event. PTSD and depression were more likely to have
Using a sample of retired World Trade Center fire- higher mental health, primary care, and emergency
fighters, they found that the relationship between care use compared to the PTSD-only group. Notable
current PTSD and alcohol abuse was mediated gender differences emerged such that women with
by depression. Similarly, World Trade Center comorbid PTSD and depression were 12.5 times
arrival time was associated with current depres- more likely to have a mental health inpatient hos-
sion, but this association was mediated by PTSD. pitalization compared to their female counterparts
The authors conclude, Our models suggest that without depression and twice as likely to have a
elevated depression and PTSD risk may be sepa- mental health hospitalization compared to men
rate constructs. After controlling for comorbidity with comorbid PTSD and depression (p.666).

Najavits, Cape zz a 103

Veterans from Operation Enduring Freedom/ the comorbidity (Brand, King, Olson, Ghaziuddin,
Operation Iraqi Freedom who had higher combat & Naylor,1996).
exposure, childhood physical assault, and accident/ In addition, men with a history of childhood
disasters were significantly more likely to have cur- sexual abuse and current comorbid PTSD and
rent PTSD and MDD compared to veterans with- depression are more likely to engage in risky sexual
out these trauma experiences (Dedert etal., 2009). behavior compared to men without an abuse his-
Among veterans from primary care settings, patients tory (Holmes, Foa, & Sammel, 2005). Also, com-
with current comorbid PTSD and depression report pared to men without a history of child abuse, men
more severe depression, lower social support, more that experienced child abuse were found to be three
outpatient health care visits, more suicidal ideation, times more likely to report high PTSD and depres-
more emotional distress, more frequent mental sion symptoms among disaster workers following
health specialty visits, and correspondingly higher the terrorist attacks on September 11, 2011 (Leck,
outpatient mental health care costs compared to Difede, Patt, Giosan, & Szkodny,2006).
depression-only patients (Campbell et al., 2007; Women who are victims of interpersonal and/
Chan, Cheadle, Reiber, Unutzer, & Chaney, 2009). or sexual trauma are also at an increased risk for
Additionally, antidepressants were prescribed to a comorbid PTSD and depression. One key com-
higher proportion of depressed patients with PTSD ponent to developing appropriate treatments and
compared to depression-only patients (61% vs interventions is to gain a better understanding
.40%; Chan etal.,2009). of various coping trajectories that women with
Veterans are at high risk for both PTSD and this comorbidity tend to use (Arriaga & Capezza,
depression, and these have known negative sequelae. 2005). While the research on adult interpersonal
However, more research is needed to understand violence described above focuses on women, men
the many factors that may account for these find- are also victims of interpersonal violence and suf-
ings. Veterans may experience trauma prior to their fer negative consequences, including depression
military service; thus, it is necessary to disentangle and PTSD (Coker, Weston, Creson, Justice, &
the impact of trauma prior to their military career Blakeney, 2005). However, we were unable to locate
as well as postdeployment experiences that impact any studies examining comorbid depression and
their mental health. PTSD among adult males who are victims of inter-
personal violence. Certainly this is an important
Interpersonal Violence topic deserving future research attention.
The comorbidity of depression and PTSD has
been found to be consistently high among victims Assessment and Intervention Strategies
of rape or physical abuse. Approximately 50% of Assessment
interpersonal violence victims report comorbid Several considerations are relevant to assess-
depression and PTSD (Nixon, Resick, & Nishith, ment of comorbid PTSD and depression. As
2004; Stein & Kennedy, 2001). Among a sample of detailed earlier, individuals with this comorbid-
Rwandan women with HIV, high levels of comor- ity are likely to have additional co-occurring Axis
bidity were found such that 82% of women with Iand Axis II disorders. Thus assessment needs to
PTSD also had depression, and of the women address the full array of clinical presentations. It
with depression, 63% also had PTSD (Cohen can be helpful to use a brief DSM-IV structured
etal.,2009). interview, such as the MINI Neuropsychiatric
Compared to only one disorder, risk factors for Interview (Sheehan etal., 1997) or the Structured
comorbid PTSD and depression among victims Clinical Interview for DSM-IV (First, Spitzer,
of interpersonal or sexual violence include suffer- Gibbon, & Williams, 1997a; First, Spitzer,
ing an adult rape or military sexual trauma, greater Gibbon, & Williams, 1997b). By assessing the
psychological abuse, and higher rates/more severe wide range of psychiatric conditions, a profile of
symptoms of PTSD and depression (Lipsky etal., comorbidity, in its most complete sense, can help
2005; Maguen, Cohen, Ren, et al., 2012; Nixon inform treatment planning.
et al., 2004). Similarly, in a sample of adolescents Also, there may be many clinically relevant areas
a history of sexual abuse was related to a higher to assess other than just diagnoses. For example,
prevalence of comorbid PTSD and depression than clinical issues that may be prominent in comorbid
adolescents without such a history. The frequency PTSD and depression may include employment
and severity of the abuse increased the likelihood of problems, family and social problems, motivational

104 Depression and Post traumatic Stress Disorder Comorbidit y

problems, and physical health problems (Evans & Center for Epidemiological Studies Depression
Hser,2004). Scale (Radloff, 1977), and the Zung Self-Rating
Another key consideration (per earlier in this Depression Scale (Zung, 1965). For PTSD, in addi-
chapter) is the significant overlap between PTSD tion to the PTSD Checklist, there are the Trauma
and depression symptoms. Disentangling which Symptom Inventory (Briere, 1995), which obtains
symptoms relate to each disorder and which are a broad range of trauma-related symptoms, and the
genuinely part of both is a challenge. Sometimes Modified PTSD Symptom Scale (Falsetti, Resnick,
a life trajectory approach can be helpful (identi- Resick, & Kilpatrick, 1993), which obtains fre-
fying which disorder arose first and which symp- quency and severity ratings for each of the 17
toms occurred in relation to onset of the disorders). DSM-IV PTSD criteria (and can be used for out-
However, for clinical purposes, the bottom line may come assessment).
simply be that the patient currently meets criteria
for both PTSD and depression and would thus need Intervention
clinical attention forboth. Several general points related to treatment of
It is essential to use validated instruments. The comorbidity are notable before we explore specific
most basic are screening tools that are brief, typi- findings on PTSDdepression comorbidity.
cally self-report, and identify the likelihood that the First, an overall framework on comorbid-
individual may have the disorder. In many settings, ity treatment can be helpful. Treatments can be
such as primary care, substance abuse treatment categorized into the following approaches (for
programs, and front-line community agencies, expanded version see; Najavits etal., 2008; Weiss
there may only be time for screening instruments, & Najavits,1997).
given time and staff limitations. For depression, the
Integrated continuous: Designed to treat both
Patient Health Questionnaire-9 (Kroenke, Spitzer,
disorders at the same time, by the same provider,
& Williams, 2001) is one of the most common
focusing on linkages between them throughout.
currently used screening instruments. The Beck
Integrated phase based: Focuses primarily
Depression Inventory-2 (Beck, Steer, & Brown,
on one disorder, then the other, by the same
1996) is also widely used. It takes longer but pro-
provider and with some attention to both disorders
vides cutoffs for different levels of depression and can
be used as an outcome measure as well as a screen-
Sequential: Treats one disorder, then the
ing tool. For PTSD, the four-item screen designed
other (may be by different providers and may be
for primary care is also widely used in a variety of
based on response to the initial treatment, such
settings (Prins et al., 2003). The PTSD Checklist
that the patient improves in one disorder and then
(Weathers, Litz, Herman, Huska, & Keane, 1993)
addresses the other).
is longer but offers the advantage of mapping onto
Parallel:Also known as concurrent (i.e., treat
DSM-IV criteria for PTSD and offering the option
each disorder but in separate treatments), often by
to have a continuous measure for outcome purposes.
different providers.
For PTSD, it is also essential to obtain a screen for
Single:Treat just one disorder.
trauma itself, for which there are various screening
tools, such as the Life Events Checklist (Gray, Litz, Second, even treatments designed for PTSD or
Hsu, & Lombardo, 2004) and the Stressful Life depression alone may have an impact on the other
Events Screening Questionnaire (Goodman, 1998). disordereven if not designed specifically to treat
If a patient screens positive on the screening tool it; many treatments have generalized impact beyond
(for depression, PTSD, or both), a trained assessor their targeted disorder (Najavits et al., 2008;
can then conduct a full diagnostic assessment using Watts,2007).
DSM-IV criteria (e.g., the MINI Neuropsychiatric Third, in PTSD treatment trials, a sizeable per-
Interview or the Structured Clinical Interview centage of patients inevitably also had comorbid
for DSM-IV, for example, per above). Following depression, even if it was not assessed, given what
assessment, it would be important to obtain out- we have reviewed above on substantial comorbid-
come measurement if the patient is entered into ity of PTSD and depression. The same holds for
treatment. Measures for outcome assessment are depression studies in terms of inclusion of PTSD
plentiful. For depression, in addition to the Beck patients.
Depression Inventory already noted, there are the At this point, we know of only one behav-
Hamilton Depression Scale (Hamilton, 1967), the ioral therapy model specifically designed a priori

Najavits, Cape zz a 105

to address PTSDdepression comorbidity. (For Overall, there is a striking dearth of treatments
information on pharmcotherapies, see Friedman, and treatment outcome trials on PTSDdepres-
2002; Sher etal., 2012; Stewart & Wrobel, 2009). sion comorbidity, considering the large number of
The behavioral model is described by Nixon and patients affected, and its notable impact. However,
Nearmy (2011) and represents a combination of the decades ahead are likely to see new developments
already existing cognitive behavioral treatment in these areas, given the increasingly prominent
components that have received evidence-based sup- focus on comorbities of all kinds and the greater
port for depression or PTSD separately:behavioral prominence of PTSD-related research specifically.
activation for depression in early sessions, followed
by exposure therapy and cognitive restructuring for
Practical Issues and Clinical Guidelines
PTSD in later sessions, for a total of 12 to 16 weeks
for Practitioners
per patient. This approach fits the integrated phase
Several practical suggestions can be offered that
based category described above. They found, in a
may help inform care, drawn from earlier work
sample of 14 patients who completed treatment, a
on comorbidity treatment (e.g., Najavits et al.,
significant decrease in PTSD and depression sever-
1997; Najavits, Ryngala et al., 2008; Weiss &
ity between pre- and midtreatment assessments
and a further decrease in PTSD from mid- to post-
treatment. Gains were maintained at three-month (a) When patients present with comorbid
follow-up; 60% of the sample no longer met PTSD PTSD and depression (or any other comorbidities),
criteria and 70% no longer met MDD criteria at they should be given care that addresses all current
that point. This treatment package is thus a highly disorders.
promising one for future research. (b) Treatment of comorbid disorders is generally
Another example of a treatment trial rel- more complex than treatment of either disorder
evant to PTSDdepression comorbidity is alone. The clinician needs to monitor symptoms
that of Dunn et al. (2007). They evaluated the of both disorders and how they interact over time.
impact of an evidence-based depression treat- Also, the clinician should provide psychoeducation
ment (self-management therapy) on patients so that the patient can recognize and chart
with depression and chronic PTSD. Specifically, symptoms of each disorder.
they randomized 101 male veterans with chronic (c) Patients may be more motivated to work
combat-related PTSD and depressive disorder to on one disorder than the other and may need
the self-management therapy condition versus an encouragement to attend to both for as complete a
active-control therapy. Their primary outcomes, recovery as possible.
using an intent-to-treat design, were subjective (d) The clinician too may feel more connection
and objective PTSD and depression scales at pre- or engagement with one disorder over the other
test, posttest, and 3-, 6-, and 12-month follow-up. and thus may lack balanced attention to the
Secondary outcomes included treatment compli- comorbidity. For example, PTSD may evoke more
ance, satisfaction, treatment-targeted constructs, sympathy as it is so clearly connected to an external
functioning, service utilization, and costs. They event (the trauma); or, alternatively, PTSD may be
found that self-management therapy had modestly more likely to be ignored as historically, at least, it
greater improvement on depression symptoms at typically has been relative to depression (Najavits
treatment completion, but these were not main- etal.,2008).
tained at follow-up. They also found that psychi- (e) Both depression and PTSD are highly
atric outpatient utilization and overall outpatient associated with additional comorbid disorders
costs were lower with self-management therapy. that have major clinical impact, such as substance
They concluded that Despite success in other use disorders and Axis II disorders. The treatment
depressed populations, self-management therapy course and strategies may need to be altered when
produced no clinically significant effect in depres- such additional comorbidities are present.
sion with chronic PTSD (p.221). However, future (f ) Depression or PTSD subtypes may also
research could potentially evaluate the impact of impact the course and treatment of the disorders,
the treatment in other PTSDdepression popu- such as depression with or without psychotic
lations, as results for male veterans with chronic features, early or delayed PTSD, andsoon.
combat-related PTSD may not generalize to other (g) The clinician should recognize that there
populations. may be various subpopulations with elevated rates

106 Depression and Post traumatic Stress Disorder Comorbidit y

of PTSD and depression, including women, the however, were in veteran and military populations,
homeless, adolescents, veterans, and individuals respectively, and so more research is needed before
who are incarcerated. These patient groups may generalizing to other populations.
need different approaches to care or adapted
treatments relevant to theirneeds. Future Directions for Research and Clinical
(h) The clinician should provide referrals to Practice
additional treatments and conduct a thorough 1. There is a need for practice guidelines that
assessment of case management needs. Patients address the comorbidity of PTSD and depression.
with depression and PTSD may need treatment of Such guidelines might address key clinical issues
physical health problems, medication consultation, such as how to address both disorders at the same
couples or family treatment, parenting skills, time and how to monitor patients for potential
and so on. Each of the disorders can additively harm to self or others (which may be increased due
impact their functioning, and, in general, the more to each disorder and their comorbidity).
treatment and the more varied the treatment, the 2. Future research and clinical practice should
better. address additional comorbidities that commonly
co-occur with PTSD and depression, such as
Also, see Campbell etal. (2007) for an excellent substance use disorders and Axis II disorders,
guide to PTSDdepression comorbidity in relation which can have major impacts on clinical
to primary care in particular but that also has rel- presentation.
evance more broadly. They observe that this comor- 3. Future treatment outcome research on PTSD
bidity is common, yet many patients are treated or depression treatments should include rates of
solely for depression. Moreover, patients with the comorbid disorders, even if the treatments are not
comorbidity, compared to those with depression designed to address comorbidity. Treatments often
alone, have a more severe clinical profile, including have impact beyond the targeted disorder, and such
poorer prognosis, delayed response to depression information could be valuable for identification of
treatment, higher likelihood of being disabled, a treatment pathways.
more persistent course of illness, increased preva- 4. Patients with PTSD and depression come
lence of suicidal ideation, and less social support. through many different treatment doors
In short, they have an increased illness burden that that is, systems of care, including primary care,
likely requires different treatment strategies than specialty care, correctional settings, schools,
depressed-only patients. They state, In summary, and veterans hospitals. It will be important to
depressed patients with positive PTSD screens pres- address how treatment may differ depending on
ent differently from those with depression alone. setting-based issues such as provider training and
Relative to what is known about effective primary workload issues, culture of the setting, costs, and
carebased depression treatment (Belnap et al., so on. For example, in primary care, the focus
2006; Kilbourne, Rollman, Schulberg, & Pincus, may be primarily on brief education and referral
2002) less is known about primary care treatment to specialty care. In specialty care, the focus
of PTSD and MDD-PTSD comorbidity (p.716). would be more intensive, such as multimodal
They emphasize the need for further treatment out- approaches (e.g., medications plus behavioral
come research on this population. therapies) as well as long-term management of the
Finally, one study found empirically that patients disorders.
with PTSD and depression, compared to depressed 5. There is a need for more widespread
patients without PTSD, had higher costs of treat- screening and assessment of both PTSD and
ment. They used specialty mental health treatments depression. Including brief self-report screening
and antidepressant medications more and had tools for each disorder as part of the intake process
overall higher mental health care costs. Similarly, may help promote early case-identification and
Fikretoglu et al. (2009) found that most military treatment.
members with PTSD did not seek out mental health 6. More research is needed on the etiology and
treatment but that, of all the predictors, comorbid presentation of comorbid PTSD and depression,
depression most increased the likelihood of seeking including order of onset of the disorders, social and
treatment. Overall, these studies suggest increased biological pathways, and subtypes of the disorders.
treatment seeking (and associated costs) among All of these may impact treatment response as well
people with PTSD and depression. These studies, as are important for scientific advancement.

Najavits, Cape zz a 107

7. Various subpopulations (e.g., women, depression:is the correlation an illusion? Journal of Anxiety
veterans, military, victims of child abuse) have Disorders, 12,2137.
Brand, E. F., King, C. A., Olson, E., Ghaziuddin, N., & Naylor,
particularly high rates of PTSDdepression M. (1996). Depressed adolescents with a history of sexual
comorbidity, and there may be a need for abuse: Diagnostic comorbidity and suicidality. Journal of
specialized or targeted treatments for them. the American Academy of Child and Adolescent Psychiatry,
For example, children and adolescents likely 35(1),3441.
need treatments that take into account their Breslau, N., Davis, G. C., Andreski, P., & Peterson, E. (1991).
Traumatic events and posttraumatic stress disorder in an urban
developmental concerns and use language, population of young adults. Archives of General Psychiatry, 48,
examples, and exercises relevant tothem. 216222.
8. More research examining comorbid Breslau, N., Davis, G. C., Peterson, E. L., & Schultz, L. (1997).
PTSD and depression among male victims of Psychiatric sequelae of posttraumatic stress disorder in
interpersonal violence is needed. women. Archives of General Psychiatry, 54,8187.
Breslau, N., Davis, G. C., Peterson, E. L., & Schultz, L. R.
9. Additional research is needed on the impact (2000). A second look at comorbidity in victims of trauma:
of comorbidity following less-studied traumatic The posttraumatic stress disorder-major depression connec-
experiences, such as life-threatening illness, tion. Biological Psychiatry, 48, 902909.
witnessing a violent death, andsoon. Breslau, N., Kessler, R. C., Chilcoat, H. D., Schultz, L. R.,
10. There is a definite need for more treatment Davis, G. C., & Andreski, P. (1998). Trauma and posttrau-
matic stress disorder in the community: The 1996 Detroit
development and treatment outcome trials focused Area Survey of Trauma. Archives of General Psychiatry, 55,
on PTSDdepression comorbidity, particularly 626632.
considering the large number of patients affected Briere, J. (1995). The Trauma Symptom Inventory (TSI):
by the comorbidity and its notable impact. Professional manual. Odessa, FL: Psychological Assessment
Bromet, E., Sonnega, A., & Kessler, R. C. (1998). Risk fac-
Summary and Conclusions tors for DSM-III-R posttraumatic stress disorder: Findings
Comorbid depression and PTSD is highly preva- from the National Comorbidity Survey. American Journal of
lent and results in many negative consequences. Epidemiology, 147, 353361.
The two disorders have similar impact, share risk Calhoun, P. S., Wiley, M., Dennis, M. F., & Beckham, J. C.
factors, and share symptoms. Given the similarities (2009). Self-reported health and physician diagnosed ill-
nesses in women with posttraumatic stress disorder and major
between the two disorders, researchers have ques- depressive disorder. Journal of Traumatic Stress, 22, 122130.
tioned whether the diagnoses of PTSD and depres- Campbell, D. G., Felker, B. L., Liu, C. F., Yano, E. M., Kirchner,
sion following a traumatic event are actually two J. E., Chan, D....Chaney, E. (2007). Prevalence of depres-
separate disorders or one general mood disorder. sionPTSD comorbidity: Implications for clinical practice
Understanding this distinction may help in develop- guidelines and primary care-based interventions. Journal of
General Internal Medicine, 22, 711718.
ing relevant intervention and treatment efforts for Carlson, E. B., & Rosser-Hogan, R. (1991). Trauma experi-
this comorbidity. The majority of research on PTSD ences, posttraumatic stress, dissociation, and depression in
and depression comorbidity has focused on traumas Cambodian refugees. American Journal of Psychiatry, 148,
related to military experience and interpersonal vio- 15481551.
lence victimization. It is unclear whether these find- Chan, D., Cheadle, A. D., Reiber, G., Unutzer, J., & Chaney, E.
F. (2009). Health care utilization and its costs for depressed
ings would generalize to other traumatic experiences. veterans with and without comorbid PTSD symptoms.
Psychiatric Services, 60, 16121617.
References Chiu, S., Niles, J. K., Webber, M. P., Zeig-Owens, R., Gustave,
American Psychiatric Association. (2000). Diagnostic and statisti- J., Lee, R.,...Prezant, D. J. (2011). Evaluating risk factors
cal manual of mental disorders (4th ed. text rev.). Washington, and possible mediation effects in posttraumatic depression
DC:Author. and posttraumatic stress disorder comorbidity. Public Health
Arriaga, X. B., & Capezza, N. M. (2005). Targets of partner vio- Reports, 126, 201209.
lence. Journal of Interpersonal Violence, 20,8999. Cohen, M. H., Fabri, M., Cai, X., Shi, Q., Hoover, D. R.,
Beck, A., Steer, R., & Brown, G. (1996). Manual for the Beck Binagwaho, A.,...Anastos, K. (2009). Prevalence and pre-
Depression Inventory-II. San Antonio, TX: Psychological dictors of posttraumatic stress disorder and depression
Corporation. in HIV-infected and at-risk Rwandan women. Journal of
Belnap, B. H., Kuebler, J., Upshur, C., Kerber, K., Mockrin, D. Womens Health, 18, 17831791.
R., Kilbourne, A. M., Rollman, B. L. (2006). Challenges of Coker, A. L., Weston, R., Creson, D. L., Justice, B., & Blakeney,
implementing depression care management in the primary P. (2005). PTSD symptoms among men and women survi-
care setting. Administration and Policy in Mental Health and vors of intimate partner violence:The role of risk and protec-
Mental Health Services Research, 33,6575. tive factors. Violence and Victims, 20, 625643.
Blanchard, E. B., Buckley, T. C., Hickling, E. J., & Taylor, A. Cougle, J. R., Resnick, H., & Kilpatrick, D. G. (2009).
E. (1998). Posttraumatic stress disorder and comorbid major PTSD, depression, and their comorbidity in relation to

