Psychological Medicine.

Monograph Supplement
http://journals.cambridge.org/PMS

Additional services for Psychological Medicine. Monograph Supplement:

Email alerts: Click here

Subscriptions: Click here
Commercial reprints: Click here
Terms of use : Click here

Schizophrenia: manifestations, incidence and course in

different cultures A World Health Organization Ten-Country
Study

A. Jablensky, N. Sartorius, G. Ernberg, M. Anker, A. Korten, J. E. Cooper, R. Day and A. Bertelsen

Psychological Medicine. Monograph Supplement / Supplement 20 / January 1992, pp 1 - 97

DOI: 10.1017/S0264180100000904, Published online: 09 July 2009

Link to this article: http://journals.cambridge.org/abstract_S0264180100000904

How to cite this article:

A. Jablensky, N. Sartorius, G. Ernberg, M. Anker, A. Korten, J. E. Cooper, R. Day and A. Bertelsen (1992).
Schizophrenia: manifestations, incidence and course in different cultures A World Health Organization
Ten-Country Study. Psychological Medicine. Monograph Supplement, 20, pp 1-97 doi:10.1017/
S0264180100000904

Request Permissions : Click here

Downloaded from http://journals.cambridge.org/PMS, IP address: 128.82.252.58 on 02 Jul 2014

SYNOPSIS In the 1960s the World Health Organization (WHO) carried out an Internationa
Study of Schizophrenia (IPSS), a transcultural investigation of more than 1200 patients ii
countries: China, Colombia, Czechoslovakia, Denmark, India, Nigeria, United Kingdom,
and USSR. The study established that large-scale international psychiatric studies are feasi
produced standardized instruments allowing reliable assessment of people suffering from r
disorders in different cultures and created a network of centres able and willing to work to
in psychiatric research.
The study established that schizophrenic disorders exist in different parts of the world an
the features of their clinical conditions were marked by their similarity rather than differeno
follow-up of the patients over a period of five years demonstrated that a significant proport
people in whom a schizophrenic syndrome had been diagnosed recovered well and th
percentage of patients with a good outcome was higher in developing than in developed cou
This finding needed confirmation because of its potential significance for public health.
A few years later WHO therefore launched a second study - this time aiming to produce est
of incidence of schizophrenia in different cultures and definitive evidence about the coun
outcome of schizophrenia in different parts of the world. Twelve centres in ten coi
participated in this work (Aarhus (Denmark), Agra and Chandigarh (India), Cali (Colo
Dublin (Ireland), Honolulu and Rochester (United States of America), Ibadan (Nigeria), M
(USSR), Nagasaki (Japan), Nottingham (United Kingdom), and Prague (Czechoslovakia)),
them (Aarhus, Agra, Cali, Ibadan, Moscow and Prague) had also taken part in the IPSS. The
was supported - as was the case with IPSS - by WHO, the United States National Instit
Mental Health and the research centres.
In each of the centres arrangements were made to identify and assess over a period of twe
all individuals who met four inclusion criteria: i.e. (i) were aged between 15 and 54 years; (
resided for at least six months in the area; (iii) had shown at least one overt syr
(hallucinations, delusions, qualitative thought or speech disorders or gross behaviour;
normalities) or at least two abnormalities suggestive of psychotic disorder (e.g. episodic
excitement, significant social withdrawal, overwhelming fear); and (iv) had recently made a f
lifetime contact with a helping agency. Clinical evidence of gross organic cerebral disor
previous contacts with psychiatric or other agencies because of mental disorders similar
current one excluded the patient from the study. Over a period of two years the centres a
standardized screening and assessment procedures according to a commonly accepted proto
seven centres it was possible to follow the protocol to the letter: Aarhus, Chandigarh, E
Honolulu, Moscow, Nagasaki and Nottingham. In the five other centres certain modificatio
to be introduced. This made it possible to obtain incidence figures in 7 of the sites and co:
information about clinical characteristics of people suffering from schizophrenic syndromes fi
the 12 sites.
Patients were examined using a set of standardized instruments selected among the av
research tools or developed especially for this study. The reliability of psychiatric assessme
tested and maintained by conducting joint interviews and by rating taped interviews. Equ
language versions of the instruments were produced for each of the centres. They dealt w
mental state, psychiatric history, social characteristics and other variables relevant for t
description of the patients.
In all, 1535 patients passed the initial screening. From that group some patients had
excluded either because they had had previous contacts with psychiatric services or for dia;
reasons. This resulted in a cohort of 1379 patients who were examined in detail and followed i
a period of two years.
The main findings of this study can be summarized as follows.
1. The study demonstrated that follow-up examinations and comparative longitudinal n
in general are feasible in different types of countries. Nearly 80% of all patients could be fo
up and assessed two years after the initial examination. The study confirmed that larg
international collaborative studies of psychiatric disorders are feasible and that psych
2 A. Jablensky and others

belonging to different cultural backgrounds can be trained to use standardized research instruments
in a reliable way.
2. The reasons for patients making these lifetime first contacts were similar in developed and
developing countries. Of all included cases, nearly 80 % were considered to constitute a broad group
of schizophrenic and related disorders. More than half of the study population showed a syndrome
of nuclear schizophrenia defined by the presence of specific symptoms (Category S+ of the
CATEGO program derived from the PSE-9 mental state examination).
3. The annual incidence of new cases of' broadly' defined schizophrenia was in the range between
1-5 and 4-2 (both sexes) per 100000 population at risk (age 15-54). The incidence of schizophrenia
defined by CATEGO class S+ was in the range between 0-7 and 1-4 per 100000. The incidence of
'broadly' defined schizophrenia was highest in India (both the rural and the urban area of
(Chandigarh). The differences between the incidence rates for the 'broad' diagnostic category of
schizophrenia in the different centres were significant. However, in every centre, the incidence rates
tended to decrease as more specific definitions of' caseness' for schizophrenia were applied and at
the level of CATEGO S +, there were no significant differences between the study areas.
4. In all the study areas, the age- and sex-specific curves of the incidence of schizophrenia
followed a similar pattern. It was demonstrated that in developed and in developing countries alike,
the onset of schizophrenia tended to occur at a later age in females as compared to males. The
similarity of age- and sex-related patterns of onset of schizophrenia across the study areas supports
the notion that under the diagnostic category of schizophrenia the same disorder has been identified
and investigated in the different cultural settings of the study.
5. The majority of the patients in the study had a remitting pattern of course over the two years
of follow-up: 50-3 % had a single psychotic episode and a further 31-1 % had two or more psychotic
episodes followed by remissions. Only 15-7% of the patients had an unremitting, continuous
psychotic illness. The more favourable outcome and remitting patterns were significantly more
common amongst the patients in the developing countries.
6. Type of onset (i.e. acute, subacute, and gradual) and setting (developing country or developed
country) were the most important predictors of several dimensions of the two-year course and
outcome. The difference in outcome between developing and developed countries remained
significant when patients with acute onset of disease and those with insidious onset were studied
separately. Other significant predictors were: clinical diagnosis on initial examination, marital
status, sex, adjustment in adolescence, frequency of contact with friends and history of the use of
'street' drugs.
7. Interpretation of the findings of this study are restricted by the comparatively small proportion
of the variation in both incidence rates and measures of outcome that can be explained by the
descriptive clinical differences among the patients. Much more social and family information was
collected than in the IPSS, but even so, no fundamentally different findings emerged. It is clear that
'culture' is confirmed as an important determinant of outcome, but exactly how the influences
implied by this general concept impinge upon the behaviour and experiences of individuals and
families requires further study. In the absence of reliable biological markers, schizophrenia can only
be described by clinical characteristics; this means that further epidemiological and cross-cultural
investigations, using an interdisciplinary approach to the study of smaller sub-groups and specific
clinical issues, are still likely to form a large part of the most productive strategy for research in the
near future.

Address for correspondence: Professor Norman Sartorius, Director, Division of Mental Health, WHO, 1211 Geneva 27, Switzerland.
 Introduction
In the late 1960s the World Health Organization occur in the different cultures represented in the
initiated the International Pilot Study of Schizo- study. A striking finding which emerged from
phrenia (IPSS), a transcultural psychiatric the follow-up stage of the project was the
investigation of 1202 patients in nine countries contrast between the initial symptomatological
- China, Colombia, Czechoslovakia, Denmark, similarity of the patients diagnosed as schizo-
India, Nigeria, USSR, United Kingdom and phrenics, both within and across the centres, and
USA. The study was coordinated by WHO and the marked variation of the forms of course and
funded jointly by WHO, the National Institute outcome of the disorders over the subsequent
of Mental Health (NIMH) of the United five years. A combined index of 'overall out-
States, and the field research centres. It was come', based on the pattern of course, the total
designed to lay scientific groundwork for future duration of the psychotic episodes, the quality
epidemiological studies of schizophrenia and of the remission (if any), and the degree of social
other psychiatric disorders. Some of the aims of impairment, demonstrated that, as a group, the
the IPSS were concerned with methodology: to patients in the three developing countries -
establish the feasibility of large-scale inter- Nigeria, India and Colombia - had a signifi-
national psychiatric studies, to develop stan- cantly better outcome than their counterparts in
dardized instruments for reliable assessment of the developed countries.
patients in different cultures, and to train Although not entirely unprecedented in
investigators in developing and developed reporting such findings (apart from clinical
countries to use such techniques to make impressions and anecdotal accounts, a more
comparable observations. Other aims of the favourable outcome of schizophrenic conditions
project were related to substantive issues about in Third World communities had been described
schizophrenia: to explore whether schizophrenic on the basis of a follow-up by Murphy &
disorders exist in different parts of the world, to Raman, 1971), the IPSS was the first large-scale
identify similarities and dissimilarities between cross-cultural study using standardized assess-
patients diagnosed as schizophrenics in the ment techniques to indicate an effect of culture
different centres, and to investigate whether the on the course of schizophrenic disorders. How-
course and outcome of schizophrenic disorders ever, the IPSS was not an epidemiological study
differ from country to country. The 1202 in the strict sense and the patients selected for it
patients, in the age range 15 to 44, were selected could not necessarily be considered representa-
in accordance with screening criteria which ex- tive of the range of syndromes and conditions
cluded cases of gross organic cerebral pathology, that might meet specified criteria for a diagnosis
chronic conditions of long duration, sensory of schizophrenia in different cultural settings.
defects and mental retardation. Each patient In view of the great potential significance of
was given a detailed clinical and social evaluation the conclusions of the IPSS, a need was felt for
at the point of intake, and the full assessment a more focused investigation of the frequency
was repeated two years, five years and in some and 'natural history' of schizophrenia and
centres 10 years after inclusion in the study. related disorders, which would be based on
The results of the IPSS have been described in more representative patient populations in dif-
a series of publications (WHO, 1973, 1979; ferent cultures. Since long-term birth cohort
Sartorius et al. 1977, 1987; Jablensky, 1987; Leff studies, or repeated census investigations of
et al. in press). In addition to advancing the entire populations were hardly possible, this
methodology of comparative research in the need could be met in two alternative ways, by
major psychiatric disorders, the project pro- selecting for study: (i) a prevalence-based sample
duced evidence that similar syndromes of schizo- or (ii) an incidence-based sample.
phrenia and other functional psychotic illnesses A prevalence sample, identified by the com-
 Table 1.1. Descriptive features of epidemiological surveys which have produced incidence or morbid
risk data on schizophrenia
Type and
Author and size of Target group Method of
year of population or sample Case-finding assessment and
publication Country studied investigated approach diagnosis Remarks

Birth cohort studies §=

Klemperer (1933) Germany Urban (city of Munich) Randomly selected birth Tracing of all probands Personal examination or
cohort attempted (44% key informant (271
I88I-I890 traced) examined)
(N = 1000)
Fremming (1947) Denmark Rural, island of Full birth cohort Tracing of all probands Personal examination •Same population
Bornholm* 1883-1887 (92 % traced) surveyed by Stromgren
(N = 46000) (AT = 5500) (1938)
Helgason, T. (1964) Iceland Mixed (total All members of the Tracing of all probands Hospital records and
Af = 85 183 in 1910) birth cohort 1895-1897 (99-4% traced) personal examination
who were alive in 1910
(N = 5400)
Census and longitudinal studies of whole populations
Brugger (1931) Germany Mixed All persons 'suspect' for Records and key Personal examination of
(Thuringia) (A7 =37561) mental disorder informants consulted a l l ' suspects' and of a
to detect 'suspects' sample of' healthy'
persons
Brugger (1933, 1938) Germany Two rural areas All inhabitants Door-to-door Personal examination
(Bavaria) (A7 = 5 4 2 5 and 3203) interviewing (semi-structured
interview)
Stromgren (1938) Denmark Rural, island of All inhabitants Key informants; door- Personal examination •Same population
Bornholm* to-door interviews in a studied by Fremming
(N = 46000) sample area (1947)
Sjogren (1948); Sweden Rural, two islands off All persons born or resi- Door-to-door Personal examination •Longitudinal data
Larsson & Sjogren the west coast dent on the islands interviewing, parish covering 45 years
(1954) records collected
Book (1953); Book Sweden Rural* All inhabitants Door-to-door Personal examination; •Genetic isolate
et al. (1978) (AT =8891 in 1949; pedigrees interviewing biochemical tests on
N= 5748 in 1974) subsample in 1974-77
Surveys based on service contacts
Odegaard (1946) Norway Whole country All first admissions Hospital diagnosis
(N= 14231) during
1926-1935
Shepherd (1957) England County of All first and Hospital diagnosis,
Buckinghamshire readmissions for grouped by author
(N = 400000) 1931-1933 and
1945-1947
Hollingshead & USA New Haven, Conn, All individuals in Census of hospitals, Diagnosis made at
Redlich (1958) (area population psychiatric treatment clinics and private facility; 6-17%
N= 174000) during a 6-month practitioners; review of re-diagnosed by authors
period in 1960 clinical records
Norris (1959) England Catchment area All persons admitted in Tracing of records, 2-
(JV= 1661000) 1947-1949 year follow-up
Kadri (1963); Singapore University students First admissions over a Screening of hospital Author's diagnosis I
Murphy (1968) 5-year period records
Wing, L. et al. (1967) England, Scotland and Three catchment areas All persons in episodes Case registers Diagnosis made at
USA (London, Aberdeen, of contact on census facility S'
Baltimore) day
Adelstein et al. (1968) England City of Salford All persons in Case register Hospital diagnosis
(N= 150000) 'inception episodes' of
contact
Walsh (1969) Ireland City of Dublin All first admissions Hospital diagnosis
(A r = 720000) (AT = 1427)
Murphy & Raman Mauritius Total island population All first admissions N = Screening of hospital Records and personal
(1971) (yv = 600000) 94) during 1956 records; follow-up examination of
interviews 12 yr later uncertain cases
Liebermann (1974) USSR District of Moscow All onsets for period Dispensary register Records; personal
(W = 248 000) 1910-1964 examination at census
Nielsen (1976) Denmark Island of Samso All service contacts over General practitioners Diagnosis made at I
(N = 6823) 18 yr case register, census in facility admitting the S3
1964 patients
Babigian (1980) USA Monroe county, NY Annual first-in-lifetime Psychiatric case register I
county population entries into psychiatric
(JV = 712000) care 3s
?
 A. Jablensky and others
munity survey method, even if statistically
representative, would include individuals dif-
fering according to previous length of illness and
duration of exposure to treatment and psycho-
social environmental influences. Also, a preva-
lence sample would miss patients who either
died early after developing a psychotic illness, or
migrated out of the study area. For these reasons,
a decision was made in favour of incidence
sampling, in the sense of identifying cases in the
early stages of the illness and evaluating them as
closely as possible to the point of their first
contact with any service or helping agency. Such
cases would then be followed up over a defined
period of time.
Apart from being an attempt at replicating
the IPSS findings on a more representative
patient population, the implementation of such
a design in a multi-culture setting was expected
to contribute data necessary for estimating
incidence rates of schizophrenia and related
disorders and their cross-cultural variation.
Knowledge in this respect is still scant, and the
absence of comparable data on incidence of
schizophrenia in various populations (especially
those in the Third World) is an obvious gap in
the epidemiology of this group of conditions.
Table 1.1 summarizes the main features of the
design of previous studies which have produced
incidence or morbid-risk data.
In most of the earlier studies, the difficulty of
developing an adequate case-finding method for
a condition of low population incidence, such as
schizophrenia, had been compounded by the
lack of specific diagnostic criteria and of the
standardized methods for collecting data on
history and psychopathology. Therefore, much
was to be gained from a multi-centre collabor-
ative study in which the same methods and
instruments, common diagnostic criteria, and
comparable case-finding procedures, would be
simultaneously applied across culturally dif-
ferent catchment areas. A special aim of the
project was to link the epidemiological esti-
mation of incidence rates to a diagnostic and
psychopathological classification of the cases. It
was explicitly recognized from the start that it
would be premature to subscribe to any one of
the alternative positions as to where the bound-
aries of schizophrenia ought to be drawn.
Therefore, diagnostic concepts of varying
' width' were to be applied to the classification of
cases, The subsequent validation of the di-
agnostic classification would take into account
the psychopathological features of the con-
ditions and a number of antecedent variables; it
would be particularly strengthened by the
supplementation of follow-up data.
In addition to such central epidemiological
and diagnostic questions that the new study was
primarily designed to tackle, there was the
secondary objective of integrating into its design
some recent research techniques for studying the
impact of environmental factors and the prog-
nosis of schizophrenic disorders which had not
been available at the time the IPSS was planned.
For example, the findings suggesting that
stressful life events may precipitate an acute
schizophrenic attack (Brown & Birley, 1968;
Schwartz & Myers, 1977), or that living with a
relative exhibiting high emotional involvement
coupled with a critical or rejecting attitude
increases the probability of relapse in schizo-
phrenic patients (Brown et al. 1972; Leff et al.
1983); or that discrepancies in the way psychotic
illness is perceived and interpreted by doctors
and by the patient's family (Katz et al. 1978),
could be of considerable relevance in explaining
culture-related differences in the prognosis of
psychotic disorders. Further refinement of the
measurement of course and outcome could be
achieved if the assessment instruments were
designed to allow a better separation of clinical
symptoms, functional impairments and social
disabilities (Jablensky et al. 1980). Some of the
results of these special studies have already been
published (Day et al. 1987; Leff et al. 1987,
1990; Wig et al. 1987); other reports are to
follow.
 Chapter 1 Design and methods
The present study, which at its inception was
given the title 'Determinants of Outcome of
Severe Mental Disorders', was designed in
1975-6, and the first patients were assessed in
August 1978. The twelve field research centres
participating in the project are: Aarhus*
(Denmark), Agra* and Chandigarh (India),
Cali* (Colombia), Dublin (Ireland), Honolulu
and Rochester (USA), Ibadan* (Nigeria),
Moscow* (USSR), Nagasaki (Japan), Notting-
ham (United Kingdom) and Prague* (Czecho-
slovakia).1 Six of these centres (indicated by an
asterisk) had taken part in the IPSS. The twelve
centres represented research settings of widely
different and contrasting cultural and socio-
economic characteristics. Like the IPSS, the
Outcome study was supported jointly by WHO,
NIMH, and the twelve field research centres.
The central coordination of the project was the
responsibility of the WHO Headquarters in
Geneva.
GENERAL DESIGN
The complex design of the Outcome study is
represented schematically in Fig. 1.1. All the
centres took part in the so-called 'core' study
which involved prospective case-finding, clinical,
diagnostic and social assessment of the cases,
and follow-up re-examinations at one year and
at two years after the initial examination. In
addition, subgroups of several centres each
carried out special exploratory studies on sub-
samples of the patients with the aim of
investigating: (i) the role of life events in the
onset of psychotic episodes; (ii) the effects of
'expressed emotion' in the family on the risk of
psychotic relapse; (iii) the nature and severity of
specific behavioural impairments and social
disabilities; and (iv) the influence of the family
members' perception of the patient's behaviour
on the course of the disorder.
' In the tables: AAR = Aarhus; AGR = Agra; CAL = Cali;
CHA/R = Chandigarh, rural area; CHA/U = Chandigarh, urban
area; DUB = Dublin: HON = Honolulu; IBA = Ibadan; MOS =
Moscow; NAG = Nagasaki; NOT = Nottingham; PRA = Prague;
ROC = Rochester.
The present report is limited to the epidemi-
ological aspects, the initial clinical assessment
findings, and the principal follow-up results of
the 'core' study; the results of the special
investigations on life events, emotional inter-
actions in the family, and family perceptions,
are the subject of separate publications (Day
etal. 1987; Wig et al. 1987; Leff etal. 1987,1990;
Katz et al. 1988); the data on levels of disability
of the patients will be reported subsequently.
Inclusion criteria and case finding
The aim of the ' core' study was to identify and
assess, in each of the 13 catchment areas2 (see
Table 1.2 for size of population), individuals
who met all of the following four inclusion
criteria and none of the exclusion criteria.
A. Inclusion criteria
(i) Age 15-54.
(ii) Residence of at least 6 months in the area
in the year preceding the initial examination.
(iii) Evidence of presence, in the preceding 12
months, of at least one of the following overt
psychotic symptoms: hallucinations or pseudo-
hallucinations in any modality; delusions; quali-
tative thought or speech disorder; qualitative
psychomotor disorder; or gross behavioural
abnormalities representing a break in the
person's previous pattern; or at least two of the
following abnormalities suggestive of psychotic
disorder: loss of interests, initiative and drive
leading to deterioration of performance; onset
of social withdrawal; episodic severe excitement,
purposeless destructiveness or aggression; epi-
sodic or persistent states of overwhelming fear
or anxiety; gross and persistent self-neglect.
(iv) First-in-lifetime contact with any' helping
agency' within the last three months, occasioned
by the symptoms and behaviours enumerated
under (iii).
2
The Chandigarh centre investigated two catchment areas, one
urban and one rural.
 A. Jablensky and others

