Diagnostic Meta Analysis Giuseppe

Biondi-Zoccai
Visit to download the full and correct content document:
https://textbookfull.com/product/diagnostic-meta-analysis-giuseppe-biondi-zoccai/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Applied Meta-Analysis with R and Stata (Chapman &

Hall/CRC Biostatistics Series) 2nd Edition Chen

https://textbookfull.com/product/applied-meta-analysis-with-r-
and-stata-chapman-hall-crc-biostatistics-series-2nd-edition-chen/

Neonatology Giuseppe Buonocore

https://textbookfull.com/product/neonatology-giuseppe-buonocore/

Effectiveness of Time Investments in Education Insights

from a review and meta analysis 1st Edition Jaap
Scheerens (Auth.)

https://textbookfull.com/product/effectiveness-of-time-
investments-in-education-insights-from-a-review-and-meta-
analysis-1st-edition-jaap-scheerens-auth/

A Legal Analysis of the Belt and Road Initiative:

Towards a New Silk Road? Giuseppe Martinico

https://textbookfull.com/product/a-legal-analysis-of-the-belt-
and-road-initiative-towards-a-new-silk-road-giuseppe-martinico/
Nonlinear Meta-Optics 1st Edition Costantino De Angelis
(Editor)

https://textbookfull.com/product/nonlinear-meta-optics-1st-
edition-costantino-de-angelis-editor/

A Guide to Qualitative Meta synthesis 1st Edition

Deborah Finfgeld-Connett

https://textbookfull.com/product/a-guide-to-qualitative-meta-
synthesis-1st-edition-deborah-finfgeld-connett/

The ABC s of Science Giuseppe Mussardo

https://textbookfull.com/product/the-abc-s-of-science-giuseppe-
mussardo/

Meta Gods War 3 1st Edition B N Miles Miles B N

https://textbookfull.com/product/meta-gods-war-3-1st-edition-b-n-
miles-miles-b-n/

Law, Development and Innovation 1st Edition Giuseppe

Bellantuono

https://textbookfull.com/product/law-development-and-
innovation-1st-edition-giuseppe-bellantuono/
Diagnostic
Meta-Analysis

A Useful Tool for Clinical

Decision-Making
Giuseppe Biondi-Zoccai
Editor

123
Diagnostic Meta-Analysis
Giuseppe Biondi-Zoccai
Editor

Diagnostic Meta-Analysis
A Useful Tool for Clinical
Decision-Making
Editor
Giuseppe Biondi-Zoccai
Department of Medico-Surgical Sciences
Sapienza University of Rome
Latina
Italy

ISBN 978-3-319-78965-1    ISBN 978-3-319-78966-8 (eBook)

https://doi.org/10.1007/978-3-319-78966-8

Library of Congress Control Number: 2018949118

© Springer International Publishing AG, part of Springer Nature 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, express or implied, with respect to the material contained herein or for any errors
or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims
in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by the registered company Springer International Publishing AG part
of Springer Nature.
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
To Vincenzo, brother, friend, and father
Foreword

When they wake up in the morning in the third millennium medical students silently
give thanks that they are no longer required to do diagnosis by tasting their patients’
urine. Truly diagnosis has come a long way since then.
The growth of interest in the scientific study of diagnosis goes back many years
and has its origins in a number of issues which emerged in the second half of the last
century. The desire to compare the health not just of people within countries but also
between countries gave rise to international classifications like that of the World
Health Organization with its successive International Classification of Diseases.
The regulation of medicinal products and their use gave rise to clinical trials of the
new medicines and these trials triggered interest in precise and operationalizable
diagnostic criteria. Finally researchers in various countries started international
comparative studies of the causes and prognosis of disease and they needed to be
sure that each team meant the same by each diagnostic label.
An interesting development in the last quarter of the last century was the realiza-
tion that patients did not always present early in the course of their illness and so
opportunities for effective intervention were being missed with resultant human
suffering. In many cases this late presentation was perfectly understandable as the
disease was often silent in its early stages. The technology of screening was rapidly
developed but also revealed that the criteria for a successful screening test were very
stringent as sensitivity and specificity needed to be very high. Low sensitivity would
mean missing many true cases of disease which rather defeated the object of screen-
ing for it. Low specificity would mean, especially when the disease was of low
community prevalence, that many true non-cases were erroneously identified and
subject to both unnecessary follow-up investigations but also equally unnecessary
anxiety and distress. Estimating the high values of both sensitivity and specificity
was going to demand very large studies.
As interest in diagnosis began to grow for the reasons we have outlined it also
became clear that many studies were far too small. There were many reasons for
this. Researchers did not usually have as their primary interest the development and
testing of their diagnostic tools and in their anxiety to proceed to the main part of
their epidemiological study or clinical trial they understandably did a good job
rather than an optimal one. Obtaining funding to develop diagnostic tools was dif-
ficult as such work often did not fit into the priorities of governments and charities.

vii
viii Foreword

Against all this background it is easy to see why researchers interested in diag-
nosis were attracted to the possibilities of taking some of the methods which their
colleagues in clinical trials had been so keen to adopt to meta-analyze study results.
It is the technology which diagnostic study meta-analysis requires which is both
similar to but also subtly different from that for clinical trials that this book sets out.

King’s College Michael Dewey

London, UK
Preface

Kindness is the language which the deaf can hear and the blind can see
Mark Twain

What is the first goal that comes to my mind when conceiving the idea of a new
book? Summarizing the evidence base on an intriguing topic? Leading an authorita-
tive team of international experts? Forcing my personal agenda in shaping the pres-
ent and future of scholarly endeavors? Actually, none of them. In all truth, I
conceived the idea of this book devoted to diagnostic meta-analyses, my third opus
on evidence synthesis after the others focusing on network meta-analyses and
umbrella reviews [1–2], with a very personal goal. I wanted to thank and provide
recognition to someone I have always felt very special to me: my brother, Vincenzo.
He has always been a father and a friend, on top of being a brother. He has always
proved a formidable mentor, not simply a technical or knowledgeable one, but a true
expert in empathy and affection. This holds even truer in these last years, which
have been among the toughest for me and my sons, Attilio Nicola, Giuseppe Giulio,
and Giovanni Vincenzo. Through his support, as well as the guidance of my mother,
Giulia, my father, Gianni, and my sisters, Erica and Gina, I have been able to over-
come several personal and professional hurdles, and luckily could complete the
unique opportunity of editing this book.
On top of my family, I really wish to testify my gratitude to my friend and men-
tor, Antonio Abbate, who, despite the distance, has been as close as ever. Enrico
Romagnoli, much closer in distance and as tight in friendship, has patiently listened
to all my personal and professional complaints and has inspired me through his
outstanding bearing as father and physician. Giacomo Frati, who has always
believed in me and continues to do so in professional and personal terms, despite
my ups and downs, is also to be thanked wholeheartedly for combining friendship
and alliance for a common success. Last but not least, Laura Gatto lighted my recent
path with her wholehearted, caring and forceful support.
Paradoxically, I also want to thank all those people who have criticized, debased,
or challenged me professionally or personally in the last few years. They have been
many and enthusiast in their ominous purpose, and this is one of the reasons I prefer
to avoid identifying them. Despite often leaving marks and bruises, and occasion-
ally scars, some quite profound, they have confirmed me that only our inner drive to
sacrifice, success, and sharing can bring forward our very best.