108 Depression and Post traumatic Stress Disorder Comorbidit y

suicidality: Cross-sectional and prospective analyses of a Holmes, W. C., Foa, E. B., & Sammel, M. D. (2005). Mens
national probability sample of women. Depression and Anxiety, pathways to risky sexual behavior: Role of co-occurring
26, 11511157. childhood sexual abuse, posttraumatic stress disorder, and
Dedert, E. A., Green, K. T., Calhoun, P. S., Yoash-Gantz, R., depression histories. Journal of Urban Health, 82(Suppl.
Taber, K. H., Mumford, M. M.,...Beckman, J. C. (2009). 1):8999.
Association of trauma exposure with psychiatric morbidity in Holtzheimer, P. E., III, Russo, J., Zatzick, D., Bundy, C., &
military veterans who have served since September 11, 2001. Roy-Byrne, P. P. (2005). The impact of comorbid post-
Journal of Psychiatric Research, 43, 830836. traumatic stress disorder on short-term clinical outcome in
Dunn, N. J., Rehm, L. P., Schillaci, J., Souchek, J., Mehta, P., hospitalized patients with depression. American Journal of
Ashton, C. M.,...Hamilton, J. D. (2007). A randomized Psychiatry, 162, 970976.
trial of self-management and psychoeducational group thera- Ikin, J. F., Creamer, M. C., Sim, M. R., & McKenzie, D. P.
pies for comorbid chronic posttraumatic stress disorder and (2010). Comorbidity of PTSD and depression in Korean
depressive disorder. Journal of Traumatic Stress, 20,221. War veterans: Prevalence, predictors, and impairment.
Elhai, J. D., Contractor, A. A., Palmieri, P. A., Forbes, D., & Journal of Affective Disorders, 125, 279286.
Richardson, J. D. (2010). Exploring the relationship between Kendler, K. S., Gardner, C. O., & Prescott, C. A. (2002). Toward
underlying dimensions of posttraumatic stress disorder and a comprehensive developmental model for major depression
depression in a national, trauma-exposed military sample. in women. American Journal of Psychiatry, 159, 11331145.
Journal of Affective Disorders, 133, 477480. Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, K.
Evans, E., & Hser, Y. (2004). Pilot-testing a statewide outcome R., & Walters, E. E. (2005). Lifetime prevalence and age-of-
monitoring system: Overview of the California Treatment onset distributions of DSM-IV disorders in the National
Outcome Project (CALTOP). Journal of Psychoactive Drugs, Comorbidity Survey Replication. Archives of General
36(Suppl. 2), 109114. Psychiatry, 62, 593602.
Falsetti, S. A., Resnick, H. S., Resick, P. A., & Kilpatrick, D. Kessler, R. C., Chiu, W. T., Demler, O., Merikangas, K. R., &
G. (1993). The Modified PTSD Symptom Scale: A brief Walters, E. E. (2005). Prevalence, severity, and comorbidity
self-report measure of posttraumatic stress disorder. Behavior of 12-month DSM-IV disorders in the National Comorbidity
Therapist, 16, 161162. Survey Replication. Archives of General Psychiatry, 62, 617627.
Fikretoglu, D., Elhai, J. D., Liu, A., Richardson, J. D., & Pedlar, Kessler, R. C., Sonnega, A., Bromet, E., Hughes, M., & Nelson,
D. (2009). Predictors of likelihood and intensity of past-year C. B. (1995). Posttraumatic stress disorder in the National
mental health service use in an active Canadian military Comorbidity Survey. Archives of General Psychiatry, 52,
sample. Psychiatric Services, 60, 358366. 10481060.
First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B.W. Kilbourne, A. M., Rollman, B. L., Schulberg, H. C., B., H.B.,
(1997a). Structured Clinical Interview for DSM-IV Axis I & Pincus, H. A. (2002). A clinical framework for depression
Disorders, Clinical Version (SCID-IV). Washington, DC: treatment in primary care. Psychiatric Annals, 32, 545553.
American PsychiatricPress. Kilpatrick, D. G., Ruggiero, K. J., Acierno, R., Saunders, B. E.,
First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. Resnick, H. S., & Best, C. L. (2003). Violence and risk of
B. W. (1997b). Structured Clinical Interview for DSM-IV PTSD, major depression, substance abuse/dependence, and
Personality Disorders, (SCID-II). Washington, DC:American comorbidity:Results from the National Survey of Adolescents.
PsychiatricPress. Journal of Consulting and Clinical Psychology, 71, 692700.
Friedman, M. J. (2002). Future pharmacotherapy for post- Koenen, K. C., Fu, Q. J., Ertel, K., Lyons, M. J., Eisen, S. A.,
traumatic stress disorder: Prevention and treatment. True, W. R.,...Tsuang, M. T. (2008). Common genetic lia-
Psychiatric Clinics of North America, 25, 427441. bility to major depression and posttraumatic stress disorder
Gill, J. M., Page, G. G., Sharps, P., & Campbell, J. C. (2008). in men. Journal of Affective Disorders, 105, 109115.
Experiences of traumatic events and associations with PTSD Koren, D., Arnon, I., & Klein, E. (1999). Acute stress response
and depression development in urban health care-seeking and posttraumatic stress disorder in traffic accident vic-
women. Journal of Urban Health, 85, 693706. tims: A one-year prospective, follow-up study. American
Goodman, L., Corcoran, C., Turner, K., Yuan, N., & Green, B. Journal of Psychiatry, 156, 367373.
(1998). Assessing traumatic event exposure:General issues and Kroenke, K., Spitzer, R. L., & Williams, J. B. (2001). The
preliminary findings for the Stressful Life Events Screening PHQ-9: Validity of a brief depression severity measure.
Questionnaire. Journal of Traumatic Stress, 11, 521542. Journal of General Internal Medicine, 16, 606613.
Hamilton, M. A. (1967). Development of a rating scale for pri- Leck, P., Difede, J., Patt, I., Giosan, C., & Szkodny, L. (2006).
mary depressive illness. British Journal of Social and Clinical Incidence of male childhood sexual abuse and psychologi-
Psychology, 6, 278296. cal sequelae in disaster workers exposed to a terrorist attack.
Gray, M. J., Litz, B. T., Hsu, J. L., & Lombardo, T. W. (2004). International Journal of Emergency Mental Health, 8, 267274.
The psychometric properties of the Life Events Checklist. Lemaire, C. M., & Graham, D. P. (2010). Factors associated
Assessment, 11, 330341. with suicidal ideation in OEF/OIF veterans. Journal of
Gros, D. F., Price, M., Magruder, K. M., & Frueh, B. C. (2012). Affective Disorders, 130, 231238.
Symptom overlap in posttraumatic stress disorder and major Lipsky, S., Field, C. A., Caetano, R., & Larkin, G. L. (2005).
depression. Psychiatry Research, 196, 267270. Posttraumatic stress disorder symptomatology and comor-
Gros, D. F., Simms, L. J., & Acierno, R. (2010). Specificity of bid depressive symptoms among abused women referred
posttraumatic stress disorder symptoms:An investigation of from emergency department care. Violence and Victims, 20,
comorbidity between posttraumatic stress disorder symp- 645659.
toms and depression in treatment-seeking veterans. Journal Maes, M., Mylle, J., Delmeire, L., & Altamura, C. (2000).
of Nervous and Mental Disease, 198, 885890. Psychiatric morbidity and comorbidity following accidental

Najavits, Cape zz a 109

man-made traumatic events: Incidence and risk factors. Roberts, B., Ocaka, K. F., Browne, J., Oyok, T., & Sondorp, E.
European Archives of Psychiatry and Clinical Neuroscience, (2008). Factors associated with post-traumatic stress disor-
250, 156162. der and depression amongst internally displaced persons in
Maguen, S., Cohen, B., Cohen, G., Madden, E., Bertenthal, D., northern Uganda. BioMed Central Psychiatry,8,38.
& Seal, K. (2012). Gender differences in health service utili- Shalev, A. Y., Freedman, S., Peri, T., Brandes, D., Sahar, T., Orr,
zation among Iraq and Afghanistan veterans with posttrau- S. P., & Pitman, R. K. (1998). Prospective study of post-
matic stress disorder. Journal of Womens Health, 21, 666673. traumatic stress disorder and depression following trauma.
Maguen, S., Cohen, B., Ren, L., Bosch, J., Kimerling, R., & Seal, K. American Journal of Psychiatry, 155, 630637.
(2012). Gender differences in military sexual trauma and men- Sheehan, D., Lecrubier, Y., Harnett-Sheehan, K., Janavs, J.,
tal health diagnoses among Iraq and Afghanistan veterans with Weiller, E., Bonora, I.,...Dunbar, G. C. (1997). Reliability
posttraumatic stress disorder. Womens Health Issues, 22,6166. and validity of the MINI International Neuropsychiatric
Meyer, R. (1986). How to understand the relationship between Interview (M.I.N.I.): According to the SCID-P. European
psychopathology and addictive disorders:Another example of Psychiatry, 12, 232241.
the chicken and the egg. In:R. Meyer (Ed.), Psychopathology Sher, L. (2005). The concept of post-traumatic mood disorder.
and Addictive Disorders (pp. 316). NewYork:Guilford. Medical Hypotheses, 65, 205210.
Mollica, R. F., McInnes, K., Sarajlic, N., Lavelle, J., Sarajlic, I., Sher, L., Oquendo, M. A., Li, S., Burke, A. K., Grunebaum,
& Massagli, M. P. (1999). Disability associated with psychi- M. F., Zalsman, G.,...Mann, J. (2005). Higher cerebro-
atric comorbidity and health status in Bosnian refugees living spinal fluid homovanillic acid levels in depressed patients
in Croatia. Journal of the American Medical Association, 282, with comorbid posttraumatic stress disorder. European
433439. Neuropsychopharmacology, 15, 203209.
Momartin, S., Silove, D., Manicavasagar, V., & Steel, Z. (2004). Sher, L., Stanley, B. H., Posner, K., Arendt, M., Grunebaum, M. F.,
Comorbidity of PTSD and depression: Associations with Neria, Y.,...Oquendo, M. A. (2012). Decreased suicidal ide-
trauma exposure, symptom severity and functional impair- ation in depressed patients with or without comorbid posttrau-
ment in Bosnian refugees resettled in Australia. Journal of matic stress disorder treated with selective serotonin reuptake
Affective Disorders, 80, 231238. inhibitors:An open study. Psychiatry Research, 196, 261266.
Najavits, L. M., Ryngala, D., Back, S. E., Bolton, E., Mueser, Simms, L. J., Watson, D., & Doebbeling, B. N. (2002).
K. T., & Brady, K. T. (2008). Treatment for PTSD and Confirmatory factor analyses of posttraumatic stress symp-
comorbid disorders:Areview of the literature. In:E. B.Foa, toms in deployed and nondeployed veterans of the Gulf War.
T. M.Keane, M. J.Friedman, & J. Cohen (Eds.), Effective Journal of Abnormal Psychology, 111, 637647.
treatments for PTSD:Practice guidelines from the International Skodol, A. E., Schwartz, S., Dohrenwend, B. P., Levav, I., Shrout,
Society for Traumatic Stress Studies (2nd ed., pp. 508535). P. E., & Reiff, M. (1996). PTSD symptoms and comorbid
NewYork:Guilford. mental disorders in Israeli war veterans. British Journal of
Najavits, L. M., Weiss, R. D., & Shaw, S. R. (1997). The link Psychiatry, 169, 717725.
between substance abuse and posttraumatic stress disorder in Stein, M. B., & Kennedy, C. (2001). Major depressive and
women: A research review. American Journal on Addictions, post-traumatic stress disorder comorbidity in female victims
6, 273283. of intimate partner violence. Journal of Affective Disorders,
Nixon, R. D., & Nearmy, D. M. (2011). Treatment of comor- 66, 133138.
bid posttraumatic stress disorder and major depressive disor- Stewart, C. L., & Wrobel, T. A. (2009). Evaluation of the effi-
der:Apilot study. Journal of Traumatic Stress, 24, 451455. cacy of pharmacotherapy and psychotherapy in treatment of
Nixon, R. D., Resick, P. A., & Nishith, P. (2004). An exploration combat-related post-traumatic stress disorder:A meta-analytic
of comorbid depression among female victims of intimate review of outcome studies. Military Medicine, 174, 460469.
partner violence with posttraumatic stress disorder. Journal of Tural, U., Onder, E., & Aker, T. (2012). Effect of depression
Affective Disorders, 82, 315320. on recovery from PTSD. Community Mental Health Journal,
North, C. S., Nixon, S. J., Shariat, S., Mallonee, S., McMillen, 48, 161166.
J. C., Spitznagel, E. L., & Smith, E. M. (1999). Psychiatric Watts, B. V. (2007). Electroconvulsive therapy for comorbid
disorders among survivors of the Oklahoma City bombing. major depressive disorder and posttraumatic stress disorder.
Journal of the American Medical Association, 282, 755762. Journal of ECT, 23,9395.
ODonnell, M. L., Creamer, M., & Pattison, P. (2004). Weathers, F. W., Litz, B. T., Herman, D. S., Huska, J. A.,
Posttraumatic stress disorder and depression following & Keane, T. M. (1993, October). The PTSD Checklist
trauma: Understanding comorbidity. American Journal of (PCL):Reliability, validity, and diagnostic utility. Paper pre-
Psychiatry, 161, 13901396. sented at the International Society for Traumatic Stress
Oquendo, M., Brent, D. A., Birmaher, B., Greenhill, L., Kolko, Studies, San Antonio,TX.
D., Stanley, B.,...Mann, J. J. (2005). Posttraumatic stress Weiss, R. D., & Najavits, L. M. (1997). Overview of treatment
disorder comorbid with major depression:Factors mediating modalities for dual diagnosis patients: Pharmacotherapy,
the association with suicidal behavior. American Journal of psychotherapy, and twelve-step programs. In:H. R.Kranzler
Psychiatry, 162, 560566. & B. J. Rounsaville (Eds.), Dual diagnosis: Substance abuse
Prins, A., Ouimette, P., Kimerling, R., Cameron, R. P., Hugelshofer, and comorbid medical and psychiatric disorders (pp. 87105).
D. S., Shaw-Hegwer, J.,...Sheikh, J. I. (2003). The primary NewYork:Marcel Dekker.
care PTSD screen (PC-PTSD):Development and operating Woodward, S. H., Friedman, M. J., & Bliwise, D. L. (1996).
characteristics. Primary Care Psychiatry, 9,914. Sleep and depression in combat-related PTSD inpatients.
Radloff, L. (1977). The CES-D Scale: A self-report depres- Biological Psychiatry, 39, 182192.
sion scale for research in the general population. Applied Zung, W. (1965). A self-rating depression scale. Archives of
Psychological Measurement, 1, 385401. General Psychiatry, 12, 6370.

110 Depression and Post traumatic Stress Disorder Comorbidit y


8 Comorbidity of Social Anxiety

Disorder and Depression

Julia K.Langer and Thomas L.Rodebaugh

Social anxiety disorder (SAD) and major depressive disorder (MDD) are prevalent disorders that exhibit
a high rate of co-occurrence. Furthermore, these disorders have been shown to be associated with each
other, suggesting that the presence of one disorder increases risk for the other disorder. In this chapter,
we discuss relevant theories that attempt to explain why SAD and MDD are related. We propose
that the available evidence provides support for conceptualizing the comorbidity of SAD and MDD as
resulting from a shared underlying vulnerability. There is evidence that this underlying vulnerability is
genetic in nature and related to trait-like constructs such as positive and negative affect. We also discuss
the possibility that the underlying vulnerability may confer tendencies toward certain patterns of thinking.
Finally, we discuss theories that propose additional causal pathways between the disorders such as direct
pathways from one disorder to the other. We advocate for a psychoevolutionary conceptualization
that links the findings on the underlying cognitions to the shared relation of lower positive affect and
the findings on peer victimization. We suggest that, in addition to a shared underlying vulnerability, the
symptoms of social anxiety and depression may function as a part of a behavior trap in which attempts
to cope with perceived social exclusion lead to even higher levels of social anxiety and depression. Finally,
we make recommendations for the best methods for assessing SAD and MDD as well as suggestions for
treating individuals with both disorders.
Key Words: social anxiety disorder, major depressive disorder, depression, comorbidity, conceptualizing
the comorbidity

Introduction rate of comorbidity between these disorders and

Social anxiety disorder (SAD) and major depres- various theories or models that attempt to explain
sive disorder (MDD) are common psychological why these disorders are related, many of the avail-
disorders, with lifetime prevalence rates estimated able theories are either difficult to falsify or not
to be 12.1% and 16.6%, respectively (Kessler, strongly supported by the data that are currently
Berglund, etal., 2005). The high rate of comorbid- available. We discuss a conceptualization that we
ity between these two disorders has been well doc- believe is most consistent with the available data as
umented (Beesdo et al., 2007; Brown, Campbell, well as propose additional explanatory factors that
Lehman, Grisham, & Mancill, 2001; Kessler, Chiu, we find plausible but require further testing. First,
Demler, & Walters, 2005; Ohayon & Schatzberg, we make the case that the specific relationship of
2010; Stein et al., 2001), and, of the anxiety dis- SAD to MDD is important to consider. Second, we
orders, SAD exhibits one of the highest degrees discuss the best methods for measuring comorbid-
of association with MDD (Kessler, Chiu, et al., ity. Third, we discuss the conceptualization of SAD
2005). Despite a large body of research on the and MDD that we find most consistent with the

data. Fourth, we review research that is consistent turn to a discussion of the best method for measur-
with this conceptualization, and, finally, we discuss ing the degree of overlap and association between
potential additions to this conceptualization that these disorders.
we find plausible. We also discuss the best methods
for assessing social anxiety and depression as well as Rates of Comorbidity
outline strategies for treating individuals who pres- In terms of the best method for capturing the
ent with both disorders. degree of association between SAD and MDD,
we find the Kessler, Chiu, etal. (2005) comorbid-
Comorbidity Between SAD and ity study to be one of the most useful because the
MDD:More than the Sum of itsParts authors report correlation results rather than merely
Why is it important to consider the comorbidity the rate of overlap between the disorders. As noted
between social anxiety and depression? It seems to by Krueger and Markon (2006), the term comor-
us that comorbidity might be reasonably ignored if bidity does not distinguish between correlation
having two disorders in one person was equivalent, and overlap. The correlation between two disorders
in terms of burden or distress, to having one disor- captures the extent to which the disorders co-occur
der each in two people. We believe that research has at higher rates than would be expected by chance
established reasonably clearly, however, that having within a population. In contrast, a high degree of
both disorders together actually creates more dis- overlap might simply represent chance (or even
tress and burden. below chance) co-occurrence of two common dis-
Essentially, the comorbidity of SAD and MDD orders. Although mere co-occurrence may be of
is associated with impairment beyond merely hav- interest (e.g., two disorders need not be associated
ing one of the disorders. For example, Bruce etal. to have compounding effects), we find correlations
(2005) reported that the addition of MDD to SAD (or associated methods, such as odds ratios) more
worsens the already low level of recovery associated compelling than mere co-occurrence. We expect
with SAD. Furthermore, researchers have reported that most researchers and clinicians are, similarly,
that individuals with both SAD and MDD are more interested in comorbidity because it indicates pos-
likely to have additional anxiety disorders, see a psy- sible correlation, not mere co-occurrence.
chiatrist, take psychotropic medications, have more It is thus of considerable interest that Kessler,
intense suicidal ideation, experience more depres- Chiu, etal. (2005) report a significant and moder-
sive symptoms, and have a longer duration of major ate tetrachoric correlation between a diagnosis of
depressive episodes (Ohayon & Schatzberg, 2010; SAD within a 12-month period and a diagnosis of
Stein etal., 2001) than individuals with either SAD MDD within a 12-month period (r=.52), which, in
or MDD. Additionally, Stein et al. reported that their analysis, is the second highest association with
individuals with both disorders (vs. only one) also MDD among the anxiety disorders. Such an associa-
had much greater odds of having attempted suicide tion implies that there is something specific about
during a follow-up period. Indeed, previous research one of these disorders itself that increases the odds of
indicates that depression, social anxiety, and their having the other during the same time period, above
interaction each predict unique variance in suicidal and beyond more general factors (e.g., SAD being an
ideation, indicating that there are both additive and anxiety disorder or internalizing disorder).
interactive effects of social anxiety and depression on Note that any method of assessing comorbidity,
suicidal ideation (Norton, Temple, & Pettit,2008). either in terms of association or overlap, must make
We find the final point mentioned above most use of a specific timeframe. Disorders can be assessed
compelling, because an interaction between the two in terms of lifetime prevalence, onset of disorder dur-
disorders in predicting suicidal ideation provides ing specific time frames (measured longitudinally),
clear evidence that comorbidity between SAD and or current diagnoses, which raises the question of
MDD is not merely the sum of its parts. The pres- which method is of most importance in understand-
ence of such interactive effects is a vital reason to ing comorbidity. Although lifetime prevalence rates
take comorbidity into account in both research and may be useful for some purposes, the documented
treatment. In this section, we have argued for the limitations of retrospective reporting (e.g., Henry,
importance of comorbidity without giving an oper- Moffitt, Caspi, Langley, & Silva, 1994) lead to ques-
ational definition for it. Given that our ability to tionable accuracy of cross-sectional lifetime preva-
consider comorbidity in research will be influenced lence estimates. Measuring current diagnoses or
by our method for assessing comorbidity, we now following participants over time have greater claims

112 Comorbidit y of Social Anxiet y Disorder and Depression

to potential accuracy and precision. Longitudinal Heimberg, 2005), researchers have sought to deter-
data are particularly important because they not mine what accounts for this high degree of over-
only side-step the issue of retrospective reporting lap. Three basic theories regarding this overlap have
but also allow researchers to test the ability of one either been articulated or implied by the available
disorder to prospectively predict another. For exam- literature:(a)SAD and MDD co-occur because of a
ple, Beesdo etal. (2007) assessed participants four shared underlying vulnerability that fully accounts
times over a 10-year period and found that 50.2% for the high rate of comorbidity; (b)in addition to
of respondents with SAD also had a depressive dis- a shared underlying vulnerability, additional causal
order at some point during the observation period. pathways exist between the two disorders; and
We believe it is obvious that longitudinal data are (c) there are intervening variables that arise from
preferable, but they are often not obtainable, and one disorder that confer risk for the other disorder.
many findings regarding MDD and SAD focus on Prior to reviewing these theories in detail, it seems
retrospective lifetime prevalence or current diagno- prudent to give an overall sense of their current sup-
sis. For example, Ohayon and Schatzberg (2010) port. Available data support statement (a)and can
reported that 19.5% of those with current SAD also be seen as consistent with (b) or (c), but a test of
met criteria for currentMDD. whether SAD and MDD continue to be associ-
We have thus far referred to MDD, but the ated above and beyond any specific shared vulner-
comorbidity between SAD and depression is often ability (e.g., genetic predispositions) has yet to be
studied in broader terms. Some studies consider completed.
diagnoses of only MDD as depression (e.g., Ohayon Speaking strictly to the available data, we believe
& Schatzberg, 2010), whereas others include MDD that the best conceptualization of the relationship
with other mood disorders such as dysthymia (e.g., between SAD and MDD is that these disorders are
Van Ameringen, Mancini, Styan, & Donison, related because of a shared vulnerability conferred
1991). MDD is characterized by depressed mood by low positive and high negative affect. Much of
or loss of interest more days than not for at least two this shared vulnerability may be genetic in origin,
weeks, whereas dysthymia is a form of depression in yet our primary assertion is that whatever leads
which the individual is not without the depressed individuals to have lower positive and higher nega-
mood for more than two months over at least a tive affect on a chronic basis will tend to confer risk
two-year period (American Psychiatric Association, for both disorders. We review the evidence for this
2000). We find it plausible that both MDD and proposition below, primarily in regard to positive
unipolar depressed mood more broadly may have affect because negative affect appears to be a risk fac-
important relationships withSAD. tor for emotional disorders in general.
In regard to future research and the conceptu- Available data can also be found that are consis-
alization of depression, we recommend measuring tent with the hypothesis that SAD, or the conse-
both dysthymia and MDD but reporting the results quences of SAD, confer additional risk for MDD,
with the disorders considered separately as well as above and beyond the shared vulnerability. For
combined to assess how these two methods influ- example, given that both disorders are related to
ence the findings. If more studies currently in the lower positive affect, it has been suggested that a
literature had followed this recommendation, we causal pathway exists between SAD, lower posi-
could have included in our review the question as tive affect, and MDD. At present, no single study
to what impact including dysthymia in comorbid- tests the plausibility of a causal pathway from SAD
ity for SAD would have. Absent a critical mass of to MDD longitudinally while simultaneously
literature in this regard, we have focused on being accounting for shared vulnerability (e.g., premorbid
clear regarding the existing literature: We use the lower positive affect). Absent such tests, we review
term MDD when a study restricted the sample the data that is consistent with additional causal
to only those with this diagnosis, but we use the pathways primarily to suggest what further testing
word depression to discuss the general condition of is needed.
lowmood. In the next sections, we examine the shared vul-
nerability and specific causal pathways theories and
Theories Regarding SAD and evidence in more detail. First, we review research
Depression Comorbidity that supports a shared underlying genetic vulnera-
Given the substantial overlap in occurrence bility as well as research that is often cited as support
between SAD and MDD (e.g., Gibb, Coles, & for a direct association but can also be explained by