Catchment area delineation; Development of instruments

identification of case-finding and research procedures;
network training of investigators

Prospective screening
of all first contacts
over a period of 2+
years (all centres)

Assessment of mental
state, history, diag-
nosis and prognosis
of included cases
(all centres)

Selection of patients
for special studies

Life events (Aar, Agr,

— Cal, Cha, Hon, Iba, Nag,
Pra, Roc)

Data processing at HQ;

Expressed emotion (Aar, assignment of a reference
Cha, Roc) CATEGO diagnostic
class

Impairments and disabilities

— (Aar, Cal, Dub, Hon, Not, Pra)

— Perception of illness
(Agr, Iba, Nag)

Follow-up examination
of mental state, and
course, treatment and
social outcome at 1
year and at 2 years
(all centres^

_L
Data analysis at WHO
HQ; data analysis in
the Centres

FIG. 1.1. Schematic representation of the design of the study.

dementia, with or without peripheral neuro-

B. Exclusion criteria3
(v) Clinical evidence of gross organic cerebral
disorder, including CNS damage due to alcohol
or drug abuse, and manifest in either delirium or
3
Severe mental retardation and communication difficulties (e.g.
deafness) were also reasons for exclusion from the Study.
pathy.
(vi) Previous contact with psychiatric, other
medical, or non-medical agencies, upon which a
mental disorder identical or similar to the current
one had been diagnosed or suspected.
The inclusion criteria were set with a view to
maximizing the probability that the study
 Schizophrenia: A World Health Organization Ten-Country Study

Table 1.2. Catchment area populations At the same time the inclusion criteria were
designed to block or restrict to a minimum the
Age groups intake of patients with primary affective dis-
Centre Total 15-54 orders.
Aarhus 574000 314344 The inclusion criteria, supplemented with
Agra 2806346 1426755 definitions and decision rules, were written in a
Cali 1347466 784009 Screening Schedule (SS) which served as the
Chandigarh/rural 103865 61642
Chandigarh/urban 348609 205786 principal instrument in the case finding. For the
Dublin 280322 149879 rare case, in which a strong suspicion of
Honolulu 357225 210020
Ibadan 931 348 635999
schizophrenia was present, even if not all of the
Moscow 392097 231866* inclusion criteria were met, the SS contained a
Nagasaki 447444 267149 provision of inclusion on the strength of a
Nottingham 380023 202214
Prague 1114809 578379
narrative not specifying the clinician's reasons
Rochester 397828 237223 for the decision. It should be noted that the
diagnosis made at the first-contact agency or the
•Age 18-54. referring facility was not one of the inclusion
criteria. The goal of case finding was to select
population would ultimately consist of patients individuals presenting with specified symptoms
with either main or alternative clinical diagnoses and behaviour abnormalities indicative of definite
of schizophrenia (ICD-9 categories 295.0-9); or possible schizophrenia; such cases were to be
paranoid psychosis (297.0-9); reactive psy- identified by the project team using the SS,
chosis, paranoid and unspecified (298.3, 298.4, regardless of any pre-existing diagnostic label.
298.8); unspecified psychosis (298.9); and other The case finding procedure involved the
conditions in which an underlying schizophrenic establishment of a list of all services and agencies
illness could not be excluded (e.g. paranoid and ('intercept points') in the area which were the
schizoid personality disorders, ICD 301.0 and likely sites of contact for potential study subjects
301.2; or paranoid and hallucinatory illness (Table 1.3); the making of the necessary notifi-
associated with alcohol or drug use, ICD 291.3, cation and referral arrangements; and the
291.5, 292.1, without evidence of gross organic continuous monitoring of these 'intercept
pathology). Since schizophrenia cannot be ruled points' with screening of all contacts for at least
out in the instance of a paranoid or hallucinatory 24 months.
psychosis without organic features in a person After the completion of case finding, the
with a history of alcohol or drug use, it was centres were required to carry out supplementary
decided to allow such cases to be included. small-scale surveys of the original network of

Table 1.3. Types of case-finding agencies cooperating in the study, by centre

Aar Agr Cal Cha/R Cha/U Dub Hon Iba Mos Nag Not Pra Roc

Psychiatric hospitals, units or X X X X X X X X X X X X

institutions
Psychiatric out-patient X X X X X X X X
departments or centres
General hospitals X X X
Polyclinics or health centres X X X X
General practitioners X X X
Private psychiatrists or X X X X
physicians
Public health nurses or social X
workers
Rural primary health care X
centres
Traditional healers X X X X
Religious healers X X X X
Police stations, prisons X X
Other X X X X X
10 A. Jablensky and others
contact sites, as well as of any other agencies
which had not been included in the study, in
order to search for missed cases (' leakage'). In
most instances this included scanning the records
of the agencies; in some centres, e.g. Aarhus and
Nottingham, the search was carried out through
the local psychiatric case register.
The variations in the demography, social
organization and geographic characteristics of
the catchment areas; the different patterns of
health care and utilization of various helping
agencies by the population; the unequal
resources of the field research centres; and, in
some instances, unanticipated events, presented
constraints which necessitated certain centre-
specific adjustments to be made in the case-
finding protocol. In some areas the monitoring
of denned contact facilities had to be supple-
mented by interviewing key informants in
various localities (Chandigarh, Ibadan); or by
limited door-to-door surveys in selected sub-
units of the catchment area (Honolulu). In other
centres (e.g. in Agra), the number of possible
contact sites far exceeded the resources available
for their monitoring; or it proved impossible to
secure the full cooperation of agencies such as
out-patient dispensaries, private practitioners or
traditional healers (e.g. in Cali). In Rochester,
unanticipated financial constraints affected the
continuity of the casefindingprocess. In Prague,
an administrative re-definition of the boundaries
of the catchment area of the research centre
took place in the course of the first year of the
study. In Ibadan, there was uncertainty about
the validity of the population census data on
which the estimation of the catchment area
population was based.
Such obstacles made it unavoidable for several
of the centres to curtail the casefindingnetwork,
or modifying in other ways the case finding
procedures, with a resulting uncertainty as to
the completeness of coverage. At the end of the
initial data collection phase, therefore, the field
research centres fell into the following two
groups as regards the intensity of case finding.
A. Centres which applied the casefindingand
assessment procedures without modification and
monitored all known or presumed contact sites
for eligible study subjects throughout the case-
finding period: Aarhus, Chandigarh, Dublin,
Honolulu, Moscow, Nagasaki and Nottingham.
B. Centres which found themselves compelled
to introduce modifications in the original study
design (e.g. excluded from monitoring some of
the known or presumed contact sites; had breaks
of the continuity of case finding; or employed
some pre-screening selection): Agra, Cali,
Ibadan, Prague and Rochester.
The modifications of the kind that occurred in
centres listed under B must have affected the
numbers of identified cases, although they would
be of little consequence for the study of the
clinical and social characteristics of the patients,
including course and outcome of their disease.
In order to base the estimation of the incidence
of schizophrenia and related disorders on the
strongest possible evidence as regards the com-
pleteness of case finding, rates are reported
separately in Chapter 3 for the two groups of
centres. However, material from all the centres
has been used for the clinical and diagnostic
analysis described in Chapter 2, and follow-up
data from all the centres (except Rochester) are
reported in Chapter 4.
Clinical and diagnostic assessment
The assessment of the subjects who had passed
the screening stage included the following steps
(the main features of the instruments used in the
core component of the Outcome Study are
summarized in Table 1.4).
(/) Mental state
Current psychopathology (symptoms occurring
in the last one month and at the point of
examination) was evaluated in a clinical inter-
view, administered by a psychiatrist using the
9th edition of the Present State Examination,
(PSE-Wing et al. 1974). For significant symp-
toms which fell outside the one-month limit, the
examiner was required to fill in the PSE
Syndrome Check List, included in the Diagnostic
and Prognostic Schedule (DPS)
(2) Past history
A new instrument, the Psychiatric and Personal
History Schedule (PPHS), was constructed
specially for the study. The PPHS is a standard-
ized guide to history data collection which can
use multiple sources of information: a relative
or other 'key' informant, the patient himself,
clinic or hospital records, etc. The PPHS enquiry
aims at a detailed reconstruction of the mode of
onset and the early manifestations of the illness
 Schizophrenia: A World Health Organization Ten-Country Study 11

Table 1.4. Main features of the research instruments

Instrument General description Source of information User

Screening Schedule A checklist containing 2 demographic, 4 history, 5 Admission records and Psychiatrist
(SS) symptomatological, and 5 behavioural items, all dichotomous case notes; brief
(yes/no). Definitions and guidelines included, as well as provisions personal interview
for recording diagnoses. Output: decision concerning patient's with patient and/or
eligibility for further study. key informant
Present State A guide to a semi-structured, standardized clinical interview Interview with the Psychiatrist
Examination (PSE), covering symptoms present in last 4 weeks. A total of 140 mental patient
9th edn state and behaviour items, rated for intensity and duration, divided
into 20 sections. Glossary definitions of symptoms and rating rules
available. Both symptom profiles and section scores obtainable.
Standard syndromes and diagnostic classes can be derived by
CATEGO computer program using PSE ratings as input.
Psychiatric and A guide to standardized psychiatric medical, social and developmen- Interviews with key Social worker,
Personal History tal history taking, with suggested probes and instructions for informants and psychologist or
Schedule (PPHS) rating. Contains sections on psychiatric history (present illness and patient; case notes psychiatrist
past episodes; onset; progression of symptoms; informant's and and other written
patient's own perception of problem; treatment); medical history; records
residence; household and family; social network; marriage;
children; sexual behaviour; parents and sibs; occupation;
education; religion; developmental history; and pre-morbid
personality.
Diagnostic and A summary of diagnostically and prognostically important All available data on Psychiatrist
Prognostic Schedule information; main, alternative and supplementary diagnosis in the patient; PSE and
(DPS) centre-specific terminology and in ICD terms; ratings of clinical PPHS
and social prognosis for next 12 months; checklist of treatments
and services with ratings of whether needed and available;
checklist of PSE syndromes; narrative summary on the case.
Follow-up Psychiatric As PPHS, but items re-formulated so as to rate change. Added As PPHS Preferably
and Personal History charts for month-to-month narrative recording and coding of psychiatrist
Schedule (FU-PPHS) symptoms, treatments, and social events
Disability Assessment A rating schedule containing 97 items grouped in 4 sections: overall Interview with key Social workers,
Schedule (DAS) behaviour, social roles, behaviour in hospital and modifying informant and/or psychologist or
factors. Performance of each social role in the last 4 weeks is rated patient psychiatrist
on a 6-point scale, with anchor points for 'no dysfunction',
'minimum dysfunction', and 'maximum dysfunction'. Manual
with guidelines and rating rules available.