ix
x Preface

Thinking about challenges, I often face people skeptical or derisive towards

meta-analysis, who ask me why I have taken so much at heart and continue to focus
attentively on a topic that many consider non-scientific at best. I have previously
recounted my personal and professional journey into meta-analysis (now totaling
more than 200 of them) [1–2]. Accordingly, I will be very brief: META-­ANALYSIS
IS FUN! Indeed, the best meta-analysis is a unique combination of clinical exper-
tise, methodological insight, and communication skills, leading together towards a
piece of original research that can guide researchers, clinicians, and patients, while
also satisfying your ego and supporting your career [3]. No other type of research
endeavor can timely combine these potent ingredients and eventually provide you
an edible product.
Despite their fun, meta-analyses are facing several challenges in recent years [4].
The first obvious challenge is that, being relatively easy to plan and perform, at least
in skilled hands, there are now several redundant if not copycat meta-analyses avail-
able [5]. In addition, in the information technology era, it is conceivable that meta-­
analyses as currently conducted will become obsolete and self-learning approaches
will automatically provide comparable results in a fraction of a second [6]. Third,
conversely, there is a booming literature on alternative methods for meta-analysis,
which creates uncertainty on which method is best, and may also undermine the
credibility of previously established methods [7]. As commonplace in research,
“prediction is very difficult, especially about the future” [8]. However, we expect
that some sort of expertise will always be required to conduct a meta-analysis, and,
even in the worst-case scenario, to interpret and apply one correctly.
And this preamble brings forward the need for the present book, aiming at boost-
ing methodological and operative knowledge in this field, with a specific focus on
diagnostic tests. Indeed, most meta-analyses published today represent systematic
reviews and pooled analyses at the study level of controlled clinical trials [1, 9, 10].
Other types of meta-analyses are available as well, but they are less prominent. One
specific type of meta-analysis is the one focusing on diagnosis rather than therapy,
i.e., a meta-analysis pooling diagnostic test accuracy studies [11]. While random-
ized trials of diagnostic strategies are also possible, they are quite uncommon, thus
leaving the evidence on diagnostic decision making mainly relying on studies of
diagnostic test accuracy [12].
As there is no book to date devoted to this interesting field of clinical and statisti-
cal research, we have aimed at filling such gap and provide clinicians and research-
ers with state-of-the-art guidance to conceive, design, conduct, report, and apply a
meta-analysis of diagnostic test accuracy studies. Readers should be aware though
that the Cochrane Library has put enormous efforts at improving the methodology
of this type of research synthesis, and draft chapters of their upcoming book on this
topic are available for free online [13]. Finally, diagnostic studies should best be
viewed in the continuum of clinical practice, from prevention, to treatment, and
rehabilitation. Accordingly, the best test of a diagnostic study remains, only appar-
ently paradoxically, a randomized trial (or a set of randomized trials summarized in
a dedicated meta-analysis) [14].
Preface xi

Despite these caveats, we hope people interested in evidence synthesis, clinical

practice, and statistical methodology will find reading this book fruitful and enjoy-
able as much as we found editing it.
Because, in all truth, meta-analysis is fun!

Latina, Italy Giuseppe Biondi-Zoccai

References
1. Biondi-Zoccai G, editor. Network meta-analysis: evidence synthesis with
mixed treatment comparison. Hauppauge, NY: Nova Science Publishers; 2014.
2. Biondi-Zoccai G, editor. Umbrella reviews. Evidence synthesis with overviews
of reviews and meta-epidemiologic studies. Cham, Switzerland: Springer
International; 2016.
3. Biondi-Zoccai G, Anderson LA. What is the purpose of launching World
Journal of Meta-Analysis? World J Meta-Anal. 2013;1:1–4.
4. Fava GA. Evidence-based medicine was bound to fail: a report to Alvan
Feinstein. J Clin Epidemiol. 2017;84:3–7.
5. Biondi-Zoccai GG, Lotrionte M, Abbate A, Testa L, Remigi E, Burzotta F,
Valgimigli M, Romagnoli E, Crea F, Agostoni P. Compliance with QUOROM
and quality of reporting of overlapping meta-analyses on the role of acetylcys-
teine in the prevention of contrast associated nephropathy: case study. BMJ.
2006;332:202–9.
6. Shekelle PG, Shetty K, Newberry S, Maglione M, Motala A. Machine learning
versus standard techniques for updating searches for systematic reviews: a
diagnostic accuracy study. Ann Intern Med. 2017;167:213–5.
7. Neupane B, Richer D, Bonner AJ, Kibret T, Beyene J. Network meta-analysis
using R: a review of currently available automated packages. PLoS One.
2014;9:e115065.
8. Niels Bohr. https://en.wikiquote.org/wiki/Niels_Bohr. Last accessed 27 June
2018.
9. Borenstein M, Hedges LV, Higgins J, Rothstein HR. Introduction to meta-­
analysis. Hoboken, NJ: Wiley; 2009.
10. Higgins JPT, Green S. Cochrane handbook for systematic reviews of interven-
tions. Hoboken, NJ: Wiley; 2008.
11. Gatsonis C, Paliwal P. Meta-analysis of diagnostic and screening test accuracy
evaluations: methodologic primer. AJR Am J Roentgenol. 2006;187:271–81.
12. Nudi F, Lotrionte M, Biasucci LM, Peruzzi M, Marullo AG, Frati G, Valenti V,
Giordano A, Biondi-Zoccai G. Comparative safety and effectiveness of coro-
nary computed tomography: systematic review and meta-analysis including 11
randomized controlled trials and 19,957 patients. Int J Cardiol. 2016;222:
352–8.
xii Preface

13. Cochrane methods—screening and diagnostic tests: handbook for DTA reviews.
http://methods.cochrane.org/sdt/handbook-dta-reviews. Last accessed 27 June
2018.
14. Biondi-Zoccai GG, Agostoni P, Abbate A. Parallel hierarchy of scientific
studies in cardiovascular medicine. Ital Heart J. 2003;4:819–20.
Contents

Part I
1 Introduction to Clinical Diagnosis����������������������������������������������������������    3
Giuseppe Biondi-Zoccai, Mariangela Peruzzi, Simona Mastrangeli,
and Giacomo Frati
2 The Evidence Hierarchy��������������������������������������������������������������������������   11
Mical Paul
3 Peculiarities of Diagnostic Test Accuracy Studies��������������������������������   19
Giuseppe Biondi-Zoccai, Simona Mastrangeli, Mariangela Peruzzi,
and Giacomo Frati
4 Meta-Analyses of Clinical Trials Versus Diagnostic
Test Accuracy Studies������������������������������������������������������������������������������   31
Michail Tsagris and Konstantinos C. Fragkos

Part II
5 Designing the Review������������������������������������������������������������������������������   43
José Mauro Madi, Machline Paim Paganella, Isnard Elman Litvin,
and Eliana Marcia Wendland
6 Registering the Review����������������������������������������������������������������������������   59
Alison Booth and Julie Jones-Diette
7 Searching for Diagnostic Test Accuracy Studies ����������������������������������   77
Su Golder and Julie Glanville
8 Abstracting Evidence������������������������������������������������������������������������������   93
Luca Testa and Mario Bollati
9 Appraising Evidence��������������������������������������������������������������������������������   99
Valentina Pecoraro
10 Synthesizing Evidence ���������������������������������������������������������������������������� 115
Paul-Christian Bürkner

xiii
xiv Contents

11 Appraising Heterogeneity������������������������������������������������������������������������ 125

Antonia Zapf
12 Statistical Packages for Diagnostic Meta-­Analysis
and Their Application������������������������������������������������������������������������������ 161
Philipp Doebler, Paul-Christian Bürkner, and Gerta Rücker
13 Network Meta-Analysis of Diagnostic Test Accuracy Studies ������������ 183
Gerta Rücker
14 Transition to Intervention Meta-Analysis���������������������������������������������� 199
Umberto Benedetto and Colin Ng
15 Updating Diagnostic Test Accuracy Systematic Reviews:
Which, When, and How Should They Be Updated? ���������������������������� 205
Ersilia Lucenteforte, Alessandra Bettiol, Salvatore De Masi, and
Gianni Virgili

Part III
16 Diagnostic Meta-Analysis: Case Study in Endocrinology�������������������� 231
Kosma Wolinski
17 Diagnostic Meta-Analysis: Case Study in Gastroenterology �������������� 249
Bashar J. Qumseya and Michael Wallace
18 Diagnostic Meta-Analysis: Case Study in Oncology���������������������������� 263
Sulbaran Marianny, Sousa Afonso, and Bustamante-Lopez Leonardo
19 Diagnostic Meta-Analysis: Case Study in Surgery ������������������������������ 285
Eliana Al Haddad, Hutan Ashrafian, and Thanos Athanasiou

Part IV
20 Avenues for Further Research���������������������������������������������������������������� 305
Yulun Liu and Yong Chen
21 Conclusion������������������������������������������������������������������������������������������������ 317
Giuseppe Biondi-Zoccai
Contributors