Langer, Rodebaugh 113

a shared vulnerability. Second, we propose a char- of SAD and depression comorbidity, this conclu-
acterization of the shared underlying vulnerability. sion would be clearer if the study were replicated
Third, we discuss propositions of additional causal with social anxiety symptoms rather than general
pathways that seem plausible but require direct anxiety symptoms.
testing. Fourth, we discuss research on treatment Given the substantial relationship between extra-
response in the context of the shared vulnerability version and neuroticism and the mood and anxi-
versus specific causal pathways theories. Finally, we ety disorders (Clark, Watson, & Mineka, 1994),
discuss recommendations for assessment and inter- researchers have investigated the genetic relation-
vention for SAD andMDD. ship between these personality traits and various
psychological disorders. Bienvenu, Hettema, Neale,
Evidence for Shared Underlying Prescott, and Kendler (2007) reported that the
Vulnerability Versus Specific Causal genetic contribution to extraversion and neuroti-
Pathways cism completely accounted for the genetic risk for
Heritability SAD. These personality traits have a strong genetic
Previous research supports a genetic influ- component (Jang, Livesley, & Vernon, 1996), sug-
ence for both SAD and MDD (Fyer, Mannuzza, gesting that they may mediate the relationship
Chapman, Liebowitz, & Klein, 1993; Kendler between genetic makeup and SAD. The authors did
et al., 2011; Ogliari et al., 2006). Additionally, not find evidence for a shared environmental liabil-
twin studies suggest a genetic association between ity or genetic risk specific to SAD (over and above
SAD and MDD (Kendler, Neale, Kessler, Heath, that related to extraversion and neuroticism).
& Eaves, 1993; Merikangas & Angst, 1995), and Findings on the genetic risk for both SAD and
relatives of people with SAD have been found to MDD support a theory that allows for shared
have an increased risk of MDD even after control- underlying genetic and environmental vulnerabili-
ling for the risk of depression in the person with ties for both disorders. However, there may be addi-
SAD (Fyer et al., 1993). These findings suggest tional unique genetic risks for each disorder as well
that the comorbidity between SAD and MDD as unique environmental contributions. The find-
is at least partly explained by shared genetic risk. ings of Bienvenu etal. (2007) suggest that any such
Kendler et al. (1993) reported that 48% of the genetic risk affects trait-like affective variables, such
observed comorbidity between SAD and MDD was as neuroticism, extraversion, or positive and nega-
explained by shared genetic factors, whereas 52% tive affect, more directly than symptoms of specific
was due to unique environmental factors that were disorders.
shared by both disorders. This finding suggests that
although shared genetics likely explains part of the Associated Risk Between the Disorders
high rate of comorbidity between SAD and MDD, Abundant evidence that SAD and depression
other shared vulnerabilities are present such as com- occur together, as well as evidence that depression
mon environmental influences that are unique to an often follows SAD (Beesdo et al., 2007; Brown
individual (not shared within families). et al., 2001) could be cited as evidence that there
Kendler, Heath, Martin, and Eaves (1986) pro- is a direct causal relationship between the two dis-
vide additional evidence supporting a shared genetic orders. However, the data available do not make
risk for both anxiety and depression as well as specific clear that any such apparent causal relationship
environmental contributions. The authors found maintains above and beyond an underlying vulner-
support for a general genetic risk for both anxiety ability. Thus, although existing studies are certainly
and depression symptoms and a more specific risk consistent with the notion that one disorder (pri-
conferred by nonshared environmental influences. marily SAD) specifically leads to the other, the same
The authors also found support for an independent studies do not rule out the possibility that a shared
pathway model in which genes had direct pathways underlying vulnerability serves as a more parsimoni-
to symptoms, rather than a pathway extending from ous explanation of thedata.
genes, to a disorder, and then to the symptoms. This For example, Stein et al. (2001) reported that
finding suggests that any existing genetic risk does participants ages 18 to 24 who qualified for SAD
not confer risk for a specific syndrome; rather, the at baseline but did not meet criteria for depres-
risk is directly related to symptoms, including both sion (current or previous) were significantly more
anxiety and depression symptomatology. Although likely than participants with no mental disorder to
we expect these results are applicable to the question meet criteria for a depressive disorder during the

114 Comorbidit y of Social Anxiet y Disorder and Depression

follow-up period. For participants who met criteria least that a specific causal pathway from SAD to
for a depressive disorder at baseline, an additional MDD is more plausible than that between most
diagnosis of SAD (either current or previous) at other disorders and MDD. There is also evidence
baseline approximately doubled the odds of devel- that the age of onset of SAD may influence the age
oping subsequent depression or depression that of onset of MDD. Dalrymple and Zimmerman
persisted to the follow-up period. However, given (2011) reported that participants with an onset
a shared vulnerability of some type between SAD of SAD in childhood were more likely to report
and MDD, it would be inevitable that people in the an onset of MDD before age 18 when compared
population with the greatest levels of vulnerability to participants with an onset of SAD in adult-
should be more likely to have at least one of the hood. This finding suggests that the age of onset of
disorders at any given time. The development of the SAD may influence the age of onset of comorbid
other disorder at another time point might there- MDD, which could provide evidence consistent
fore depend merely on the level of the vulnerability with a causal relationship between SAD and MDD.
rather than a specific pathway from one disorder to However, these results do not provide an adequate
another. test of whether there is an association between SAD
and MDD above and beyond a shared vulnerability,
Temporal Relationship because no proposed shared vulnerability was mea-
The tendency for SAD to predate the onset sured or accounted for in these studies. We thus do
of MDD in individuals with both disorders has not know if the age of onset of SAD would con-
been well documented in the literature (Brown tinue to predict the age of onset of MDD above and
et al., 2001; deGraaf, Bijl, Spijker, Beekman, & beyond a shared vulnerability. It is plausible that
Vollebergh, 2003) and could be interpreted as evi- greater vulnerability leads to earlier age of onset of
dence for a direct causal pathway. SAD in general both disorders, rather than earlier age of onset for
tends to begin at an early age (e.g., Dalrymple & SAD specifically leading to earlier onset of MDD.
Zimmerman, 2011; deGraaf et al., 2003; Kessler, Theories proposing a specific pathway from SAD to
Berglund, etal., 2005), especially for the generalized MDD are therefore consistent with available evi-
(nonspecific) subtype (Wittchen, Stein, & Kessler, dence, but theories proposing a shared vulnerability
1999). Most reports of the typical age of onset of leading to both are just as consistent and arguably
SAD range from childhood to the middle adoles- more parsimonious.
cent years, with the highest estimates in the early
twenties (Brown et al., 2001; Kessler & Berglund Cognitions
etal., 2005), whereas for MDD, the estimates tend In addition to evidence that SAD and MDD
to be the late twenties to the early thirties (Kessler, share genetic risk, literature on the cognitions
Berglund, etal.,2005). associated with each disorder suggests common
Rapee and Spence (2004) suggested one cau- cognitive features that may represent either addi-
tion to the interpretation of these data as mean- tional shared vulnerability or a different symp-
ing that SAD causes depression:SAD may predate tom of the same vulnerability. For example, both
depression solely because of its earlier age of SAD and MDD have been linked to negative
onset, rather than because of a causal relationship. self-statements and cognitive discrepancies (Dozois
Indeed, SAD is associated with the earliest age & Frewen, 2006; Strauman, 1989). These findings
of onset of the Axis Idisorders (age at 50th per- are described in more detail below in the Theorized
centile of age-of-onset distribution=13; Kessler, Characterization of the Underlying Vulnerability
Berglund, et al., 2005) and has been reported section. Additionally, it appears that SAD and
to be the disorder that most frequently precedes MDD interact to influence cognitive features such
other disorders in a clinical sample (Brown etal., as attentional bias and interpretationstyle.
2001). These findings suggest that SAD occurs Here, we evaluate whether the interactive effect
before other disorders because the early age of of the disorders on cognitions provides support
onset is a feature of the disorder, rather than an for the shared vulnerability versus the direct causal
indication that SAD is causing the other disorders pathway theory. Grant and Beck (2006) investi-
that tend to followit. gated the influence of comorbidity on attentional
On the other hand, when MDD is secondary to biases in social anxiety and dysphoria using the
other disorders, it most often follows specific pho- emotional Stroop task. The authors compared
bia and SAD (deGraaf et al., 2003), suggesting at responses to this task between three groups: a

Langer, Rodebaugh 115

group with higher social anxiety, a group with these findings within the context of theories for the
higher dysphoria, and a group with both higher comorbidity of SAD andMDD.
social anxiety and higher dysphoria. Based on pre- It appears that in some instances the addition of
vious research, the authors hypothesized that par- depression to social anxiety leads to a decrease on
ticipants in the group with higher social anxiety an index of severity (attentional bias toward threat),
would show an attentional bias to socially threat- whereas in other cases, the addition of depression
ening stimuli. They predicted that the groups with to social anxiety leads to an increase on an index of
higher dysphoria would show no attentional biases severity (negative interpretations). These findings
to any threat stimuli. To justify their hypotheses, provide further evidence that the combination of
the authors cited previous research that suggests social anxiety and depression within an individual
that when anxiety and depression are comorbid, has an influence on variables such as cognitive bias
depression may counteract the attentional biases and interpretation that is different than either alone.
associated with anxiety (Bradley, Mogg, Miller, However, these results do little to strengthen support
& White, 1995; Mogg, Bradley, Williams, & for the shared underlying vulnerability theory versus
Mathews, 1993). The authors found that the group the direct causal pathway theory; both theories could
with higher social anxiety showed longer latencies be seen as consistent with the data. For example,
to color name the social anxiety and depression these findings could be interpreted as evidence that
threat words relative to the group with higher dys- MDD has a direct influence on SAD because the
phoria and the group with higher social anxiety addition of MDD to SAD resulted in a reduction of
and higher dysphoria. Based on these results, social the attention bias toward threat. However, we find it
anxiety and depression may interact to alleviate more plausible that the results are explained by both
the bias toward social and depressive threat stimuli disorders being caused by greater levels of a shared
associated with higher social anxiety. Why this underlying vulnerability. The underlying vulnerabil-
would be the case remains unclear. These results ity to certain cognitions may lead to more attention
suggest that attentional bias toward threat stimuli bias at moderate levels of the vulnerability but less
may be a feature specific to anxiety. However, in attention bias at higher levels of the vulnerability.
previous research depression has been associated Someone who has SAD without MDD (moder-
with negative attention biases (e.g., Joormann, ate level of vulnerability) might be biased to detect
Talbot, & Gotlib, 2007), so it is clearly not the social threat in order to avoid it, whereas someone
case that depression never confers attention bias with SAD and MDD (higher level of vulnerabil-
toward negative stimuli. ity) may have given up trying to avoid social threat
In addition to associations with attentional because of their more extreme negative interpreta-
biases, both social anxiety and depression have been tions. The interpretation of social threat as not only
linked to negative interpretative biases. Wilson and catastrophic but also unavoidable is more negative
Rapee (2005) investigated the relationship between than viewing it as catastrophic but avoidable, yet an
depression, social anxiety, and attributional style attention bias might only be promoted by the belief
in a sample of treatment-seeking individuals with that the catastrophe can be avoided. Therefore we
SAD. The participants were divided into two would expect the comorbidity of SAD and MDD to
groups: those with and those without a comorbid be associated with more negative interpretations but
mood disorder. Based on a self-report measure of reduced attention bias toward threat. Our expecta-
interpretative style, the authors concluded that both tions on this point are largely driven by our charac-
SAD and depression were associated with more neg- terization of the underlying vulnerability, which we
ative interpretations. Furthermore, the group with expound on in the next section.
both SAD and a mood disorder reported believing
their negative interpretations significantly more Characterization of the Underlying
than the group with SAD but no mood disorder. Vulnerability
These results suggest that social anxiety and depres- How can we characterize the shared underlying
sion may have an additive quality in producing neg- vulnerability of SAD and MDD and how does this
ative interpretations. Notably, this finding (in regard vulnerability, or set of vulnerabilities, give rise to
to interpretation bias) contrasts with the Grant and the respective symptom patterns that we observe?
Beck (2006) finding, in which depression seemed We believe that the common vulnerability may be
to reduce attention bias toward social threat. In the the presence of an excessive belief that one is not
following section we describe one explanation for good enough. The proposition that seeing oneself

116 Comorbidit y of Social Anxiet y Disorder and Depression

as not good enough is related to depression should prevent further defeats (Allen & Badcock, 2003;
be uncontroversial: Feelings of worthlessness and Price, Sloman, Gardner, Gilbert, & Rhode, 1994).
excessive guilt are criteria for the disorder (American In a recent review, Johnson, Leedom, and Muhtadie
Psychiatric Association, 2000). Some theorists have (2012) found support for a link between feelings
proposed that SAD is not related to a trait-like ten- of subordination or submissiveness within a domi-
dency to view oneself as not good enough (Clark nance hierarchy and anxiety (especially social anxi-
& Wells, 1995), proposing that people with SAD ety) and depression across a variety of assessment
only experience such thoughts in social situations. modalities (self-report, observational, biological).
However, subsequent research has failed to sup- Weeks, Rodebaugh, Heimberg, Norton,
port this assertion, instead finding that people with and Jakatdar (2009) investigated a submissive
higher social anxiety tend to view themselves more cognitions model of social anxiety to test the
negatively in domains apparently unrelated to social psycho-evolutionary theory of social anxiety. The
situations (e.g., Moscovitch, Orr, Rowa, Reimer, & authors proposed that, based on evolutionary theo-
Antony,2009). ries of social anxiety and depression, social anxiety
On the level of theory, we find the belief of being should be associated with measures related to sub-
not good enough to be a plausible shared vulnera- missive cognitions. The authors found support for
bility because both disorders are related to ones sta- a model in which measures related to submissive
tus within social hierarchies, specifically feeling that cognitions loaded on a higher order submissive cog-
one is of lower status compared to others (Gilbert, nitions factor. Furthermore, based on factor scores
Allan, Brough, Melley, & Miles, 2002; Wilhelm & derived from this factor, submissive cognitions were
Parker, 1989). This belief would then function to found to be more related to a measure of social
influence ones response to a social defeat (percep- interaction anxiety than to either of two measures
tion of loss of social status), which is theorized to of generalized anxiety symptoms. Submissive cogni-
be a key event that can lead to a either a submis- tions also mediated the relationship between social
sive (i.e., socially anxious) or defeat (i.e., depressive) comparison and submissive behaviors. These find-
response. We first review the psychoevolutionary ings suggest that social anxiety is indeed related to
theory of social anxiety and depression because it submissive cognitions, which are related to submis-
provides the basis for our hypothesis that a belief sive behaviors. The fact that submissive cognitions
in being not good enough provides the shared vul- mediated the relationship between social compari-
nerability between the disorders. We then review son and submissive behaviors could indicate that
a study that links social anxiety and depression to only some individuals who are faced with a lower
social defeats. Finally, we draw on relevant research social rank will tend to respond with submissive
from the cognitive literature that is consistent with behaviors. These individuals may have an underly-
the hypothesis that SAD and MDD are associated ing vulnerability to submissive cognitions, which
with cognitions related to negative views of theself. may be a shared underlying vulnerability to social
Trower and Gilbert (1989) have proposed an anxiety and depression.
evolutionary theory that conceptualizes both social Gilbert etal. (2002) presented evidence that is
anxiety and depression as strategic responses to a consistent with the theory that anxiety and depres-
competitive social environment. According to this sion are related to defeats within a social hierarchy.
theory, individuals who have higher social anxi- The authors investigated the relationship between
ety feel that they are of lower status and are, thus, social comparison, defeat, and scales of anhedonic
unlikely to achieve a dominant status within a social depression and anxious arousal. All variables were
hierarchy. These individuals may employ a submis- measured through self-report questionnaires.
sive strategy to avoid competition and communicate Structural equation modeling results suggested a
to others that they are not a threat. If this strategy higher association of defeat with anhedonic depres-
is unsuccessful, an individual may use a strategy of sion in comparison to anxious arousal. Social com-
escape or avoidance. If these strategies are still not parison was also related to defeat and was nearly
successful, the individual may enter a depressive significantly related to anhedonic depression. The
state of resignation or despair, in which the strug- authors suggested that it may be evolutionarily
gle for survival is abandoned. Psycho-evolutionary adaptive for defeats to lead to lower positive affect
theorists have suggested that depression may be an (measured by the anhedonic depression scale in
adaptive strategy to social threat because it is associ- this study) so as to lower activity and explorative
ated with reduced exploratory behaviors, which may behavior, thus helping the individual to avoid

Langer, Rodebaugh 117

further defeating interactions. Though this study of how others think they should be), whereas par-
was cross-sectional and does not provide informa- ticipants diagnosed with MDD reported a greater
tion on causality, plausible causal relationships can discrepancy between their actual personal attri-
be considered. It seems plausible that defeats could butes and their ideal attributes (state to which
lead to a depressive-type response that would be individual personally aspires). Additionally, par-
associated with lower positive affect and reduced ticipants with MDD showed increases in dejected
exploratory behaviors such as social contact, as mood following ideal self-discrepancy priming,
has been suggested by authors such as Allen and whereas participants with SAD showed increases
Badcock (2003) and Price et al. (1994). We also in agitated mood following ought self-discrepancy
theorize that submissive cognitions or a general priming. In evaluating the relation of this study
belief in being less than others may mediate the to comorbidity, it is important to note that no
relationship between defeats and a depressive-type SAD and depression comorbidity was present in
response. Such mediation would suggest that cer- the sample.
tain individuals are more prone to develop depres- In a study that attempted to replicate the
sion when a defeat occurs; however, this theory findings of Strauman (1989) while account-
needs to be tested against the alternative theory ing for comorbidity, Weilage and Hope (1999)
that certain individuals are more prone to defeats included individuals with both SAD and dys-
in general. It could be that one plausible route from thymia to assess the influence of comorbidity on
social anxiety to depression is that social anxiety self-discrepancies. The authors found that indi-
arises out of perceived low-rank status, which may viduals with both SAD and dysthymia reported
make defeats more likely, thus leading to increased higher actualideal and higher actualought dis-
depression and decreased positive affect. crepancies compared to normal controls; however,
Findings from research on self-statements individuals with dysthymia alone reported equiva-
(positively and negatively valenced adjectives that lent levels of actualought discrepancy relative to
are rated in terms of applicability to the self ) in those with SAD alone. Furthermore, individuals
relation to SAD and MDD are consistent with with dysthymia alone did not differ from normal
the hypothesis that both disorders share a vulner- controls in actualideal discrepancy. These results
ability to negative beliefs about the self. It appears partially conflict with the results of Strauman.
that both negative and positive self-statements Overall, however, the picture provided by these
are associated with social anxiety and depression data is that both SAD and depression or dysthy-
but that the symptom clusters differ in the degree mia are related to increased cognitive discrepancy.
to which each type of self-statement is related. It also appears that SAD plus dysthymia confers
Dozois and Frewen (2006) measured the dis- more cognitive discrepancy than either condi-
tance between self-statements as an indication of tion alone. These results are consistent with the
the cognitive availability of these types of state- hypothesis that both disorders share a vulnerabil-
ments. The authors found that both participants ity to certain cognitions such as a belief that one
with MDD and participants with SAD reported is inferior or less than others. This shared vulner-
more distance among positive self-statements and ability may explain the high rate of comorbidity
less distance among negative self-statements as between the disorders, and individuals who are
compared to nonpsychiatric and anxiety control comorbid for both conditions may be individu-
groups (diagnosed with an anxiety disorder other als who have a greater cognitive vulnerability (i.e.,
than SAD). The decreased distance between nega- greater cognitive discrepancy, for example).
tive self-statements was interpreted as an indica-
tion that negative statements are more readily Potential Causal Pathways
available and seen as more related to the self for from SADtoMDD
those with either MDD orSAD. Although we cannot rule out that the following
Studies on self-discrepancies by Strauman data are better explained by a shared underlying
(1989) and Weilage and Hope (1999) provide vulnerability to both disorders, we feel it is impor-
more evidence for negative views of the self in tant to present the following discussion of theories
those with SAD and MDD. In Strauman partici- for potential causal pathways between the disorders
pants diagnosed with SAD reported greater dis- (including proposed intervening variables), because
crepancy between their actual personal attributes they have received attention in the literature. We
and their ought personal attributes (perceptions first discuss three theories that have at best modest

118 Comorbidit y of Social Anxiet y Disorder and Depression

support and have not been tested rigorously with partial support from the testing completed thus far.
comparison to competing theories (helplessness More specifically, a direct test of whether helpless-
hopelessness theory, exhaustion response theory, ness and hopelessness play a causal role in the devel-
and behavioral avoidance theory). We then discuss opment of social anxiety and depression symptoms
two theories that describe potential intervening is sorely needed.
variables between SAD and MDD (positive affect
theory and early life experiences theory); we find Exhaustion ResponseTheory
these latter two potential pathways as more plau- Some authors have proposed that chronic anxi-
sible than those proposed by the first three theories. ety may lead to an exhaustion response that even-
More precisely, it seems to us that the latter two tually results in depression (e.g., Akiskal, 1990).
theories might better explicate the same phenomena Kessler, Stang, Wittchen, Stein, and Walters (1999)
the first three theories attempt to capture. investigated this theory, predicting that if this
were true in regard to SAD the rates of depression
HelplessnessHopelessnessTheory would increase as the duration of SAD increases.
Mineka, Watson, and Clark (1998) developed Additionally, the authors predicted that if SAD
a model that integrates the hopelessness theory of were a true risk factor, only current SAD, but not
depression with the perceived uncontrollability remitted, would predict depression (if SAD were a
associated with anxiety (Wells & Carter, 2001). marker of vulnerability rather than a specific risk
Although this theory does not relate to social anxiety factor for depression, remitted SAD might continue
specifically, we still find the theory relevant, because to relate to the risk of depression). Kessler et al.
fears about social situations may behave in a simi- (1999) reported that current SAD predicted mood
lar fashion to fears in general, though a direct test disorders overall with an increased odds ratio for
of this assumption is needed. This model suggests SAD with three or more social fears compared to
that individuals who are uncertain about their ability one fear. For individuals with only one social fear
to control important outcomes will be most likely (e.g., public speaking), SAD was not related to the
to experience anxiety without depression, whereas severity or course of mood disorders. Remitted SAD
individuals who are certain of both helplessness and had no significant effect in predicting the first onset
occurrence of negative outcomes will experience the of either mood disorders or dysthymia and mod-
hopelessness of depression (Alloy, Kelly, Mineka, & est effects in predicting major depression. However,
Clements, 1990). This model is supported by evi- the relationship between SAD and mood disorders
dence that anxiety is associated with threat or dan- did not vary significantly with time since the first
ger and is thus related to uncontrollability of events, onset other than for those with same-year onsets
whereas depression is often associated with a loss (i.e., both disorders started within the same year).
(e.g., Brown, Harris, & Eales, 1993). This model These findings provide partial support for the above
could explain the temporal relationship between the predictions, though the lack of relationship between
disorders because helplessness could come first (from the duration of SAD and rates of depression casts
a vulnerability to anxiety) and hopelessness could some doubt on the exhaustion theory. Additionally,
develop after a specific negative outcome occurs. the exhaustion theory appears to relate only to more
In an experience-sampling study of the helpless- severe or generalized cases of SAD and may not be
nesshopelessness theory, Swendsen (1997) assessed able to explain co-occurrence with depression in less
participants five times per day for a week regard- severe cases or in the case of specific SAD. We spec-
ing experience of negative events, attributions, and ulate that SAD does not produce exhaustion per se
anxiety and depression. Swendsen found partial but rather that, for reasons including SAD severity
support for the helplessnesshopelessness theory; but not exclusive to SAD severity, some people with
increases in hopeless attributions predicted increases SAD give up avoiding social exclusion and thus
in residual depressed mood following negative life become more vulnerable to depression. We elabo-
events. However, control attributions (helplessness) rate on this idea further after concluding our review
did not predict residual anxiety scores. We were of specific theories of how SAD leads to depression.
able to find only two additional specific tests of this
theory (Swendsen, 1998; Waikar & Craske, 1997), Behavioral AvoidanceTheory
both of which found only partial support for the It has been proposed that behavioral avoid-
theory. The hopelessnesshelplessness theory is a ance may be the primary link between social anxi-
promising theory with little direct testing and only ety and depression (Moitra, Herbert, & Forman,

Langer, Rodebaugh 119

2008). According to this theory, social anxiety to that of reduced social contact and positive affect,
leads to behavioral avoidance, which in turn leads as well as the psychoevolutionary theory of SAD
to depression. Moitra etal. tested this theory in a and MDD, as reviewed above. Specifically, the
sample of treatment-seeking individuals assessed at belief that one is not good enough or of low status
baseline, six weeks into treatment, and at the end could lead to behavioral avoidance as a strategy to
of treatment. They found that behavioral avoid- avoid further defeats.
ance partially mediated the relationship between
social anxiety and depression symptoms. Changes PositiveAffect
in behavioral avoidance from pretreatment to Lower positive affect, once thought to be a
midtreatment were significantly correlated with unique feature associated with depression, has also
midtreatment to posttreatment changes in depres- been linked to higher social anxiety. We propose that
sive symptoms. lower positive affect functions as both an underly-
This theory provides support for an intermedi- ing vulnerability as well as a consequence of social
ate variable or condition along the path between the anxiety and depression. First, we review evidence
two disorders. Behavioral avoidance may mediate that supports the shared relation to lower positive
the relationship, or, alternatively, the consequences affect. Second, we consider the possibility that both
of behavioral avoidance could activate an underly- disorders relate to lower positive affect because of
ing vulnerability to depression. Finally, there are deficits in social activities. Third, we consider other
likely other contributing factors besides behavioral explanations for the shared relation, such as a shared
avoidance because it explained only part of the vari- underlying vulnerability to lower positive affect.
ance in depression in Moitra etal. (2008). Notably, Figure 8.1 depicts the theoretical pathways that we
the concept of behavioral avoidance is very similar consider during this discussion.