(including the responses of the patient's social
environment), the past medical and psychiatric
history, the family background, social func-
tioning and circumstances, and the development
of the pre-morbid personality.
(3) Diagnosis
The diagnostic assessment of each case, made by
the centre investigators, was recorded in a DPS.
In seven of the centres the diagnosis was based
on a consensus between two or more investi-
gators (in Nottingham, the chief investigator
reviewed all the diagnoses). In five centres the
DPS was filled in by an individual investigator
and no case by case review by the research team
was involved.
The DPS requires: (i) a diagnosis of the
current mental state (i.er a diagnosis of the
psychopathological syndrome manifest at the
time of the initial examination); (ii) a ' main'
diagnosis of the condition (i.e. a nosological
diagnosis taking into account the previous
history and all other information available); and
(iii) alternatives to the main diagnosis and sup-
plementary diagnoses of any accompanying
conditions or personality disorder. The diag-
nostic statements could be made in the format
and terminology customary in each centre, but
the appropriate ICD-9 numerical codes had to
be assigned to them in accordance with the ICD-
9 glossary (WHO, 1978). The correctness of the
assignment was verified at the study Head-
quarters. In the same schedule, a number of
ratings were made of the psychiatrist's judge-
ment about the prognosis of the case for the
next 12 months, and of the estimated need for
12 A. Jablensky and others

Table 1.5. Cases screened, cases included and cases missed in the data collection phase
Aar Agr Cal Cha/R Cha/U Dub Hon Iba Mos Nag Not Pra Roc All

Patients who passed initial 128 95 157 66 198 83 80 144 198* 117 99 112 58 1535
screening
Subsequently excluded: 26 4 2 12 43 3 10 — — 7 5 1 2 116
diagnostic reasons
Subsequently excluded: 3 2 1 13 2 1 1 — 2 1 1 26
previous contacts
Subsequently excluded: — 8 1 — 2 — 3 — 14
outside case finding
period
Final size of 2-year 99 81 154 54 155 67 68 142 197 108 92 107 55 1379
incidence cohort
Possibly eligible cases 21 100 + 1- — 1 1 1-2 NK NK 4 257 NK
missed, identified (estimate) (estimate) (estimate)
retrospectively

•Collected over three years of case finding.

NK = Not known.

each of a number of checklist items of specific
treatment and management measures. A detailed
narrative summary of the case was also included
in the DPS.
In addition to the clinical diagnostic decision
made in the field research centres, there was a
central diagnostic classification of the cases,
using the CATEGO computer program (Wing,
1976). The program first assigns each case to one
of eight graded levels (index of definition, ID),
depending on whether the information recorded
on the PSE is sufficient and specific enough; the
CATEGO then orders the PSE material into
syndromes and broader diagnostic classes in
accordance with hierarchical decision making
rules. Since the CATEGO version used up to
date in the analysis of the project data was based
on the PSE only, the resulting CATEGO classes
cannot be regarded as equivalent or alternative
to the clinician's diagnosis, but as its com-
plement.
Yet another level of control of the correctness
of the application of the criteria was imple-
mented at the study Headquarters where two
of the authors (A. J. and J. E. C.) independently
reviewed the computerized records of all patients
and jointly made decisions concerning any
doubtful inclusions or exclusions.
TOTAL STUDY POPULATION
In order to increase the total number of cases
available for analysis, and to reduce the effects of
possible seasonal fluctuations on the detection
of new cases (which might be due to an uneven
propensity of people to seek help in the different
seasons), or of any other random interference,
the case finding was extended over two years.
The total number of subjects who passed the
initial screening and were given a clinical and
social assessment was 1538 (Table 1.5). After the
processing and checking of all the records and
schedules at Headquarters for consistency with
the agreed criteria for inclusion, 156 cases were
excluded, the majority of them (116) for di-
agnostic reasons. These were cases in which
neither the CATEGO assigned class was S, P, or
O, nor the clinical diagnosis (main or alternative)
corresponded to the ICD rubrics selected to
represent a broad class of schizophrenic and
related disorders. In most instances, these were
patients with affective disorders who had passed
the screen because of psychotic and suspected
schizophreniform symptoms which later were
not confirmed by the clinical examination.4 In
the remaining instances, the reasons for ex-
clusion were either evidence of previous treat-
ment for the current condition or a date of
screening falling outside the agreed 24-month
period.5 With these additional exclusions, the
final size of the study population was 1379.
4
The case-finding procedure was over-inclusive by design, in
order not to miss at that stage cases in which a schizophrenic
diagnosis could not be ruled out.
8
Case finding in Moscow was extended to 48 months, after it was
discovered that full coverage of the contact sites was not achieved in
the first 12 months. Incidence estimates for Moscow are based on the
cases collected in the 3 years following the adjustment in the case
finding procedure.
 Schizophrenia: A World Health Organization Ten-Country Study 13

Centre Ratio highest

beginning 30
to lowest N
Aar 1 Dec. 78 20

10

30
20
20
Cha 1 Oct. 78
10 - 4-3

30 -,

20
Dub 1 Nov. 78 2-6
10

30 -,

20
Hon 1 Dec. 78 30

10

30 -i

Mos 1 Nov .79* 20 1-6

10

30

Nag 1 Jan. 79 20
30
10 -

1 2
30 -,

Not 1 Aug. 78 20 1-7

10

FIG. 1.2. Number of included cases by trimester (only centres with complete coverage of catchment area).
(*, Excluded in the first year.)
14 A. Jablensky and others

As pointed out above, not all of the centres instrument used in the IPSS. In order to monitor
could achieve or maintain the full and con- inter-rater agreement on PSE assessment within
tinuous coverage of the catchment area which is those centres where more than one rater inter-
required for epidemiological analysis. As expec- viewed patients, the investigators were required
ted, most of the centres which curtailed the case to carry out joint interviews with every 10th
finding procedure later reported high numbers patient. Agreement on PSE ratings between the
of potential cases which might have been centres was tested by joint rating of audio- or
included but were detected or estimated only video-tapes during meetings of investigators. In
retrospectively, through the 'leakage' studies addition, PSE audio-tapes were circulated to all
undertaken after the completion of the pro- the centres for rating.
spective case finding. The retrospective search The reliability of other assessments was also
yielded small numbers of missed cases in the tested in specially designed exercises. Brief case
centres which had established a comprehensive histories were circulated to the centres and
case finding network (see the bottom line in scored for the screening inclusion/exclusion
Table 1.5). criteria prior to the beginning of case finding.
For the latter group of centres, the number of Similar to the PSE reliability exercise, audio-
included cases by trimester is shown on Fig. 1.2. tapes of interviews using the PPHS, FU-PPHS,
The ratio from the highest to the lowest quarterly and DAS were supplied to the centres for
number of included cases, which can be taken as training and inter-rater agreement evaluation. A
an index of the fluctuation of the case finding synopsis of the reliability exercises is presented
rate, did not exceed 3 (except in the rural area of in Table 1.6, and the results of the analysis of the
Chandigarh where the population size was reliability data are reported below.
small).
Table 1.6. Number of cases assessed in intra-
FOLLOW-UP STUDY centre reliability exercises, by centre and research
instrument
All the centres participated in a follow-up of
the included cases which involved the re- Screening DPS DPS
interviewing of patients and informants at one Centre schedule PSE PPHS 1-U-PPHS (initial) (FU)
year and at two years following the date of the Aar 0 16 2 2 11 0
initial screening. The instruments used at such Agr 18 34 4 13 16 17
follow-up examinations included the PSE, a Cal 0 8 0 1 0 5
Cha 3 29 17 0 8 0
follow-up version of the PPHS (FU-PPHS) Dub NA 6 0 1 0 3
which allowed the recording and coding of Hon 0 3 22 0 0 0
Iba 10 7 4 0 5 2
information about symptoms, treatments and Mos 0 0 0 0 0 0
social variables on a month-by-month basis, and Nag 23 29 8 0 2 0
a modified DPS. At the second year follow-up to Not 0 7 1 6 9 0
Pra 7 9 7 0 0 0
this package of instruments was added a Roc 0 0 0 0 0 0
Disability Assessment Schedule (DAS), designed All 61 148 63 23 51 27
to give a social role performance profile of each
patient. NA = Not applicable.

TRAINING OF INVESTIGATORS AND INTRA-CENTRE RELIABILITY

MONITORING OF THE RELIABILITY OF
RESEARCH PROCEDURES
All the psychiatrists who participated in patient
assessment in the different centres had been
formally trained to use the PSE (either by one of
the originators of the technique or by an
experienced user). Many had extensive previous
experience with the earlier version of the
(a) Screening Schedule (SS)
A total of 61 potential inclusion cases were
assessed jointly by two or more investigators in
5 of the centres. Of these cases, 64 % were rated
as meeting the inclusion criteria and 36% as
failing to meet them. Within each centre there
was a perfect agreement on the inclusion/
exclusion of the cases.
 Schizophrenia: A World Health Organization Ten-Country Study 15

(b) Present State Examination (PSE) (c) PPHS and FU-PPHS

Joint rating of cases by two investigators The data from the joint rating with the PPHS of
(alternating in the roles of an active interviewer a total of 63 cases in 8 centres indicate a high
and a passive rater) took place in all the centres level of inter-rater agreement (PAR values higher
except Moscow and Rochester. The number of than 0-89 and ICC values higher than 0-70) on
study subjects rated in this way varied and, all of the 25 history items which were selected
while four of the centres did not attain the for reliability evaluation. The follow-up version
recommended target of 10% of the patients of the schedule, the FU-PPHS, was used in
being assessed in reliability exercises, another reliability exercises in 5 of the centres (a total of
four centres significantly exceeded this number. 23 cases), with similarly high levels of agreement
The results (Table 1.7) indicate that in all the (Table 1.8).
centres which provided reliability data the PSE
was used at an acceptable level of inter-rater
agreement, both at initial examination and at Table 1.8. Intra-centre reliability {pairwise
the follow-up examinations. agreement rate, PAR) of selected history items
assessed with PPHS and FU-PPHS
Item PAR
Table 1.7. PSE intra-centre reliability data (a) PPHS (63 subjects rated)
Type of onset 0-90
At initial examination At follow-up Alcohol use 0-97
Drug use 0-96
Number of Number of Mental illness in family 0-97
symptoms % symptoms % Overall adjustment in childhood 0-93
Overall adjustment in adolescence 0-89
(a) Symptom distribution according to levels of the intra-class Pre-morbid personality traits 0-94
correlation coefficient (ICQ (108 cases rated)
(b) FU-PPHS (23 subjects rated)
ICC
Remission since initial examination 0-93
<0-0 1 0-7 17 12-3
Relapses since initial examination 0-94
0-0-0-1 0 00 5 3-6
Pattern of course 1 -00
> 0-1-0-2 1 0-7 4 2-9
Alcohol use 100
> 0-2-0-3 1 0-7 5 3-6
Drug use 0-96
> 0-3-0-4 10 7-2 10 7-2
Socioeconomic level 0-86
> 0-4-0-5 9 6-5 13 9-4
> 0-5-0-6 22 15-9 21 15-2
> 0-6-0-7 35 25-4 18 13-0
> 0-7-0-8 38 27-5 15 10-9
17 12-3 13 9-4
(d) DPS
> 0-8-0-9
> 0-9-1-0 3 2-2 7 51 The reliability assessment of the DPS involved a
NV* 1 0-7 10 7-2
Total 138 100-0 138 1000
determination of the diagnostic agreement on
*No variation in the ratings : ICC impossible to calculate.
ICD-9 three-digit codes by pairs of raters, each
reviewing independently all the history and
(b) Symptom distribution according toi levels of the pairwise mental state data collected on the subject. In the
agreement rate (PAR) (108 cases rated)
PAR course of the initial examinations, a total of 51
0-75-0-80 0 00 1 0-7 cases were rated by 2 or more investigators in 6
> 0-80-0-85 8 5-8 4 2-9 of the centres. There were only two disagree-
> 0-85-0-90 35 25-4 31 22-5
> 0-90-0-95 59 42-8 50 36-2 ments (one involving a diagnosis of paranoid
> 0-95-1-00 36 260 52 37-7 schizophrenia versus paranoid state, and another
Total 138 1000 138 1000 involving a diagnosis of' other' chronic organic
psychosis versus 'other' transient organic psy-
chosis). On follow-up, 27 cases were rated with
the follow-up version of the DPS by 27 pairs of
The intra-class correlation coefficient (ICC) raters in 4 centres, with only one disagreement
was significantly different from zero for all PSE (neurotic depression versus reactive depressive
symptoms but one (conversion), and for 137 of psychosis).
the 138 PSE symptoms the pairwise agreement The data of the intra-centre reliability exer-
rate (PAR) was higher than 0-80. cises indicate fairly high overall levels of inter-
16 A. Jablensky and others

rater agreement in the application of the 'core' Table 1.9. PSE inter-centre reliability data (3
instruments of the study. However, no reliability audiotapes rated by an average of 14 raters per
data on any of the instruments were obtained case)
from two of the centres, and those centres which
did carry out such exercises differed according to (a) PSE symptom and CATEGO syndrome distribution
the number of cases assessed with the different according to levels of the pairwise agreement rate
instruments. In view of the satisfactory inter- PSE CATEGO
rater agreement demonstrated by all the centres PAR symptoms syndromes
which reported reliability exercise data, and the
0-50-0-59 4 4
involvement of experienced clinical investigators 0-60-0-69 13 8
in the two centres which did not, it would appear 0-70-0-79 22 7
unlikely that the hiatuses in the intra-centre 0-80-0-89 39 13
0-90-100 60 6
reliability data would have major implications
Total 138 38
for the validity of intra- and inter-centre com-
parisons. (b) PSE symptoms with PAR < 0-70

PSE symptoms PAR

INTER-CENTRE RELIABILITY Worrying 0-61

Complaints of inefficient thinking 0-53
Inter-centre (comparing ratings done by raters Poor concentration 0-66
Delayed sleep 0-66
from different centres) reliability exercises were •Early waking 0-64
carried out with the SS and the PSE. Verbal hallucinations based on depression 0-68
•Delusional misinterpretation 0-62
(a) Inter-centre reliability of the SS •Primary delusions 0-60
•Acting out delusions 0-57
A total of 34 case summaries prepared by the Incoherence of speech 0-63
different centres were circulated for rating as Misleading answers 0-68
Adequacy of interview 0-65
regards inclusion/exclusion, using the SS cri- Social impairment 0-56
teria. Altogether, 40 raters in all the centres Lack of self-confidence 0-68
except Dublin and Rochester took part in the •Delusions of persecution 0-59
Situational autonomic anxiety 0-69
exercise. The returns were distributed as follows: •Delusions of grandiose identity 0-68
included, 63-9%; excluded, 31-5%; more in-
formation needed, 4-6%. •Included in the 44-symptom profile (see Fig. 2.5).
The mean PAR across all the 34 cases was
0-73, and the mean ICC was 0-82 (significant at pairwise and group exercises involving more
P < 0001). This result can be taken as an than 2 raters, a discrepant rating of a symptom
indication that the inclusion/exclusion criteria is counted as a disagreement, even if in the case
written in the SS, which are critical to the study of a group exercise a single participant rates at
as a whole, were interpreted in a consistent and variance with the majority. Assuming (arbi-
comparable manner by investigators responsible trarily) that PAR should be higher than 0-70 for
for the implementation of the case finding a PSE symptom to be regarded as being reliably
procedures in the different centres. rated in the different centres, it can be seen
(Table 1.9Z>) that only 16 out of 136 PSE
(b) Inter-centre reliability of the PSE symptoms did not satisfy this requirement. Of
The assessment was based on the rating of 3 these, only 4 are included in the selective list of
English-language audio-tapes of PSE interviews 44 symptoms used to construct a symptomato-
which were circulated to the centres. The three logical profile of schizophrenia. The conclusion
tapes were prepared by the Nottingham centre; from the inter-centre reliability exercises using
the total number of raters was 16. A summary the PSE is that the instrument was applied in a
of the results is presented in Table 1.9a and b. sufficiently consistent and reliable manner in the
The PAR was higher than 0-70 for 121 out of centres.
136 PSE symptoms rated. The agreement on
CATEGO syndromes was somewhat lower, due
to the scoring rules which require that in both
 Schizophrenia: A World Health Organization Ten-Country Study
DATA PROCESSING
Duplicates of the research schedules, or specially
designed scoring sheets, were regularly mailed to
Geneva throughout the data collection period.
After the data had been transferred to magnetic
tape, they were checked for completeness,
validity and consistency. Errors were corrected
by project staff, when necessary after com-
munication with the centres concerned. Most of
the computational work was carried out using
standard statistical packages (SAS) but some
computer programs had to be written specially
for the study. Once the basic data were tabulated,
the case records from each centre were again
reviewed by Headquarters investigators and
17
consultants, before proceeding with more com-
plex statistical analyses. This process was signifi-
cantly aided by the final evaluation sheets, filled
in on each patient by the centre investigators. In
these sheets, the research workers were asked to
supplement the information contained in the
schedules by providing global judgements on
some items and by re-rating items which after
initial scrutiny at Headquarters had been found
to lack sufficient data or suffer from errors.
In addition to the statistical analyses which
were performed centrally at Headquarters, most
of the field research centres are analysing and
reporting their own material, focusing especially
on centre-specific details and issues.
 Chapter 2 Sociodemographic, clinical and
diagnostic description of the study population
A unique feature of the present study is its disorder similar to, or continuous with, the
having collected a large cross-cultural sample of current disorder, he or she was not eligible for
patients in the early stages of their illnesses and the study.
with minimal previous exposure to treatment
and the social effects of being under psychiatric
care. For these reasons, the descriptive charac- INTERVAL BETWEEN FIRST CONTACT
teristics of the study population, assessed at the WITH SERVICES AND INCLUSION
initial examination, are of particular epidemi- Fig. 2.1 shows the percentages of the study
ological and clinical interest. population falling within different intervals
between a first contact and intake (the date of
screening). Altogether 93 % of all screened and
GATEWAYS OF ENTRY INTO THE included patients for whom the first contacts
STUDY could be accurately dated, met the criterion of
As stated in chapter 1, the aim of the case having had no contacts with ' helping agencies'
finding procedure was to identify individuals prior to the last three months. In the remaining
aged 15-54 who in the past 12 months had either 7 % a history of earlier contacts was ascertained
experienced one or more clearly psychotic upon initial examination but it was decided to
symptoms1 or shown at least two specified retain such patients in the analysis, if at the time
behavioural abnormalities2 suggestive of a psy- of the first contact the disorder had only
chotic illness. To be eligible for the study, such manifested prodromal signs, had not been
individuals had to be residents of the catchment
area and to be making for the first time in their
lives a contact with a 'helping agency' because 70-
of psychiatric disturbance. By definition, the
first contact was the first visit to a help-giver of 60-
any kind which could be followed by other
referrals related to the same problem within the
50-
three months preceding the 'detection' of the
case by the study team.
In other words, thefirstcontact with a helping 8> 4 0 -
agency must be distinguished from the first S3
contact with the study. The latter is referred to 30
as 'inclusion', 'intake' or 'screening', and the i, ^
interval between the first contact and the 20-
inclusion into the study ought to be three
months or less. If it were established that the
person had had earlier contacts of a 'helping 10-
agency' (i.e. prior to the three-month period
referred to above) because of a psychiatric -q>-?-9
1
4 5 6 7 8 9 10 11 12 12+
Hallucinations or pseudohallucinations; delusions; qualitative
thought disorder; qualitative psychomotor disorder. Months
2
Incidence
Reduction or loss of interests, initiative and drive; deterioration count
of performance; conspicuous social withdrawal; excitement, de-
structiveness or aggression; persisting states of pervasive fear; gross FIG. 2.1. Time between first contact and initial examination
self-neglect. (N= 1288).
18
 Schizophrenia: A World Health Organization Ten-Country Study 19