Sousa Afonso Surgical Division, Gastroenterology Department, Clinics Hospital,

University of Sao Paulo School of Medicine, Sao Paulo, Brazil
Eliana Al Haddad The Division of Cardiac Surgery, Department of Surgery,
Columbia University, New York, NY, USA
Hutan Ashrafian The Department of Surgery and Cancer, Imperial College
London, London, UK
Thanos Athanasiou The Department of Surgery and Cancer, Imperial College
London, London, UK
Department of Cardiac Surgery, Imperial College Healthcare NHS Trust,
London, UK
Umberto Benedetto Bristol Heart Institute, University of Bristol, Bristol, UK
Alessandra Bettiol Department of Neurosciences, Psychology, Drug Research
and Child Health (NEUROFARBA), University of Florence, Florence, Italy
Giuseppe Biondi-Zoccai Department of Medico-Surgical Sciences and
Biotechnologies, Sapienza University of Rome, Latina, Italy
Department of AngioCardioNeurology, IRCCS Neuromed, Pozzilli, Italy
Mario Bollati Department of Cardiology, IRCCS Pol. S. Donato, S. Donato
Milanese, Milan, Italy
Alison Booth Department of Health Sciences, University of York, York, UK
Paul-Christian Bürkner Institute of Psychology, Faculty of Psychology and
Sport Sciences, University of Münster, Münster, Germany
Yong Chen Department of Biostatistics, Epidemiology and Informatics, Perelman
School of Medicine, University of Pennsylvania, Philadelphia, PA, USA
Salvatore De Masi Clinical Trial Office, University Hospital “Azienda Ospedaliero-­
Universitaria Meyer”, Florence, Italy
Philipp Doebler Department of Statistics, TU Dortmund University, Dortmund,
Germany

xv
xvi Contributors

Konstantinos C. Fragkos University College London, London, UK

Giacomo Frati Department of Medico-Surgical Sciences and Biotechnologies,
Sapienza University of Rome, Latina, Italy
Department of AngioCardioNeurology, IRCCS Neuromed, Pozzilli, Italy
Julie Glanville York Health Economics Consortium, University of York, York, UK
Su Golder Department of Health Sciences, University of York, York, UK
Julie Jones-Diette Centre for Reviews and Dissemination, University of York,
York, UK
Bustamante-Lopez Leonardo Surgical Division, Gastroenterology Department,
Clinics Hospital, University of Sao Paulo School of Medicine, Sao Paulo, Brazil
Isnard Elman Litvin Faculdade de Medicina, Universidade de Caxias do Sul
(UCS), Caxias do Sul, Brazil
Yulun Liu Department of Biostatistics, Epidemiology and Informatics, Perelman
School of Medicine, University of Pennsylvania, Philadelphia, PA, USA
Ersilia Lucenteforte Department of Clinical and Experimental Medicine,
University of Pisa, Pisa, Italy
José Mauro Madi Faculdade de Medicina, Universidade de Caxias do Sul (UCS),
Caxias do Sul, Brazil
Sulbaran Marianny Gastrointestinal Endoscopy Service, Gastroenterology
Department, Clinics Hospital, University of Sao Paulo School of Medicine, Sao
Paulo, Brazil
Simona Mastrangeli Superior School of Advanced Studies, Sapienza University
of Rome, Rome, Italy
Colin Ng Singapore General Hospital, Singapore, Singapore
Machline Paim Paganella Laboratório de Pesquisa em HIV/AIDS, Universidade
de Caxias do Sul (UCS), Caxias do Sul, Brazil
Mical Paul Rambam Heath Care Campus and The Ruth and Bruce Rappaport
Faculty of Medicine, Technion—Israel Institute of Technology, Haifa, Israel
Valentina Pecoraro Laboratory of Toxicology, Department of Laboratory
Medicine and Pathological Anatomy, Azienda USL of Modena, Modena, Italy
Mariangela Peruzzi Department of Medico-Surgical Sciences and
Biotechnologies, Sapienza University of Rome, Latina, Italy
Bashar J. Qumseya Division of Gastroenterology and Hepatology, Archbold
Medical Group/Florida State University, Thomasville, GA, USA
Contributors xvii

Gerta Rücker Faculty of Medicine, Institute for Medical Biometry and Statistics,
Medical Center – University of Freiburg, Freiburg im Breisgau, Germany
Luca Testa Department of Cardiology, IRCCS Pol. S. Donato, S. Donato Milanese,
Milan, Italy
Michail Tsagris Department of Computer Science, University of Crete, Heraklion,
Greece
Gianni Virgili Department of Surgery and Translational Medicine (DCMT),
University of Florence, Florence, Italy
Michael Wallace Division of Gastroenterology and Hepatology, Mayo Clinic,
Jacksonville, FL, USA
Eliana Marcia Wendland Departamento de Saúde Coletiva, Universidade Federal
de Ciências da Saúde de Porto Alegre (UFCSPA), Porto Alegre, Brazil
Kosma Wolinski Department of Endocrinology, Metabolism and Internal
Medicine, Poznan University of Medical Sciences, Poznań, Poland
Antonia Zapf Department of Medical Statistics, University Medical Center
Göttingen, Göttingen, Germany
Department of Medical Biometry and Epidemiology, University Medical Center
Hamburg-Eppendorf, Hamburg, Germany
Part I
Introduction to Clinical Diagnosis
1
Giuseppe Biondi-Zoccai, Mariangela Peruzzi,
Simona Mastrangeli, and Giacomo Frati

Art is long, life is short

Hippocrates

Since its initial formalizing attempts, medical practice has focused on the precise
characterization of a patient’s ailments and their cure. Whereas cure may often
be beyond reach, no constructive management plan can be envisioned if the dis-
ease has not been identified and characterized in a formal and reproducible fash-
ion, through the process of clinical diagnosis. Diagnosis, which has been
formalized thousands of years ago in Ancient Greece, means indeed knowing
through [1]. It is based on the iteratively approximating process of identifying
the actual cause(s) of a complex maze of pathological signs and symptoms from
beginning to late consequences, pinpointing management and providing an
opportunity to improve health while minimizing the risk of adverse effects of
clinical intervention [2].
Diagnostic practice, thus, is based on comparative analysis and probability
concepts, in as much as a given diagnosis appears more likely than a competing
one, eventually toppling in the practitioner’s mind the less plausible alternatives,
albeit using the falsification framework formalized by Karl Popper [3]. However,
establishing a diagnosis requires the application of a rule or test, or set of them,
that definitely or likely identify a given condition. Absolute certainty is not always

G. Biondi-Zoccai (*) · G. Frati

Department of Medico-Surgical Sciences and Biotechnologies, Sapienza University of Rome,
Latina, Italy
Department of AngioCardioNeurology, IRCCS Neuromed, Pozzilli, Italy
e-mail: giuseppe.biondizoccai@uniroma1.it
M. Peruzzi
Department of Medico-Surgical Sciences and Biotechnologies, Sapienza University of Rome,
Latina, Italy
S. Mastrangeli
Superior School of Advanced Studies, Sapienza University of Rome, Rome, Italy

© Springer International Publishing AG, part of Springer Nature 2018 3

G. Biondi-Zoccai (ed.), Diagnostic Meta-Analysis,
https://doi.org/10.1007/978-3-319-78966-8_1
4 G. Biondi-Zoccai et al.

Fig. 1.1 Impact of

different diagnostic levels Symptoms Common cold
in clinical practice, with
suitable examples of Signs Herpes zoester
clinical scenarios

Accuracy
Clinical examination Arterial hypertension

Simplicity
Non-invasive test Stroke

Invasive test Breast cancer

Molecular test Fabry disease

Diagnostic Clinical
level example

required for clinical diagnosis, as a careful balance between efficiency and plau-
sibility is always sought (Fig. 1.1) [4, 5]. In the absence of a pathological or post-
mortem characterization, any diagnostic test requires comparative support from
another test (typically called reference standard or gold standard), thus defining
the peculiarity of studies appraising diagnostic tests [6, 7]. Once the features of
the index and reference test have been defined and data collected, dimensions of
comparative accuracy can be appraised quantitatively and their yield used to guide
decision-making.