1. Lower positive affect represents a deficit in the ability to come in contact with pleasurable

Depression or Decrease in
Lower PA
social anxiety social contact

2. Lower positive affect represents an enduring deficit in the ability to experience pleasure.

Lower positive

3. Lower positive affect is both a cause and a consequence of social anxiety and depression.

of social anxiety to lower
positive affect
and depression

Increased risk
for social
anxiety and

Figure8.1 Potential causal pathways among social anxiety, depression, and lower positive affect.

120 Comorbidit y of Social Anxiet y Disorder and Depression

In proposing alternatives to the Diagnostic and A study by Kashdan and Steger (2006) on the
Statistical Manual of Mental Disorders (fourth edi- relationship between trait social anxiety, daily social
tion, text revision; American Psychiatric Association, anxiety, and positive events provides some support
2000) categorical system, researchers have modeled for the idea that social anxiety is related to reduced
shared relationships with basic underlying traits positive affect because of a lack of contact with posi-
such as positive and negative affect (e.g., Watson, tive affect-inducing activities. The authors found
2009). For example, the tripartite model of anxiety that trait social anxiety was inversely associated with
and mood disorders models the common relation- daily positive emotions and daily positive events,
ship to higher negative affect that is characteristic above and beyond variance explained by trait nega-
of the anxiety and mood disorders, as well as the tive affect. The relationship between trait social anx-
unique relationship to lower positive affect that iety and daily positive events remained significant
is characteristic of depression (Clark & Watson, above and beyond variance explained by depression,
1991). This model has also been adapted to account whereas the relation to daily positive emotions was
for accumulating research that shows that general- reduced to nonsignificance. The authors also found
ized SAD is unique from the other anxiety disorders that individuals in the group with higher social
in that it also shows a consistent relation to lower anxiety reported 39% fewer daily positive events
positive affect (Brown, Chorpita, & Barlow,1998). than those in the group with lower social anxiety.
The shared relation of social anxiety and depres- Furthermore, individuals with higher social anxiety
sion to positive affect has been a consistent find- with greater daily social anxiety (defined as at least
ing in the literature (Brown et al., 1998; Hughes one standard deviation above the mean) reported
et al., 2006; Kashdan, 2004; Kashdan & Steger, 24% fewer daily positive events than individu-
2006; Naragon-Gainey, Watson, & Markon, 2009; als with higher social anxiety with less daily social
Watson, Clark, & Carey, 1988), both at the symp- anxiety. These findings provide some support for
tom level (Watson etal., 1988) and the diagnostic the hypothesis that higher social anxiety is related
level (Brown et al., 1998). Watson et al. reported to lower positive affect because of decreases in social
that negative affect was related to the majority of activities. However, the nonsignificant association
the anxiety and depression symptoms measured, between trait social anxiety and daily positive emo-
whereas positive affect was related to 11 of 20 tions above and beyond depression is unexpected
depression symptoms but only 3 of 33 anxiety given the documented independent relationship
symptoms. These three symptoms (nervousness, between social anxiety and positive affect.
being in a crowd, and speaking in public) were all These findings suggest that within the context
related to social anxiety. Indeed, positive affect cor- of higher social anxiety and depression, lower posi-
related with MDD and dysthymia diagnoses as tive affect could result from the deficits in social
well as SAD diagnoses. Brown etal. found similar contact associated with these symptoms. If this
results; the authors reported that higher negative were the case, a causal pathway could link these
affect related to generalized anxiety disorder, SAD, symptoms in the following order: (1) social anxi-
panic disorder, and MDD, whereas lower positive ety, (2)reduced social contact, (3)reduced positive
affect related to SAD andMDD. affect, (4)increased depression. Once an individual
In an attempt to explain why SAD and MDD was depressed, positive affect would likely be further
are related to lower positive affect, researchers have reduced due to the association between depression
suggested that the deficits in social contact that and reduced social contact. If this were the case, this
are characteristic of these disorders may produce relationship could at least partly explain the high
decreases in positive affect. Both higher social anxi- rate of association between the disorders.
ety (Rodebaugh, 2009; Schneier et al., 1994) and Although it seems likely that deficits in social
higher depression (Joiner & Coyne, 1999) have been contact lead to reductions in positive affect for those
associated with reduced social contact. Additionally, with higher social anxiety and depression, lower
sociability and social interactions have both been positive affect may function as an underlying vulner-
shown to relate to positive affect (Brown, Silvia, ability that confers risk to both disorders, rather than
Myin-Germeys, & Kwapil, 2007; Costa & McCrae, as a consequence of either disorder (second model
1980; Vittengl & Holt, 2000), suggesting that social in Figure 8.1). As noted above, research on the heri-
contact could mediate the relationship between these tability of extraversion and neuroticism (traits that
disorders and lower positive affect. The first model in are strongly related to positive and negative affect,
Figure 8.1 depicts this relationship. respectively) suggests that ones genetic makeup

Langer, Rodebaugh 121

confers risk for certain levels of positive and negative depression and social anxiety because of further loss
affect that then lead to various forms of psychopa- of social bonds. It could be that certain individuals,
thology (Bienvenu etal., 2007). Therefore, a suscep- in certain contexts or cultures, are able to use the
tibility to lower positive affect (or extraversion) may submissive or depressive techniques in a temporary,
exist as both an enduring underlying trait as well effective manner but others tend to fall into a behav-
as a feature that is exacerbated by consequences of ioral trap that leads to chronic depression or social
the disorders such as reductions in social activities anxiety, which would be equivalent to MDD, SAD,
(depicted in the third model in Figure8.1). or their combination. See Figure 8.2 for a visual
Though not described in terms specific to posi- depiction of this theory. We believe that this theory
tive affect, a combination of the evolutionary and is compatible with the hypothesis that SAD and
behavioral theories described above can be used MDD share a vulnerability to certain beliefs such as
to explain how social anxiety, depression, reduced the belief that one is inferior to others; we propose
social contact, and positive affect might have a ten- that this vulnerability may determine whether one
dency to influence each other. In the face of a social is able to use a submissive or withdrawal response
threat or the threat of social exclusion, a depressive effectively or whether one is drawn into a behavior
withdrawal or submissive social anxiety response trap. We suggest that those with a cognitive vulner-
may have been evolutionarily adaptive because such ability would be more likely to be drawn into the
responses would help one to avoid further exclusion behavior trap because they would be more likely
and promote self-reflection designed to avoid fur- to interpret an interaction as a social exclusion or
ther difficulty (Allen & Badcock, 2003; Price etal., defeat.
1994). This response would be associated with a
reduction in the reward system such that explor- Early Life Experiences
atory and approach behaviors would be reduced. In addition to positive affect, early life experi-
The reduction in these behaviors would be associ- ences, such as bullying, have been explored as a
ated with lower positive affect, which would likely possible cause of social anxiety and depression and
lead to temporary further decreases in positive affect have also been explored as a mediator between the
as a result of reduced social contact. If such a sys- two conditions. We suspect that such a relationship
tem were to be adaptive, it would need an exit point would be related to the constellation of social con-
leading to more normative behaviors:For example, tact, positive affect, social anxiety, and depression
a person might resume exploration and seeking of noted above. Specifically, peer exclusion may func-
reward once it was clear that social exclusion was tion as one type of social defeat that could lead to
no longer likely, perhaps because current social rank the behavior trap described above. In one instance
was clearly defined and defensible. of related research, Gazelle and Rudolph (2004)
In the absence of clear evidence that social exclu- investigated the role of the environment in the rela-
sion is no longer a danger, the relationship between tionship between anxious solitude and peer exclu-
perceived social exclusion, reduced exploratory sion in fifth- and sixth-graders, theorizing that peer
behaviors, and reduced positive affect could be exclusion may mediate the relationship between
seen as a behavioral trap. The term behavior trap anxiety and depression. The authors found that
describes a situation in which an individuals natu- for anxious boys with high anxious solitude, high
rally occurring behavior is reinforced and this rein- peer exclusion was associated with increased levels
forcement leads to even higher rates of the original of depression over time, whereas low exclusion was
behavior, which is then further reinforced (Baer & associated with decreasing levels of depression over
Wolf, 1970). The result of this cycle from a negative time. For girls, high anxious solitude and high peer
reinforcement perspective is that the individual will exclusion predicted elevated depressive symptoms.
end up receiving more of the unpleasant stimulus The relationship between anxious solitude and
he or she was trying to avoid. In this example, an depression did not differ depending on high versus
attempt to cope with a threat of social exclusion low exclusion as it did in boys. It may be that social
leads to reduced contact, which might paradoxically anxiety leads to depression only under certain cir-
increase the threat of further exclusion (e.g., because cumstances, such as when a stable negative event of
the individual has not maintained or strengthened a specific type (e.g., peer exclusion) happens to a
existing ties). Once an individual has responded person with elevated tendencies toward social anxi-
with a submissive or withdrawal technique, that ety, leading to depression. These findings are consis-
person may be more likely to experience further tent with an intervening variable that is related to

122 Comorbidit y of Social Anxiet y Disorder and Depression

Underlying Objective
vulnerability to social exclusion
perceive oneself as (not required)
not good enough

Temporary exit when

perceived rank is Higher
sensitivity to Perceived
clear, defensible, and social
adequate for survival exclusion

when social
seen as
Reduced inevitable
exploratory and Increased
Increased social anxiety

Lower Social
positive avoidance

Figure8.2 Depiction of the behavioral trap between lower positive affect, social anxiety, and depression.

social anxiety and has a direct path to depression; vulnerability. We review research from the treat-
however, further testing is required to determine ment literature that could be interpreted as support
whether this additional causal pathway is not better for either the shared underlying vulnerability the-
accounted for by an underlying shared vulnerability. ory or the direct causal pathway theory. We suggest
It might be that peer exclusion mediates the rela- that, as it stands, the treatment literature does little
tionship between social anxiety and depression, for to advance one theory over theother.
example; yet it seems equally plausible that a vul- There are four main findings from the treat-
nerability that increases a persons tendency to view ment literature that are relevant to our discussion
peer exclusion as catastrophic might confer vulner- of theories for the comorbidity between SAD and
ability to social anxiety to begin with and promote MDD: (a) Cognitive behavior therapy (CBT) for
the development of depression when such exclusion SAD has been shown to alleviate comorbid depres-
occurs. sion (Erwin, Heimbrg, Juster, & Mindlin, 2002);
We propose that the potential intervening vari- (b)individuals with both SAD and MDD have been
ables of lower positive affect and peer exclusion rated as having more severe social anxiety by clinician
deserve further testing to determine whether they rating (Joormann, Talbot, & Gotlib, 2007); (c)indi-
function as mediating variables in a causal pathway viduals with both disorders have shown an increase in
from SAD to MDD or whether they are related to social anxiety symptoms posttreatment, and baseline
an underlying vulnerability. To continue our discus- scores on a depression measure have predicted the
sion of whether a shared underlying vulnerability worsening of social anxiety symptoms from termi-
versus direct causal pathways better explains the nation to follow-up (Marom, Gilboa-Schechtman,
comorbidity between SAD and MDD, we now turn Aderka, Weizman, & Hermesh, 2009); and
to research on treatment response. (d)Decreases in social anxiety symptoms mediated
decreases in depression but decreases in depres-
Treatment sion did not mediate decreases in social anxiety
Research on treatment response that considers (Moscovitch, Hofmann, Suvak, & In-Albon, 2005).
the comorbidity between SAD and MDD has the In relation to finding (d), it is important to note that
potential to inform whether additional causal path- the participants in this study were recruited specifi-
ways exist above and beyond a shared underlying cally for SAD, not MDD. Findings (a)and (d)could

Langer, Rodebaugh 123

be seen as evidence for a direct pathway from SAD to a common question is how to assess and treat
MDD, because treatment for SAD alleviated depres- this comorbidity. For the assessment of SAD and
sion symptoms or predicted decreases in depression MDD we recommend the use of a few self-report
symptoms. However, we suggest that what is needed instruments as well as a structured interview. For
is a test that includes the measurement of the shared self-report questionnaires, we recommend the Social
underlying vulnerability to see whether decreases in Interaction Anxiety Scale (SIAS) and the Social
SAD symptoms or treatment of SAD continue to Phobia Scale (Mattick & Clarke, 1998) for SAD and
predict decreases in depression symptoms above and the Beck Depression Inventory-II (Beck, Steer, &
beyond the underlying vulnerability. It could be that Brown, 1996) for MDD. We recommend omitting
treatment for SAD addresses underlying tendencies the reverse-scored items when calculating a total for
for both disorders. Regarding findings (b) and (c), the SIAS because previous research indicates that
although these results could be seen as support for these items decrease the overall validity of the scale
depression having a direct influence on social anxi- (Rodebaugh etal., 2011). We recommend a score
ety, we believe that it is just as likely that the pres- of 28 as a cutoff on the straightforward SIAS items
ence of depression in those with SAD is a marker to indicate probable SAD; this score is comparable
for greater levels of a shared underlying vulnerability. to the existing cutoff of 34 for the original total of
the SIAS (Brown etal., 1997; Heimberg, Mueller,
Conclusions Holt, Hope, & Liebowitz, 1992; Rodebaugh etal.,
We believe that evidence exists to suggest that 2011). Importantly, utilizing both the SIAS and the
the comorbidity of SAD and MDD is at least partly Social Phobia Scale together will give a comprehen-
explained by a common underlying vulnerability sive picture of the social anxiety symptom severity.
that is likely largely inherited. There are likely other For the Beck Depression Inventory-II, the authors
common vulnerabilities beyond genetics and poten- recommend the following cutoffs:zero to 9 (mini-
tially direct causal pathways between the disorders. mal depression), 10 to 18 (mild depression), 19 to 29
The research reviewed above suggests to us that a (moderate depression), and 30 to 63 (severe depression;
common underlying vulnerability to both disorders Beck etal.,1996).
is conferred by genetics and enduring heritable traits In addition to self-report instruments, we
such as tendencies toward negative and positive strongly recommend the use of a structured interview
affect. These heritable traits may influence the cogni- such as the MINI International Neuropsychiatric
tive styles that relate to social anxiety and depression. Interview, Version 5.0 (Sheehan etal., 1998) or the
An individual whose genetic makeup predisposes her Structured Clinical Interview for DSM-IV (Spitzer,
to experience more negative affect and less positive Williams, & Gibbon, 1995). Astructured interview
affect may be at increased risk for the development will be helpful in determining whether the client
of a depressive disorder or SAD. Tendencies toward meets criteria for a Diagnostic and Statistical Manual
higher negative affect and lower positive affect will for Mental Disorders (fourth edition) diagnosis of
likely influence her cognitions and level of approach SAD and/or MDD as well as helping to determine
and exploratory behaviors. These tendencies may the duration of the disorders. The self-report instru-
increase the risk for social threats such as bullying ments are helpful for a continuous measurement
or social exclusion. Under conditions of perceived of the symptomatology, which can indicate level of
social threat, she may respond with a submissive severity.
strategy and a further reduction in reward respon- For intervening with clients who meet criteria
siveness and exploratory behaviors. These behaviors for both MDD or dysthymia and SAD, we rec-
may lead to further reductions in social rank and ommend beginning with a cognitive behavioral
decreases in the experience of positive affect. This intervention for SAD. CBT for SAD is an effective
cycle may lead to the withdrawal tactic that is char- treatment for alleviating social anxiety symptoms
acteristic of a depressive state. When these tactics are (Clark etal., 2003, 2006; Zaider, Heimberg, Roth,
employed in a chronic, maladaptive fashion, they Hope, & Turk, 2003) and has also been shown to
may be characteristic of clinical levels of social anxi- alleviate comorbid depression (Erwin etal., 2002).
ety and/or depression. Instead of attempting to determine which disorder
is primary, which is often difficult or impossible,
Assessment and Intervention Strategies we recommend beginning with CBT for SAD for
Given that comorbidity between SAD and the majority of clients and beginning with a treat-
MDD is both important and relatively prevalent, ment for depression only when a clients depression

124 Comorbidit y of Social Anxiet y Disorder and Depression

is so severe as to demand immediate attention. For for dimensional constructs underlying diagnoses
targeting depression symptoms either before begin- (e.g., degree of social anxiety, degree of depressed
ning CBT for SAD or treating residual depression mood) might allow greater power for distinguishing
after a course of CBT for SAD, we recommend cog- unique genetic variance. Along similar lines, further
nitive therapy, behavioral activation, or other CBT research on unique environmental influences on
treatments designed specifically for depression. each disorder would also help to explicate any dif-
Cognitive therapy or CBT for depression should ferences in etiology between the disorders.
include cognitive restructuring as well as behav- Given the high degree of association between
ioral activation interventions. Behavior activation these disorders and the theorized causal pathways
alone has produced promising results for alleviating between the two, one of the most interesting areas
depression (Dimidjian etal., 2006; Dobson etal., for future research might be to investigate cases in
2008). Fortunately, there is substantial overlap which the disorders occur independently. These cases
between behavioral activation and CBT for SAD in may tell us more about the relationship between
the behaviors that are recommended to the client. these two disorders by hinting at conditions that
For example, in behavioral activation the primary need to be present for both to occur. Additionally,
objective is to help the client become more active, we found many instances in the literature in which
which nicely mirrors the primary goal of CBT for MDD was theorized to be secondary or a conse-
SAD of having the client participate in exposure to quence of SAD but hardly any suggestion of SAD
social activities. In this way, we expect that, for many being a consequence of MDD. Nevertheless, the
clients, treatment for SAD will alleviate comorbid treatment literature provided some suggestion of a
depression either directly through the increase in bidirectional causal link between the two disorders.
activity or indirectly as the decrease in social anxiety Further investigation of when SAD may be sec-
symptoms leads to an improved sociallife. ondary to MDD should be undertaken by future
Another important point to consider is that clients researchers to test whether there is a causal pathway
with both SAD and MDD may have more severe from MDD to SAD. Finally, researchers in each of
symptoms than clients with either disorder by itself the literature areas investigated should consider how
(Joormann, Talbot, & Gotlib, 2007). Additionally, their results may explain the association between
comorbid clients may be more likely to show increases social anxiety and depression. For example, research
in social anxiety symptoms posttreatment, and this in the cognitive area provides evidence on shared
risk may be greater for those with higher levels of and unique cognitive features but little description
depression at the beginning of treatment (Marom of how the cognitive style associated with one disor-
etal., 2009). However, previous research suggests that der may relate to the symptomatology of the other
comorbid depression does not interfere with SAD disorder. For example, it could be that the negative
treatment and vice versa (Erwin etal., 2002; Kashdan cognitions associated with depression may make
& Roberts, 2011). Clinicians should be encouraged social anxiety more likely because persons who
that treatment for SAD is likely to be effective for feel that they are worthless may be more likely to
clients with and without comorbid depression but try to hide their self from others and believe that
may want to consider the severity of depression when they are more likely to be rejected. Attempts to
planning for the length of treatment and planning for move research on the comorbidity between SAD
relapse prevention with comorbid clients. and MDD in this direction will help to reveal the
specific mechanisms that explain the high degree of
Future Directions association between these disorders.
There are many questions that remain unan-
swered in relation to the high degree of overlap References
Akiskal, H. S. (1990). Toward a clinical understanding of the
between SAD and MDD. Much of the research
relationship of anxiety and depressive disorders. In: J.
we reviewed here focuses on shared features of D.Maser & C. R.Cloninger (Eds.), Comorbidity of mood and
the disorders, but research on which features may anxiety disorders (pp. 597607). Washington, DC:American
be unique to the disorders could be further devel- Psychiatric Association.
oped. For example, we found very little research Allen, N. B., & Badcock, P. B.T. (2003). The social risk hypoth-
esis of depressed mood: Evolutionary, psychosocial, and
that addresses whether either disorder has a unique
neurobiological perspectives. Psychological Bulletin, 129,
genetic risk above and beyond shared genetic risk, 887913.
but this may be an area that warrants further test- Alloy, L. B., Kelly, K. A., Mineka, S., & Clements, C.
ing. We propose that research regarding genetic risk M. (1990). Comorbidity of anxiety and depressive

Langer, Rodebaugh 125

disorders: A helplessnesshopelessness perspective. In: J. placebo-controlled trial. Journal of Consulting and Clinical
D.Maser & C. R.Cloninger (Eds.), Comorbidity of mood and Psychology, 71, 10581067.
anxiety disorders (pp. 499543). Washington, DC:American Clark, L. A., & Watson, D. (1991). Tripartite model of anxi-
Psychiatric Association. ety and depression: Psychometric evidence and taxonomic
American Psychiatric Association. (2000). Diagnostic and statisti- implications. Journal of Abnormal Psychology, 100, 316336.
cal manual of mental disorders (4th ed., text rev.). Washington, Clark, L. A., Watson, D., & Mineka, S. (1994). Temperament,
DC:Author. personality, and the mood and anxiety disorders. Journal of
Baer, D. M., & Wolf, M. M. (1970). The entry into natural Abnormal Psychology, 103, 103116.
communities of reinforcement. In: R. Ulrich, T. Stachnik, Clark, D. M., & Wells, A. (1995). A cognitive model of social
& J. Mabry (Eds.), Control of human behavior (Vol. 2, pp. phobia. In:R.G. Heimberg, M. R.Liebowitz, D. A.Hope,
319324). Glenview, IL:Scott, Foresman. & F. R.Schneier (Eds.), Social phobia:Diagnosis, assessment,
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Beck Depression and treatment (pp. 6993). NewYork:GuildfordPress.
Inventory-II manual. San Antonio, TX: Psychological Costa, P. T., & McCrae, R. R. (1980). Influence of extraver-
Corporation. sion and neuroticism on subjective well-being: Happy and
Beesdo, K., Bittner, A., Pine, D. S., Stein, M. B., Hofler, M., unhappy people. Journal of Personality and Social Psychology,
Dipl-Stat, R. L., Wittchen, H.-U. (2007). Incidence of 38, 668678.
social anxiety disorder and the consistent risk for secondary Dalrymple, K. L., & Zimmerman, M. (2011). Age of onset of
depression in the first three decades of life. Archives of General social anxiety disorder in depressed outpatients. Journal of
Psychiatry, 64, 903912. Anxiety Disorders, 25, 131137.
Bienvenu, O. J., Hettema, J. M., Neale, M. C., Prescott, C. A., deGraaf, R., Bijl, R. V., Spijker, J., Beekman, A. T. F., &
& Kendler, K. S. (2007). Low extraversion and high neuroti- Vollebergh, W. A.M. (2003). Temporal sequencing of life-
cism as indices of genetic and environmental risk for social time mood disorders in relation to comorbid anxiety and
phobia, agoraphobia, and animal phobia. American Journal substance use disorders: Findings from the Netherlands
of Psychiatry, 164, 17141721. Mental Health Survey and Incidence Study. Social Psychiatry
Bradley, B. P., Mogg, K., Miller, N., & White, J. (1995). and Psychiatric Epidemiology, 38,111.
Cognitive processing of negative information:Effects of clin- Dimidjian, S., Hollon, S. D., Dobson, K. S., Schmaling, K. B.,
ical anxiety, concurrent depression, and awareness. Journal of Kohlenberg, R. J., Addis, M. E., Jacobson, N. S. (2006).
Abnormal Psychology, 104, 532536. Randomized trial of behavioral activation, cognitive ther-
Brown, T. A., Campbell, L. A., Lehman, C. L., Grisham, J. R., apy, and antidepressant medication in the acute treatment
& Mancill, R. B. (2001). Current and lifetime comorbidity of adults with major depression. Journal of Consulting and
of the DSM-IV anxiety and mood disorders in a large clinical Clinical Psychology, 74, 658670.
sample. Journal of Abnormal Psychology, 110, 585599. Dobson, K. S., Hollon, S. D., Dimidjian, S., Schmaling, K. B.,
Brown, T. A., Chorpita, B. F., & Barlow, D. A. (1998). Structural Kohlenberg, R. J., Gallop, R. J., Jacobson, N. S. (2008).
relationships among dimensions of the DSM-IV anxiety Randomized trial of behavioral activation, cognitive therapy,
and mood disorders and dimensions of negative affect, and antidepressant medication in the prevention of relapse
positive affect, and autonomic arousal. Journal of Abnormal and recurrence in major depression. Journal of Consulting and
Psychology, 107, 179192. Clinical Psychology, 76, 468477.
Brown, G. W., Harris, T. O., & Eales, M. J. (1993). Aetiology Dozois, D. J.A., & Frewen, P. A. (2006). Specificity of cognitive
of anxiety and depressive disorders in an inner-city popu- structure in depression and social phobia:Acomparison of
lation: II. Comorbidity and adversity. Psychological interpersonal and achievement content. Journal of Affective
Medicine: A Journal of Research in Psychiatry and the Allied Disorders, 90, 101109.
Sciences, 23, 155165. Erwin, B. A., Heimberg, R. G., Juster, H., & Mindlin, M.
Brown, L. H., Silvia, P. J., Myin-Germeys, I., & Kwapil, T. R. (2002). Comorbid anxiety and mood disorders among
(2007). When the need to belong goes wrong. Psychological persons with social anxiety disorder. Behavior Research and
Science, 18, 778782. Therapy, 40,1935.
Brown, E. J., Turovsky, J., Heimberg, R. G., Juster, H. R., Brown, Fyer, A. J., Mannuzza, S., Chapman, T. F., Liebowitz, M. R., &
T. A., & Barlow, D. H. (1997). Validation of the Social Klein, D. F. (1993). A direct interview family study of social
Interaction Anxiety Scale and the Social Phobia Scale across phobia. Archives of General Psychiatry, 50, 286293.
the anxiety disorders. Psychological Assessment, 9,2127. Gazelle, H., & Rudolph, K. D. (2004). Moving toward and away
Bruce, S. E., Yonkers, K. A., Otto, M. W., Eisen, J. L., Weisberg, from the world: social approach and avoidance trajectories
R. B., Pagano, M.,...Keller, M. B. (2005). Influence of in anxious solitary youth. Child Development, 75, 829849.
psychiatric comorbidity on recovery and recurrence in Gibb, B. E., Coles, M. E., & Heimberg, R. G. (2005).
generalized anxiety disorder, social phobia, and panic dis- Differentiating symptoms of social anxiety and depression
order: A 12-year prospective study. American Journal of in adults with social anxiety disorder. Journal of Behavior
Psychiatry, 162, 11791187. Therapy, 36, 99109.
Clark, D. M., Ehlers, A., Hackman, A., McManus, F., Fennell, Gilbert, P., Allan, S., Brough, S., Melley, S., & Miles, J. N.V. (2002).
M., Grey, N.,...Wild, J. (2006). Cognitive therapy versus Relationship of anhedonia and anxiety to social rank, defeat,
exposure and applied relaxation in social phobia: A ran- and entrapment. Journal of Affective Disorders, 71, 141151.
domized controlled trial. Journal of Consulting and Clinical Grant, D. M., & Beck, J. G. (2006). Attentional biases in social
Psychology, 74, 568578. anxiety and dysphoria:Does comorbidity make a difference?
Clark, D. M., Ehlers, A., McManus, F., Hackmann, A., Fennell, Anxiety Disorders, 20, 520529.
M., Campbell, H.,...Louis, B. (2003). Cognitive therapy Heimberg, R. G., Mueller, G. P., Holt, C. S., Hope, D. A.,
versus fluoxetine in generalized social phobia:Arandomized & Liebowitz, M. R. (1992). Assessment of anxiety in