diagnosed as psychotic, and no appropriate

treatment had been initiated.
Most of the included cases (68 %) entered the " 2
study within a month of their first contact (38 % CN — OO I OO \O >O ON
within a week). Only 12 % and 7 % were included 2 8
during the second and the third month, re-
spectively, after a first contact. Therefore, the rp cp I I I O\ sp
study population can be described, by and large, 17
as having had no previous exposure to psy-
chiatric institutions and services.
The distribution of intervals between first
contact and intake was similar in all centres.
r- —• I r- op i
There was a tendency for patients in developing In
countries to enter the study after slightly longer
intervals between the first contact and screening
(66% of the cases in the developing countries
were included within the first month, compared
to 70% in the developed countries), but the O fp I IT) I I 00 •*
difference is not significant.
Reasons for making the first contact
There was a remarkable similarity in the events
and social reactions leading to help-seeking and
o ww |
eventually to a formal ascertainment and initi- 6 fN •- '
ation of treatment of a psychotic illness in the "a
different cultures. The reasons for seeking help s:
in the form of advice, admission or referral, as
stated by the 'key informants',3 were analysed
to establish their rank order and frequency. — (N | rp I I \O
Onset or recent exacerbation of odd behaviour, 6l - 8
appearance or talk, was the most frequently
given reason for seeking psychiatric treatment in r, g
all the centres (in 90 % of the cases), followed by
reports of actual or feared violent behaviour
towards self or others. Acts or threats of violence
to others, or to property, were more frequently
reported as motives for seeking psychiatric help
in the developing countries (in 27%) than in
developed countries (11%). Attempted suicide
or self-injury were reported with similar fre-
quency in the developed countries (10 %) and in
the developing countries (6 %). «5
Types of contacts prior to intake
S3 §
The actually occurring sequences of different 11
contacts leading to the inclusion of patients in e i- u. -s s
3
A ' key informant' was interviewed with the PPHS in 69 % of the 2 8 P>J>
cases; for 31 % of the patients no such person was available and the
main source of PPHS information was the patient himself and/or ' "H .5
previous case notes. Where a 'key informant' was involved, the 8 8 f -S c _
person was patient's mother in 41 % of all the instances, spouse in •C a. " g °
O o
24%, and father in 15%. In 1 1 % of the cases more than one 'key
informant' was interviewed for the same patient. i 3 2 •£ 3 o - -S
lZ H£ £ U O
20 A. Jablensky and others
the study are shown in Table 2.1. The top six
rows give the frequencies of simple pathways of
entry, consisting either of one step only (a
psychiatrist or another mental health profes-
sional being the first and only contact prior to
inclusion) or of two steps (other agent in the
community being the first contact, and the
psychiatrist the second). In addition there were
more complex sequences (next to last row in the
table), involving three or more (up to six) steps
before reaching the mental health professional
collaborating with the project team.
It is clear that the different types of organ-
ization of health care (including formal and
informal systems) existing in the catchment
areas leave their imprint on the pattern of
contacts leading to the inclusion of a subject into
the study. However, it is worth noting that the
largest proportion (39 %) of the study patients
made their first contact with a psychiatrist. The
percentage of patients contacting a psychiatrist
directly, without a referral by any primary care
agency, was particularly high in Cali, Ibadan,
Moscow, and Nagasaki, but low in Nottingham,
Aarhus, and Agra. The second most frequent
source of referral across the centres was the
general practitioner or other physician in a non-
psychiatric service. Other community agents,
such as nurses, social workers, or police, did not
appear to play a significant role in the first
contacts and referrals of psychotic patients in
any centre.
A great variety of traditional and religious
healers (e.g. herbalists, homeopaths, prac-
titioners of Ayurvedic or Unani medicine, or
yoga teachers in India; babalawo and aladura
healers in Nigeria, priests and other religious
persons) exist in many centres and are known to
play an important role in the management of
mental health problems in developing countries.
Table 2.1 shows that such practitioners were the
primary source of referral for 28 % of the cases
in Agra, 28 % in Ibadan, and for lower percent-
ages in several of the remaining centres. In
addition to the simple patterns of contacts,
traditional or religious healers were involved in
more than one half of all the complex patterns of
referral. If the complex patterns of contacts are
also considered, the percentage of patients who
had seen traditional or religious practitioners
prior to inclusion would increase from 7 to 16 in
the total study population. In Agra, 73% had
had such contacts before being detected by the
project team; the corresponding percentage for
Ibadan is 35, and for Chandigarh 36. This
underscores the observation that in many
developing countries it is a common practice for
people with mental health problems to seek help
from, and utilize simultaneously, the two systems
of care, the ' Western' and the traditional one. It
also highlights the importance of ensuring the
cooperation of informal and traditional systems
of care for psychiatric patients in any epidemi-
ological research involving communities where
such practitioners may be a preferred or parallel
resource. Although in this study the attempts to
involve traditional and religious healers in case
finding were not entirely successful, a total of
200 eligible cases of psychotic illness would have
been missed, had such attempts not been
undertaken.
Varying proportions of patients in the dif-
ferent centres had been prescribed some kind of
treatment by the 'helping agencies' visited by
them before their inclusion into the study.
Considerable numbers (between 22 and 46 %) of
patients in Agra, Chandigarh and Ibadan had
used traditional remedies or unspecified 'drugs'.
Neuroleptics had been prescribed to more than
10% of the patients in all centres except Ibadan,
Moscow, Nottingham and Rochester. In Agra,
17% had had one or more electroconvulsive
treatments in the psychiatric hospital prior to
inclusion, and in Honolulu, 65 % of the patients
reported about having had some form of
psychotherapy.
SOCIODEMOGRAPHIC
CHARACTERISTICS
As no resources were available in the present
study for a systematic statistical comparison
between the study population and the general
population in the different catchment areas, the
data reported below are no more than a
descriptive summary of the main sociodemo-
graphic features of the study samples.
Age and sex
The size and the age distribution of the study
population in each centre is given in Table 2.2,
and the male/female ratio for each age group in
Table 2.3. Within the total study population of
1379 subjects (745 men and 634 women), males
 Schizophrenia: A World Health Organization Ten-Country Study 21

predominate in the younger age groups and

(V= 1379
All
Si— ff,q 1-00
^ > constitute 58 % of the patients aged less than
\b •- rn
35.4 Women predominate in the age groups
35-54, representing 62 % of this population. In
Moscow, Nottingham, and Aarhus, particularly
Roc

II high proportions (50, 42, and 39 % respectively)

of all female patients are in age groups 35-54.
The male/female ratios shown in Table 2.3
N= 107

must not be interpreted epidemiologically (age-

Pra

£^ 8 and sex-specific rates are discussed in chapter 3).

Nevertheless, they suggest a pattern. If Prague is
excluded because of an over-selection of female
Not

II patients (due to a sampling bias), then an excess

of males in the age group 15-24 is found in all
centres except one. In the age group 25-34 a
N= 108

similar excess of males is observed in 9 out of the

Nag

V-) ~ ~ ^ - O
w-i f*^ — O 12 centres. In the age group 35—44 there is an
excess of males in two, equal numbers in another
I o II oo o fp ao
three, and an excess of females in the remaining
centres. In the age group 45-54 the pattern is
I ~ v~i fN (
<N m r^ <
reversed, and in 10 out of the 12 centres the
percentage of females exceed that of males. This
V = 142

pattern, and also the excess of females in the

Iba

older age groups, tallies well with the curves and

the age- and sex-specific incidence rates, dis-
•2
X cussed in chapter 3.
Hon

II
.3
a Urban/rural residence
r^ The majority of the study patients are urban
f
Dub

II \—ON ON residents. If urban areas are defined so as to

OO
include suburbs and periurban conurbations,
the percentage of patients living in urban
Cha/U

communities would be between 68 in Agra and

q
II
fe: (N 4 o 100 in Moscow. In Honolulu, 57% of the
OS — O patients were living in residential areas outside
the city; the same was the case in somewhat
Cha/R

II lower proportions (43, 20, 18 and 15%) of the

patients in Nagasaki, Rochester, Dublin, and
Aarhus. In Ibadan, 59% of the patients were
in
described as residents of the poor periurban
Cal

o ^ nm
II 6 ~ « - areas which had mushroomed around the old
town in recent years and contained mostly slum
00
dwellings. Slum conditions are also the main
feature of the barrios in Cali, which accom-
Agr

CO OS — (N
II
o
modate a large proportion of the study
patients.
The rural catchment area of Chandigarh was
Aar

-^- rn rg —
II
' ' 2 2 8 the only traditional village area specially selected
4
In Moscow 33 % of all male patients were in the age group
35-54. This relatively high proportion is probably due to the
Age

truncated distribution of the younger age groups: the lowest age set
in IA v> in for case finding in that centre was 18.
cs m TT
22 A. Jablensky and others

for study. With 98 % of the patients there being

classified as rural residents, it provided a
contrasting environment to the Chandigarh
urban study area. The only other centre with a
sizeable rural population was Agra where 32 %
of the cases came from the villages of the rural
district surrounding the town.
The socioeconomic level of the patients'
§2S~ S neighbourhoods was rated by the investigators
on a simple scale as 'average', 'higher than
average', or' lower than average', in comparison
with the catchment area as a whole. With the
exception of Ibadan, Cali, Nottingham, and
Chandigarh (rural area), in all the centres a
Is majority of the patients (51-97%) were rated as
— TT o o — J living in 'average' neighbourhoods. In Ibadan,
V Cali, and Chandigarh (rural area), the field
-o
investigators reported that more than 50% of
s
the patients were living in socioeconomically de-
« o o o o prived localities.' Below average' socioeconomic
1 level of the neighbourhood was also rated for
46 % of the patients in Nottingham, 33 % of the
f •i - o - -
patients in Rochester, 28 % in Agra, and 26 % in
Chandigarh (urban area). Honolulu was the
only centre in which a considerable proportion
(19%) of the patients were residents of affluent,
£ above-average residential communities.
Marital status
o — — o —
The marital status of the patients is shown in
Table 2.4. There are marked differences among
the centres. A much higher proportion (68 %) of
the males, than of the females (39 %), are single.
The only centres with relatively low proportions
of never-married men are Chandigarh (rural
rn area) (42%), Agra (45%), and Moscow (50%).
In the latter centre, this is partly due to the
X> absence of subjects aged less than 18. The
^2 proportion of the married among the female
patients (46 % in the total sample), ranges from
22 % in Nagasaki to 76 % in Ibadan. In most of
the centres in developed countries, between 15
and 24 % of the female patients are divorced or
separated.
Agr

»— ro o »-*
V

Type of household
oo oo ON r~- In all the centres except Chandigarh (rural
i-66
area) and Ibadan, the nuclear family (con-
taining at most two generations with conjugal,
parent - child, and/or sibling relationships
among its members) was the predominant type
of family unit (49 to 87% in the different
 Schizophrenia: A World Health Organization Ten-Country Study 23

— \O
centres). In Chandigarh (rural area) and in
< II Ibadan, the extended family (including more
than two generations and more distantly related
kin) or the joint family (composed of two or
i: more brothers with their wives and children
living together), was a more frequent type of
s
social unit than the nuclear family. The pro-
a. portion of patients living alone varied from zero
in Agra and Chandigarh (rural area) to 10-15 %
in Moscow, Nottingham and Prague, and to as
high as 35 % in Aarhus.
Overall, the proportions of patients living
alone were considerably lower in the centres in
developing countries (where they ranged from
0% in Agra and Chandigarh, rural area, to
2- g 6-5% in Chandigarh, urban area) than in the
centres in developed countries (between 6-0 % in
CO ° ° Dublin and 35-4% in Aarhus). In contrast,
£ I'
household arrangements of the extended family
type were much more common in the developing
S1
a <»>
•O VI O I O
countries (frequency between 14-8 % in Chandi-
**; OO — I O
garh, urban area, and 44-4% in Chandigarh,
rural area) as compared with the developed
—CT\I I O countries (between 1-0% in Aarhus and 11-8%
so m I I o
in Honolulu).
5.3
Education
6%
The level of completed education varied con-
siderably among the centres. Illiterate subjects
5 fe (who never attended school) constitute 56, 40
and 30 % of the patient samples in Chandigarh/
rural, Agra, and Ibadan respectively. Women
I
OO OO (N (N O
00 O
outnumber men in this category in every centre.
At the other extreme, the study population in
$£* I 8 Moscow, Chandigarh (urban area), Prague, and
Honolulu, includes 23, 15, 11 and 10% re-
spectively, university graduates. In the study
r- r- ^t CN o
population as a whole, the majority (58 %) had
completed either primary or secondary edu-
cation. Relatively large proportions, between 15
and 27 %, of the subjects in Rochester, Aarhus,
Prague, Chandigarh (urban area), and Nagasaki,
were students at the point of entry into the
study.
1 Employment
MT3 5 In the total study population, 25% of the
C « ^
•r! ^ i-i
patients had never had a gainful employment for
various reasons (housewives, students, unpaid
1^i_ workers in family household, or general un-
"S 8 °° e2 employment). The majority of the patients (74%
fill fill of the total sample) had some work record, and
inlOO S5S OO
24 A. Jablensky and others
their proportion varied from 43 % in Chandigarh
(rural area) to 97 % in Honolulu and Moscow.
ONSET AND EARLY MANIFESTATIONS
OF ILLNESS
In the design of the study, and throughout the
data collection period, special attention was
given to ensuring that accurate estimates were
made of the beginning of the illness and its
length up to the moment of screening and
inclusion. During the interview with the key
informant or the patient, the investigators
explored the occurrence and dated each one of
25 specified early behavioural manifestations of
the disorder, taking into account all of the
informant's statements but using their own
judgement in estimating the number of months
since onset. The investigator's estimate had to be
supported by a narrative summary of the early
symptoms and signs, mode of onset, the pro-
gression of the symptoms, and any relevant
circumstances. The mode and timing of onset
were established through detailed questioning
and cross-examining the informant or the
patient. The following excerpts from the PPHS
illustrate the content and wording of this part of
the history interview.
'You have told me about the reason why X had to
come to the hospital at this point in time, and about
the kind of problems he has now. I should like now to
ask you about things which happened in the past,
mainly in the last year and maybe even earlier. What
was it that made you aware for the first time ever that
X was not behaving like his usual self? Did other
people notice anything unusual about X's behaviour
around that time, or maybe even earlier than you did?
(Allow the informant to think and reply, then cross-
examine): 'Was there nothing of the sort before that?
Did that happen before or after... (use as a reference
point a fact that the informant has already mentioned,
or an event which should be known locally)'. (Write
a narrative note using the informant's own words to
describe thefirstabnormality that he recollects and its
approximate timing.)
'You told me about some unusual things that X did
or said, which made you think that he was not
behaving like his former self... Did this change
develop suddenly, say within days, or slowly, over a
longer period of time, maybe in weeks or even
months?'
Following the initial data collection period,
all information on the mode of onset was re-
rated by the centre investigators, to achieve
greater uniformity of the criteria applied. The
considerable emphasis given to these variables in
the study design led to much effort being put
into their validation during data collection and
data processing. The results, therefore, can be
regarded with confidence as to their validity,
inspite of the well-known difficulties associated
with the retrospective dating and evaluation of
the onset of psychotic illnesses.
Mode of onset
The onset of the disorder was defined as the
beginning of the first psychotic episode, mani-
fested in the emergence of the following signs
and symptoms.
A. At least one overt psychotic symptom or
sign:
(i) hallucinations or pseudohallucinations
(in any modality);
(ii) delusions;
(iii) thought and speech disorder (incoher-
ence, irrelevance, blocking, neologisms,
incomprehensibility of speech);
(iv) qualitative psychomotor disorder
(negativism, mutism or stupor; cata-
tonic excitement; constrained attitudes
and postures);
(v) bizarre or grossly inappropriate behav-
iour; or
B. the simultaneous presence of two or more
' suggestive' signs or symptoms:
(vi) marked reduction of interests, initiative,
and drive leading to a deterioration of
performance;
(vii) marked social withdrawal;
(viii) severe excitement, purposeless destruc-
tiveness or aggression (frequent episodes
or continuous);
(ix) persistent, pervasive fear or anxiety;
(x) gross self-neglect.
Any of the latter would be regarded as a
prodromal phenomenon, if it appeared in iso-
lation prior to the outbreak of overt psychotic
symptoms.
The frequencies of the different types of onset
by centre are given in Table 2.5. The definitions
of the categories are as follows.
(a) Acute
A florid psychotic state developing within days
(up to a week); mild ('suggestive', non-psy-
 Schizophrenia: A World Health Organization Ten-Country Study 25

chotic) prodromal signs or symptoms may have

i2§ *> g been absent (sudden onset) or present (pre-
cipitous onset).