1.1 Index Test

The index test is the diagnostic test which is of interest to the clinician, as a novel
diagnostic tool to identify a given condition. Several reasons may make it more
appealing than the reference test. For instance, it may be less expensive, less inva-
sive, and ultimately safer [8, 9]. Otherwise, it may be performed earlier than a refer-
ence test, and thus provide ampler means to prevent or treat the condition of interest
[10]. Finally, it may serve as an add-on to refine diagnostic clustering after a pre-
liminary test has been completed [11].
Typically, any diagnostic test, including thus the index test, provides a range of
results, which may have a varying range of strengths of association with the condi-
tion of interest. There may be key exceptions to this paradigm (e.g., genetic tests
aimed at identifying a specific mutation), but most tests used in clinical practice do
not provide per se a yes or no answer. Accordingly, a diagnostic threshold is applied,
such that test results lower than the threshold are considered negative and those
higher than the threshold positive (Fig. 1.2) [12]. This continuum of possible
1 Introduction to Clinical Diagnosis 5

Fig. 1.2 Graphical Threshold

representation of different
results of the index test in
non-diseased and diseased
subjects (according to a
dichotomous interpretation
of the reference test),
highlighting the choice of a Non-diseased Diseased
threshold which balances
sensitivity and specificity.
FN false negatives, FP
false positives, TN true TN TP
negatives, TP true positives

FN FP

Fig. 1.3 Impact of a lower Threshold

diagnostic threshold when
interpreting different
results of the index test in
non-diseased and diseased
subjects (according to a
dichotomous interpretation
of the reference test), Non-diseased Diseased
highlighting the increase in
sensitivity offset by a
decrease in specificity
(compared to the previous TN TP
figure). FN false negatives,
FP false positives, TN true
negatives, TP true positives

FN FP

diagnostic thresholds can be formally appraised in a comparative analysis with spe-

cific dimensions of diagnostic test accuracy [13]. Usually, there is indeed an obvi-
ous trade-off between a lower and a higher threshold. For instance, using a lower
threshold to define the presence of diabetes based on fasting blood glucose testing
will lead to more patients being considered diabetics (Fig. 1.3). This could result in
more subjects being offered promising preventive means but could also increase the
risk of side effects and inappropriate disability labeling. Conversely, using a higher
threshold when interpreting the results of fine-needle aspiration biopsy for a breast
lump might limit chemotherapy and radiotherapy to the few patients with large
malignant masses but could also prevent other subjects with borderline lesions to
receive timely and potentially life-saving treatment (Fig. 1.4).
6 G. Biondi-Zoccai et al.

Fig. 1.4 Impact of a Threshold

higher diagnostic threshold
when interpreting different
results of the index test in
non-diseased and diseased
subjects (according to a
dichotomous interpretation
of the reference test), Non-diseased Diseased
highlighting the increase in
specificity offset by a
decrease in sensitivity
(compared to the previous TN TP
two figures). FN false
negatives, FP false
positives, TN true
negatives, TP true positives
FN FP

1.2 Reference Test

Any diagnostic test which is considered clinically established can be used as a refer-
ence test for a comparative diagnostic test accuracy study. In most clinical and
research settings, a reference test is a testing procedure which is established and
reliable, despite its inherent limitations. Obviously, in most cases, what is currently
considered a reference test might well have been labeled in the past an index test
requiring external validation. Thus, the arguments proposed beforehand to index
tests may also apply to reference tests. Of course, in some cases the ultimate refer-
ence test is the self-evident clinical diagnosis itself, either considered clinically or
appraised pathologically, e.g., post-mortem. Paradoxically, limitations in the accu-
racy of pathologic and post-mortem diagnosis can also apply. In any case, waiting
for disease onset or even fatal disease is clinically inefficient and unethical, thus the
need for more timely diagnosis. Typical reference tests can be considered, for
instance, a diagnosis of myocardial infarction based on a complex set of symptoms,
permanent electrocardiographic abnormalities, and changes in serum levels of car-
diac biomarkers, or the pathognomonic features of brain magnetic resonance imag-
ing shortly after and long after an ischemic stroke.

1.3 Dimensions of Diagnostic Accuracy

The comparison of two diagnostic tests, such as the index and the reference tests,
can be considered as an appraisal of the consistency of the association between
two variables [14]. The simplest scenario is when both provide a yes and no
answer, leading to the creation of a two-by-two table with the accompanying mea-
sures of association (Tables 1.1 and 1.2). When one variable is continuous and the
other dichotomous, different thresholds can be considered, or a continuous
1 Introduction to Clinical Diagnosis 7

Table 1.1 Appraisal of the diagnostic accuracy of an index test in comparison to a reference test
(used to define disease status) using a 2×2 table
Diseased Non-diseased
Positive test TP FP
Negative test FN TN
Prevalence = (TP + FN)/(TP + FN + TN + FP)
Sensitivity = TP/(TP + FP)
Specificity = TN/(TN + FP)
Other dimensions of diagnostic accuracy can be appraised formally using similar data, such as
likelihood ratios, predictive values, and diagnostic odds ratios (see the Sect. 1.3)
FN false negatives, FP false positives, TN true negatives, TP true positives

Table 1.2 Example of the Diseased Non-­diseased

appraisal of the diagnostic Positive test 40 5
accuracy of an index test in
Negative test 10 45
comparison to a reference test
(used to define disease status) Prevalence = 50%
using a 2×2 table, assuming a Sensitivity = 80%
total of 100 patients tested, Specificity = 90%
with 50% diseased, Other dimensions of diagnostic accuracy can be appraised for-
sensitivity of 80%, and mally using similar data, such as likelihood ratios, predictive val-
specificity of 90% ues, and diagnostic odds ratios (see the Sect 1.3)
FN false negatives, FP false positives, TN true negatives, TP true
positives

threshold approach can be used. If both variables are continuous, then different
thresholds for both tests can be used. Otherwise, standard approaches to appraise
correlation, regression, and bias (e.g., the Bland-Altman method) can be employed
[15, 16].
The most common and clinically relevant scenario is however one in which both
tests provide a yes and no answer, leading to the definition of four groups of patients:
true negatives, false negatives, true positives, and false positives, with accompany-
ing results for prevalence, sensitivity, and specificity.

Conclusion
Diagnosis is based on testing, and thus the precise purpose and context of the test
must be borne in mind. Eventually, the test must be considered capable of
improving patient health or, at the very minimum, provide societal utility, [17,
18] thus requiring proof of benefit from a randomized controlled trial [19]. More
pragmatically, the applicability of the test, for screening, as an add-on, or a
replacement of an existing test, impacts on its comprehensive evaluation and
adoption [20].

Funding/Disclosure None.
8 G. Biondi-Zoccai et al.