126 Comorbidit y of Social Anxiet y Disorder and Depression

social interaction and being observed by others:The Social Krueger, R. F., & Markon, K. E. (2006). Reinterpreting comor-
Interaction Anxiety Scale and the Social Phobia Scale. bidity:Amodel-based approach to understanding and classi-
Behavior Therapy, 23,5373. fying psychopathology. Annual Review of Clinical Psychology,
Henry, B., Moffitt, T. E., Caspi, A., Langley, A., & Silva, P. A. 2, 111133.
(1994). On the remembrance of things past: A longitu- Marom, S., Gilboa-Schechtman, E., Aderka, I. M., Weizman, A.,
dinal evaluation of the retrospective method. Psychological & Hermesh, H. (2009). Impact of depression on treatment
Assessment, 6, 92101. effectiveness and gains maintenance in social phobia:Anatu-
Hughes, A. A., Heimberg, R. G., Coles, M. E., Gibb, B. E., ralistic study of cognitive behavior group therapy. Depression
Liebowitz, M. R., & Schneier, F. R. (2006). Relations of the and Anxiety, 26, 289300.
factors of the tripartite model of anxiety and depression to Mattick, R. P., & Clarke, J. C. (1998). Development and vali-
types of social anxiety. Behavior Research and Therapy, 44, dation of measures of social phobia scrutiny fear and social
16291641. interaction anxiety. Behaviour Research and Therapy, 36,
Jang, K. L., Livesley, W. J., & Vernon, P. A. (1996).Heritability 455470.
of the big five personality dimensions and their facets:Atwin Merikangas, K. R., & Angst, J. (1995). Comorbidity and
study. Journal of Personality, 64, 577591. social phobia: Evidence from clinical, epidemiologic, and
Johnson, S. L., Leedom, L. J., & Muhtadie, L. (2012). The dom- genetic studies. European Archives of Psychiatry and Clinical
inance behavioral system and psychopathology: Evidence Neuroscience, 244, 297303.
from self-report, observational, and biological studies. Mineka, S., Watson, D., & Clark, L. A. (1998). Comorbidity
Psychological Bulletin, 138, 692743. of anxiety and unipolar mood disorders. Annual Review of
Joiner, T., & Coyne, J. (1999). The interactional nature of depres- Psychology, 49, 377412.
sion: Advances in interpersonal approaches. Washington, Mogg, K., Bradley, B. P., Williams, R., & Mathews, A. (1993).
DC:American Psychological Association. Subliminal processing of emotional information in anxi-
Joormann, J., Talbot, L., & Gotlib, I. H. (2007). Biased process- ety and depression. Journal of Abnormal Psychology, 102,
ing of emotional information in girls at risk for depression. 304311.
Journal of Abnormal Psychology, 116, 135143. Moitra, E., Herbert, J. D., & Forman, E. M. (2008). Behavioral
Kashdan, T. B. (2004). The neglected relationship avoidance mediates the relationship between anxiety and
between social interaction anxiety and hedonic defi- depressive symptoms among social anxiety disorder patients.
cits: Differentiation from depressive symptoms. Anxiety Journal of Anxiety Disorders, 22, 12051213.
Disorders, 18, 719730. Moscovitch, D. A., Hofmann, S. G., Suvak, M. K., & In-Albon,
Kashdan, T. B., & Roberts, J. E. (2011). Comorbid social anxi- T. (2005). Mediation of changes in anxiety and depression
ety disorder in clients with depressive disorders. Behaviour during treatment of social phobia. Journal of Consulting and
Research and Therapy, 49, 875884. Clinical Psychology, 73, 945952.
Kashdan, T. B., & Steger, M. F. (2006). Expanding the topogra- Moscovitch, D. A., Orr, E., Rowa, K., Reimer, S. G., & Antony,
phy of social anxiety:An experience-sampling assessment of M. M. (2009). In the absence of rose-colored glasses:Ratings
positive emotions, positive events, and emotion suppression. of self-attributes and their differential certainty and impor-
Psychological Science, 17, 120128. tance across multiple dimensions in social phobia. Behaviour
Kendler, K. S., Aggen, S. H., Knudsen, G. P., Rysamb, Research and Therapy, 47(1),6670.
E., Neale, M. C., & Reichborn-Kjennerud, T. (2011). Naragon-Gainey, K., Watson, D., & Markon, K. E. (2009).
The structure of genetic and environmental risk fac- Differential relations of depression and social anxiety symp-
tors for syndromal and subsyndromal common DSM-IV toms to the facets of extraversion/positive emotionality.
Axis I and all Axis II disorders. American Journal of Journal of Abnormal Psychology, 118, 299310.
Psychiatry,168,2939. Norton, P. J., Temple, S. R., & Pettit, J. W. (2008). Suicidal
Kendler, K. S., Heath, A., Martin, N. G., Eaves, L. J. (1986). ideation and anxiety disorders: Elevated risk or artifact
Symptoms of anxiety and depression in a volunteer twin of comorbid depression? Journal of Behavior Therapy, 39,
population:The etiologic roles of genetic and environmental 515525.
factors. Archives of General Psychiatry, 43, 213221. Ogliari, A., Citterio, A., Zanoni, A., Fagnani, C., Patriarca,
Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & V., Cirrincione, R.,...Battaglia, M. (2006). Genetic and
Eaves, L. J. (1993). Major depression and phobias: The environmental influences on anxiety dimensions in Italian
genetic and environmental sources of comorbidity. twins evaluated with the SCARED questionnaire. Journal of
Psychological Medicine, 23, 361371. Anxiety Disorders, 20, 760777.
Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, Ohayon, M. M., & Schatzberg, A. F. (2010). Social phobia
K. R., & Walters, E. E. (2005). Lifetime prevalence and and depression: Prevalence and comorbidity. Journal of
age-of-onset distributions of DSM-IV disorders in the Psychosomatic Research, 68, 235243.
national comorbidity survey replication. Archives of General Price, J., Sloman, L., Gardner, R., Gilbert, P., & Rhode, P.
Psychiatry, 62, 593602. (1994). The social competition hypothesis of depression.
Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. (2005). British Journal of Psychiatry, 164, 309315.
Prevalence, severity, and comorbidity of 12-month DSM-IV Rapee, R. M., & Spence, S. H. (2004). The etiology of social
disorders in the National Comorbidity Survey Replication. phobia: Empirical evidence and an initial model. Clinical
Archives of General Psychiatry, 62, 617627. Psychology Review, 24, 737767.
Kessler, R. C., Stang, P., Wittchen, H.-U., Stein, M., Walters, Rodebaugh, T. L. (2009). Social phobia and perceived friendship
E. E. (1999). Lifetime comorbidities between social phobia quality. Journal of Anxiety Disorders, 23, 872878.
and mood disorders in the US National Comorbidity Survey. Rodebaugh, T. L., Heimberg, R. G., Brown, P. J., Fernandez, F.
Psychological Medicine, 29, 555567. C., Blanco, C., Schneier, F. R., & Liebowitz, M. R. (2011).

Langer, Rodebaugh 127

More reasons to be straightfoward: Findings and norms Waikar, S. V., & Craske, M. G. (1997). Cognitive correlates of
for two scales relevant to social anxiety. Journal of Anxiety anxious and depressive symptomatology: An examination
Disorders, 25, 623630. of the helplessness/hopelessness model. Journal of Anxiety
Schneier, F. R., Heckelman, L. R., Garfinkel, R., Campeas, Disorders, 11,116.
R., Fallon, B. A., Gitow, A.,...Liebowitz, M. R. (1994). Watson, D. (2009). Rethinking the anxiety disorders in DSM-V
Functional impairment in social phobia. Journal of Clinical and beyond:Quantitative dimensional models of anxiety and
Psychiatry, 55, 322331. related psychopathology. In:D. McKay, J. S.Abramowitz, S.
Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, Taylor, & G. J.G. Asmundson (Eds.), Current perspectives on
P., Janavs, J., Weiller, E.,...Dunbar, G. C. (1998). The the anxiety disorders:Implications for DSM-V and beyond (pp.
Mini-International Neuropsychiatric Interview (M.I.N.I): 275302). NewYork:Springer.
The development and validation of a structured diagnostic Watson, D., Clark, L. A., Carey, G. (1988). Positive and negative
psychiatric interview for DSM-IV and ICD-10. Journal of affectivity and their relation to anxiety and depressive disor-
Clinical Psychiatry, 59(Suppl. 20),2233. ders. Journal of Abnormal Psychology, 97, 346353.
Spitzer, R. L., Williams, J. B., & Gibbon, M. (1995). Structured clin- Weeks, J. W., Rodebaugh, T. L., Heimberg, R. G., Norton, P.
ical interview for DSM-IV patient version. NewYork: Biometrics J., & Jakatdar, T. A. (2009). To avoid evaluation, with-
Research Department, NewYork State Psychiatric Institute. draw: Fears of evaluation and depressive cognitions lead to
Stein, M. B., Fuetsch, M., Muller, N., Hofler, M., Lieb, R., & social anxiety and submissive withdrawal. Cognitive Therapy
Wittchen, H.-U. (2001). Social Anxiety Disorder and the risk Research, 33, 375389.
of depression. Archives of General Psychiatry, 58, 251256. Weilage, M., & Hope, D. A. (1999). Self-discrepancy in social
Strauman, T. J. (1989). Self-discrepancies in clinical depression phobia and dsythymia. Cognitive Therapy and Research, 23,
and social phobia: Cognitive structures that underlie emo- 637650.
tional disorders? Journal of Abnormal Psychology, 98,1422. Wells, A., & Carter, K. (2001). Further tests of a cognitive model
Swendsen, J. D. (1997). Anxiety, depression, and their comor- of generalized anxiety disorder:Metacognitions and worry in
bidity: An experience sampling test of the helplessness GAD, panic disorder, social phobia, depression, and nonpa-
hopelessness theory. Cognitive Therapy and Research, 21, tients. Behavior Therapy, 32, 85102.
97114. Wilhelm, K., & Parker, G. (1989). Is sex necessarily a risk factor
Swendsen, J. D. (1998). The helplessnesshopelessness theory for depression? Psychological Medicine:AJournal of Research
and daily mood experience:An idiographic and cross-situa- in Psychiatry and the Allied Sciences, 19, 401413.
tional perspective. Journal of Personality and Social Psychology, Wilson, J. K., & Rapee, R. M. (2005). The interpretation of
74, 13981408. negative social events in social phobia with versus without
Trower, P., & Gilbert, P. (1989). New theoretical conceptions of comorbid mood disorder. Anxiety Disorders, 19, 245274.
social anxiety and social phobia. Clinical Psychology Review, Wittchen, H.-U., Stein, M. B., & Kessler, R. C. (1999). Social
9,1935. fears and social phobia in a community sample of adolescents
Van Ameringen, M., Mancini, C., Styan, G., & Donison, D. and young adults:Prevalence, risk factors, and co-morbidity.
(1991). Relationship of social phobia with other psychiatric Psychological Medicine, 29, 309323.
illness. Journal of Affective Disorders, 21,9399. Zaider, T. I., Heimberg, R. G., Roth, D. A., Hope, D. & Turk,
Vittengl, J. R., & Holt, C. S. (2000). Getting acquainted:The C. L. (2003, November). Individual cognitive-behavioral
relationship of self-disclosure and social attraction to therapy for social anxiety disorder:Preliminary findings. Paper
positive affect. Journal of Social and Personal Relationships, presented at the annual meeting of the Association for
17,5366. Advancement of Behavior Therapy, Boston.

128 Comorbidit y of Social Anxiet y Disorder and Depression


9 Important Issues in Understanding

Comorbidity Between Generalized Anxiety
Disorder and Major Depressive Disorder
Susan Mineka, Deepika Anand, and Jennifer A.Sumner

The comorbidity of anxiety and mood disorders has been of great interest to psychopathology
researchers for the past 25years. One topicthe comorbidity of generalized anxiety disorder (GAD)
and major depressive disorder (MDD)has received considerable attention, in part because it has raised
fundamental nosological issues regarding whether GAD should continue to be categorized as an anxiety
disorder or whether it should be recategorized as a mood disorder. We review the logic for reclassifying
GAD with the mood disorders as well as what we believe to be even more compelling reasons for why
it should be retained as an anxiety disorder. In doing so, we review three different kinds of comorbidity
cross-sectional, cumulative (lifetime), and sequential. We also discuss overlaps and distinctions in what
is known about the etiology of GAD and MDD and how their somewhat different cognitive and
neurobiological profiles bear on these issues of classification. Finally, we briefly discuss what some of the
treatment implications may be for individuals with comorbid GAD andMDD.
Key Words: generalized anxiety disorder, major depressive disorder, comorbidity, nosological issues,
mood disorders, anxiety disorders, treatment implications

Background reviewed later, studies that have modeled the asso-

Decades of epidemiological research have dem- ciations between symptoms and diagnoses of emo-
onstrated that high levels of comorbidity are the tional disorders have quite consistently found that
norm rather than the exception within the realm of MDD and GAD load most strongly on a single
the emotional disorders (e.g., Krueger & Markon, higher-order distress factor rather than on separate
2006; Mineka, Watson, & Clark, 1998). For mood and anxiety disorder factors (e.g., Krueger
example, high rates of comorbidity between major & Markon, 2006; Watson, 2005). Together, these
depressive disorder (MDD) and generalized anxiety results suggest that the overlap between MDD and
disorder (GAD) have become a topic of especially GAD is substantial. Consequently much research
great interest for psychopathology researchers in has been conducted in an effort to understand why
recent years; this topic is the focus of the current such high rates of comorbidity between MDD and
chapter. MDD and GAD not only share a number of GAD occur and what the implications are for how
phenotypic features (e.g., insomnia, fatigue, repeti- these two disorders should be conceptualized.
tive thought) but also have been shown to exhibit In this chapter, we review highlights of the lit-
considerable co-occurrence in both epidemiologi- erature on comorbidity between MDD and GAD.
cal and clinical samples (e.g., Brown, Campbell, Throughout the chapter, we discuss various ways
Lehman, Grisham, & Mancill, 2001; Kessler, in which this comorbidity manifests itself and
Chiu, Demler, & Walters, 2005). Furthermore, as describe different etiological models that have been

proposed to explain why this substantial comorbid- Liverant, Rosellini, & Brown, 2009). First intro-
ity occurs. We also review some of the nosological duced in DSM-III-R and retained in DSM-IV,
issues that have been raised surrounding the high the diagnostic hierarchy rule stipulates that GAD
rates of comorbidity between GAD and MDD. cannot be diagnosed if it occurs exclusively dur-
Furthermore, we briefly discuss several important ing MDD or another mood disorder. This rule
issues regarding the somewhat different cognitive was implemented in order to foster diagnostic par-
and neurobiological profiles of these two disorders. simony, but it may lead to an underestimation of
Finally, we also touch on some possible therapeutic the true rates of comorbidity between GAD and
implications of the comorbidity between MDD and MDD. For example, in a clinical outpatient sample,
GAD. Throughout we highlight questions for fur- Brown etal. (2001) found that GAD was comor-
ther research that are important to address in order bid with a current principal diagnosis of MDD in
to advance thefield. only 5% of cases when the diagnostic hierarchy rule
was applied. However, when the diagnostic hier-
Understanding the Apparent Rise in archy rule was ignored, GAD was comorbid with
Comorbidity Between MDD andGAD a current principal diagnosis of MDD in 67% of
In order to fully comprehend the current rates of cases. Furthermore, research suggests that individu-
comorbidity between MDD and GAD, it is impor- als with symptoms of GAD that occur only dur-
tant to first take into consideration the changes in ing the course of a major depressive episode exhibit
the diagnostic criteria for GAD that have occurred similar levels of psychopathology and functional
between the third (revised) and fourth editions of impairment as do those with comorbid MDD and
the Diagnostic and Statistical Manual of Mental GAD based on the diagnostic hierarchy rule (e.g.,
Disorders (DSM). The criteria for GAD under- Lawrence etal., 2009). Together, these findings have
went major revisions between DSM-III-R (1987) led some researchers to question the validity and
and DSM-IV (1994), and these changes provide utility of the diagnostic hierarchy rule. Consistent
a key context for understanding the high levels of with this notion, in recent years many studies of the
comorbidity between MDD and GAD. One criti- comorbidity between MDD and GAD, including
cal change with DSM-IV was the elimination of the a number of those cited in this chapter, have not
autonomic hyperactivity symptoms from the diag- applied the diagnostic hierarchy rule in diagnosing
nostic criteria for GAD. Even though this change GAD (e.g., Kendler, Gardner, Gatz, & Pedersen,
resulted in increased reliability in the diagnosis of 2007; Kessler et al., 1996; Moffitt, Harrington,
GAD and somewhat reduced rates of comorbidity etal.,2007).
between GAD and other anxiety disorders (Mennin,
Heimberg, Fresco, & Ritter, 2008), it also served Cross-Sectional, Cumulative, and
to increase the overlap in the diagnostic criteria for Sequential Comorbidity
GAD and MDD. For example, in DSM-IV, a major- Cross-Sectional Comorbidity
ity of the physical symptoms associated with GAD In referring to the co-occurrence of two or more
(e.g., restlessness, fatigue, and difficulty sleeping and disorders, comorbidity can actually manifest in a
concentrating) are also included in the symptom list number of ways depending on the time course con-
for MDD. Not surprisingly, the increase in overlap- sidered. Here we present the results for three forms
ping diagnostic criteria for MDD and GAD made of comorbiditycross-sectional, cumulative, and
it more likely that individuals would be diagnosed sequential comorbidityin order to provide a fairly
with both disorders. At the same time, this symptom comprehensive overview of issues in the comor-
overlap somewhat obscured the meaning of comor- bidity between MDD and GAD. Cross-sectional
bidity between MDD and GAD. As we discuss comorbidity is defined as meeting criteria for two
below, these complicating factors have led some to disorders within the same time-frame (e.g., the past
seriously question the diagnostic validity of the dis- year) (Moffitt, Harrington, et al., 2007). Overall,
tinction between GAD and MDD as well as whether there is evidence of relatively high cross-sectional
GAD should be placed in the depressive disorder comorbidity between MDD and GAD using
rather than the anxiety disorder category. DSM-III-R diagnoses. For example, in the National
Another diagnostic feature of GAD that may Comorbidity Survey (NCS), a nationally repre-
have obscured our understanding of the extent to sentative survey of individuals between the ages of
which GAD co-occurs with MDD is the diagnostic 15 and 54 in the United States using DSM-III-R
hierarchy rule for these two disorders (Lawrence, based diagnoses, the odds ratio (OR) for MDD

130 Generalized Anxiet y Disorder and Major Depressive Disorder

and GAD being comorbid within the past year was addition, there were also substantial associations
8.2 (Kessler etal., 1996). Thus an individual with between lifetime diagnoses of MDD and GAD at
a diagnosis of MDD within the past year had 8.2 the 20012003 NCS follow-up assessment that
times the risk of also having a diagnosis of GAD occurred approximately 10years later after the base-
within the same 12-month period versus an individ- line NCS (OR=6.6) (Kessler etal.,2008).
ual without a diagnosis of MDD. More specifically, Similarly, in the Dunedin study, Moffitt,
in the original NCS, 15.4% of individuals with Harrington, et al. (2007) examined comorbidity
12-month MDD also met the criteria for GAD between lifetime diagnoses of MDD and anxiety dis-
within the same time period. orders that were made between the ages of 11 and 32.
The prospective birth cohort design of Of those participants with a lifetime history of MDD,
the Dunedin Multidisciplinary Health and 48% also had a lifetime history of an anxiety disorder
Development Study permitted a more fine-grained (defined as a diagnosis of GAD at ages 18, 21, 26,
examination of cross-sectional and sequential and 32 or as a juvenile diagnosis of overanxious dis-
comorbidity between MDD and GAD at differ- order, separation anxiety disorder, or phobias at ages
ent ages. Participants in this study were all born in 11, 13, and 15 because GAD was not diagnosed prior
Dunedin, New Zealand, between April 1, 1972, to age 18 in this sample). Among individuals with a
and March 31, 1973; they were assessed every two lifetime anxiety disorder (again defined as a diagnosis
to three years since birth (the latest assessment of GAD at ages 18, 21, 26, and 32 or as a juvenile
phase was at age 38). When participants were 18, diagnosis of overanxious disorder, separation anxiety
21, 26, and 32 years of age, Moffitt, Harrington, disorder, or phobias at ages 11, 13, and 15), 72% had
et al. (2007) examined the issue of cross-sectional a lifetime diagnosis of MDD. This was done because
MDD-GAD comorbidity by comparing rates of GAD cannot be diagnosed before age 18. However,
comorbid diagnoses of MDD and GAD (using by including phobias, this may well be an overesti-
DSM-IV diagnoses) within the same year. With mate of comorbidity with GAD per se. Furthermore,
increasing age, rates of MDD with comorbid GAD these patterns of comorbidity were similar in both
increased, namely from 8% at age 21 to 30% at age men and women (an issue rarely examined in most
32. However, rates of GAD with comorbid MDD studies on this topic). Together, these findings sug-
decreased slightly with age, from 88% at age 18 to gest that MDD and GAD frequently both occur dur-
63% at age 32. In other words, at age 18, nearly all ing an individuals lifetime.
individuals with GAD (88%) had comorbid MDD;
but at age 32, only 63% of individuals with GAD Sequential Comorbidity
had comorbid MDD. Across all ages, rates of GAD Evidence of elevated cross-sectional and cumu-
with comorbid MDD were greater than rates of lative (lifetime) comorbidity between MDD
MDD with comorbidGAD. and GAD (as well as other anxiety disorders) has
prompted researchers to investigate whether one dis-
Cumulative (Lifetime) Comorbidity order is more likely to precede the other (e.g., Alloy,
Comorbidity can also manifest as cumulative Kelly, Mineka, & Clements, 1990). For example,
comorbidity, in which both disorders occur but not does GAD generally have its onset first, followed by
necessarily within the same period of time (Moffitt, MDD? Or does MDD tend to predate the onset of
Harrington, et al., 2007). Assessments of cumula- GAD? Sequential comorbidity refers to the degree to
tive (lifetime) comorbidity provide a more accurate which one disorder is reliably followed by the onset
estimate of comorbidity across the lifetime than do of another disorder (Moffitt, Harrington, et al.,
those of cross-sectional comorbidity because, by 2007). Information on sequential comorbidity can
focusing on a limited time frame, cross-sectional not only help to inform understanding of etiology
assessments may significantly underestimate the but also has potential implications for interven-
degree to which an individual experiences multiple tion and prevention. For example, if one disorder is
disorders during his or her life. In general, there is found to be followed by another disorder at higher
strong evidence of cumulative comorbidity between levels than what would be expected by chance, then
MDD and GAD. For instance, in the NCS, 17.2% treating the first disorder might have the potential
of individuals with lifetime MDD also had a life- to prevent onsets of the second disorder, although
time diagnosis of GAD (OR = 7.5 for lifetime only if the first disorder somehow plays a causal role
comorbidity of MDD and GAD at the original for (or serves as an important vulnerability factor
NCS baseline assessment; Kessler et al., 1996). In for) the second disorder.