i: (b) Subacute
Symptoms appearing and developing into a
clear-cut psychotic state over a period of up to
£n one month.
(c) Gradual
NSS "> 8
Slow, incremental development of psychotic
symptoms over a period exceeding one month;
If prodromal signs or symptoms (if any) cannot be
I clearly distinguished from overt psychotic symp-
toms as regards their timing because of a gradual
transition from one to the other.
(d) Insidious
-I No clear demarcation can be made between
I premorbid personality and mental illness, and
aow u-i onset as such cannot be rated; included are also
cases in which no overt psychotic symptoms
were present at the time of examination but the
investigator had a strong suspicion of an
underlying psychotic illness.
In order to simplify the presentation of the
I data, the gradual and the insidious types of
onset have been collapsed. For a total of 78
patients (most of them in Honolulu,
!=2 I 8 Chandigarh/urban area, and Nagasaki) no
— 0 0 OO
reliable information was available on the mode
of onset. Assuming that the cases with missing
data were not significantly different from those
on whom adequate assessment could be made, it
— ( N W-l <N can be concluded that, overall, the gradual/
"S:
insidious type of onset was a more common type
.8 of beginning of psychotic illnesses5 than either
a?5^ I 8 I the acute or the subacute type. It should be
noted, however, that the frequency of the acute
type of onset was not much lower than that of
the gradual onset. The frequency of subacute
onset was less than one half of the frequencies of
each of the two other types.
Table 2.6 shows that there are highly signifi-
cant differences in the frequencies of different
modes of onset in developing and developed
countries. The proportions of the acute and the
3 Sj gradual/insidious types are inverted in the two
o o a

o? 5
Even if all cases with missing values were at one extreme of the
§ o 1 distribution according to the mode of onset, this conclusion would
E e o remain valid because the number of missing cases in all centres
(except for Honolulu) was small.
s<,
26 A. Jablensky and others

100-
90 •

80-
70-
60 •

50-

« •

30-
20
10-

0-
i i • • • i

1 2 4 8 16 32 64 128 256 512

Number of months since onset
FIG. 2.2. Length of illness prior to inclusion in the study (cumulative percentage distribution based on data on 1218 patients).

Table 2.7. Length of illness prior to inclusion into the study (numbers and percentages)
Less than 6 months 6 months and over Total

Developing countries 443 (80-4) 108 (19-6) 551

Developed countries* 374 (79-4) 97 (20-6) 471
Total 817 (75-9) 205 (201) 1022

•Excluding Moscow.

settings (the former predominates in developing
countries and the latter in developed countries)
while the frequencies of the subacute onset are
almost identical.
Length of illness prior to intake
The intervals between the estimated onset and
the point of inclusion into the study, i.e. the
length of previous illness, and the percentages of
cases falling into each interval, are shown in Fig.
2.2.
Patients on whom no adequate information
was available have been excluded from this
graph; this explains the reduction of the total to
1218 cases. The vast majority (86 %) had entered
the study within the first year after the onset of
illness, and in 61 % of the patients the inclusion
into the study had occurred within three months
of the onset of psychotic symptoms. This was
the case in every centre, and in each of the two
groups of centres, in developing and in developed
countries respectively. If six months is arbitrarily
chosen as a cut-off between short- and long-
duration of illness prior to inclusion, then as
Table 2.7 shows, no differences can be found
between the two groups: the proportions of
patients with length of illness less than six
months and with length of illness equal to, or
over six months, are practically identical in
developing and in developed countries.8
Even if we assume that all the cases without a
clear demarcation of onset (i.e. with an insidious
beginning), which were more common in the
developed countries, would fall into the category
of patients with more than six months of
previous illness, their number would not be large
enough to change significantly the distribution
shown in Table 2.7. It can be concluded,
therefore, that patients in developing and in
developed countries were not significantly dif-
ferent from each other at the point of entry into
the study with regard to the length of previous
illness.
6
Moscow was excluded from this tabulation because of a difference
from the other centres in the manner in which onset was dated.
 Schizophrenia: A World Health Organization Ten-Country Study 27

r
Early manifestations of illness
In order to identify and date the earliest mani-
festations of the disorder, the key informant
was asked specific questions about 25 different
behaviours that could occur as a result of an
incipient mental illness. The questions were
worded in lay terms, e.g. ' Did X, at any time in
the past, say that he was being persecuted,
harmed, or bewitched by other people?' The
content of the items was chosen so as to cover a I-4
— o ooo — oo

broad range of behavioural manifestations of

psychotic disorders. 1
The 10 most common manifestations of the
beginning of the illness are listed in Table 2.8 in
the rank order of the frequency with which they
were reported for the total study population, 3
and the percentages of patients in whom such
behaviour had occurred, are given for each a
centre. In spite of the cultural differences, there •S
is a great similarity across the centres in the ways
in which psychotic illnesses present themselves
to local lay observers in the immediate social
environment of the patients. It is noteworthy
•s-g i oo r- r-- ci \D rs t
that 'negative' manifestations, such as neglect
of usual activities and loss of appetite, sleep, or
interest in sex, are even more conspicuous in the •B-s
eyes of the community in most of the centres,
than the more dramatic signs of psychotic Is
disturbance, for example talk of persecution,
harm or bewitchment, or behaving as if hearing
voices. The key informants in the community
who were approached for data appeared to be
il
sensitive enough observers to serve as sources
for case finding in epidemiological surveys of
mental disorders in all the cultural settings in
which the study was done.
s
DIAGNOSTIC CLASSIFICATION OF THE
STUDY POPULATION
I
Issues concerning diagnosis and classification
are given considerable attention in the present .s-8
report because of the continuing debate on what |
constitutes a ' valid' diagnosis of schizophrenia, s o-S-2
and where the boundaries of the diagnostic S 3 -a o . l
concept should be drawn. —« g
3 rt o Si
In the present study diagnostic classification •3 " i, s
E &
was carried out in two different and independent o g J3 £ o h3 o •
ways. First, the investigators in each centre
assigned a clinical diagnosis to each included 8 "§>-^ 3 . S w w (« •
<T2

subject. This was done, in the large majority of 3

28 A. Jablensky and others

cases, after a full assessment of the history and during the four weeks before examination. A list
present mental state (using PPHS and PSE). In of the principal CATEGO classes, including the
a minority of cases (80 in total, 33 of them in constituent syndromes and symptoms, is given
Chandigarh), a PSE interview could not be in Tables 2.9, 2.10, 2.12 and 2.13.
administered for various reasons; however, in A satisfactory classification for use in this
most of these cases clinical notes from the centre study should be: (i) inclusive, in the sense of not
were available and it was possible to make at leaving out cases that could be considered
least a tentative diagnosis. schizophrenic on grounds of any one of the
The diagnostic assessment of each case in the different influential interpretations of the con-
field research centres was recorded in the cept ; and (ii) analytical, in the sense of allowing
Diagnostic and Prognostic Schedule (DPS) in the identification and comparison of subtypes or
terms of: (i) main diagnosis of the condition; (ii) subgroups of cases. For such purposes, a
alternative to the main diagnosis; and (iii) combination of the two methods of classifi-
supplementary diagnosis of any associated cation, i.e. ICD-9 clinical diagnosis and
problem - psychiatric physical, or neurological. CATEGO computer classification was thought
The narrative recording of the diagnosis was to be a suitable approach. Generally, the
done in the terminology normally used in the ' inclusiveness' of the clinical diagnosis was
centre, but the investigators were requested to expected to be wider than that of the CATEGO
enter the 'best fitting' ICD-9 code for each classes designed to represent schizophrenic and
diagnosis, in accordance with the ICD glossary paranoid disorders. However, as it was known
(WHO, 1978). The correspondence between that some centres adhered to a very restrictive
centre diagnosis and ICD code was checked at concept of schizophrenia and would apply non-
Headquarters, and any discrepancies or in- schizophrenic diagnostic categories to a pro-
accuracies were removed. Special 5-digit codes portion of the patients with 'schizophrenic'
were created for the diagnostic categories based CATEGO classes, the minimum diagnostic
on pattern of course which were specific to the criteria adopted for the inclusion of patients into
classification used in Moscow and were not the clinical study were as follows.
available in ICD-9.
The second method of arriving at a diagnostic Criterion I
classification involved the application of the Main clinical diagnosis of the patient made at
CATEGO diagnostic program (Wing et al. 1974) the centre being one of the following:
to the PSE data on all the patients, once their (1) Schizophrenia (ICD 295);
records had been received and checked at (2) Paranoid state (ICD 297);
Headquarters. The CATEGO system produces (3) Acute paranoid reaction (ICD 298.3) or
a reference classification based on standard psychogenic paranoid psychosis (ICD 298.4),
inference rules. The last stage of the program or other and unspecified reactive psychosis
applies hierarchical criteria in order to reach a (298.8);
50-part classification which has been used in (4) Alcoholic hallucinosis (ICD 291.3) or al-
reporting most of the present study. Its main coholic jealousy (ICD 291.5);
advantage is in providing a 'yardstick' for (5) Paranoid and/or hallucinatory states in-
comparisons of psychopathology between dif- duced by drugs (ICD 292.1);
ferent patient populations. (6) Paranoid (ICD 301.0) and schizoid (ICD
It should be noted that the authors of the 301.2) personality disorder;
PSE-CATEGO system do not recommend a (7) Unspecified pychosis (298.9).
straightforward application of diagnostic labels
to CATEGO classes, particularly when history Criterion II
data have not been included. Other techniques In the absence of any one of the above, the
for recording clinical information (e.g. the patient might still be included, if his mental state
Aetiology Schedule) in the CATEGO system is assigned by the CATEGO program to one of
were not used in this project, so that the the following classes:
CATEGO classification of cases reported here is (1) Schizophrenic psychosis (S+ or S?);
based solely on the symptoms or signs present (2) Paranoid psychosis (P+ or P?);
 Table 2.9. Diagnostic distribution of the study population according to ICD-9 clinical diagnosis of the field research centre and
according to CAT EGO class assigned at headquarters
Aar Agr Cal Cha/R Cha/U Dub Hon Iba Mos Nag Not Pra Roc All

Schizophrenia (295) N = 37 76 143 40 118 63 54 129 197 101 68 79 46 1151 Co

1 =S + 22 36 98 13 35 26 28 111 84 54 48 55 24 634
2 = S?, P, O + 7 23 26 7 28 8 4 10 55 15 9 13 9 214
3-
K'
3 = O? 3 11 10 13 18 6 1 2 17 13 3 5 5 107
4 = Other 2 4 9 4 12 12 1 5 40 13 7 6 7 122
5 = No CATEGO 1 — — — 2 3 — 1 1 1 1 — — 10
6 = No PSE 2 2 — 3 23 8 20 — — 5 — — 1 64
Paranoid states (297) N = 3 0 4 0 1 0 2 3 0 2 1 16 0 32
1 =S + — — 4 — 1 — 1 3 — — — 12 — 21
2 = S?,P, O + 1 — — 1 — 2 1 2 7
4 = Other 1 — — — — — — — — — 2 — 3
\
Psychogenic psychosis and unspecified psychosis /
(298.3-298.9) N = 35 2 1 / ~ 4 0 1 0 0 4 4 11 5 69
1 =S + 11 — 1 ' 1 — — — — — — — 8 4 25 3-
2 = S?, P, 0 + 4 2 — 3 — — 1 2 1 — 13 O
3 = O? 3 — — — — — — — — 1 — 1 1 6
4 = Other 7 — — 1 1 — — — — 2 2 1 — 14
6 = No PSE 10 — — — — 1 — — — — — — 11
Alcohol/drug psychosis (291.3, 291.5, 292.1) iV = 9 0 0 0 0 0 2 1 0 0 0 0 0 12 5.
1 = S+ 5 — — — — — — 1 — — — — — 6
2 = S?, P, O + 1 1
4 = Other 2 — — — — — — — — — — — — 2 §
6 = No PSE 1 — — — — — 2 — — — — — — 3
Personality disorder (301.0, 301.2) N = 0 0 0 0 0 0 1 0 0 0 1 0 0 2 0
1 =S + — — — — — — 1 — — — — — — 1 3
2 = S?, P, O + — — — — — — — — — — 1 — — 1 V;
Other ICD diagnosis N = 15 3 6 12 32 4 8 9 0 1 18 1 4 113 00
1 =S + 8 1 2 — 2 1 7 7 — — 10 — 2 40 s?
2 = S?, P, O + 1 2 — 5 11 — — 1 — — 3 1 — 24
3 = O? 1 1 5 6 2 — — — 1 — 16
4 = Other 2 — 3 2 6 — — 1 — 1 4 — 2 21
6 = No PSE 3 — — — 7 1 1 — — — — — — 12
Total 99 81 154 54 155 67 68 142 197 108 92 107 55 1379
30 A. Jablensky and others

CATEGO SPO — Clinical diagnosis

(ICD 295 and other
specified rubrics

Clinical diagnosis ICD 295; 297; 298.

3, 4, 8, 9; 291. 3, 5; 292. 1; 301. 0, 2 1266
Clinical diagnosis, other 113
CategoSPO 1116
Catego S+ 727
F.G. 2.3. Interrelations between clinical diagnosis and CATEGO classification in the total study population (N J . H, Clinical
diagnosis- M CATEGO SPO; B, CATEGO S+ . Thirty-three patients with clinical diagnosis Other did not have CATEGO S +
O T P O and therefore do no fppear in the diagram, although they were included in the total study population of 1379 patients.