References
1. Ackerknecht EH, Rosenberg CE, Hausshofer L. A short history of medicine. Baltimore, MD:
Johns Hopkins University Press; 2016.
2. Adeleye GG, Acquah-Dadzie K, Dadzie KA, Sienkewicz TJ, McDonough JT. World dictionary
of foreign expressions: a resource for readers and writers. Mundelein, IL: Bolchazy-Carducci;
1999.
3. Popper K. The logic of scientific discovery. London, UK: Routledge; 2002.
4. Montalescot G, Sechtem U, Achenbach S, Andreotti F, Arden C, Budaj A, Bugiardini R, Crea
F, Cuisset T, Di Mario C, Ferreira JR, Gersh BJ, Gitt AK, Hulot JS, Marx N, Opie LH, Pfisterer
M, Prescott E, Ruschitzka F, Sabaté M, Senior R, Taggart DP, van der Wall EE, Vrints CJ.
2013 ESC guidelines on the management of stable coronary artery disease: the task force on
the management of stable coronary artery disease of the European Society of Cardiology. Eur
Heart J. 2013;34:2949–3003.
5. Smith SC Jr, Allen J, Blair SN, Bonow RO, Brass LM, Fonarow GC, Grundy SM, Hiratzka L,
Jones D, Krumholz HM, Mosca L, Pearson T, Pfeffer MA, Taubert KA, AHA; ACC; National
Heart, Lung, and Blood Institute. AHA/ACC guidelines for secondary prevention for patients
with coronary and other atherosclerotic vascular disease: 2006 update endorsed by the National
Heart, Lung, and Blood Institute. J Am Coll Cardiol. 2006;47(10):2130–9.
6. Cochrane collaboration: handbook for diagnostic test accuracy reviews. http://methods.
cochrane.org/sdt/handbook-dta-reviews. Accessed 27 June 2018.
7. EUnetHTA guideline: meta-analysis of diagnostic test accuracy studies. Available at: http://
www.eunethta.eu/sites/default/files/sites/5026.fedimbo.belgium.be/files/Meta-analysis%20
of%20Diagnostic%20Test%20Accuracy%20Studies_Guideline_Final%20Nov%202014.pdf.
Accessed 27 June 2018.
8. Bossuyt PM, Irwig L, Craig J, Glasziou P. Comparative accuracy: assessing new tests against
existing diagnostic pathways. BMJ. 2006;332:1089–92.
9. Nudi F, Iskandrian AE, Schillaci O, Peruzzi M, Frati G, Biondi-Zoccai G. Diagnostic accuracy
of myocardial perfusion imaging with CZT technology: systemic review and meta-­analysis of
comparison with invasive coronary angiography. JACC Cardiovasc Imaging. 2017;10:787–94.
10. Pucci S, Bonanno E, Sesti F, Mazzarelli P, Mauriello A, Ricci F, Zoccai GB, Rulli F, Galatà
G, Spagnoli LG. Clusterin in stool: a new biomarker for colon cancer screening? Am J
Gastroenterol. 2009;104:2807–15.
11. D’Ascenzo F, Barbero U, Cerrato E, Lipinski MJ, Omedè P, Montefusco A, Taha S, Naganuma
T, Reith S, Voros S, Latib A, Gonzalo N, Quadri G, Colombo A, Biondi-Zoccai G, Escaned J,
Moretti C, Gaita F. Accuracy of intravascular ultrasound and optical coherence tomography
in identifying functionally significant coronary stenosis according to vessel diameter: a meta-­
analysis of 2,581 patients and 2,807 lesions. Am Heart J. 2015;169:663–73.
12. Deeks JJ, Macaskill P, Irwig L. The performance of tests of publication bias and other sample
size effects in systematic reviews of diagnostic test accuracy was assessed. J Clin Epidemiol.
2005;58:882–93.
13. Dinnes J, Deeks J, Kirby J, Roderick P. A methodological review of how heterogeneity has
been examined in systematic reviews of diagnostic test accuracy. Health Technol Assess.
2005;9:1–113.
14. Irwig L, Tosteson AN, Gatsonis C, Lau J, Colditz G, Chalmers TC, Mosteller F. Guidelines for
meta-analyses evaluating diagnostic tests. Ann Intern Med. 1994;120:667–76.
15. Garrone P, Biondi-Zoccai G, Salvetti I, Sina N, Sheiban I, Stella PR, Agostoni P. Quantitative
coronary angiography in the current era: principles and applications. J Interv Cardiol.
2009;22:527–36.
16. Novara M, D'Ascenzo F, Gonella A, Bollati M, Biondi-Zoccai G, Moretti C, Omedè P, Sciuto
F, Sheiban I, Gaita F. Changing of SYNTAX score performing fractional flow reserve in mul-
tivessel coronary artery disease. J Cardiovasc Med (Hagerstown). 2012;13:368–75.
1 Introduction to Clinical Diagnosis 9

17. Biondi-Zoccai G, editor. Network meta-analysis: evidence synthesis with mixed treatment
comparison. Hauppauge, NY: Nova Science Publishers; 2014.
18. Biondi-Zoccai G. Umbrella reviews. Evidence synthesis with overviews of reviews and meta-­
epidemiologic studies. Springer International: Cham, Switzerland; 2016.
19. Nudi F, Lotrionte M, Biasucci LM, Peruzzi M, Marullo AG, Frati G, Valenti V, Giordano A,
Biondi-Zoccai G. Comparative safety and effectiveness of coronary computed tomography:
systematic review and meta-analysis including 11 randomized controlled trials and 19,957
patients. Int J Cardiol. 2016;222:352–8.
20. Siontis KC, Siontis GC, Contopoulos-Ioannidis DG, Ioannidis JP. Diagnostic tests often fail to
lead to changes in patient outcomes. J Clin Epidemiol. 2014;67:612–21.
The Evidence Hierarchy
2
Mical Paul

Systematic reviews typically provide quantitative estimates of the pooled evidence.

A crucial part of the systematic review is to appraise and report the quality of the
evidence for the pooled estimates. This ranking then goes on to guideline panels and
recommendation statements to support the strength of recommendations for or
against an intervention or a test.
The classical evidence-based medicine hierarchy ranked evidence from system-
atic reviews of randomized controlled trials (RCTs) and randomized controlled tri-
als at the top to cohort studies, followed by lower degrees of evidence and ending
with expert opinion (Fig. 2.1) [1]. The underlying understanding was that unbiased
comparisons can be achieved only through adequate, unbiased randomization.
While this remains true, GRADE has taken a step further in appraising the evidence
[2–5]. GRADE starts with the study design and the classical risk of bias assessment
that appraises if and what internal biases exist in the studies and asks further:

1. Whether the evidence derived from different sources is consistent or not

2. Whether the evidence addresses directly the clinical question that was posed
3. The importance of the outcome assessed from the patient’s perspective
4. How precise and how large the effect estimate is, considering the totality of the
evidence
5. The existence of publication bias
6. Once again the risk of bias is addressed, specifically addressing plausible
confounding
7. Whether a dose-response gradient exists that strengths our belief in the plausibil-
ity of an effect

M. Paul
Rambam Heath Care Campus and The Ruth and Bruce Rappaport Faculty of Medicine,
Technion—Israel Institute of Technology, Haifa, Israel
e-mail: paulm@technion.ac.il

© Springer International Publishing AG, part of Springer Nature 2018 11

G. Biondi-Zoccai (ed.), Diagnostic Meta-Analysis,
https://doi.org/10.1007/978-3-319-78966-8_2
12 M. Paul

SR of RCT

Randomised controlled
trials

Observational cohort studies

Case reports, case series, expert opinion

Fig. 2.1 The classical evidence hierarchy for intervention studies

Thus, GRADE expands risk of bias assessment of individual studies to an

appraisal of the quality of the evidence for specific interventions, considering the
totality of evidence, its relevance to the clinical scenario, its importance to patients,
and its plausibility. Evidence is summarized in four grades from high to very low.
GRADE has been largely accepted into evidence-based recommendations.
GRADE can be directly applied to diagnostic tests or strategies that were tested
in RCTs. Such trials randomize patients to receive a new diagnostic test or strategy
vs. the standard or older diagnostics and examine the impact of the test choice on
patient-relevant outcomes. These constitute the real and only high-level evidence
for diagnostic tests. A systematic survey of RCTs examining diagnostic tests pub-
lished in MEDLINE up to December 2013 described a random sample of 103 such
RCTs (out of an assumed total of 781 RCTs) [6]. Trials were rare prior to 2000, and
since then their number seems to increase. Most trials (70%) assessed diagnostic
imaging, and the leading fields were cardiology and general surgery. The review
included only trials that examined at least one patient-important outcome. Of these,
41 (39.8%) reported on mortality, 63 (61.2%) on morbidities, and 14 (13.6%) on
symptoms, quality of life, or functional status. A significant difference in patient
outcomes was reported only in 12 (11.6%) RCTs. Similarly, diagnostic tests can be
assessed in observational studies examining the association between diagnostic
2 The Evidence Hierarchy 13

testing and patient outcomes. These can be evaluated by GRADE, receiving a priori
lower quality of evidence due to unavoidable selection bias.
However, the large majority of the evidence on diagnostic tests relies on diagnos-
tic studies examining test accuracy alone (sensitivity and specific, positive, and
negative predictive values). A GRADing scheme has been devised to place such
studies in the evidence hierarchy [7, 8]. The principle of this scheme is to proceed
from the evaluation of the quality of diagnostic test accuracy studies to inferences
regarding the effect of the test of patient-important outcomes, lacking clinical stud-
ies that have tested this. This is a multistage process. First, risk of bias of the diag-
nostic test accuracy study or studies in a systematic review is appraised using
available tools such as the QUADAS-2 score (described in Chaps. 9 and 11) [9].
Then, assumptions have to be made regarding the effects that the diagnostic accu-
racy will have on patient outcomes. These outcomes should be defined. The effect
on patient outcomes depends on the availability of treatment for the diagnosis and
patients’ prognosis with and without treatment. The consequnces of false positives
and false negatives to the patient have to be considered; these include superfluous or
missed treatment, further testing, and psychological consequences [10].
The same criteria used for GRADing interventions can be applied with slightly
different definitions for diagnostic studies (Table 2.1) [7]:

Table 2.1 Quality of the evidence grading in systematic reviews of diagnostic test accuracy
studies
N Factors GRADE classificationa
1 Study design Studies assessing consecutive representative patients and using an
appropriate reference standard should be ranked as no serious bias
2 Risk of bias Use the QUADAS-2 tool to perform quality assessment of the diagnostic
studies included in the systematic review. Summarize the QUADAS-s score
of studies included in the specific comparison to rank the overall risk of bias
3 Inconsistency Inconsistency in sensitivity, specificity, or likelihood ratios. Very frequent in
diagnostic systematic reviews
4 Indirectness Deduct from results the presumed influences on patient-important
outcomes, considering the implications of true and false positives and
negatives and the potential complications of the test. Accuracy studies
typically qualify as serious or very serious indirectness
5 Imprecision Wide confidence intervals for estimates of test accuracy or true and false
positive and negative rates results in serious to very serious risk. Consider
the 95% confidence intervals of the hierarchical summary curve. Large
confidence intervals typical in diagnostic systematic reviews
6 Other Consider the risk of publication bias
considerations
* Importance Consider the overall importance of the index test assessment, considering
its applicability and potential implications to patient management given the
assumed outcomes
a
GRADE classifies each factor as not serious, serious, and very serious risk of bias. From factors
1–6, a single GRADE score is generated of high, moderate, low, or very low quality of the evi-
dence. In addition, importance of the question on 1 (not important) to 9 (critical) scale. Guidance
on classification is available at https://gradepro.org/
14 M. Paul

1. Study design: When comparing outcomes for patients undergoing different diag-
nostic tests (interventional diagnostic studies), RCTs will be graded highest fol-
lowed by cohort studies. Diagnostic test accuracy studies are scored as no serious
risk of bias if assessing an index test on consecutive and representative patients
(i.e., those patients for whom the diagnostic test is relevant) and selecting a valid
reference standard.
2. Risk of bias: The QUADAS-2 scoring system assesses four domains referring to
the patient selection, index test, reference standard, and study flow [9]. A higher
score is given to studies assessing the test among consecutive patients for which
the test is intended (rather than case-control studies), when the index test and
reference standard are evaluated independently of each other, when the reference
standard adequately separates patients with and without the condition and does
not define the condition, and when all patients undergo the index test and refer-
ence standard. RCTs and cohort studies are assessed using different domains.
The most transparent strategy is that of The Cochrane Collaboration using the
individual domain approach and available through the open access Cochrane
Library.
3. Whether the evidence derived from different sources is consistent or not:
Heterogeneity is very frequent in diagnostic meta-analyses, relating to different
populations, tests, and reference standards. Even when assessing the same index
tests and reference standard, local differences in test performance and interpreta-
tion may cause heterogeneity. Heterogeneity that can be explained by varying
thresholds for test positivity is acceptable. Unfortunately, all too frequent thresh-
old effects do not explain all heterogeneity, and the quality of the evidence must
be downgraded for inconsistency.
4. Whether the evidence addresses directly the clinical question that was posed:
Evidence that relies only on test accuracy studies is scored low on directness
since the effects on patient-relevant outcomes were not measured but have to be
inferred. Similarly, the directness of the patient population studied and testing is
evaluated; was the relevant patient population included in the studies? Were the
index test and reference standard applied in the studies as they would be applied
in clinical practice? If competing index tests are evaluated, were they directly
compared to each other?
5. How precise and how large the effect estimate is, considering the totality of
the evidence: Similar to intervention effects, the confidence intervals sur-
rounding test accuracy estimates determine the quality of evidence related to
precision.
6. The existence of publication bias.

These six criteria are summarized to generate a GRADE score from high to
very low quality of evidence. In addition to the quality of the evidence, the impor-
tance of the specific question addressed is ranked on a nine-point scale from criti-
cal to not important (Table 2.1) considering its impact from the patient’s
2 The Evidence Hierarchy 15

perspective. In clinical studies evaluating diagnostic strategy effects, the impor-

tance of the outcomes assessed in the studies can be appraised. In test accuracy
studies, the overall importance of the index test assessment should be assessed,
considering the need for the test, its applicability, and potential implications to
patient management.
Clearly, the large majority of the evidence on diagnostic tests is ranked as low-­
quality evidence. When only studies evaluating sensitivity and specificity are avail-
able, evidence will be typically downgraded for heterogeneity, indirectness, and
imprecision. The importance of ranking test accuracy studies as low-quality evi-
dence in systematic reviews was demonstrated in a review of RCTs assessing the
impact of different diagnostic tests on important patient outcomes, where only a
small minority (12/103, 11%) of trials showed a significant difference in one or
more of the outcomes [6]. Thus, excellent test performance should not be automati-
cally assumed to change patients’ outcomes.
GRADE has simplified the reporting of bias in systematic reviews, classify-
ing the evidence on interventions or diagnostic tests into four simple categories:
high, moderate, low, and very low quality of evidence. An open-access software
(https://gradepro.org/) allows to perform evidence GRADing and provides guid-
ance and easy creation of evidence tables. Many regulatory bodies, including
the World Health Organization, have endorsed this strategy to summarize evi-
dence for guidelines [11, 12]. GRADE simplifies reading through systematic
reviews and guidelines. However, its disadvantage is common to risk of bias
scores, where the final grade summarizes different domains that are not trans-
parent. Thus, “low” quality evidence might derive from RCTs with risk of bias
concerns (e.g., open trials), while “high” quality evidence can be derived from
cohort studies that were considered to be well-conducted showing a large asso-
ciation and a dose-response gradient. The GRADE system has not undergone
strict validation and the inter-reviewer agreement in scoring was not strong [12,
13]. Thus, systematic reviewers are encouraged to provide explicitly data on the
study designs that have contributed to the final GRADE, their internal risk of
bias, and the criteria that led to downgrading or upgrading the level of evidence.
Readers are encouraged to look into the risk of bias assessment in systematic
reviews further than the final GRADE, to better understand the quality of the
evidence.
Evidence-based medicine places the patient at the center. At the top of the evi-
dence hierarchy is research that is devoid of internal bias and assesses outcomes
relevant to patients. The current understanding with regard to the evidence hierarchy
is plotted in Fig. 2.2. Diagnostic test accuracy studies are rarely ranked as high-­
quality evidence, since the impact of patients’ outcomes can only be assumed.
Systematic reviews of diagnostic test accuracy most commonly will not result in
higher quality evidence than the original studies because heterogeneity will pre-
clude firm conclusions. Diagnostic systematic review has yet to find its place in the
evidence hierarchy for diagnostic tests.
Another random document with
no related content on Scribd:
learnt their lessons by using the memories he had given them,
because he had given them minds by which they understood them.
They loved their parents, and relations, and companions, because
their Friend had given them affections.”
“It seems to me,” interrupted Charley, “that Friend gave them
everything. It must be God, mother, for I know he gives us everything
we have.”
“Yes, my dear Charley; and I am sorry to say these two children
neglected their Friend. They had often been told by their mother
never to get into bed without first kneeling and thanking him for all
his gifts; but they did not think of him. They used and enjoyed his
gifts, but they sometimes forgot the Giver.”
Ellen laid her head on her mother’s bosom,—
“Mother,” she said, “you mean us.”
“My dear Ellen,” replied her mother, “your conscience is like the
ring in the fairy tale. Yes, I did mean you and Charles. I was sorry,
when I came into the room to-night, to see you getting into bed
without saying your prayers. God has given you a voice to speak, my
children. Your dog, Dash, Charles, cannot speak to thank God for
anything he receives; but you can.”
“And I will!” exclaimed the good little boy, ashamed that he had
been ungrateful and thoughtless. “Come, Ellen, we will jump up and
say our prayers; and,” he added in a whisper, “we’ll speak for Dash
too.”
Cromwell at Croyden palace.
 Oliver Cromwell.