Mineka, Anand,Sumner 131

For many years, it was generally believed that was reported). In addition, the magnitude of these
GAD (and the other anxiety disorders) was much predictive relationships decreased as the amount of
more likely to precede a diagnosis of MDD than time between the onsets of the initial and subse-
MDD was likely to precede a subsequent diagno- quent disorders increased. In other words, although
sis of GAD (e.g., Mennin et al., 2008). However, still elevated, the risk of experiencing a subsequent
this conclusion was based mostly on studies that first onset of either MDD or GAD became attenu-
had only examined whether a history of GAD ated as more time passed since the onset of the ini-
predicted subsequent MDD; very few studies on tial disorder (either GAD or MDD, respectively).
sequential comorbidity had considered whether a Nevertheless, lifetime diagnoses of MDD and GAD
history of MDD also predicted later GAD (Moffitt, at baseline were still significant predictors of the first
Harrington, et al., 2007). Findings from more onset of both GAD and MDD, respectively, for up
recent studies that assessed the sequential comor- to a decade or more after the onset of the initial
bidity of MDD and GAD in a more balanced way disorder.
have actually provided support for a more bidirec- In addition to predicting the first onsets of MDD
tional sequential relationship. For example, in the and GAD, Kessler et al. (2008) examined predic-
Dunedin study, 37% of cases of MDD diagnosed tion of the persistence of MDD and GAD over the
between ages 11 and 32 were associated with prior follow-up period. In these analyses, prior GAD was
or concurrent diagnoses of GAD or juvenile anxiety found to predict the persistence of MDD, but prior
(Moffitt, Harrington, et al., 2007). An additional MDD was not a significant predictor of the persis-
10% of the MDD cases experienced onsets of GAD tence of GAD. Together, these findings suggest that
or juvenile anxiety after the onset of MDD. Among GAD may be a somewhat stronger predictor of sub-
those individuals with a diagnosis of GAD or a sequent first onset and persistence of MDD than
juvenile anxiety disorder made between the ages of MDD is of subsequent GAD onset and persistence.
11 to 32, a total of 32% had an onset of MDD Nevertheless, prior MDD has been found to sig-
prior to or concurrent with the onset of the anxi- nificantly increase vulnerability to developing GAD
ety disorder and an additional 41% had an onset later on. Although this may in part reflect their
of MDD after the onset of the anxiety disorder. In common genetic diathesis, the different relative
sum, the percentage of cases of MDD that were strengths of their predictive relationships (i.e., GAD
preceded by GAD (37%) was comparable to the preceding MDD or vice versa) cannot be explained
percentage of cases of GAD that were preceded by simply by their shared genetic relationship.
MDD (32%). Furthermore, at least 63% of cases of Further research is needed to better understand
MDD were not preceded by either GAD or juvenile what accounts for these patterns of sequential
anxiety. These findings demonstrate that MDD can comorbidity. Some researchers have posited that
predateGAD. unremitting symptoms of an initial anxiety disor-
Further support for bidirectional sequential der may serve as a stressor that then precipitates
comorbidity between MDD and GAD was observed the onset of subsequent depression (e.g., Akiskal,
in the NCS and NCS follow-up assessment, which 1990; Alloy et al., 1990). For example, individ-
occurred approximately 10 years later (Kessler uals with GAD who perceive their worry to be
etal., 2008). Participants with a lifetime diagnosis uncontrollable might experience chronic feelings
of MDD at the NCS baseline assessment had an of helplessness that, after sufficient time, might
increased risk of experiencing a first onset of GAD contribute to feelings of hopelessness (e.g., Alloy
over the years leading up to the NCS follow-up as etal., 1990). This sense of hopelessness might then
compared with those without MDD (OR = 2.7), put individuals at risk for developing depression.
perhaps reflecting their shared genetic risk (see sec- Based on this theory, it would be predicted that
tion 4 below). Furthermore, a baseline diagnosis the longer and/or more severe the course of GAD,
of lifetime GAD also predicted the first onset of the greater the risk of developing subsequent
MDD over the follow-up period (OR=3.2). Even depression. Longitudinal, prospective research
though both relationships were statistically signifi- designs that assess the duration and severity of
cant, the predictive relationship between GAD and GAD at multiple points in time and track subse-
subsequent first onset of MDD was slightly stron- quent onsets of MDD would be needed to fully
ger than the predictive relationship between MDD address this issue. It is also of interest for future
and subsequent first onset of GAD (although no research to investigate what might account for
test of the statistical significance of this difference the observed patterns of bidirectional sequential

132 Generalized Anxiet y Disorder and Major Depressive Disorder

comorbidity. In addition to sequential comorbid- between MDD and GAD (e.g., Kendler, 1996;
ity being explained by the effects of one disorder Kendler etal., 1992, 2007; Roy, Neale, Pedersen,
on another, researchers have posited that these Math, & Kendler, 1995). The results have been
observed patterns might also be due to unmea- quite consistent across investigations in suggest-
sured common causes of MDD and GAD. As we ing that shared genetic vulnerability primarily
discuss below, both some common genetic and accounts for the co-occurrence of MDD and
environmental causes of MDD and GAD have GAD in these twin samples. Indeed, the genetic
been the subject of research (e.g., Kendler etal., correlations between MDD and GAD (both in
2007), and more work on this topic is clearly terms of lifetime and past-year diagnoses) gener-
needed. ally have been estimated to be unity (i.e., 1.0)
(Kendler, 1996; Kendler etal., 1992; Roy etal.,
Etiological Models of Comorbidity 1995). The one exception is a study by Kendler
Between MDD andGAD et al. (2007) in which the genetic correlation
As mentioned above, common causes of MDD between lifetime MDD and lifetime GAD was
and GAD have been posited as one explanation for estimated to be 1.00 for female twins but only
the high rates of comorbidity between these two .74 for male twins. In summary, vulnerability to
disorders. Indeed, models positing common causes MDD and vulnerability to GAD are influenced
are not unique to the co-occurrence of MDD and largely by the same genetic factors, although pos-
GAD but rather have been explored within the sibly slightly less so formen.
context of a variety of clinical presentations (e.g.,
associations between personality and depression) Environmental Contributions
(Klein, Durbin, & Shankman, 2009). According to Comorbidity
to this model, two clinical manifestations originate Perhaps not surprisingly, in contrast to the find-
from the same or at least an overlapping set of causal ings of very high genetic similarity, there has been
processes. much less evidence of sizable overlap in the environ-
Over the past few decades, researchers in pur- mental contributions to MDD and GAD in twin
suit of common causes of MDD and GAD have studies. For example, as for most disorders, there
explored the extent to which common genetic and/ has been little support for a role for familial/shared
or environmental risk factors might account for environmental factors in contributing to risk for
the high rates of MDD and GAD co-occurrence MDD or GAD. Furthermore, there has been evi-
(e.g., Kendler, Neale, Kessler, Heath, & Eaves, dence of more modest overlap between the nonfa-
1992). In this research, genetic vulnerability has milial/individual-specific environmental risk factors
been estimated with a genetic correlation, namely for MDD and GAD, with most estimates of the
the extent to which the same genes increase risk for nonfamilial/individual-specific environmental cor-
both disorders. With respect to common environ- relation between MDD and GAD ranging between
mental factors, two kinds of environmental influ- .2 and .6 (Kendler, 1996; Kendler et al., 1992,
ences have generally been distinguished: (1) those 2007; Roy etal., 1995). These findings suggest that
that are shared by members within the same fam- although there is some overlap in the nonfamilial/
ily (familial/shared environmental factors, such as individual-specific environmental risk factors for
socioeconomic status) and (2)those that are unique MDD and GAD, they are not entirely overlapping.
to individuals and not shared by all family mem- Together, the results of these twin studies indicate
bers (nonfamilial/individual-specific environmental that MDD and GAD are characterized by a highly
factors, such as the breakup of a romantic relation- similar (if not identical) genetic substrate, leading to
ship). Genetically informative samples, such as twin the high likelihood that environmental influences
studies, can help to identify the degree to which primarily determine whether individuals develop
all three of these factors (genetic, familial/shared MDD and/orGAD.
environment, nonfamilial/individual-specific envi-
ronment) contribute to comorbidity between two Neuroticism as a Common
disorders. Genetic Vulnerability?
In light of these findings, researchers have been
Genetic Contributions to Comorbidity interested in identifying what may underlie this
To date, a number of twin studies have exam- shared genetic diathesis as well as which environ-
ined genetic contributions to the comorbidity mental factors may contribute to the development

Mineka, Anand,Sumner 133

of MDD versus GAD. One proposed candidate associated with neuroticism were found to account
for the common genetic vulnerability has been for a considerably more modest proportion of the
neuroticism, a broad personality trait with sub- genetic overlap between MDD and GAD compared
stantial heritability that is thought to reflect emo- with the study of Hettema et al. (2006), where
tional instability, proneness to negative affect, and closer to 50% of the genetic correlation between
distress (e.g., Kendler, Neale, Kessler, Heath, & MDD and GAD was shared with neuroticism.
Eaves, 1993). Neuroticism is a strong predictor of However, it is worth noting several important
both MDD and GAD (e.g., Moffitt, Caspi, etal., methodological differences that existed between
2007), and there is evidence of genetic correlations the Kendler etal. (2007) and Hettema etal. (2006)
between neuroticism and each of these disorders studies that may account for the apparent discrep-
as well (e.g., Hettema, Neale, Myers, Prescott, & ancy in their findings. For example, Kendler etal.
Kendler, 2006). Furthermore, a personality trait (2007) used a relatively uncommon measure of neu-
that reflects individual differences in experiencing roticism (a small number of items selected from the
general distress is a particularly promising candidate EPI), whereas Hettema et al. (2006) used a more
given that in structural models, MDD and GAD frequently employed neuroticism measure:the short
have both been found to load strongly on a gen- form of the Eysenck Personality Questionnaire
eral distress factor (e.g., Krueger & Markon, 2006; (EPQ; Eysenck & Eysenck, 1975). Perhaps even
Watson,2005). more important is that Kendler etal. assessed neu-
To date, at least two studies have examined the roticism 25 years prior to the assessment of psy-
degree to which the genetic factors underlying neu- chopathology. In contrast, Hettema et al. (2006)
roticism account for the genetic correlation between measured neuroticism and lifetime history of MDD
MDD and GAD. In the population-based Virginia and GAD much more closely in time. Neuroticism
Adult Twin Study of Psychiatric and Substance Use is only moderately stable over long periods of time
Disorders, neuroticism was measured with the short (one estimate from a meta-analysis by Roberts &
form of the Eysenck Personality Questionnaire, and DelVecchio [2000] was about .5). Thus Kendler
lifetime diagnoses of MDD and GAD were assigned et al.s (2007) measure of neuroticism may not
using diagnostic interviews (Hettema etal., 2006). have accurately reflected the participants neuroti-
Consistent with prior research, there was almost cism levels at the time of psychopathology assess-
complete overlap in the genetic factors associated ment 25 years later. Nevertheless, together these
with MDD and GAD (the genetic correlation results suggest that there is sizablebut certainly
between the two disorders was .98). This associa- not completeoverlap between the genetic factors
tion was also broken down into genetic correlations underlying neuroticism and the shared genetic risk
that did and did not overlap with the genetic factors for MDD and GAD. Thus it is of interest for future
underlying neuroticism. In the Hettema etal. (2006) research to continue to explore candidates which
study, a substantial portion of the genetic variance will help to explain the genetic overlap between
underlying comorbidity between MDD and GAD MDD andGAD.
was accounted for by the genetic factors associ-
ated with neuroticism (genetic correlation = .47). Stressful Life Events as Determinants
However, a sizable percentage of the genetic risk of MDD and/orGAD
common to MDD and GAD was also independent Although a common genetic diathesis appears
of neuroticism (genetic correlation=.51). to characterize MDD and GAD, the results of twin
Somewhat similar conclusions were drawn from studies suggest that different environmental risk
a study of participants from the Swedish Twin factors determine whether this underlying vulner-
Registry (Kendler et al., 2007). In this investiga- ability becomes manifested as MDD and/or GAD.
tion, neuroticism was measured with nine items One environmental factordifferent types of
from the short form of the Eysenck Personality stressful life eventshas been the subject of research
Inventory (EPI), and DSM-IVbased lifetime diag- investigating such potential disorder-specific deter-
noses of MDD and GAD were assigned approxi- minants. For example, Kendler, Hettema, Butera,
mately 25years after the assessment of neuroticism. Gardner, and Prescott (2003) examined the diag-
In this sample, genetic factors underlying neuroti- nostic specificity of different kinds of stressful life
cism accounted for 23% and 25% of the genetic events within the population-based Virginia Twin
comorbidity between MDD and GAD in male and Registry. Stressful life events were classified as loss
female twins, respectively. Thus in this study, genes events (e.g., death of a loved one), humiliation

134 Generalized Anxiet y Disorder and Major Depressive Disorder

events (e.g., rejection by a significant other), discussed thus far raise important issues that ques-
entrapment events (e.g., an event that prolongs tion the validity of their current classification in the
and worsens current difficulties), or danger events DSM-IV. The following section will address noso-
(e.g., threat of potential future loss). Events were logical concerns raised by the comorbidity between
identified that were associated with pure MDD ver- GAD and MDD and the implications of these con-
sus pure GAD episodes (although GAD episodes cerns to their DSM classification.
needed to be only two weeks in duration in order
to make the duration requirement comparable to Should MDD and GAD Be Reclassified
that for MDD) that had a minimum duration of Within the Same Diagnostic Category?
two weeks (rather than 6 months). Humiliation At the descriptive level, researchers have argued
events predicted pure MDD but not pure GAD that the current diagnostic classification system
episodes. In contrast, danger events predicted pure assumes stronger relationships between disorders
GAD episodes but not pure MDD episodes. Loss within a diagnostic category (e.g., social anxiety dis-
events were not specific to either disorder, but they order and GAD) and weaker relationships between
did predict the onset of pure MDD to a greater disorders across different categories (e.g. dysthymia
degree than they did pure GAD episodes. Perhaps and GAD) (Watson, 2005). However, this assump-
not surprisingly, onsets of comorbid episodes that tion seems to be violated to a certain extent by
were characterized by both depressive and general- the evidence on both the cross-sectional and the
ized anxiety symptoms were predicted by stressful sequential comorbidity between GAD and MDD
life events that, in general, were a sum of the pre- reviewed above. One of the main limitations of the
dictors of pure MDD (loss and humiliation events) current classification system that may be responsible
and pure GAD episodes (danger events). However, for this is that the DSM-IV is a rationally derived
there was some evidence that entrapment events system wherein disorders are grouped together
also predicted these mixed episodes. Overall, these based on shared phenomenological features rather
findings support the notion that, with a common than empirical evidence that includes the rates of
genetic diathesis, particular environmental factors comorbidity between disorders (Watson,2005).
(specifically different types of stressful life events), One important nosological challenge is raised
can differentially predict the extent to which indi- by the argument that MDD and GAD represent
viduals develop MDD or GAD. Furthermore, a single diagnostic entity and that the relationship
episodes of mixed depression-generalized anxiety between them extends beyond overlapping symp-
symptoms were predicted by the sum of these more tom criteria. In support of this view, Watson (2005)
disorder-specific factors. cited several lines of research pointing to their struc-
tural overlap beyond their phenotypic similarities.
Nosological Issues Raised by the He also pointed out that the significant genetic cor-
Comorbidity of MDD andGAD relation between them suggests that they result from
The descriptive comorbidity results described similar etiological mechanisms (Kendler, Prescott,
above may be interpreted in different ways. For Myers, & Neale,2003).
example, they might mean that our measures of In addition, as alluded to above, factor analytic
these disorders merely lack discriminant valid- investigations of the latent structures underlying
ity due to overlapping symptoms that spuriously GAD and MDD diagnoses revealed that these two
inflate the correlation between these two disorders. disorders were best represented as belonging to
On the other hand, high comorbidity rates might a new diagnostic category which Watson (2005)
mean that the two disorders represent a common called distress disorders. This category would
underlying clinical syndrome. This latter claim incorporate disorders that were predominantly
would imply that the co-occurring disorders are best characterized by high levels of negative affect and
conceptualized as a single diagnostic entity and that negative affectivity. Distress disorders would be
the distinction made between them is not clinically distinguished from the fear disorders category
or theoretically meaningful. Both these possibilities (comprising panic disorder, agoraphobia, social
raise important challenges to a diagnostic classifica- phobia, and specific phobia) and the bipolar disor-
tion system that categorizes these co-occurring dis- ders category (comprising bipolar I, bipolar II, and
orders separately without accounting for their high cyclothymia). Other structural modeling studies
rates of comorbidity (Watson, 2005). Therefore the have also reached the conclusion that GAD should
high comorbidity rates between GAD and MDD be recategorized along with the mood disorders.

Mineka, Anand,Sumner 135

For example, using data from the original National between them is not informative because both are
Comorbidity Survey, Krueger (1999) showed that driven by common pathological mechanisms and
GAD cohered with MDD and dysthymia, forming hence any distinction between them is spurious.
a common factor. Consistent with this, Vollebergh Researchers who argue against the reclassification
etal. (2001) fit a three-dimensional model to data of GAD as a mood disorder primarily cite evidence
from the Netherlands Mental Health Survery and that questions each of these three assumptions (e.g.,
Incidence Study (NEMESIS). This epidemiological Hettema, 2008; Mennin etal.,2008).
study on DSM-III-R diagnoses found that GAD
was best captured by the mood disorders dimen- Co-occurrence of MDD with Other
sion, which also contained MDD and dysthymia. Anxiety Disorders
In addition to this mood dimension, the investi- Mennin etal. (2008) reviewed evidence suggest-
gators identified an anxiety disorders dimension ing that both GAD and MDD also share complex
consisting of specific phobia, agoraphobia, social relationships with other depressive and anxiety
phobia, and panic disorder and a third substance disorders and that these relationships would not
use disorders dimension with alcohol dependence be accounted for if GAD were recategorized with
and drug dependence. In further support of this MDD. In support of this argument, they pointed
view, Watson (2005) noted that symptoms of both out that although high rates of comorbidity exist
GAD and MDD respond to similar pharmacologi- between GAD and MDD, MDD has been found to
cal treatment, such as serotonin and norepinephrine co-occur at moderately high rates with other anxiety
reuptake inhibitors, which was interpreted as fur- disorders as well. For example, Kessler, Chiu, etal.
ther evidence of their overlap (but see section on (2005) studied the relationships between MDD
treatment implications of comorbidity below). and several anxiety disorders using the NCS-R.
In light of the evidence regarding the overlap They reported a tetrachoric correlation of .62
between GAD and MDD, the limitations of the between MDD and GAD but also found moderate
current (DSM-IV) classification of these disorders correlations between MDD and other anxiety disor-
have been at the forefront of much discussion, par- ders that were also high and statistically significant,
ticularly during the development of the DSM-5. including panic disorder (.48), agoraphobia (.52),
Specifically, some researchers have recommended specific phobia (.43), social phobia (.52), posttrau-
that GAD and MDD be classified within the same matic stress disorder (.50), and obsessive compul-
diagnostic category in order to reflect their pheno- sive disorder (.42). Asimilar pattern of correlations
typic, etiological, and structural overlap. ranging from .50 to .56 was also found between
MDD and anxiety disorders by Slade and Watson
Should GAD and MDD Be Retained in (2006) using an epidemiological sample from the
Separate Diagnostic Categories? Australian National Survey of Mental Health and
Despite the evidence used by some to argue for Well-Being. Finally, using a clinical sample, Brown
grouping GAD and MDD into the same category, etal. (2001) found rates of co-occurrence as high as
several important arguments that we tend to agree 46% between MDD and panic disorder, which did
with have been raised against this view. These argu- not differ substantially from the 57% co-occurrence
ments suggest that such a decision would reflect between MDD and GAD. It should be noted that
certain assumptions which we believe are premature these studies unfortunately did not report testing
given current evidence (e.g., Mennin etal., 2008). the statistical significance of difference of the cor-
First, categorizing GAD along with the mood disor- relations between MDD and the different anxiety
ders would seem to assume that GAD bears a stron- disorders. Nonetheless, all of these findings raise an
ger relationship to MDD than to the other anxiety important point, which is that high comorbidity
disorders and that this would justify its being taken with MDD is not unique to GAD by any means
out of the anxiety disorders category. Second, it but extends to other anxiety disorders as well (see
would seem to assume that the reasons justifying also Hettema,2008).
the recategorization of GAD do not apply to the
other anxiety disordersthat is, that MDD shares Co-occurrence of GAD with Other
a uniquely strong relationship with GAD compared Anxiety Disorders
with the other anxiety disorders. Third, the decision In discussing issues regarding the current classifi-
to subsume GAD into the MDD diagnosis would cation of GAD as an anxiety disorder, Mennin etal.
seem to assume that the current distinction made (2008) pointed out that removing the autonomic