(3) Borderline and doubtful psychoses (O + or
O?).
A total of 1379 patients from the 13 catchment
areas were included in the clinical study. Of
them, 1266 had one of the clinical diagnoses
shown above as a main diagnosis and could
have been included on the strength of Criterion
I only. It should be noted, however, that 1036
(82 %) of these patients also met Criterion II. A
total of 113 patients failed to meet Criterion I
but were included either because of meeting
Criterion II (80 cases), or because of other
reasons for considering them eligible after all the
data available had been reviewed at Head-
quarters (33 cases, including 12 without a PSE,
who all showed features suggesting a schizo-
phrenic illness, and had one of the Criterion I
rubrics as an alternative clinical diagnosis).
The distribution of the included cases from
the different centres according to the two
diagnostic systems is shown in Tables 2.9 and
2.10. The relationship of the ICD-9 and the
CATEGO classifications of the cases is illus-
trated in Fig. 2.3, where the sum of the numbers
in the different area of the Venn diagram equals
the total study population of 1379.
Concordance between the two diagnostic
approaches
It should be noted that while in 10 centres
(Aarhus, Agra, Cali, Chandigarh/urban,
Dublin, Honolulu, Moscow, Nagasaki, Prague
and Rochester) the diagnostic group denned by
the ICD rubrics listed under Criterion I was
numerically larger than the CATEGO group
consisting of classes S, P, and O, in the other 3
centres (Chandigarh/rural, Ibadan and Notting-
ham) the CATEGO S, P, O group was larger in
scope than the ICD group. However, in all
centres high proportions of the patients (between
60 and 95%) who fell into one of the ICD
rubrics of Criterion I were also classified as
CATEGO S, P, or O (Fig. 2.4).
The proportions of patients with a clinical
diagnosis of schizophrenia (ICD 295) who were
classified as CATEGO S+ ('nuclear' schizo-
phrenia) are somewhat lower - between 33 % in
Chandigarh (rural area) and 86% in Ibadan
(Fig. 2.4). There were no obvious differences in
that respect between centres in developing
countries and centres in developed countries; in
fact, both the highest and the lowest concordance
 Schizophrenia: A World Health Organization Ten-Country Study 31

rates between a clinical diagnosis of schizo-

•o)
0)
0)
_ 88 8 phrenia and CATEGO classification as S + were
£ Si 25 2™ 8 o w
observed in developing countries.

1151
LIS
The analysis of the 221 discrepant cases, i.e.
1
S those in which there was a disagreement between
»t = © :;s

100 Criterion I and Criterion II, showed that: (a) in

•4)
Os
rs m - 80 patients the ICD main diagnosis was other
c^ ^o so oo c*
than one of the rubrics qualifying for automatic
100

inclusion, but the CATEGO class was S, P, or

14-2
op O; and (b) in 141 patients the ICD diagnosis
s corresponded to the agreed rubrics but the
so — ^
§ CATEGO class was not S, P, or O. A total of
101 cases could not be assessed for concordance
between the two diagnostic approaches, either
(6-!
«?
295-

rs QO T
because of lack of PSE data, or because of too
100

sO *N OO
SO
few positive PSE items to allow a CATEGO
classification. There were also 21 cases in which
the ICD main diagnosis did not qualify for
) (4-0)

»(2-3)
i (0-9)

cs oo a O —
100

rs
?5-6

is
inclusion and the CATEGO class was other
95)

than S, P, or O. As mentioned above these cases

w
as were included with the others on grounds of
100

00 "" rsi 'c

clinicians' 'strong suspicion' of schizophrenic
3-7;
hrei

V,

a.
oN
illness.
The majority of the ICD diagnoses in cases
Schi

rs O
001

oo —
Q which did not meet Criterion I but were included
274 (23-8
(10-9
(40-3

y ^> because of eligible CATEGO class or other

reason, were of affective disorders (66 of the 113
29:
11. Subt
100

patients). Aarhus, Nottingham and Chandigarh,

which contributed 46 of these discrepant cases,
oTG^ OO
rs r*i were the centres where a number of patients with
(28

© --a- as
8 rs m
U fN
so so <N a centre diagnosis of affective psychosis (ICD
^ rs
3 296) were classified by CATEGO as S, P, or O.
Tt O
H On the other hand, in 151 of the cases where the
r3 is m 8 ICD diagnosis corresponded to Criterion I but
;5-2

rs
£ the computer diagnosis was discrepant, the
29^

rs oo
S
oo r- r~ OO O
CATEGO class was D, M, N, or R - i.e. one of
100

SO — the classes describing affective sympomatology.

(0-E

Of these patients, 71 were in Moscow,

(£•

Os^
Chandigarh and Nagasaki. Most of the discrep-
106

$ S 2m o CN 00
ancies were thus related to the diagnostic
100

boundary between the broadly defined group of

schizophrenic illnesses and the affective dis-
27 (2-3)
8(1-6)
19 (2-9)

i
295-0

orders.
100

ICD subtypes of schizophrenia

Within the subset of patients who had a clinical
diagnosis of schizophrenia (ICD 295), the
tries
ries

O
a-' §c
distribution of the diagnostic subtypes snowed
c/l crt O 8§ great variation from centre to centre. For all
OOO _ centres combined, each of the ten ICD subtype
OOO '£&
73
o
o o
•313
s rubrics had been assigned to no less than 25
<<< 5 o H patients, but the frequency of the use of
uuu
OZ
II
32 A. Jablensky and others

Aarhus - ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ H

Agra J ^ ^ ^ ^ l ^ ^ ^ l
Cali -HHM^HH
Chandigarh/R ~~H|HjHj^^H
Chandigarh/U -fl|^|^^H|flHl WmEWYs":\ i
Dublin -^^^^^^^^H ^•1 J
Honolulu - ^ ^ ^ ^ H ^ ^ ^ ^ H

Ibadan -fl^^^^J^^^^H HHBK

Moscow - ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ f l • ' 1
Nagasaki -^^^^^^^^^^^^M
• "i
Nottingham ~ H | | ^ ^ H | j ^ ^ ^ |

^^BBM
Prague - ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ H 1 i
Rochester -^^^^^^^^^^H — | I
1 i i i 1
20 40 60 80 100 120
Percentage

Aarhus ~fl^^^|H •••1

Agra ~fl^H|H ^ M I |
Cali - H U HHHMT~;: :
Chandigarh/R ~ H ^ ^ B |

Chandigarh/U - J ^ B M

Dublin -JI^H^I
Honolulu - ^ ^ ^ ^ ^ ^ ^ H

Ibadan -^^^^^^H

Moscow - ^ ^ ^ ^ ^ ^ ^ H
IBi 1 i
Nagasaki - ^ ^ ^ ^ ^ ^ ^ H ••;:;,:;:
Nottingham -^^^^^^H
MB 1 1
Prague -^^^^^^H
^^^^^^^^
Rochester - - ^ ^ ^ ^ ^ ^ H
1 1 1 1
20 40 60 80 100 120
Percentage
FIG. 2.4. (a) ICD/CATEGO diagnostic concordance. All patients meeting ICD inclusion criteria and who were administered a PSE.
(b) ICD/CATEGO diagnostic concordance. All patients with ICD diagnosis of schizophrenia (295) and who were administered a
PSE. • , S+ patients; • , SPO not S+ ; U, other CATEGO.
 Schizophrenia: A World Health Organization Ten-Country Study
20
FIG. 2.4. (c) ICD/CATEGO diagnostic concordance. All patients meeting ICD inclusion criteria. • , SPO patients; M, other
CATEGO; 13, no PSE.
individual subtype rubrics varied from 0 to 65 %
of the cases in the different centres (Table 2.11).
In the study population as a whole, the largest
subgroup (N = 322) was that of paranoid schizo-
phrenia (295.3), followed (JV=273) by that
of acute schizophrenic episode (295.4). In the
developing countries, however, the latter subtype
was diagnosed almost twice as often (in 40 % of
the cases) as paranoid schizophrenia (in 23 % of
the cases). Catatonic schizophrenia (295.2) was
diagnosed in 52 cases (10%) of the study
population in the developing countries, but only
in a negligible number of cases in the developed
countries. In contrast, the hebephrenic subtype
(295.1) was diagnosed in 13 % of the patients in
developed countries and in only 4 % of the
patients in developing countries. Simple schizo-
phrenia (295.0), latent schizophrenia (295.5),
and residual schizophrenia (295.6) were rarely
diagnosed in either developing or developed
countries. More surprising and difficult to
explain, however, is the rare use, in either
developing (3 %) or developed (8 %) countries,
of the rubric of the schizoaffective subtype.
33
40 60 80 100 120
Percentage
SYMPTOMATOLOGY AND
BEHAVIOURAL MANIFESTATIONS
The symptomatology of the cases included in
the study was analysed, first in terms of
CATEGO subclasses (there are 50 such sub-
classes, each composed of one or more syn-
dromes (see Tables 2.9, 2.10, 2.12 and 2.13)
and, secondly, in terms of individual PSE
symptoms.
CATEGO subclasses
The most frequent CATEGO subclasses and
their rank order in the total population and in
each centre are listed in Table 2.12. In the
different centres, the mental state of 81-100 % of
the patients (90 % for the total sample) could be
classified under one of the 13 most frequent
CATEGO subclasses. In all centres, except
Chandigarh (rural area), the patients with the
syndrome of' nuclear' schizophrenia (subclasses
NS + ) represent the largest single subgroup,
comprising between 22% (Chandigarh/urban
area) and 82% (Ibadan) of the cases. In
comparison, the remaining 12 CATEGO sub-
 Table 2.12. Rank order and frequency {percentage, in brackets) of CATEGO syndromes in the total sample of patients meeting
the inclusion criteria who had a PSE assessment at intake {N = 1288)
CATEGO Aar Agr Cal Cha/R Cha/U Dub Hon Iba Mos Nag Not Pra Roc All
syndrome iV=82 AT = 79 N= 154 AT = 5 1 N= 125 N= 58 Af = 44 N= 142 Af= 197 N= 103 Af = 92 N= 107 AT = 5 4 N= 1288

NS + 1(46) 1(41) 1(54) 2(16) 1(22) 1(36) 1(75) 1(82) 1(35) 1(47) 1(46) 1(53) 1(43) 1(46)
DP? 7(4) 4(6) 6(3) 5(8) 2(13) 4(7) 4(5) 2(2) 4(7) 2(8) 2(9) 2(8) 0 2(6)
PD + 2(7) (0) 6(3) (0) 7(4) 2(10) (0) (0) 3(13) 4(5) 4(7) 5(6) 4(6) 3(5)
UP? 13(1) 3(8) 4(5) 1(20) 4(8) 2(10) (0) 6(1) 8(3) 3(7) 6(4) 6(4) 2(7) 3(5)
DP + 3(5) 12(1) 6(3) 6(6) 6(5) (0) (0) 6(1) 2(15) 5(4) 8(3) (0) 4(6) 3(5)
DS + 7(2) 12(1) 3(7) 4(10) 9(3) 4(7) (0) 2(2) 8(3) 13(1) 3(8) 9(2) 4(6) 6(4)
DP?/AP? 3(5) 3(8) (0) 7(4) 3(10) 6(4) (0) 2(2) 11(1) 13(1) 10(2) 6(4) 7(4) 7(3)
NSPD 7(4) 7(3) 4(5) (0) 20(1) 10(2) 2(9) (0) 11(1) 7(2) 5(5) 2(8) 7(4) 7(3)
MN + 3(5) 12(1) 9(2) 7(4) 12(2) 8(3) 5(2) 2(2) 6(5) 13(1) 8(3) (0) 2(7) 7(3)
CS + 7(4) 2(16) (0) 7(4) 9(3) (0) (0) 6(1) 10(2) 7(2) 12(1) 8(3) 7(4) 7(3)
NS? 7(2) 7(3) 9(2) (0) 7(5) (0) (0) 6(1) 7(4) 13(1) 6(4) 4(7) 7(4) 7(3)
DS? (0) (0) 2(11) 12(2) 12(2) 10(2) 3(7) 6(1) (0) 7(2) (0) 10(1) (0) 12(2)
SS 13(1) 7(3) (0) 3(10) 5(6) (0) 5(2) (0) 5(5) 7(2) (0) (0) (0) 12(2)
Percentage cases (86) (91) (95) (84) (84) (81) (100) (95) (94) (83) (92) (96) (91) (90)
presenting with the 1
13 top ranking
syndromes
1
o

Table 2.13. Rank order and frequency {percentage, in brackets) of CATEGO syndromes in patients allocated to CATEGO
class S+{N= 727)
CATEGO syndrome Aar Agr Cal Cha/R Cha/U Dub Hon Iba Mos Nag Not Pra Roc All

NS + 1(83) 1(87) 1(79) 1(57) 1(71) 1(77) 1(89) 1(96) 1(82) 1(88) 1(72) 1(76) 1(76) 1(82)
DS + 3(4) 4(3) 2(19) 2(36) 3(10) 2(15) (0) 2(2) 3(7) 4(2) 2(12) 4(3) 2(10) 2(7)
NSPD 2(7) 2(5) 3(7) (0) 4(3) 3(4) 2(11) (0) 4(3) 2(4) 3(9) 3(12) 3(7) 3(5)
NS? 3(4) 2(5) 4(3) (0) 2(13) (0) (0) 3(1) 2(8) 2(4) 4(7) 3(9) 3(7) 4(5)
NSMN/DS 5(2) (0) 5(1) 3(7) 4(3) 3(4) (0) 3(1) (0) 4(2) (0) (0) (0) 5(1)
Total (100) (100) (100) (100) (100) (100) (100) (100) (100) (100) (100) (100) (100) (100)

Legend: NS = nuclear schizophrenia; DS = schizophrenia without first rank symptoms; PD = psychotic depression; MN = mania or mixed affective psychosis.
(In combinations of more than two letters, the first pair indicates which type of symptom was predominant. The symbols + and ? refer to different degrees of certainty in syndrome
identification).
 Schizophrenia: A World Health Organization Ten-Country Study 35

classes include relatively small numbers of Table 2.14. PSE symptoms (44) chosen to
patients, except for 'possible borderline psy- construct psychopathology profiles of subgroups
chosis' (UP?) in Chandigarh/rural area (20%); of patients
catatonic schizophrenia (CS + ) in Agra (16%);
paranoid psychosis (DP + and DP?) in Moscow Affect Delusions
(22%) and in Chandigarh/urban area (18%); 23 Depressed mood •71 Control
27 Morning depression 72 Reference
and psychotic depression (PD + ) in Moscow (rating 2 only) 73 Delusional
(13%). misinterpretation
Among the 727 patients who were assigned 37 Early waking 74 Persecution
41 Expansive mood 75 Assistance
CATEGO class S + , the majority (82%) exhi- 42 Ideomotor pressure 76 Grandiose abilities
bited the syndrome of nuclear schizophrenia 43 Grandiose ideas and actions 77 Grandiose identity
78 Religious
(NS +), defined by the presence of one or more Subjective thought disorder 79 Paranormal
of Schneider's first-rank symptoms (Table 2.13). •55 Thought insertion 80 Physical forces
*56 Thought broadcast •81 Alien forces
PSE symptoms •57 Thought echo •82 Primary delusions
*58 Thought withdrawal 83 Subcultural
At the level of individual symptoms, those which 59 Thoughts being read 84 Morbid jealousy
ranked highest in frequency by being present in 49 Delusional mood 86 Sexual
87 Fantastic
the majority of the patients, were primarily non- Hallucinations 88 Guilt
specific symptoms indicative of a generally *62 Voices in third person 89 Appearance
disturbed mental state, e.g. restlessness, poor 63 Voices speaking to subject 90 Depersonalization
64 Dissociative hallucinations 91 Hypochondriacal
concentration, social withdrawal, or subjective 66 Visual hallucinations 92 Catastrophe
feeling of nervous tension. In order to obtain a 68 Olfactory hallucinations 93 Systematization of
delusions
clearer description of the diagnostically signifi- 69 Delusion of smell 94 Evasiveness
cant aspects of the mental state of the study 70 Other hallucinations 95 Preoccupation with
population, such non-specific symptoms were delusions or
hallucinations
removed from the analysis and a total of 44 PSE 96 Acting out of delusions
symptoms (listed in Table 2.14) were selected to
construct a psychopathology profile for each of •Symptom automatically allocating the patient to CATEGO class
S + . In the absence of any of these, the case may still be classified as
the various subgroups of the study population, S + if symptom 70 is present together with any one of 72-81 or
using the percentage of patients with positive 84-92.
rating (1 or 2) as a score on a given symptom.
The 44 PSE symptoms were supplemented by a among patients in developed countries. This,
selection of 9 behavioural items taken from the however, is true only for symptoms representing
PPHS, which give a concise description of the subjectively experienced thought disorder
main abnormalities in the early stage of the (thought insertion or thought broadcast) or
disorder as perceived by key informants. changed quality of the experience of reality (e.g.
Together, the 44 PSE symptoms and 9 PPHS primary delusions), but not for symptoms such
items correspond to most of the descriptive as characteristic hallucinations (voices discussing
items contained in recent lists of operational the subject in the third person) or delusions of
diagnostic criteria, including DSM-III (APA, control, which are either more common in
1980). patients in developing countries, or show the
The symptomatological profiles of all included same frequency in the two settings. This finding
patients are shown on Fig. 2.5, separately for the should recall Schneider's (1959) view that first
centres in developing countries and the centres rank symptoms are a heterogeneous group as
in developed countries. The two profiles are regards their causation. Although it is tempting
generally similar, but there are three aspects in to consider cultural variation (including
which the differences should be noted. language) as a possible explanation of such
First, the frequency of affective symptoms, differences, no definitive interpretation could be
and especially of depression, is higher among the given of this observation in the absence of
patients in developed countries. knowledge concerning the pathophysiological
Secondly, Schneiderian first-rank symptoms basis of first-rank symptoms.
also appear with somewhat greater frequency Thirdly, patients in developing countries score
36 A. Jablensky and others

00
95
90
85
80
75 / )
70
65 /\\
60 1/ <\
55 \\
50
45
k A ' \
40 1
35 1

30
i\
i \ / ,
,'• A
A ( >' 1
A 1 '< 1
A
/\ ;
25
20
15'
\1 //V \' •A' // \\/;/'
\ i \ \i A / \\ t
10 'TV I -\\ / V U-si /^v u /V4 1
5 \\f
0
i i i i i rr pp i i ii ii ii ii II i n i i i r p i i i i i i i i i i i i i i i i i i i i i i 11 i
23 27 37 41 42 43J55 56 57 58 59 62 71 8ll63 64 66 68 69J49 70 72 73 74 75 76 77 78 79 80 82*83 84 86 87 88 89 90 91 92 93 95 96,94
i ' i i
, i | ^ ^ i

Affective First-rank Hallucinations Delusions

symptoms
Symptom
FIG. 2.5. Profiles on 44 selected PSE items of 551 patients in developing countries ( ) and 737 patients in developed countries
( ), all meeting 'broad' diagnostic criteria for schizophrenia and related disorders. *, Primary delusion is a first-rank symptom
but is included among delusions.