This individual was one of the most wonderful men that ever
lived. He was born at Huntingdon, in England, April 28, 1599. It is
related of him, that, when an infant, a large ape seized him, and ran
with him up to the top of a barn; there the creature held him, and
refused, for a long time, to give him up, frightening the people with
the idea that he should let him fall. It is said that, while he was still
young, a gigantic female figure appeared at his bedside, and foretold
his future greatness.
Cromwell was well educated; but, after quitting the university, he
became very dissipated. At twenty-one, he married Elizabeth
Bouchire, from which time he became regular in his life.
In 1625 he was chosen to parliament; and thus began, at twenty-
six years of age, that public career which ended in his becoming the
sole ruler of England, and one of the most energetic and powerful
sovereigns of Europe. He was soon distinguished as a speaker in
parliament, always taking part against the court and the established
church. In 1642, when civil war was about to commence, he raised a
troop of horse, and seizing the plate of the university of Cambridge,
appropriated it to the paying of the expenses of the army. He was
engaged in several battles, where he displayed the utmost skill and
courage. In 1645, the famous battle of Naseby was won by his valor
and good management; and, in consideration of his services,
parliament voted him the annual sum of £25,000 during his life.
King Charles I., against whom Cromwell and his party were
acting, was betrayed into their hands by the Scotch. By the intrigues
of Cromwell, he was tried, condemned, and beheaded. Cromwell
himself became, soon after, the ruler of the kingdom, under the title
of Lord Protector of the Commonwealth of England, Scotland and
Ireland. Though he had obtained his power by a series of violent
acts, and by the practice of every species of hypocrisy, Cromwell
now set himself about promoting the strength, power, and prosperity
of his kingdom. Though this was done harshly, yet it was with
wisdom and energy. The country flourished at home, and the name
of England was much respected abroad.
But though Cromwell had risen to the utmost height of honor and
power, he was a miserable man. He was perpetually haunted with
superstitious fears, the promptings of a conscience ill at ease. The
death of the king, which was effected by his management, weighed
upon his spirit like a murder. He went constantly armed, and yet he
was constantly in fear. At last, when Col. Titus wrote a book, entitled,
Killing no Murder, in which he attempted to prove that it was a duty
of the citizens to kill Cromwell, he was thrown into a fever, and died,
Sept. 3, 1658, leaving his weak brother, Richard, to wield the sceptre
for a few years, and then surrender it to a son of the murdered
Charles I. Cromwell was buried in Westminster Abbey; but, after
Charles II. came to the throne, his body was dug up and hung on a
gibbet, beneath which it was buried!
 Musings.
I wandered out one summer night—
’Twas when my years were few:
The breeze was singing in the light,
And I was singing too.
The moonbeams lay upon the hill,
The shadows in the vale,
And here and there a leaping rill
Was laughing at the gale.

One fleecy cloud upon the air

Was all that met my eyes;
It floated like an angel there,
Between me and the skies.
I clapped my hands and warbled wild
As here and there I flew;
For I was but a careless child,
And did as children do.

The waves came dancing o’er the sea

In bright and glittering bands:
Like little children wild with glee,
They linked their dimpled hands.
They linked their hands—but ere I caught
Their mingled drops of dew,
They kissed my feet, and, quick as thought,
Away the ripples flew.

The twilight hours like birds flew by,

As lightly and as free;
Ten thousand stars were in the sky,
 Ten thousand in the sea;
For every wave with dimpled cheek
That leaped upon the air,
Had caught a star in its embrace,
And held it trembling there.

The young moon too, with upturned sides,

Her mirrored beauty gave;
And as a bark at anchor rides,
She rode upon the wave.
The sea was like the heaven above,
As perfect and as whole,
Save that it seemed to thrill with love,
As thrills the immortal soul.

The leaves, by spirit-voices stirred,

Made murmurs on the air—
Low murmurs, that my spirit heard,
And answered with a prayer:
For ’twas upon the dewy sod,
Beside the moaning seas,
I learned at first to worship God,
And sing such strains as these.

The flowers, all folded to their dreams,

Were bowed in slumber free,
By breezy hills and murmuring streams,
Where’er they chanced to be.
No guilty tears had they to weep,
No sins to be forgiven;
They closed their eyes, and went to sleep,
Right in the face of heaven.

No costly raiment round them shone,

No jewels from the seas,
Yet Solomon upon his throne
Was ne’er arrayed like these:
And just as free from guilt and art
 Were lovely human flowers,
Ere sorrow set her bleeding heart
On this fair world of ours.

I heard the laughing wind behind,

A playing with my hair—
The breezy fingers of the wind,
How cool and moist they were!
I heard the night bird warbling o’er
Its soft, enchanting strain—
I never heard such sounds before,
And never shall again.

Then wherefore weave such strains as these,

And sing them day by day,
When every bird upon the breeze
Can sing a sweeter lay?
I’d give the world for their sweet art,
The simple, the divine;
I’d give the world to melt one heart,
As they have melted mine.
Sou. Lit. Mess.

Anecdote of an Atheist.—An atheist on his death-bed was

addressed by his son,—“Father, the physician says you can live but
a few hours.” “I know it, my son. Have you anything to say to me?”
“My father, you and my mother have held different creeds; my
mother is a Christian—you believe there is no God. Shall I follow her
faith or yours?” “My son,” said the dying parent, “believe in the God
of your mother.”
Thus it is in the hour of sickness, at the moment when the frail
supports of pride and passion are wrecked, that the sinking atheist
clutches at the plank of the Christian. Thus it is that the atheist, when
he is brought upon the stand before his Maker, confesses that his
creed is not one that he would wish to bequeath to his children.
 Who made this?
Here is a picture of the bones or skeleton of a horse. What a
wonderful piece of mechanism it is! How many bones and joints, and
how they are all fitted to each other!
Now, every horse has such a skeleton or frame-work of bones:
and who contrives and makes them? Can men make such curious
machinery? Certainly not. Men may make steamboats, and ships,
and cotton-factories, but they cannot make the bones of an animal;
nor can they put muscles and life to these bones. Now, if man cannot
do these things, who can? God only: he only can do these wonderful
things.

Wisdom of the Creator.—The happy proportioning of one thing

to another shows the wisdom of the Creator. Man, for instance, is
adapted to the size and strength of a horse. If men were giants, they
could not ride horses. If men were either pigmies or giants, they
could not milk cows, mow grass, reap corn, train vines, or shear
sheep, with anything like the conveniency they do now. If men were
pigmies, they would be lost in the grass and rushes, and their
children would be carried off by birds of prey. Every one can see,
that, other things being as they are, man would suffer by being either
much larger or smaller than he is.

Yankee Energy.—A few days since, a gentleman of the city of

New York was standing near the canal, at Albany, when he saw a
small yawl-boat approaching him, propelled by a lad about
seventeen years of age. The boat contained also the boy’s mother,
six sisters, and a small brother. Our friend asked him where he was
from, and where bound, and was answered, in substance, as
follows:
“We are from Ohio. My father died there, and as we were nearly
destitute, mother thought we had better go back to Saybrook, Conn.,
where we used to live; so we raised money enough to get this boat,
and started from Ohio last fall. We came through Lake Erie, and got
into the canal, where we were stopped by the ice. During the winter
we hauled our boat up by the side of the canal, where we remained
till the ice broke up. Sometimes we were considerably cold, and at
times were sick a little, but on the whole we all got along right smart.
We shall go down the North river, and up the sound to Saybrook.”
During this conversation, our friend was walking along the margin
of the canal; our noble Yankee boy, being unwilling to lose any time,
kept constantly propelling his boat forward, the younger brother, a
lad of only seven or eight years of age, steering the craft. It was
Sunday morning, and the mother and daughters were clad in their
Sabbath suits, and engaged in reading. A small furnace was
standing on the deck of the boat, and a sail, snugly stowed, was
lying fore and aft. The few cooking utensils, bedding, and clothing
belonging to this poor family, were securely placed under the deck.
Here is an instance of industry and perseverance, which
commends itself to the notice of the rising generation—ay, and the
present one too. No doubt, if this boy lives, he will yet make a stir in
the world; and if we knew his name, we would publish it.

Who Made Man?—Look at the foot—how ingeniously is this

contrived! Look at the arm: what piece of mechanism can compare
with it? But of all parts of the body, the eye is perhaps the most
wonderful. It has in it a lens, like that of a telescope, through which
the rays of light pass; and at the back of the eye a little picture of
whatever comes before the eye is formed. This picture falls upon a
nerve which lines the interior of the eye, and thus it is we see. All this
contrivance is very ingenious. And observe how the eye itself is
placed in the head. See how easily it turns this way and that!
Consider these things, and tell me, who but a Superior Being, one
who contrives, one who thinks, could have made man?

Power of God.—The sun is as large as three hundred and

thirty-seven thousand of our worlds. Jupiter is as large as one
thousand two hundred and eighty-one of our worlds. Mercury flies
along in its path at the rate of twenty miles in a second. Uranus is
seventeen times as large as our world, one billion eight hundred
milions of miles from the sun, and flies along at the rate of two
hundred and forty miles every minute!
Here, then, is the power of God! A world, with all its mountains,
and oceans, and kingdoms, is but a pebble in the hands of the
Almighty!
 THE BIRD’S ADIEU!
the words and music composed for
merry’s museum.