136 Generalized Anxiet y Disorder and Major Depressive Disorder

symptoms from GAD diagnostic criteria from and risk factor for, multiple psychological disorders
DSM-III-R for DSM-IV may have made it easier to including other anxiety disorders (Clark, Watson,
differentiate GAD from the other anxiety disorders. & Mineka, 1994; Weinstock & Whisman, 2006;
However, GAD continues to correlate highly with Zinbarg, Mineka, etal., 2013). Second, the anxiety
fear disorders such as panic disorder (.61), agora- and depression-specific factors point to important
phobia (.58), social anxiety disorder (.62), and post- ways in which GAD and MDD might differ from
traumatic stress disorder (.63), which are almost as each other. Reduced positive affect has been impli-
high as the correlation between GAD and MDD cated in MDD through multiple factor analytic
(.65) (Slade & Watson, 2006). Such findings imply studies (e.g., Watson etal., 1995). In addition, the
that GAD may relate to other anxiety disorders in role of reduced positive affect in MDD has been
important ways that we should take into consider- further supported through experimental studies
ation before removing it from this category (see also that show reduced responsiveness and sensitivity to
Hettema, 2008). Therefore high rates of comorbid- positive stimuli (e.g., Rottenberg, Kasch, Gross, &
ity alone do not seem to justify reclassifying GAD Gotlib, 2002). By contrast, to date GAD has not
as a mood disorder. been linked with reduced positive affect. Indeed,
Brown et al. (1998) found that MDD and social
Structural Analyses of Dimensional anxiety disorder (but not GAD) were both associ-
Symptom Measures ated with low positive affectivity. This finding sug-
Another point that should be considered before gests that MDD might share certain important
making classification decisions based on comorbid- features with certain other anxiety disorders that it
ity results is that, as mentioned earlier, these rates does not share withGAD.
of comorbidity are influenced by the evolution of Further, Brown et al. (1998) reported that
symptom criteria for MDD and GAD that has GAD was negatively related to autonomic arousal,
occurred across the DSM-III-R and the DSM-IV. suggesting autonomic suppression in individuals
Therefore we must be cautious in interpreting results with GAD. Borkovec, Alcaine, and Behar (2004)
from structural analyses that make use of diagnostic proposed that this may be because individuals
results from DSM-III-R versus DSM-IV because with GAD engage in worry (a predominantly ver-
clearly the results from these studies are influ- bal activity) that inadvertently serves as a means
enced by the nature of the inputs to these models. to avoid more elaborate processing of emotional
However, structural analyses of dimensional data material. This, in turn, may serve to suppress auto-
would not be as susceptible to changing diagnostic nomic hyperarousal, which might arise from more
criteria. Such early structural analyses of depression image-focused processing of threatening informa-
and anxiety symptoms measures revealed that these tion (Borkovec, Lyonfields, Wiser, & Deihl, 1993).
symptoms load onto three affective dimensions Consistent with this avoidance theory of worry (for
high negative affect, low positive affect, and high a review see also Mineka, 2004), more recent neuro-
autonomic arousal (Watson et al., 1995). These biological research supports the view that worry may
findings were the basis for the tripartite model of be associated with cognitive disengagement from
anxiety and depression (Clark & Watson, 1991). aversive imagery (Schienle, Schfer, Pignanelli, &
This model defined negative affect as a nonspecific Vaitl, 2009). These findings of autonomic suppres-
factor common to both depression and anxiety, sion have so far been demonstrated only in GAD.
low positive affect as a depression-specific factor, Indeed, MDD does not seem to be associated with
and high autonomic arousal as unique to anxiety suppressed autonomic arousal (Brown etal., 1998),
(although it was later found that anxious arousal thus further differentiating the two disorders.
related primarily to panic disorder and not to the
other anxiety disorders) (Brown, Chorpita, & Differences in Cognitive Correlates
Barlow, 1998; Mineka etal.,1998). Associated with GAD andMDD
Structural analyses of symptoms reveal two In addition to these overlapping but partially dis-
points relevant to the nosology of GAD and MDD. tinct affective dimensions of GAD and MDD, other
First, although both GAD and MDD share the points of difference between these two disorders
negative affect dimension (Brown et al., 1998), involve cognitive and biological mechanisms that are
this may not be a good basis on which to reclas- at least somewhat specific to each. One of the key
sify GAD with the mood disorders because nega- points of difference is with regard to biases in infor-
tive affect is considered a nonspecific feature of, mation processing of emotional material. Individuals

Mineka, Anand,Sumner 137

with GAD have consistently been found to direct hand, individuals with GAD have sometimes been
their attention toward threatening (or other negative) found to show implicit memory biases toward
information presented in the form of words (Mogg, threatening information, reflecting the unconscious
Bradley, Williams, & Mathews, 1993) or faces influence of previously presented information
depicting negative emotions (e.g., Bradley, Mogg, (e.g., Coles, Turk, & Heimberg, 2007). However,
White, Groom, & De Bono, 1999). In GAD this a meta-analysis by Mitte (2008) of many such
bias has been found with different forms of negative studies of implicit memory biases found no overall
stimuli (e.g., threat words, depression-related words, effect across studies. Additionally, individuals with
and angry faces), suggesting a relative nonspecificity MDD have consistently been found to show a bias
in the bias toward negative emotional information toward recalling overly general autobiographical
in GAD, possibly due to the diverse content of wor- memories when asked to recall specific memories
ries that characterize this disorder (Mogg & Bradley, to a positive or negative cue word (Park, Goodyear,
2005). Importantly, this attentional bias is evident & Teasdale, 2002; Williams etal., 2007). To date,
when negative stimuli are presented for brief periods however, this bias has not been found in individuals
of time (e.g., 300 to 1,000 ms). Also, nonconscious with GAD (Burke & Mathews, 1992; Williams et
attentional biases toward negative stimuli have been al., 2007), although these authors did not use the
demonstrated with stimuli presented for intervals Autobiographical Memory Test, which is the stan-
as small as 14 ms (e.g., Bradley, Mogg, Millar, & dard test used in the vast majority of research on
White,1995). overly general autobiographical memory in depres-
Nevertheless, studies have also found that when sion. Unfortunately very little research has yet
the stimulus duration is increased to allow more addressed this question; therefore any conclusions
elaborative processing of the information (over regarding overly general memory and GAD would
1,000 ms stimulus duration), individuals with be premature.
GAD fail to show this attentional bias (e.g., Gotlib, Other cognitive processes that have been dis-
Krasnoperova, Yue, & Joormann, 2004; Mogg, cussed in the context of GAD and MDD are two
Millar, & Bradley, 2000). This suggests that GAD is related forms of repetitive thinking. MDD is thought
associated with an attentional bias that occurs only to be associated with rumination, which is a form
at a very early stage of processing, which is not as of repetitive and passive focus on ones depressive
consistently found when more elaborative processing symptoms and negative affect (Nolen-Hoeksema,
occurs. By contrast, a different pattern of attentional 1991). Worry, on the other hand, is one of the core
bias is found in individuals with MDD such that features of GAD and is a form of repetitive thinking
these individuals show an attentional bias toward focused primarily on the possibility of future nega-
negative information, but only when the stimuli are tive outcomes (Borkovec, Robinson, Pruzinsky, &
presented for a longer duration of 1,000 ms or more DePree, 1983). Research suggests substantial over-
(e.g., Gotlib etal., 2004). These findings then sug- lap between the processes of rumination and worry.
gest that GAD (but not MDD) is associated with an Specifically, the self-report measures of rumination
automatic attentional bias when initially orienting and worry tend to be highly correlated (e.g., Fresco,
to negative environmental information. In contrast, Frankel, Mennin, Turk, & Heimberg, 2002; Muris,
MDD (but not GAD) seems to be associated with Roelofs, Meesters, & Boomsma, 2004). Moreover,
biased elaborative processing of negative informa- both rumination and worry may represent forms of
tion, making it difficult for these individuals to dis- unconstructive repetitive thinking that is abstract,
engage attention from negative information (for a negative, and self-focused (Watkins, 2008). Both
review see Mogg & Bradley,2005). have been associated with symptoms of depression
Furthermore, differences have also been found and of anxiety (e.g., Segerstrom, Tsao, Alden, &
in terms of memory biases for negative informa- Craske, 2000) and with negative outcomes such as
tion. Whereas MDD has been fairly consistently poor social problem solving and inadequate solu-
associated with both explicit and implicit memory tion implementation (e.g., Davey, 1994; Watkins &
biases for negative information (e.g., Rinck & Moulds,2005).
Becker, 2005), the bias in explicit memory (involv- However, there are differences between worry
ing conscious, effortful recall) has not been found and rumination. Factor analyses of self-report mea-
consistently in individuals with GAD (Coles & sures of rumination and worry indicate that the two
Heimberg, 2002; for reviews, see also Mineka, constructs form separate factors (although they are
Rafaeli & Yovel, 2003; Mitte, 2008). On the other correlated about .46), suggesting that they may be

138 Generalized Anxiet y Disorder and Major Depressive Disorder

statistically distinguishable (Fresco et al., 2002). is clearly required to definitively conclude whether
As already noted, they are also thought to differ GAD is associated with HPA axis dysregulation.
based on the content of repetitive thought, with Furthermore, much more research is required to
worry being focused on future negative outcomes test the specificity of other neurobiological corre-
and rumination involving repetitive focus on past lates of MDD by comparing groups that have pure
negative events and the causes and meaning of these diagnoses of MDD and GAD. However, space
events. Furthermore, at least based on self-report, prohibits our covering other known or suspected
researchers have suggested that worry and rumi- differenceshere.
nation may serve different functions for the indi-
vidual. When ruminating, individuals report that Are There Different Etiological Pathways for
this helps them gain better insight into their prob- MDD andGAD?
lems. On the other hand, people who engage in As discussed earlier, another important point that
worrying believe that it helps them to anticipate has been used as an argument for grouping GAD and
and prepare for future threat in the face of uncer- MDD together is the high degree of genetic over-
tainty (for a review see Nolen-Hoeksema, Wisco, lap between them. Nevertheless, Hettema (2008)
& Lyubomirsky, 2008). In addition, as mentioned reviewed evidence showing that genetic overlap is
earlier, Borkovec etal. (2004) suggested that worry not unique to MDD and GAD. For example, Lieb,
serves to dampen autonomic arousal and thereby Isensee, Hofler, Pfister, and Wittchen (2002) stud-
permits nonconscious avoidance of aversive images ied rates of illness in children of depressed parents,
and negative affect. Consequently worry prevents controlling for comorbid conditions in the parent.
individuals from experiencing the likely increase Parental MDD was significantly related to elevated
in autonomic arousal that would follow from con- rates of GAD in the offspring. However, parental
fronting such negative stimuli. Furthermore, rumi- MDD was also significantly associated with elevated
nation has been associated with experiencing higher rates in their offspring of panic disorder, obsessive
levels of negative imagery compared with thoughts compulsive disorder, and posttraumatic stress dis-
(McLaughlin, Borkovec, & Sibrava, 2007). In con- order. Twin studies also indicate that panic disorder
trast, worry has been associated with higher levels shares a strong genetic relationship with GAD (e.g.,
of negative thought compared to imagery. Another Scherrer etal.,2000).
cognitive process thought to be unique to GAD Moreover, other studies indicate that there is
is intolerance of uncertainty, which is defined as a only a modest overlap in environmental causal fac-
characteristic difficulty with ambiguous and uncer- tors for GAD and MDD (Kendler, Neale, et al.,
tain outcomes (e.g., Dugas, Buhr, & Ladouceur, 1992). As reviewed earlier, studies examining stress-
2004). Although MDD has not been associated ors that precede onsets of MDD and GAD have
with especially high intolerance of uncertainty, indi- found that different types of stressors may be linked
viduals with MDD do seem to have higher certainty to the development of each of these disorders (e.g.,
that negative events will occur than do individuals Kendler, Hettema, etal., 2003). Furthermore, along
with GAD (Miranda & Mennin,2007). the lines of tracing etiological pathways leading to
GAD and MDD, Beesdo, Pine, Lieb, and Wittchen
Are There Differences in the Neurobiological (2010) studied developmental features associated
Correlates of GAD andMDD? with these two disorders. In this 10-year prospec-
With respect to neurobiological correlates, tive longitudinal study, participants were between
some studies have found differences between GAD the ages of 14 and 24 at baseline and between
and MDD. As one example, GAD and MDD the ages of 21 and 34 years at the last published
are thought to differ with regard to the function- follow-up. This investigation examined the inci-
ing of the hypothalamic-pituitary-adrenal (HPA) dence, comorbidity, and risk patterns for GAD in
axis. A number of studies have shown that MDD relation to depressive (MDD, dysthymia) and other
is associated with several aspects of HPA axis dys- anxiety disorders (specific phobia, social phobia,
regulation (e.g., Goodyer, Park, & Herbert, 2001; agoraphobia, and panic disorder). These analyses
Vreeburg et al., 2009). However, several studies revealed higher hazard ratios for GAD predicting
have failed to find evidence of such abnormalities other anxiety disorders (4.14) and for other anxi-
in individuals with GAD (Hoehn-Saric, McLeod, ety disorders predicting GAD (5.05) as compared
Lee, & Zimmerli, 1991; Pomara, Willoughby, with the association of GAD with depressive disor-
Sidtis, Cooper, & Greenblatt, 2005). More research ders (2.25). Further, the authors compared putative

Mineka, Anand,Sumner 139

risk factors for GAD to those found to be specific role in discussions surrounding revisions to the
to two groupsthose with depression alone and diagnostic criteria for GAD in DSM-5. The current
those with other anxiety disorders alone (without proposal for diagnosing GAD in the next version of
depression or GAD). These analyses revealed that the DSM continues to list GAD as an anxiety dis-
factors associated with GAD overlapped with all the order rather than as a mood disorder. However, the
specific risk factors implicated in other anxiety dis- DSM-5 Task Force did recommend some changes
orders. However, GAD shared only one specific risk in the diagnostic criteria for GAD. For example,
factor (i.e., history of parental depression) in com- one recommendation was to reduce the duration
mon with depression. Thus the authors concluded of anxiety and worry from at least six months to at
that in terms of developmental features and risk fac- least three months. This proposed change was based
tors, GAD overlapped more closely with the other on findings that a lower duration threshold (e.g.,
anxiety disorders than with the depressive disorders. one to five months) identifies individuals with simi-
lar symptom severity and impairment as the cur-
Are Different Outcomes Associated with rent six-month minimum duration requirement in
GAD andMDD? DSM-IV (e.g., Kessler, Brandenburg, etal.,2005).
Points of difference between GAD and MDD The DSM-5 proposed criteria for GAD also
have also been found in terms of outcomes such included changes to the somatic symptoms that are
as the course of illness and response to treatment required for diagnosis. At present, three or more
(Hettema, 2008). Studies show that the two disor- of six symptoms are required in order to receive a
ders often have somewhat different courses of ill- DSM-IV-based diagnosis of GAD. However, the
ness. For example, Bruce et al. (2005) found that new scheme in DSM-5 would have required at least
GAD, like the other anxiety disorders, had a more one of only two symptoms:(1)restlessness, or feel-
chronic course compared with MDD, with a prob- ing keyed up or on edge, and (2) muscle tension
ability of recovery (.58) at 12-year follow-up similar (Andrews et al., 2010), which are most uniquely
to that of several other anxiety disorders (e.g., recov- characteristic of GAD. Although these proposed
ery rates for panic disorder and social phobia were changes were approved by the DSM-5 Task Force,
.48 and .37, respectively) On the other hand, being the American Psychiatric Associations Board of
an episodic disorder, MDD had a somewhat higher Trustees ultimately vetoed these changes and stated
rate of recovery (.73) thanGAD. that they were in need of further study before being
Also, although there is some overlap in response implemented.
to pharmacological treatments for MDD and
GAD, this overlap is only partial. Whereas both Effects of Comorbidity
disorders tend to respond to most antidepressants, As discussed above, studies indicate that MDD
this is not the case for anxiolytic medications, and GAD may differ in terms of several cognitive
which are mostly effective for GAD but not MDD correlates. Given their high rates of comorbid-
(e.g., Hettema, 2008; Levine, Cole, Chengappa, ity, it is important to discuss the implications of
& Gershon, 2001). Furthermore, antidepressant co-occurring MDD and GAD on these correlates
medications have been found to be efficacious not which, as discussed above, have so far largely been
only for GAD but also for other anxiety disorders linked to each disorder separately. This can be accom-
(e.g., Ravindran & Stein, 2010). This suggests once plished by comparing comorbid individuals with
again that the overlap with MDD is not unique to those who have pure diagnoses of GAD or MDD
GADalone. to see whether correlates linked with one disorder
manifest differently when an individual also has a
Implications of These NosologicalIssues comorbid diagnosis of the other. Most research on
In summary, despite the different aspects of over- MDD and GAD has viewed comorbidity in com-
lap between GAD and MDD, evidence suggests parison groups as a limitation that hinders the ability
that both MDD and GAD relate to other anxiety to draw conclusions about each disorder separately.
disorders in important ways that must also be taken Therefore these studies rarely draw conclusions that
into consideration before making reclassification focus on the implications of comorbidity. However,
decisions. Further, the points of difference between some studies (particularly in the area of cognitive
MDD and GAD also suggest that the current dis- biases) have discussed findings using this perspec-
tinction between them in the DSM-IV may be valid tive by viewing comorbidity not as a methodologi-
and informative. Such evidence has likely played a cal hindrance but as having meaningful mechanistic

140 Generalized Anxiet y Disorder and Major Depressive Disorder

and theoretical significancea perspective that we MDD. Particularly, although most studies to date
think will be likely to make important contributions have included only two groupspure GAD and a
to understanding these issues. comorbid group (owing to the high rates of GAD
diagnoses in individuals with depression)more
Effects of Comorbidity on CognitiveBiases studies need to include three groups: one with
As discussed earlier, an important point dis- comorbidity, one with pure GAD, and another with
cussed by Mogg and Bradley (2005) relates to the pure MDD. This would better allow researchers to
observation that whereas GAD has been extensively disentangle the mechanisms at play in individual
associated with early attentional biases for negative versus comorbid diagnoses.
stimuli, this finding has not generally been found Oehlberg, Revelle, and Mineka (2012) proposed
in individuals with MDD (particularly for shorter another useful approach for studying attentional
stimulus durations of less than 1,000 ms). This in biases, taking into account the relationships between
spite of the fact that in several studies, most indi- anxiety and depression. They pointed out that most
viduals in the MDD groups also have co-occurring studies have adopted a disorder model of studying
GAD and/or show elevations in anxiety scores, attentional biases, in which different types of biases
comparable to levels in the pure GAD group (e.g., are associated with observed anxiety or depression.
Bradley et al., 1995). Moreover studies that have This approach has presupposed that each emotional
directly compared a GAD/MDD comorbid group disorder may be associated with a different pattern
with a pure GAD group have found early atten- of processing emotional information. On the other
tional biases toward threat information in the GAD hand, it could also be useful to study how atten-
group but failed to find these biases in the comorbid tional biases map onto underlying common and
group (e.g., Bradley etal., 1995; Mogg etal.,2000). specific dimensions of anxiety and depression that
These observations imply that certain depressive have been identified by structural analyses of their
mechanisms that are at play in co-occurring MDD symptoms. Such an approach would allow us to test
seem to somehow mask the attentional bias seen whether biases are associated with negative affect
in individuals with pure GAD. Attempting one (general factor model) or the specific factors associ-
possible explanation for these findings, Mogg and ated with each disorder (specificity model).
Bradley (2005) articulated a cognitive-motivational In line with this approach, Oehlberg etal. (2012)
analysis of anxiety. According to this model, indi- found that early attentional biases toward angry
viduals with GAD are thought to interpret a wide faces at a stimulus duration of 300 ms were associ-
range of minor environmental cues as being threat- ated with a general tendency toward negative affec-
ening, which in turn prompts goal engagement tivity rather than being specifically associated with
mechanisms that direct processing resources to symptoms of anxiety or dysphoria per se. Further,
these threatening cues. However, although indi- anxious and depressive symptoms were associated
viduals with comorbid MDD might similarly over- with different patterns of responding to sad faces at
estimate the threat value of environmental stimuli, a stimulus duration of 1,000-ms. Depressive symp-
their co-occurring depression may lead to relatively toms were associated with an attentional bias toward
unresponsive goal engagement processes so that the sad faces, and anxious symptoms were associ-
less attention is directed toward external nega- ated with a bias away from sad faces. However, both
tive cues unless the stimuli have high motivational these latter two effects were only observed when the
salience for the individual (such as a high degree of effects of both anxious as well as depressive symp-
self-relevance). Therefore, according to this expla- toms were considered together in the analyses. The
nation, reduced responsiveness of goal engagement authors interpreted these results to suggest that the
processes may be responsible for the general lack of common negative affect factor primarily accounted
findings of early attentional biases toward negative for attentional biases toward threatening stimuli at
information among individuals with co-occurring 300 ms. However, it was the specific dimensions
depression. of anxiety and depression that accounted for their
The findings discussed above suggest that differential patterns of responding to sad stimuli
researchers should look more closely at the implica- at longer stimulus durations, but this effect was
tions of comorbid diagnoses. Obviously, however, uncovered only when controlling for the overlap-
there is a need for more studies designed especially ping variance between anxiety and depression.
to investigate the differences in cognitive biases in Similarly, with regard to GAD and MDD, it may
comorbid versus pure manifestations of GAD and be useful to move beyond DSM-defined diagnoses

Mineka, Anand,Sumner 141

and focus instead on core symptom dimensions Implications of Comorbidity for
that are common or unique to the two disorders. By Impairment
this approach, we may be able to disentangle what Current debates regarding the nosology of GAD
specific and nonspecific symptom dimensions drive and MDD have leaned toward retaining their sepa-
attentional biases within individuals with comorbid rate classifications in the DSM-5 as well as retain-
GAD and MDD as well as study the interactive ing GAD in the anxiety disorder category. However,
influence of each of these dimensions. These effects it is still important to acknowledge the impact of
would not be evident if one were to conduct only high comorbidity rates between these two disorders.
group comparisons using individuals with comor- Given that each disorder alone is associated with sig-
bid GAD andMDD. nificant impairment, it is not surprising that comor-
bidity would be associated with a higher degree of
Effects of Comorbidity on Other impairment and cost compared with the unique
Cognitive Variables diagnosis of either alone. For example, Hunt, Slade,
With regard to repetitive thinking, most stud- and Andrews (2004) found that among individuals
ies indicate that pure GAD has been associated with GAD, comorbidity with MDD had implica-
with increased levels of worry and pure MDD with tions for higher levels of disability than comorbidity
increased levels of rumination. However, those with with other anxiety disorders. In addition, Carroll,
a comorbid diagnosis of both show elevations in Phillips, Gale, and Batty (2010) reported findings
both forms of repetitive thought, with levels of worry from the Vietnam Experience Study suggesting that
comparable to elevations seen in pure diagnoses of when the presence of GAD, the presence of MDD,
GAD and rumination levels comparable to those and the comorbid occurrence of GAD and MDD
with pure MDD (e.g., Chelminski & Zimmerman, were included as predictors in a fully adjusted model
2003; Hofmann, Schulz, Heering, Muench, & predicting the incidence of hypertension, only
Bufka, 2010; Yook, Kim, Suh, & Lee, 2010). Studies comorbidity emerged as a significant predictor. In
have also compared levels of intolerance of uncer- terms of the cost implications, findings suggest that
tainty in comorbid individuals with those in persons the comorbid diagnosis of MDD in individuals with
with GAD or MDD alone. Unfortunately research GAD is linked with higher annual inpatient costs,
on this topic has been inconsistent. For example, total annual costs, and a higher likelihood of being
Yook etal. (2010) and Depuy and Ladouceur (2008) hospitalized than is the case for individuals with a
both found that individuals with comorbid GAD diagnosis of pure GAD (Zhu, Zhao, Ye, Marciniak,
and MDD had higher levels of intolerance of uncer- & Swindle, 2009). These findings point to the clini-
tainty than those who were diagnosed with pure cal significance of the comorbidity between GAD
GAD or MDD. On the other hand, Aldao, Mennin, and MDD and the need to take comorbidity into
Linardatos, and Fresco (2010) reported similar lev- consideration in treating these disorders.
els of intolerance of uncertainty in those with pure
and comorbid diagnoses of GAD. Therefore more Treatment Implications of Comorbidity
research is needed to determine the implication of Given the high degree of impairment and sever-
comorbidity for levels of intolerance of uncertainty. ity associated with the co-occurrence of MDD and
Another approach through which to investigate GAD, an important point to consider is the psycho-
the implications of comorbidity is to look at the therapeutic treatment implications of this comor-
biological correlates of these cognitive variables. bidity. That is, to what extent should a comorbid
For example, Hofmann et al. (2010) found that diagnosis of either GAD or MDD be taken into
among individuals with GAD who were engaged consideration in treating either of these disorders?
in a worry task, those who had MDD had greater Treatment studies suggest that cognitive behav-
high-frequency heart rate variability during the ioral treatments targeting one disorder are only
worry task compared with those without depres- sometimes effective in also reducing symptoms
sion, suggesting that depression may moderate the and maintaining gains with regard to the comor-
effects of worrying on physiological arousal. This bid diagnosis. For example, Newman, Przeworski,
finding provides some very preliminary evidence Fisher, and Borkovec (2010) found that 14 weekly
that different physiological correlates may under- sessions of cognitive behavioral therapy target-
lie the cognitive manifestations of comorbid GAD ing GAD reduced co-occurring MDD symptoms
and MDD as compared with a pure diagnosis of immediately after treatment. Unfortunately these
eitherone. gains for MDD symptoms were not retained at