80 The two groups of patients differ very little

with regard to the frequency of various kinds of
70
delusions.
Schneiderian first-rank symptoms have often
ICATEGOS?, P+, P?, O+ been in the centre of debate in recent years, in
60 view of their potential value for the development
of explicit diagnostic criteria for the schizo-
50 phrenic disorders (e.g. Carpenter et al. 1973;
Koehler, 1979; Hoenig, 1984; Marneros, 1984).
For this reason, the characteristics of the large
40 subset of the study population (727 patients),
assigned to CATEGO class S +, are of special
30 interest. Class S + is composed mainly, but not
exclusively, of patients who manifest one or
more out of eight specified PSE symptoms:
20 - thought insertion, thought broadcast, thought
echo or commentary, thought block or with-
10 - drawal, voices discussing the patient in third
person, delusions of control, delusions of alien
Total study population "
I I I L
forces penetrating and controlling the mind or
0 1 2 3 4 5 6+
body, and primary delusions.
Number of first-rank symptoms present The frequency of these symptoms in the total
FIG. 2.6. Percentage of patients with different numbers of study population shows a J-shaped distribution
Schneiderian first-rank symptoms in three diagnostic groups. (Fig. 2.6), with 36% of the cases manifesting no
first-rank symptom, 15% having one, and
higher than patients in developed countries on diminishing proportions displaying increasing
auditory hallucinations (other than the first- numbers of such Schneiderian symptoms. In the
rank type) and, particularly, on visual hallucin- CATEGO S + group of patients, the frequency
ations. of the first-rank symptoms shows a skewed bell-
 Schizophrenia: A World Health Organization Ten-Country Study 37

100
95
90
85- / ^,

80- 1 1

75- J 'l
I UIO

70-
65-
Cu 60-
i
i
/\ >
j 1
sym

55- JI

50-
45-
i i
',/ Y
1 /' \ '• i

s 40- I i ii \
35- (
tients

30-
25-
A
1 X 1
\/ 1 A /\ I \
1 1 I
20- >i
<2 15-
\ \I
10-
V \
' \—
0-
_yv\ n i i i i i i i i
23 27 37 41 42 43J55 56 57 58 59 62 71 8lj63 64 66 68 69|49 70 72 73 74 75 76 77 78 79 80 82*83 84 86 87 88 89 90 91 92 93 95 96J94
1 i i i i r y

Affective First-rank Hallucinations Delusions

symptoms
Sympton
FIG. 2.7. Profiles of 727 CATEGO S + patients ( ) and of 389 CATEGO S?PO patients ( ) on 44 selected PSE symptoms
(all centres). *, Primary delusion included among the delusions although a first-rank symptom.

shaped distribution, with 7% of the patients that, rather than being a symptomatologically
manifesting none, 51 % having one or two; 27 % discrete group, the S+ patients are merely the
having three or four; 8 % having five; and 5 % more severely ill subjects (in the sense of having
having six or more such symptoms. more psychotic symptoms) among the total
A comparison of the 44-symptom profiles of patient population defined by the inclusion
the CATEGO S + patients and the rest of the criteria. The presence of Schneiderian first-rank
patients falling within the other classes repre- symptoms, therefore, appears to be an indication
senting schizophrenic disorders (i.e. S?, P, and of a florid psychotic mental state characterized
O) demonstrates two important differences by multiple 'positive' disturbances in different
between these major subsets of the study areas of psychopathology.
population (Fig. 2.7). The cross-cultural differences within the group
The first difference is of a qualitative nature of CATEGO S+ patients (presented in a
and stems from the definitions of the two subsets simplified way as differences between patients in
in the CATEGO program. Patients in the S?PO developing and in developed countries - Fig.
group score 0 on 7 of the 8 symptoms defining 2.8) are of the same type as the differences
'nuclear' schizophrenia (indicated by asterisk) between the total samples of included patients in
simply because cases with positive scores are the two settings: S+ patients in developed
automatically allocated to S +. countries score higher on affective symptoms
The second difference, however, is a quan- and on most of the first rank symptoms, while
titative one and concerns the magnitude of the patients in developing countries have higher
scores on the 44 selected PSE symptoms in the scores on auditory and visual hallucinations.
two groups. With the exception of the affective The two groups are almost indistinguishable in
symptoms, which occur with the same frequency the area of delusions.
in S+ and non-S-|- patients, the S+ patients
score higher on almost every psychotic symptom. Behaviour profiles
The two symptom profiles tend to be parallel in The above observation about the psychopatho-
shape and, except in the area of affective logical differences between the CATEGO S +
disturbances, do not intersect. This suggests and CATEGO S?, P, O subgroups of patients is
38 A. Jablensky and others

100
95
90
85
~ 80

O 65
g- 60
fl 55
™ 50
« 45
S 40

g 30

^ 20
15
10
5
0
1 I I I I 1. 1 I I i i 11 i I i I l I I I I l l I I I I I T T
23 27 37 41 42 431 55 56 57 58 59 62 71 8l! 63 64 66 68496970 72 73 74 75 76 77 78 79 80 82*83 84 86 87 88 89 90 91 92 93 95 96] 94

Affective First-rank Hallucinations Delusions

symptoms
Sympton
FIG. 2.8. Profiles of CATEGO S+ patients in developing ( ) countries (JV=316) and in developed ( ) countries
(N = 411) on 44 selected PSE items. *, Primary delusions included among delusions although a first-rank symptom.

to be ' negative' manifestations (neglect of usual

activities, loss of interest in own appearance and
cleanliness, avoidance of other people), the S +
patients were considerably more often perceived
by the key informants as showing behaviour
recognized as psychotic in most cultures: hearing
voices, feeling persecuted, bewitched, or other-
wise harmed, claiming impossibilities, or per-
sistently complaining of strange bodily sen-
sations.

1 2 3 4 5 6 7 BACKGROUND AND ANTECEDENT

Symptom number
FIG. 2.9. Behaviour profiles (selected PPHS items) of patients during
early stage of illness. (1, Neglect of usual activities; 2, loss of interest
in own appearance and cleanliness; 3, excitement; 4, sad, hopeless
look; 5, avoidance of other people; 6, behaves as if hearing voices; 7,
feels persecuted, bewitched; 8, claims impossibilities; 9, complains
of strange sensations; , CATEGO S + ; , remainder of
patients.)
corroborated by the comparison of PPHS
behaviour profiles of the CATEGO S + patients
and the rest of the included patients (Fig. 2.9).
These profiles are constructed from 9 history
items assessed in key informant interviews which
were totally independent of the PSE assessment.
The graph shows that while the two groups did
not differ in respect of what could be construed
FACTORS
Mental illness in the family
The collection of genetic data was not among
the primary goals of the study. Nevertheless, the
history interview with a key informant included
a number of detailed questions about the
occurrence of mental disorders in first-degree
relatives, and every attempt was made to record
the information elicited in terms of six cate-
gories: psychosis; neurosis or personality dis-
order; chronic alcohol or drug abuse; mental
subnormality; other or unspecified mental dis-
order; and no mental disorder. Whenever
positive data emerged, the investigator was
required to make a narrative note, giving a
 Schizophrenia: A World Health Organization Ten-Country Study 39

description of the symptoms, to the extent that

was possible. There could be no certainty as to + + o

All

All
00 <N t> oo
(N rs
the completeness of ascertainment of secondary o
cases in all the centres and the results of this O
S o
enquiry should be treated with caution. ^^ U,
(N
o
o£
©
rs rs
Out of all the cases meeting the minimum of

oup: C
diagnostic criteria for inclusion, mental health
a

Pra
oo

Pra
- OO
rn rs
problems in first-degree relatives were reported
in 362 cases (27 % of the 1321 patients who were bo o
.« bb
assessed with the PPHS). Relatives with mental o —, rs
"o eg
00

disorder were identified in a larger proportion of o Z •~

s;
cases in the centres in developed countries (32 %) bo

diagn
Nag

Nag
than in developing countries (21 %), a difference |
=
00 ©

which is likely to be due to a lower probability

of reporting secondary cases in the developing
o o rs
countries. 1 1
Of the 128 fathers with reported mental 1
health problems, in 39% the problem was

hik
r-
1
Iba

Iba
<s ©
psychotic illness; in 28% alcohol or drug

ion
dependence; and in 20 % neurosis or personality •S ion •S
disorder. Of the 126 mothers affected, psychosis co
3 -2 g
1
o oo ro m m
H
/ problei

was recorded in 51 % and neurosis or personality

disorder in 25 %. A total of 182 patients reported 1 -Si
•o
o
X
Dub

having siblings with a history of mental disorder, (N

rs
r-
a. Q
a\ oo rs

and 82 of them had one or more siblings s "5

s
suffering from psychotic disorder.
1
otio
•2 D

a/U
00 m ON rs
The comparison between CATEGO S +
patients and all the remaining patients meeting 1 U

OL
1 6
the diagnostic criteria of inclusion (the 'non- a\
re.

rs -o
S + ' cases) failed to demonstrate any difference
in the strength of vertical transmission between
3
^3
U 1 rs

•s
CO)
'CO,

the two groups. The only feature which distin- rs

s
Cal

Cal

guishes the S + patients is that they tend more <5^

1 r-
IU
tN rs

often to have a psychotic mother (in 52 % of the

cases where at least one parent was ill, compared > __ 00 VO

•a
to 36 % in the ' non-S + ' cases). This difference, -s: •a
however, did not reach statistical significance. -s:
-s
Aai

C^ O o
The available data, therefore, do not show for rs

CATEGO S+ patients a frequency of family •2

'I
history of psychiatric disorder different from
t
riteria

riteria
subj

that in the rest of the study population.

u u

Adjustment problems in childhood and .2 ^^^^ c

o
lus

adolescence bp
^^
3
^. o
c Jr-Q •00— 1
The history interview (PPHS) contained check- V)
bO
II c
lists of 17-items each, for rating specific behavi- Id
D
oural and adjustment problems in childhood 6 rs

and adolescence. In addition, an 'overall im- II 1 II

pression ' rating was made on the basis of all the + +
information available. An exploratory analysis
CN 00
C3
• c ^--
able

Alln;maini

able

1g a «1
EGO
CAT EGO

of this information was carried out with full »s

All rs

All p
(AT

acknowledgement of the limitations inherent in H <t.

I.
retrospective assessment of behaviour.
40 A. Jablensky and others
In total, 18 % of the patients were rated as
having had conduct or emotional problems in
childhood, and 25 % were rated as having had
such problems in adolescence. However, there
were varying percentages of cases on whom no
sufficient information was obtained and this
precluded direct comparisons between the
centres. Nevertheless there was a slight tendency
for CATEGO S + cases to have higher rates of
such disturbances (Tables 2.15 and 2.16). For
adjustment problems in childhood, the rate for
S+ is higher in 8 out of the 13 patient samples;
for problems in adolescence, it is higher in 7 out
of 12 samples with complete data sets. In order
to see whether such differences were related
to any specific disturbances in childhood and
adolescence, the rank order and percentage
scores were determined for all the items included
in the two checklists.
Traits such as 'extreme shyness', 'preference
for playing alone' and being a 'model child'
were the three most common items for childhood
problems in the study population as a whole,
appearing in 10-19 % of the patients. CATEGO
S + patients had higher scores than ' non-S + '
subjects on two items: 'marked fears' (11%
compared to 4 % in' non-S + ' cases) and' unable
to sit still, restless' (6% versus 2%). On all the
other items the scores of the two diagnostic
groups were very similar.
As regards adolescence, the most frequent
items in the total study population were: 'very
reserved and quiet' (30%); 'very sensitive'
(22%); and ' suspiciousness' (10%). On the
latter two items, the scores of S + subjects were
higher than the scores of 'non-S+' subjects:
31 % v. 21 %, and 13% v. 5%.
History of convulsions
A total of 36 cases had a definite history of
convulsions in childhood; in 20 of them these
were described as repeated febrile convulsions
while in 14 seizures had occurred without any
association with a febrile illness. It is noteworthy
that 13 of the febrile convulsion cases, and 13 of
the non-febrile seizure cases, were CATEGO
S + . History of seizures after the age of 10 was
reported in 26 cases; 18 of them were also
CATEGO S + . Of the 4 patients in whom a
history of seizures during the year preceding
inclusion in the study was revealed, 3 turned out
to be CATEGO S + . CATEGO S+ patients,
therefore, were 2 to 3 times more likely to have
a history of convulsive disorder than the rest of
the study patients (P < 0-01). This unexpected
finding is difficult to explain; no references to an
association between convulsions in childhood
and first-rank symptoms of schizophrenia were
found in the literature.
Pre-morbid personality traits
A total of 32 traits were rated as either present
or absent in the history interview (PPHS) on the
basis of the informant's description. In order to
investigate the possibility that these traits may
represent an underlying structure involving a
smaller number of personality types, factor
analysis was performed on the 32 individual
traits. The most satisfactory solution was
obtained using a principal components analysis,
retaining five factors which together accounted
for 44% of total variation (Table 2.17). Every
trait loaded on at least one factor, with factor
loadings 0-3 and above. Six of the 32 variables
loaded on more than one factor. The factors
were clearly mutually independent and the
rotated factor pattern seems to correspond to
recognizable personality types. The first and
fourth factors, Fl and F4, correspond to the
conventional description of the paranoid and
schizoid personality types. Factor 3 could be
seen as describing a mildly eccentric and passive
type of character while factors 2 and 5 refer to
rather idealized versions of positive' pre-morbid
personality traits frequently given by informants,
an adult counterpart of the 'model child'.
In order to investigate the possible link
between pre-morbid personality and diagnosis,
five factor-based scales were constructed. The
number of component traits which a subject
manifested was the score on a given scale. The
association between these scores and assignment
to CATEGO S+ or to the non-S + group was
found to be significant for each of the individual
factors except for factor 4 (' schizoid' type). This
was in spite of the large numbers of patients who
scored zero on one or more factors (for 23 % of
the patients no pre-morbid personality trait was
rated at all, and 71 % had no more than five
traits recorded as present).
Patients who scored high on any of the factor
scales were more likely to be included in the S +
than the non-S + group. A logistic regression
was conducted to see in which way member-
 Schizophrenia: A World Health Organization Ten-Country Study 41

Table 2.17. Pre-morbid personality traits: results offactor analysis

Rotated factor pattern

Fl F2 F3 F4 F5

2 = Picked on 0-45 002 -008 0-43 0-24

3 = Jealous 0-45 003 0-17 011 0-26
4 = No self criticism 0-60 018 001 0-22 -015
5 = Pessimistic 0-38 014 016 0-30 0-17
7 = Changeable moods 0-51 011 010 012 0-06
11 = Loses temper 0-67 004 003 011 002
12 = Shouts, argues 0-72 006 003 -004 009
13 = Anger with violence 0-59 007 0-24 007 004
14 = Exaggerated feelings 0-63 -001 014 -003 0-24
16 = Likes audience 0-41 003 -004 -015 0-33
20 = Hurt others 0-56 004 000 011 -012
22 = Violent, no anger 0-51 007 012 -002 0-16
28 = Confabulates 0-48 001 0-28 004 0-22
10 = Shows feelings 015 0-49 008 006 0-30
21 = Shows sympathy 005 0-69 005 011 014
29 = Optimistic -001 0-67 010 -011 009
30 = Endures stress 009 0-53 008 006 006
31 = Loyal, reliable 002 0-80 007 009 009
32 = Autonomous decisions Oil 0-76 -007 004 -003
15 = Clinging 014 010 0-73 011 018
17 = Easily dominated -002 013 0-69 006 013
18 = Needs pushing 018 001 0-70 016 -006
19 = Cannot cope 0-38 002 0-47 0-30 -003
26 = Eccentric 0-27 010 0-37 015 004
1 = Suspicious 0-45 009 -003 0-50 011
8 = Prefers to be alone 000 -013 0-22 0-68 001
9 = Hard to find friends 0-04 003 0-26 0-65 010
27 = Cool and withdrawn 015 0-25 011 0-59 006
6 = Excitable, energetic 0-21 0-35 011 0-18 0-41
23 = Too tidy 002 0-21 009 013 0-64
24 = Too high standards 011 0-20 014 007 0-60
25 = Fixed habits 013 003 -002 016 0-67

Principal components factor analysis, 5 factors retained, varimax rotation.