Farewell to the meadow,

For summer is past;
Farewell, for its leaves
Now whirl o’er the blast.
Farewell to the bough
Where my cradle was swung,
And the song of my mother
Was joyously sung.
2
How sweet was that song
Of the light-hearted bird;
No other I’ll sing—
’Twas the first that I heard.
And though to far lands
I must hasten away,
Wherever I roam,
I will carry that lay.
3
How sweet are these scenes,
For my birth-place is here;
And I know that in absence,
They’ll be but more dear.
I’ll sing of them there,
In the land where I roam,
And, winter departed,
I’ll return to my home.
MERRY’S MUSEUM.
VOLUME II.—No. 4.
 Merry’s Life and Adventures.

CHAPTER XII.
Raymond’s story of the School of Misfortune—concluded.

“It was several hours after his arrival at the city before R. had fully
recovered his senses. When he was completely restored, and began
to make inquiries, he found that all his ship companions had
perished. He, who probably cared least for life—he, who had no
family, no friends, and who was weary of existence—he only, of all
that ship’s company, was the one that survived the tempest!
“There was something in this so remarkable, that it occupied his
mind, and caused deep emotions. In the midst of many painful
reflections, he could not, however, disguise the fact, that he felt a
great degree of pleasure in his delivery from so fearful a death.
Again and again he said to himself, ‘How happy, how thankful I feel,
at being saved, when so many have been borne down to a watery
grave!’ The loss of his property, though it left him a beggar in the
world, did not seem to oppress him: the joy of escape from death
was to him a source of lively satisfaction; it gave birth to a new
feeling—a sense of dependence on God, and a lively exercise of
gratitude towards him. It also established in his mind a fact before
entirely unknown, or unremarked—that what is called misfortune, is
often the source of some of our most exquisite enjoyments. ‘It seems
to me,’ said R., in the course of his reflections, ‘that, as gems are
found in the dreary sands, and gold among the rugged rocks, and as
the one are only yielded to toil, and the other to the smelting of the
fiery furnace,—so happiness is the product of danger, suffering, and
trial. I have felt more real peace, more positive enjoyment from my
deliverance, than I was able to find in the whole circle of voluptuous
pleasures yielded by wealth and fashion. I became a wretch,
existence was to me a burthen, while I was rich. But, having lost my
fortune, and experienced the fear of death, I am happy in the bare
possession of that existence which I spurned before.’
“Such were the feelings and reflections of R. for a few days after
his escape; but at length it was necessary for him to decide upon
some course of action. He was absolutely penniless. Everything had
been sunk with the ship. He had no letters of introduction, he had no
acquaintances in New York; nor, indeed, did he know any one in all
America, save that a brother of his was a clergyman in some part of
the United States; but a coldness had existed between them, and he
had not heard of him for several years. R. was conscious, too, that
this coldness was the result of his own ungenerous conduct; for the
whole of his father’s estate had been given to him, to the exclusion
of his brother, and he had permitted him to work his own way in life,
without offering him the least assistance. To apply to this brother
was, therefore, forbidden by his pride; and, beside, he had every
reason to suppose that brother to be poor.
“What, then, was to be done? Should he return to England? How
was he to get the money to pay his passage? Beside, what was he
to do when he got there? Go back to the village where he carried his
head so high, and look in the faces of his former dashing
acquaintances—acknowledging himself a beggar! This was not to be
thought of. Should he seek some employment in America? This
seemed the only plan. He began to make inquiries as to what he
could find to do. One proposed to him to keep school; another, to go
into a counting-room; another, to be a bar-keeper of a hotel. Any of
these occupations would have given him the means of living; but R.’s
pride was in the way;—pride, that dogs us all our life, and stops up
almost every path we ought to follow, persuaded R. that he, who was
once a gentleman, ought to live the life of a gentleman; and of
course he could not do either of the things proposed.
“But events, day by day, pressed R. to a decision. His landlord, at
last, became uneasy, and told him that for what had accrued, he was
welcome, in consideration of his misfortunes; but he was himself
poor, and he begged him respectfully to make the speediest possible
arrangements to give up his room, which he wanted for another
boarder. ‘I have been thinking,’ said R. in reply to this, ‘that I might
engage in the practice of physic. In early life I was thought to have a
turn for the profession.’ This suggestion was approved by the
landlord, and means were immediately taken to put it in execution.
Dr. R., late of England, was forthwith announced; and in a few weeks
he was in the full tide of successful experiment.
“This fair weather, however, did not continue without clouds. Many
persons regarded Dr. R. only as one of the adventurers so frequently
coming from England to repay the kindness and courtesy of the
Yankees with imposition and villany. Various inquiries and stories
were got up about him; some having a sprinkling of truth in them,
and, for that reason, being very annoying. R., however, kept on his
way, paying little heed to these rumors, fancying that, if left to
themselves, they would soon die. And such would, perhaps, have
been the result, had not a most unfortunate occurrence given
matters another turn.
“In the house where R. boarded, several small sums of money,
and certain ornaments of some value, were missed by the boarders,
from time to time. Suspicions fell upon a French servant in the
family; but as nothing could be proved against him, he was retained,
and a vigilant watch kept over his actions. Discovering that he was
suspected, this fellow determined to turn the suspicion against R.;
he, therefore, in the dead of night, took a valuable watch from one of
the rooms, and laid it under the pillow of R.’s bed. This was done
with such address, that neither the gentleman from whom the watch
was stolen, nor R. himself, saw anything of it at the time. The watch
was missed in the morning, and the French servant was arrested.
But as soon as the chambermaid began to make up R.’s bed,
behold, the pilfered watch was there! The French servant was at
once released, and R. was arrested, briefly examined, and thrown
into prison.
“The circumstances in which he had come to the country now all
made against him. The unfavorable rumors that had been afloat
respecting him were revived; all the stories of swindlers that had
visited the country for twenty years back, were published anew, with
embellishments. In short, R. was tried and condemned by the public,
while he lay defenceless in prison, and long before his real trial came
on. The subject became a matter of some notoriety; the
circumstances were detailed in the newspapers. A paragraph
noticing these events met the eye of R.’s brother, who was settled as
a minister of the gospel in a country parish not far distant, and he
immediately came to the city. Satisfying himself by a few inquiries
that it was indeed his brother who was involved in difficulty and
danger, he went straight to the prison, with a heart overflowing with
sympathy and kindness. But pride was still in the way, and R.
haughtily repulsed him.
“The pious minister was deeply grieved; but he did not the less
seek to serve his brother. He took care to investigate the facts, and
became persuaded that the French servant had practised the
deception that has been stated; but he was not able to prove it. He
employed the best of counsel; but, in spite of all his efforts, and all
his sympathy, R. was found guilty, condemned, and consigned to
prison.
“Up to this time, the pride of R. had sustained him; but it now gave
way. He had borne the loss of fortune, but to be convicted of a low,
base theft, was what his spirit could not endure. His health sunk
under it, and his reason, for a time, departed. His sufferings during
that dark hour, God only knows. He at last recovered his health and
his senses, and then he heard, that, on his death-bed, the French
servant had confessed his iniquity. It was from the lips of his brother,
and under his roof, where he had been removed during his insanity,
that R. learnt these events. He was released from prison, and his
character was cleared of the imputation of crime.
“From this period R. was an altered man. His pride was effectually
quelled; no longer did that disturber of earth’s happiness,—the real
serpent of Eden,—remain to keep him in a state of alienation from
his brother. The two were now, indeed, as brothers. But there were
other changes in R.; his health was feeble, his constitution was
broken; his manly beauty had departed, and he was but the wreck of
former days. But, strange as it may seem, he now, for the first time,
found peace and happiness. He had now tasted of sorrow, and was
acquainted with grief. This enabled him to enter into the hearts of
other men, to see their sorrows, and to desire to alleviate them. A
new world was now open to him; a world of effort, of usefulness, of
happiness. In the days of prosperity, he had no cares for anybody
but himself; and mere selfishness had left him a wretch while in
possession of all the supposed means of bliss. He had now made
the discovery,—more important to any human being than that of
Columbus,—that pride is the curse of the human race, and humility
its only cure; that trial, sorrow, and misfortune are necessary, in most
cases, to make us acquainted with our own hearts, and those of our
fellow-men; and that true bliss is to be found only in a plan of life
which seeks, earnestly and sincerely, the peace and happiness of
others.”
Here ended R.’s story of the School of Misfortune; and I had no
difficulty in discovering that he had been telling the story of his own
life, though he had, in some respects, as I had reason to suppose,
departed from its details.
(To be continued.)