142 Generalized Anxiet y Disorder and Major Depressive Disorder

follow-up. In fact, the clinical severity ratings for response to negative thoughts and emotions believe
MDD at a 24-month follow-up were no longer that such forms of thinking are useful in solving
significantly different from pretreatment levels. their problems. They also tend to carry negative
Nevertheless, these findings are perhaps not surpris- beliefs that their thoughts and emotions are uncon-
ing given the high rates of relapse and recurrence trollable and carry great significance.
observed in depression and the challenges that they Metacognitive therapy aims to reduce repetitive
pose for treatment of the disorder (e.g., Richards & thinking in the form of rumination in depression
Perri,2010). and worry in GAD by (1) building awareness of
Given the high rates of comorbidity and the these negative forms of repetitive thought, (2)chal-
implications of this comorbidity for treatment lenging positive metacognitive beliefs about the
outcomes, it is important to study and implement usefulness of such forms of thinking, and (3)chal-
specific strategies in the clinical setting that may be lenging negative metacognitive beliefs about the
useful with comorbid cases. Belzer and Schneier uncontrollability and significance of negative
(2004) made several clinical recommendations for thoughts and feelings. Some research suggests that
addressing co-occurring depression and anxiety that this approach may be effective in treating individu-
may still be useful to consider in the specific context als with GAD (Wells & King, 2006) and MDD
of GAD and MDD. Specifically, it is important that (Wells et al., 2009). However, more research is
individuals presenting with one of the disorders be needed to determine its efficacy in treating individ-
evaluated for co-occurring symptoms of the other, uals with co-occurring GAD andMDD.
even if they occur at subsyndromal levels, because
this may impact treatment response. It would also Unified Protocols for Treating Comorbid
be useful to obtain a longitudinal perspective on Depression and Anxiety
symptom manifestation by determining the age of Along similar lines, some researchers have pro-
onset and temporal course of each set of symptoms. posed a unified treatment approach specifically for
For example, anxiolytic medication might be more the purpose of treating multiple diagnoses of mood
beneficial in cases where a mild depressive episode and anxiety disorders. Such an approach would
occurs within the context of high levels of chronic make use of the fact that these two classes of dis-
worry compared with patients who experience orders share certain underlying vulnerabilities and
increased worry primarily during a depressive epi- expressions of pathological emotional responding.
sode. Therefore it is important to tailor treatment For example, given that high negative affect is a
to an individual, taking into account the diverse nonspecific characteristic of most emotional disor-
profiles of MDD and GAD symptoms with which ders (including GAD and MDD), addressing core
individuals present. processes that lead to increased levels of negative
As mentioned in an earlier section, SSRIs are affect may be a useful strategy in treating individu-
known to be helpful in treating symptoms of both als with co-occurring anxiety and depression. Other
anxiety and depression. Such medication may be nonspecific targets for therapeutic change would
useful in treating individuals with comorbid GAD include altering biases in cognitive processing that
and MDD. However, more research is needed to are shared by co-occurring disorders (e.g., overes-
determine the efficacy of specific SSRIs in treating timating the probability of negative outcomes) and
individuals with comorbidity. Also, the efficacy of reducing behavioral avoidance that may arise out
an SSRI in reducing co-occurring anxiety symp- of attempts to reduce or avoid negative affect. For
toms may differ across different anxiety disorders example, Moses and Barlow (2006) proposed a uni-
given that the efficacy of SSRIs varies across anxi- fied approach that would begin with psychoeduca-
ety disorders (for a more detailed account on issues tion about the nature and function of emotions and
with medication treatment of anxiety and depres- emotional distortions and continue by focusing on
sion see Belzer & Scheier,2004). three main aspects:modifying antecedent cognitive
One useful approach in treatment may be to appraisals, changing behaviors or action tenden-
reduce pathological processes that are common to cies that occur in response to emotional states, and,
GAD and MDD. One such approach is metacog- finally, preventing emotional avoidance.
nitive therapy, rooted in the metacognitive model Testing the preliminary efficacy of such a treat-
of emotional disorders (Wells & Matthews, 1996). ment plan based on such a unified protocol, Ellard,
According to this model, individuals who are prone Fairholme, Boisseau, Farchione, and Barlow (2010)
to engage in maladaptive repetitive thinking in presented results from two open clinical trials using

Mineka, Anand,Sumner 143

a heterogeneous clinical sample of participants is warranted, and the DSM-5 Task Force seems to
with multiple anxiety and depressive diagnoses. have come to the same conclusion. We briefly dis-
They demonstrated that the unified protocol was cussed the known genetic overlap between GAD and
effective in treating both the principal diagnosis MDD but also the relative paucity of our knowledge
of an anxiety disorder (including GAD) and also at this point regarding the different environmental
in reducing symptoms of the comorbid diagnosis precursors of GAD versus MDD. We also reviewed
(which included MDD) both at posttreatment and the effects on cognitive biases and on impairment of
at six-month follow-up. Additional initial empirical having comorbid GAD and MDD. Finally, we dis-
support for the unified protocol as a transdiagnos- cussed the clinical implications of comorbid GAD
tic treatment for anxiety disorders and co-occurring and MDD and approaches that may be useful in
MDD was obtained in at least one study to date by addressing this comorbidity in a clinical setting.
Farchione etal. (2012). These investigators demon-
strated the efficacy of this treatment in comparison to References
wait list controls in achieving subclinical responder Akiskal, H. S. (1990). Toward a clinical understanding of the
relationship between anxiety and depressive disorders. In J.
or high end-state functioning status in individuals
D.Maser & C. R.Cloninger (Eds.), Comorbidity of mood and
with co-occurring MDD or dysthymia. Nevertheless, anxiety disorders (pp. 597610). Washington, DC:American
it will be very important to replicate such findings PsychiatricPress.
using more rigorous control conditions, particularly Aldao, A., Mennin, D.D., Linardatos, E., & Fresco, D. M.
because the number of individuals with GAD and (2010). Differential patterns of physical symptoms and sub-
jective processes in generalized anxiety disorder and unipolar
co-occuring depression was quite small in thisstudy.
depression. Journal of Anxiety Disorders, 24, 250259.
Another integrative approach that is more behav- Alloy, L. B., Kelly, K. A., Mineka, S., & Clements, C. M.
ioral was also proposed by Weersing, Gonzalez, (1990). Comorbidity of anxiety and depressive disor-
Campo, and Lucas (2008) for treating comorbid ders:Ahelplessness-hopelessness perspective. In J. D.Maser
depression and anxiety in a youth population. & C. R. Cloninger (Eds.), Comorbidity of mood and anxi-
ety disorders (pp. 499544). Washington, DC: American
Their graded engagement technique integrates
behavioral activation techniques for depression and Andrews, G., & Hobbs, M. J. (2010). The effect of the draft
exposure techniques for anxiety. Therefore, given DSM-5 criteria for GAD on prevalence and severity. Australian
the commonalities in the vulnerability towards and and New Zealand Journal of Psychiatry, 44, 784790.
mechanisms underlying GAD and MDD, such Andrews, G., Hobbs, M. J., Borkovec, T. D., Beesdo, K.,
Craske, M. G., Heimberg, R. G.,...Stanley, M. A. (2010).
unified approaches to treatment could be very ben-
Generalized worry disorder: A review of DSM-IV general-
eficial with individuals with comorbid GAD and ized anxiety disorder and options for DSM-V. Depression and
MDD. However, much more research is required to Anxiety, 27, 134147.
test the extent of the efficacy (and effectiveness) of Beesdo, K., Pine, D. S., Lieb, R., & Wittchen, H. (2010).
unified treatment protocols specifically in treating Incidence and risk patterns of anxiety and depressive dis-
orders and categorization of generalized anxiety disorder.
individuals with comorbid GAD andMDD.
Archives of General Psychiatry, 67,4757.
Belzer K., Schneier F.R. (2004). Comorbidity of anxiety and
Conclusions depressive disorders:Issues in conceptualization, assessment,
In conclusion, this chapter has reviewed many and treatment. Journal of Psychiatric Practice, 10, 296306.
of the major findings and controversies that have Borkovec, T. D., Alcaine, O., & Behar, E. (2004). Avoidance
theory of worry and generalized anxiety disorder. In
emerged in the past 25 years since DSM-III-R
R. G. Heimberg, C. L. Turk, & D. S. Mennin (Eds.),
(1987) regarding the comorbidity of GAD and Generalized anxiety disorder:Advances in research and practice
MDD. GAD is certainly not the only anxiety disor- (pp. 77108). NewYork:GuilfordPress.
der to have high comorbidity with MDD. However, Borkovec, T. D., Lyonfields, J. D., Wiser, S. L., & Deihl, L.
among the different anxiety disorder/MDD comor- (1993). The role of worrisome thinking in the suppression
of cardiovascular response to phobic imagery. Behaviour
bidities, the GAD/MDD comorbidity topic has
Research and Therapy, 31, 321324.
been perhaps the most controversial, given that Borkovec, T. D., Robinson, E., Pruzinsky, T., & DePree, J. A.
the DSM-5 Task Force was charged with evaluat- (1983). Preliminary exploration of worry:Some characteris-
ing the evidence regarding whether GAD should tics and processes. Behaviour Research & Therapy, 21,916.
actually be removed from the anxiety disorder cat- Bradley, B. P., Mogg, K., Millar, N., & White, J. (1995). Selective
processing of negative information:Effects of clinical anxiety,
egory and placed in the unipolar mood disorder cat-
concurrent depression, and awareness. Journal of Abnormal
egory. Our review has led us to conclude that they Psychology, 104, 3, 532536.
should remain in separate categories, as they have Bradley, B. P., Mogg, K., White, J., Groom, C., & De Bono, J.
been. Clearly, however, more research on this topic (1999). Attentional bias for emotional faces in generalized

144 Generalized Anxiet y Disorder and Major Depressive Disorder

anxiety disorder. British Journal of Clinical Psychology, 38, Fresco, D. M., Frankel, A. N., Mennin, D. S., Turk, C. L., &
267278. Heimberg, R. G. (2002). Distinct and overlapping features
Brown, T. A., Campbell, L. A., Lehman, C. L., Grisham, J. R., of rumination and worry:The relationship of cognitive pro-
& Mancill, R. B. (2001). Current and lifetime comorbidity duction to negative affective states. Cognitive Therapy and
of the DSM-IV anxiety and mood disorders in a large clinical Research, 26, 179188.
sample. Journal of Abnormal Psychology, 110, 585599. Goodyer, I. M., Park, R. J., & Herbert, J. (2001). Psychosocial
Brown, T. A., Chorpita, B. F., & Barlow, D.H. (1998). Structural and endocrine features of chronic first-episode major depres-
relationships among dimensions of the DSM-IV anxiety sion in 816year olds. Biological Psychiatry, 50, 351357.
and mood disorders and dimensions of negative affect, Gotlib, I. H., Krasnoperova, E., Yue, D. N., & Joormann, J.
positive affect, and autonomic arousal. Journal of Abnormal (2004). Attentional biases for negative interpersonal stimuli
Psychology, 107, 179192. in clinical depression. Journal of Abnormal Psychology, 113,
Bruce, S. E., Yonkers, K. A., Otto, M. W., Eisen, J. L., Weisberg, 127135.
R. B., Pagano, M.,...Keller, M. B. (2005). Influence of Hettema, J.M. (2008). The nosological relationship between
psychiatric comorbidity on recovery and recurrence in generalized anxiety disorder and major depression. Depression
generalized anxiety disorder, social phobia, and panic dis- and Anxiety, 25, 300316.
order: A 12-year prospective study. American Journal of Hettema, J. M., Neale, M. C., Myers, J. M., Prescott, C. A.,
Psychiatry, 162, 11791187. & Kendler, K. S. (2006). A population-based twin study of
Burke, M. & Mathews, A. (1992). Autobiographical memory the relationship between neuroticism and internalizing disor-
and clinical anxiety. Cognition and Emotion, 6,2335. ders. American Journal of Psychiatry, 163, 857864.
Carroll, D., Phillips, A. C., Gale, C. R., & Batty, G. D. (2010). Hoehn-Saric, R., McLeod, D. R., Lee, Y. B., & Zimmerli, W.
Generalized anxiety and major depressive disorders, their D. (1991). Cortisol levels in generalized anxiety disorder.
comorbidity and hypertension in middle-aged men. Psychiatry Research, 38, 313315.
Psychosomatic Medicine, 72,1619. Hofmann, S. G., Schulz, S. M., Heering, S., Muench, F., &
Chelminski, I., & Zimmerman, M. (2003). Pathological Bufka, L. F. (2010). Psychophysiological correlates of gener-
worry in depressed and anxious patients. Journal of Anxiety alized anxiety disorder with or without comorbid depression.
Disorders, 17, 533546. International Journal of Psychophysiology, 78,3541.
Clark, L. A. & Watson, D. (1991). Tripartite model of anxi- Hunt, C., Slade, T., & Andrews, G. (2004). Generalized anxi-
ety and depression: Psychometric evidence and taxonomic ety disorder and major depressive disorder comorbidity
implications. Journal of Abnormal Psychology, 100, 316336. in the National Survey of Mental Health and Well-Being.
Clark, L. A., Watson, D., & Mineka, S. (1994). Temperament, Depression and Anxiety, 20,2331.
personality and the mood and anxiety disorders. Journal of Kendler, K. S. (1996). Major depression and generalised anxiety
Abnormal Psychology, 103, 103116. disorder. Same genes, (partly) different environmentsrevis-
Coles, M. E., & Heimberg, R. G. (2002). Memory biases in the ited. British Journal of Psychiatry, 168 (Suppl.),6875.
anxiety disorders:Current status. Clinical Psychology Review, Kendler, K. S., Gardner, C. O., Gatz, M., & Pedersen, N. L.
22, 587627. (2007). The sources of co-morbidity between major depres-
Coles, M. E., Turk, C. L., & Heimberg, R. G. (2007). Memory sion and generalized anxiety disorder in a Swedish national
bias for threat in generalized anxiety disorder: The poten- twin sample. Psychological Medicine, 37, 453462.
tial importance of stimulus relevance. Cognitive Behaviour Kendler K. S., Hettema J. M., Butera F., Gardner C. O., &
Therapy, 36,6573. Prescott C. A. (2003). Life event dimensions of loss, humili-
Davey, G.C.L. (1994). Worrying, social problem solving abili- ation, entrapment, and danger in the prediction of onsets of
ties, and problem-solving confidence. Behaviour Research and major depression and generalized anxiety. Archives of General
Therapy, 32, 327330. Psychiatry, 60, 789796.
Dupuy, J., & Ladouceur, R. (2008). Cognitive processes of gen- Kendler K.S., Neale M.C., Kessler R.C., Heath A.C., & Eaves
eralized anxiety disorder in comorbid generalized anxiety L.J. (1992). Major depression and generalized anxiety disor-
disorder and major depressive disorder. Journal of Anxiety der: Same genes, (partly) different environments? Archives of
Disorders, 22, 505514. General Psychiatry, 49, 716722.
Dugas, M. J., Buhr, K., & Ladouceur, R. (2004). The role of Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., &
intolerance of uncertainty in etiology and maintenance. Eaves, L. J. (1993). A longitudinal twin study of personal-
In R. G. Heimberg, C. L. Turk, & D. S. Mennin (Eds.), ity and major depression in women. Archives of General
Generalized anxiety disorder:Advances in research and practice Psychiatry, 50, 853862.
(pp. 142163). NewYork:GuilfordPress. Kendler K. S., Prescott C. A., Myers J., & Neale M. C. (2003).
Ellard, K. K., Fairholme, C. P., Boisseau, C. L., Farchione, T. The structure of genetic and environmental risk factors for
J., & Barlow, D. H. (2010). Unified protocol for the trans- common psychiatric and substance use disorders in men and
diagnostic treatment of emotional disorders:Protocol devel- women. Archives of General Psychiatry, 60, 929937.
opment and initial outcome data. Cognitive and Behavioral Kessler, R. C., Brandenburg, N., Lane, M., Roy-Byrne, P., Stang,
Practice, 17, 88101. P. D., Stein, D. J., & Wittchen, H-U. (2005). Rethinking
Eysenck, H. J. & Eysenck, S. B.G. (1975). Eysenck Personality the duration requirement for generalized anxiety disor-
Questionnaire:Manual. London:Hodder and Stoughton. der: Evidence from the National Comorbidity Survey
Farchione, T. J., Fairholme, C. P., Ellard, K. K., Boisseau, C. L., Replication. Psychological Medicine, 35,110.
Thompson-Hollands, J., Carl, J. R,...Barlow, D. H. (2012). Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. (2005).
Unified protocol for transdiagnostic treatment of emotional Prevalence, severity, and comorbidity of 12-month DSM-IV
disorders: A randomized controlled trial. Behavior Therapy, disorders in the National Comorbidity Survey Replication.
43, 666678. Archives of General Psychiatry, 62, 617627.

Mineka, Anand,Sumner 145

Kessler, R. C., Gruber, M., Hettema, J. M., Hwang, I., Sampson, Moffitt, T. E., Harrington, H., Caspi, A., Kim-Cohen, J.,
N., & Yonkers, K. A. (2008). Comorbid major depression and Goldberg, D., Gregory, A. M., & Poulton, R. (2007).
generalized anxiety disorders in the National Comorbidity Depression and generalized anxiety disorder: Cumulative
Survey follow-up. Psychological Medicine, 38, 365374. and sequential comorbidity in a birth cohort followed pro-
Kessler, R. C., Nelson, C. B., McGonagle, K. A., Liu, J., Swartz, spectively to age 32years. Archives of General Psychiatry, 64,
M., & Blazer, D. G. (1996). Comorbidity of DSM-III-R 651660.
major depressive disorder in the general population:Results Mogg, K., & Bradley, B. P. (2005). Attentional bias in gener-
from the US National Comorbidity Survey. British Journal of alized anxiety disorder versus depressive disorder. Cognitive
Psychiatry, 168,1730. Therapy and Research, 29,2945.
Klein, D. N., Durbin, C. E., & Shankman, S. A. (2009). Mogg, K., Bradley, B. P., Williams, R., & Mathews, A. (1993).
Personality and mood disorders. In I. H. Gotlib & C. Subliminal processing of emotional information in anxi-
L.Hammen (Eds.), Handbook of depression and its treatment ety and depression. Journal of Abnormal Psychology, 102,
(2nd ed., pp. 93112). NewYork:GuilfordPress. 304311.
Krueger, R.F. (1999). The structure of common mental disor- Mogg, K., Millar, N., & Bradley, B. P. (2000). Biases in eye
ders. Archives of General Psychiatry, 56, 921926. movements to threatening facial expressions in generalized
Krueger, R. F., & Markon, K. E. (2006). Reinterpreting comor- anxiety disorder and depressive disorder. Journal of Abnormal
bidity:Amodel-based approach to understanding and classi- Psychology, 109, 695704.
fying psychopathology. Annual Review of Clinical Psychology, Moses, E. B., & Barlow, D. H. (2006). A new unified treatment
2, 111133. approach for emotional disorders based on emotion science.
Lawrence, A. E., Liverant, G. I., Rosellini, A. J., & Brown, T. Current Directions in Psychological Science, 15, 146150.
A. (2009). Generalized anxiety disorder within the course Muris, P., Roelofs, J., Meesters, C., & Boomsma, P. (2004).
of major depressive disorder: Examining the utility of Rumination and worry in nonclinical adolescents. Cognitive
the DSM-IV hierarchy rule. Depression and Anxiety, 26, Therapy and Research, 28, 539554.
909916. Newman, M. G., Przeworski, A., Fisher, A. J., & Borkovec, T. D.
Levine, J., Cole, D. P., Chengappa, R. K., & Gershon, S. (2001). (2010). Diagnostic comorbidity in adults with generalized
Anxiety disorders and major depression, together or apart. anxiety disorder: Impact of comorbidity on psychotherapy
Depression and Anxiety, 14, 94104. outcome and impact of psychotherapy on comorbid diagno-
Lieb, R., Isensee, B., Hofler, M., Pfister, H., & Wittchen ses. Behavior Therapy, 41,5972.
H. U. (2002). Parental major depression and the Nolen-Hoeksema, S. (1991). Responses to depression and their
risk of depression and other mental disorders in off- effects on the duration of depressive episodes. Journal of
spring: A prospective-longitudinal community study. Abnormal Psychology, 100, 569582.
Archives of General Psychiatry, 59, 365374. Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008).
McLaughlin, K., Borkovec, T. D., & Sibrava, N. J. (2007). The Rethinking rumination. Perspectives on Psychological Science,
effect of worry and rumination on affective states and cogni- 3, 400424.
tive activity. Behavior Therapy, 38,2338. Oehlberg, K. A., Revelle, W., & Mineka, S. (2012). Time-course
Mennin, D. S., Heimberg, R. G., Fresco, D. M., & Ritter, M. R. of attention to negative stimuli:Negative affectivity, anxiety,
(2008). Is generalized anxiety disorder an anxiety or mood or dysphoria? Emotion, 12, 943959.
disorder? Considering multiple factors as we ponder the fate Park, R. J., Goodyer, I. M., & Teasdale, J. D. (2002). Categoric
of GAD. Depression and Anxiety, 25, 289299. overgeneral autobiographical memory in adolescents with
Mineka, S. (2004). The positive and negative consequences major depressive disorder. Psychological Medicine, 32,
of worry in the etiology of generalized anxiety disor- 267276.
der: A vicious circle? In J. Yiend (Ed.), Cognition, emotion Pomara, N., Willoughby, L. M., Sidtis, J. J., Cooper, T. B., &
and psychopathology:Theoretical, empirical and clinical direc- Greenblatt, D. J. (2005). Cortisol response to diazepam:Its
tions (pp. 2948). NewYork:Cambridge UniversityPress. relationship to age, dose, duration of treatment, and pres-
Mineka, S., Rafaeli, E., & Yovel, I. (2003) Cognitive biases ence of generalized anxiety disorder. Psychopharmacology,
in emotional disorders: Information processing and 178,18.
social-cognitive perspectives. In R. Davidson, H. Goldsmith, Ravindran, L. N., & Stein, M. B. (2010). The pharmacologic
& K. Scherer (Eds.), Handbook of affective science (pp. 976 treatment of anxiety disorders: A review of progress. The
1009). NewYork:Oxford UniversityPress. Journal of Clinical Psychiatry, 71, 839854.
Mineka, S., Watson, D., & Clark, L. A. (1998). Comorbidity Richards, C. S., & Perri, M. G. (Eds.). (2010). Relapse preven-
of mood and anxiety disorders. Annual Review of Psychology, tion for depression. Washington, DC:American Psychological
49, 377412. Association.
Miranda, R., & Mennin, D. S. (2007). Depression, generalized Rinck, M., & Becker, E. S. (2005). A comparison of attentional
anxiety disorder, and certainty in pessimistic predictions biases and memory biases in women with social phobia and
about the future. Cognitive Therapy and Research, 31,7182. major depression. Journal of Abnormal Psychology, 114,6274.
Mitte, K. (2008) Memory bias for threatening information Roberts, B. W. & Del Vecchio, W. F. (2000). The rank-order
in anxiety and anxiety disorders: A meta-analytic review. consistency of personality traits from childhood to old age:
Psychological Bulletin, 134, 886911. A quantitative review of longitudinal studies. Psychological
Moffitt, T. E., Caspi, A., Harrington, H., Milne, B. J., Melchior, Bulletin, 126,325.
M., Goldberg, D., & Poulton, R. (2007). Generalized anxi- Rottenberg, J., Kasch, K. L., Gross, J. J., & Gotlib, I. H. (2002).
ety disorder and depression: Childhood risk factors in a Sadness and amusement reactivity differentially predict con-
birth cohort followed to age 32. Psychological Medicine, 37, current and prospective functioning in major depressive dis-
441452. order. Emotion, 2, 135146.

146 Generalized Anxiet y Disorder and Major Depressive Disorder

Roy, M. A., Neale, M. C., Pedersen, N. L., Math, A. A., & depression in student, adult, and patient samples. Journal of
Kendler, K. S. (1995). A twin study of generalized anxiety Abnormal Psychology 104,1525.
disorder and major depression. Psychological Medicine, 25, Weersing, M. M., Gonzalez, A., Campo, J. V. & Lucas, A. N.
10371049. (2008). Brief behavioral therapy for pediatric anxiety and
Scherrer, J. F., True, W. R., Xian H., Lyons M.J., Eisen S.A., depression: Piloting an integrated treatment approach.
Goldberg J.,...Tsuang M. T. (2000). Evidence for genetic Cognitive and Behavioral Practice, 15, 129139.
influences common and specific to symptoms of generalized Weinstock, L. M., & Whisman, M. A. (2006). Neuroticism
anxiety and panic. Journal of Affective Disorders, 57,2535. as a common feature of the depressive and anxiety disor-
Schienle, A., Schfer, A., Pignanelli, R. & Vaitl, D. (2009). ders:Atest of the revised integrative hierarchical model in a
Worry tendencies predict brain activation during aversive national samp