Total variances explained: 44%.
Final communality estimates: 0-33-0-50.

ship of the S + or non-S + group could be
predicted on the basis of the five scales. This
analysis showed that, when all factors were
considered together, only the second factor was
significant, i.e. the 'positive' personality type,
This was backed by the results of a second
logistic regression using all the 32 pre-morbid
personality traits together to investigate the
association between pre-morbid personality and
membership of the S + group. The conclusion
from this analysis was that positive pre-morbid
personality traits tend to be associated with the
development of first-rank symptoms.
Medical problems in the past year
A total of 25% of the study patients were
reported to have had somatic medical problems
in the year preceding intake into the study. The
percentage was highest in Agra and Ibadan. The
problems described were varied and lacked any
specificity, ranging from unexplained fevers to
parasitic and nutritional disease.
Alcohol and drug use
A history of alcohol use in the year preceding
intake was given for a total of 57 % of the male
patients. A definite or suspected alcohol problem
was reported on high proportions of the study
subjects in Aarhus (26 %), Dublin (24 %),
Nottingham (18 %) and Rochester (18 %). Drug
abuse was described in a lower proportion of the
t o t a l s t u d y p o p u i a t i o n ( 14 o /o ) b u t t h e pro blem
was heavily concentrated in a few centres:
Honolulu (41 %), Rochester (36 %), and Aarhus
42 A. Jablensky and others

(24 %). The extent to which abuse of street drugs disorder, could not be determined at this stage,
(mostly marijuana and cocaine) may have An analysis of the follow-up of these cases
contributed to the onset of psychotic illness, or, (chapter 4) contributed further information on
conversely, may have been a sequel of the this problem.
 Chapter 3 Incidence in different geographical areas
One of the chief aims of the study was to collect
series of cases of recent inception, which would
allow an estimation to be made of the rates of
incidence of schizophrenic disorders in the
different catchment areas. In contrast to the
large number of psychopathological, course and
outcome, diagnostic, treatment, and family
genetic studies, there have been relatively few
large-scale epidemiological investigations on
schizophrenia, and only a handful of them have
addressed the issue of incidence. An overview of
the design and methods of a selection of such
studies in given in Table 1.1.
An essential step in the epidemiological study
of any disease is the determination of the
frequency of its occurrence in the two sexes and
in different age groups, of the geographical
variation in its frequency, and of its associations
with other disorders and environmental factors.
Incidence, i.e. the number of new cases occurring
in a given population over a specified period per
1000, 10000, or 100000 persons at risk, is a
particularly valuable index in the study of
schizophrenia. Incidence rates are better suited
than prevalence rates for comparisons between
different populations, because they are less
affected by differential mortality, migration, and
other demographic factors. The study of series
of patients of recent onset is important also in
view of the possibility that pathogenetic or
triggering factors which are active in the period
preceding thefirstmanifestations of the disorder
may cease to operate at later stages of its
evolution. Age- and sex-specific incidence rates,
in comparison to prevalence rates, offer a better
basis for the estimation of morbid risk (or
disease expectancy), an index expressing the
probability of developing a disease, provided
that a given individual survives until a specified
age.
METHODS OF ESTIMATION OF
INCIDENCE AND MORBID RISK
The screening and case finding procedures
employed in the field research centres, and the
43
steps taken to ensure that no great numbers of
eligible cases are missed, are described in chapter
1 of this report. Chapter 2 contains an analysis
of the sociodemographic, clinical and diagnostic
features of the included cases and outlines the
rationale for presenting the data both in terms
of a 'broad' diagnostic definition (CATEGO
classes S, P, and O, or one of the above
mentioned ICD clinical diagnoses), and in terms
of a more specific and restrictive criterion,
CATEGO class S + .
In determining incidence, another two group-
ings were added to these alternative case
definitions: (i) the cases falling into classes S, P,
or O only, excluding patients who had a
CATEGO class different from these, regardless
of the clinical diagnosis; and (ii) the cases falling
into classes S, P, or O + only, excluding the O?
class which may be assigned to 'borderline' or
doubtful clinical pictures 1. Separate rates were
calculated for these four levels of clinical and
diagnostic specificity.
As pointed out in chapter 1, the analysis of
incidence is based on data from the eight
catchment areas in which fairly complete cover-
age was achieved of the various 'helping
agencies' that were likely to serve as first-contact
sites for psychotic patients. In six of these areas
it was possible to explore, after the end of data
collection, whether any 'leakage' of potential
cases had occurred (see Table 1.5). The results of
this additional enquiry indicated that few eligible
patients had been missed. In the other two areas,
a retrospective search for missed cases has not
taken place, but, following site visits during
which Headquarters and centre investigators
reviewed in detail the casefindingprocedures, it
was considered unlikely that a significant num-
ber of patients would have remained there
undetected.
The data from the remaining five areas, in
which the case finding was not complete, or
where doubts existed about the population bases,
were also treated as rates for the purpose of
comparison. Because of the incomplete coverage,
these rates were expected to be lower than the
44 A. Jablensky and others

rates in the first group of areas. These rates were ms = annual incidence for males aged 50-
indeed considerably lower than the rates found 54,
in the first group of centres.
The population data which provided the then the morbid risk for females is:
denominator values were taken from the census 5(7^ +/ 2 +... +/ 8 ) and for males:
data for the year that was closest to the period
of casefindingin each centre. This corresponded
to the 1980 round of censuses in most of the Effectiveness of the case-finding methods
centres, for which detailed age- and sex- A question which arises in connection with the
distributions of small areas had yet to be data reported below is: to what an extent could
published. The denominator values used in the the method of case-finding employed in this
present publication were therefore usually pro- study result in a valid approximation of the
vided directly by the appropriate national census 'true' incidence of schizophrenia and related
office which performed any necessary extrapo- disorders?
lations or estimations to cover the period of Deciding what constitutes the ' true' incidence
case-finding 1978-80. Where detailed age and of schizophrenia is bound to involve a good deal
sex distribution was not available (e.g. for the of arbitrariness, unless a pathophysiological
rural area in Chandigarh), estimates were made marker of disease onset is available. In the
by a professional demographer on the basis of absence of such a marker, the ascertainment of
national figures. Adjustments also had to be onset on the basis of clinical history will be
made in the case of Moscow, where the youngest unreliable in a substantial proportion of cases in
age group of study subjects was 18-19, and not which the early manifestations of the disorder
15-19 as in the other centres. represent a gradual accentuation of long-stand-
Morbid risk estimates were obtained directly ing impairments, or of pre-morbid personality
from the age- and sex-specific incidence rates. traits. A prospective monitoring of' true' onsets
Morbid risk may be estimated by the general is hardly ever feasible in epidemiological re-
formula for an approximation of the cumulated search. Most studies up to date have instead
incidence rate, CIR: relied on the 'social onset' (Odegaard, 1946) of
the disorder, which is defined by an event, such
as a first admission.
The case-finding methods used in previous
The summation from age group 1 to age group research fall into four broad groups: (i) the so-
n of ij t}, where i, is the annual incidence of age called genealogical random sample method
groupy and t} is the span of years included in age (genealogischer Stichprobentest), introduced by
group j , e.g. 5 for 5-year age-groups. This Riidin (1916), which requires the collection of
formula is applicable for cumulated incidence psychiatric morbidity data on the biological
rates about or below 0-1. If it is assumed that the relatives of a random sample of' propositi' and
age-specific incidence rates are constant over allows the estimation of morbid risk; (ii) the
time, the morbid risk of a condition may be birth cohort method (Klemperer, 1933; Frem-
approximated by the sum of rates for each age, ming, 1947; Helgason, 1964) which presupposes
i.e. if: the successful tracing and diagnosing of a
sufficiently large percentage of the members of
f-y = annual incidence for females aged the cohort, once they have passed through the
15-19 entire period of risk; (iii) the population survey
mx = annual incidence for males aged 15-19 or census method (Brugger, 1931, 1933;
f2 = annual incidence for females aged Stromgren, 1938; Sjogren, 1948; etc.), which
20-24 can produce incidence estimates from prevalence
m2 = annual incidence for males aged 20-24 data, assuming that onsets can be dated retro-
spectively, or involves a re-examination of the
same population after some time during which a
fs = annual incidence for females aged number of new cases have become manifest; and
50-54 (iv) thefirstadmission method (Odegaard, 1946),
 Schizophrenia: A World Health Organization Ten-Country Study
which depends on the monitoring and regis-
tration of hospitalizations, assuming that few
cases fail to pass through the door of the
psychiatric hospital. Each of these methods has
advantages and inherent limitations; none was
considered fully adequate for a cross-cultural
comparative study.
The method of first-in-lifetime contact,
employed in this study, comes close to the first
admission method and represents an extension
of this technique to non-hospital facilities and
services which are rarely monitored in psy-
chiatric epidemiology (e.g. the practitioners of
traditional medicine, religious healers, and vari-
ous social agencies). It is conceptually similar to
the 'first ever contact' method of determining
disease inception described by Wing & Fryers
(1976) as part of the techniques used in
psychiatric case register studies. As applied in
the Outcome study, thefirst-in-lifetimecontact
method does not eliminate the possibility that
some schizophrenics may never make contact
with any service and would therefore be missed.
However, it is assumed that the ratio between
such cases and patients who seek treatment
would not vary widely across the different study
areas.
This assumption may appear questionable,
especially when areas with different density of
services and different cultural patterns are being
compared. There is no direct method for its
verification, short of a door-to-door search and
counting of cases who have never made a
contact.
Nevertheless, some attempts at verifying this
assumption were made in this study during the
so-called 'leakage' surveys, which were small-
scale enquiries about missed cases, carried out
with community agencies (such as private medi-
cal practitioners, hostels and shelters for the
homeless, etc.) that had not been part of the
original case-finding network. Although the
primary goal of such enquiries was the identifi-
cation of cases who actually had made a contact
but were missed by the project team, key
informants were also asked whether they knew
about any other similar cases in the area. Except
for the rural area of Chandigarh, where a small
number of the study cases were identified by key
informants in the community and the first
contact with them was initiated by the project
psychiatrist, hardly any symptomatologically
45
manifest cases who had not previously sought
help could be found in the other areas. Thus,
although no strict test could be given to the
proposition that few psychotic patients remain
'invisible' to the helping agencies in a com-
munity, no evidence to the contrary was pro-
duced by the 'leakage' surveys.
A second, indirect way of estimating the
magnitude of the ' hidden morbidity' problem is
provided by the statistical distributions of the
length of previous illness in the series of included
cases. If in a given area the threshold for seeking
treatment is much higher than in another area,
one would expect to find in the former that the
mean intervals between the time of onset of
symptoms and the first contact with a helping
agency are longer. As pointed out in chapter 2,
there were no differences whatsoever between
the developing countries and the developed
countries as regards the proportions of patients
who made a contact with a 'helping agency'
within six months of the onset of symptoms and
those who made a contact after six months. In
all the areas for which incidence rates are
estimated (except Chandigarh, rural area, and
Dublin), small percentages of cases could be
found in which thefirstcontact had been delayed
by two years or more after the onset of
symptoms. These percentages ranged from 3 %
in Aarhus to 10% in Nagasaki, but no centre
had an excess of such cases. The data, therefore,
do not indicate any marked differences between
the study areas in respect of referral thresholds.
This conclusion finds further support in the
observation (see chapter 2) that the reasons and
events which lead to the index referral in
schizophrenics are very similar across the
centres. Therefore, Odegaard's (1952) conten-
tion that most schizophrenics in the community
eventually make a contact with a service does
not seem to be contradicted (at least as regards
predominantly urban areas), and there is no
reason to suppose that in any of the study
settings the incidence rates reported here rep-
resent only a fraction of an ' iceberg' of hidden
schizophrenic morbidity.
MAIN FINDINGS ABOUT INCIDENCE
Annual incidence rates for age 15-54
The one-year rates, calculated per 10000 popu-
lation at risk in all the age groups between 15
 Table 3.1. One-year incidence rates per 10000 population at risk, age 15-54, by four different levels of case
definition for schizophrenia
Clinical diagnosis or CATEGO S, P, O CATEGO S, P, O CATEGO S, P, O + CATEGO S +

Centre M F M-F M F M-F M F M-F M F M-F

Aar rate 1-8 1-3 1-6* 1-3 0-8 11 1-2 0-7 0-9 0-9 0-5 0-7
S.D. 0-34 0-28 0-26 0-29 0-23 0-22 0-27 0-21 0-20 0-24 016 0-18
Cha/R rate 3-7 4-8 4-2 3-0 4-1 3-5 1-9 2-3 2-1 1-3 0-9 1-1
S.D. 105 1-31 0-83 114 1-43 0-90 0-91 1-07 0-70 0-62 0-57 0-42
Cha/U rate 3-4 3-5 3-5 2-4 2-5 2-5 1-7 2-1 1-9 0-8 11 0-9 §-
S.D. 0-53 0-24 0-41 0-45 0-54 0-35 0-38 0-50 0-30 0-26 0-36 0-21
Dub rate 2-3 2-1 2-2 1-5 1-3 1-4 1-3 10 1-1 1-0 0-8 0-9
S.D. 0-55 0-53 0-38 0-45 0-42 0-31 0-42 0-37 0-27 0-31 0-33 0-25
Hon rate 1-8 1-4 1-6 1-0 0-8 0-9 10 0-8 0-9 10 0-8 0-9
S.D. 0-42 0-36 0-28 0-30 0-27 011 0-30 0-27 0-11 0-31 0-27 0-21 a.
Mos rate 2-5 31 2-8 21 2-4 2-2 1-7 2-3 20 10 1-4 1-2
S.D. 0-48 0-50 0-35 0-42 0-43 0-30 0-38 0-42 0-28 0-30 0-34 0-23
Nag rate 2-3 1-8 2-0** 1-7 1-5 1-6 1-4 1-3 1-3 11 0-9 10
S.D. 0-43 0-36 0-27 0-37 0-33 0-25 0-33 0-30 0-22 0-30 0-25 019
Not rate 2-8 1-5 2-2*** 2-2 1-5 1-8 2-0 1-5 1-8 1-7 1-2 1-4
S.D. 0-52 0-39 0-33 0-46 0-39 0-30 0-44 0-39 0-30 0-41 0-35 0-26
Total x2 24-2 73-1 24-4 66-4 11-8 35-2 9-9 16 3
0001 00001 0001 00001 NS 0001 NS NS

If rate corrected for missed cases: * 1-8;*• 2 1 ; »«» 2.4.

 Schizophrenia: A World Health Organization Ten-Country Study 47

The further inclusion of the CATEGO class

Aarhus O? with borderline and dubious cases mainly
diagnosed clinically as acute schizophrenic epi-
Chandigarh/R sodes or as unspecified subtype of schizophrenia,
increased considerably the variation particularly
Chandigarh/U because of a steep rise of the rates in the
Chandigarh rural sample whereas the other
Dublin rates varied by a factor of 2 to 2-50 only.
Further broadening the definition by including
Honolulu patients with a clinical diagnosis of schizo-
phrenia and supposedly related disorders, but
Moscow with other or no CATEGO classifications, the
rates and variations between centres further
Nagasaki increased, again with highest rates in the
developing centre samples from Chandigarh
Nottingham rural and urban areas and the lowest rates in the
Aarhus and Honolulu centres. The variation is
two- and three-fold compared to the Chandigarh
FIG. 3.1. Annual incidence rates per 100000 population aged 15-54 samples but less than two-fold when compared
(both sexes) for the 'broad' (*) definition (clinical diagnosis or
CATEGO S, P, O) and for the' restrictive' (•*) definition (CATEGO
to centres in developed countries except for
S + ) of schizophrenia. • , Series 1*; E3, series 2**. females in the Moscow centre which already by
the SPO+ level showed an unexplained high
and 54, are shown in Table 3.1 and in Fig. 3.1, rate of incidence.
together with their approximate confidence The variation observed for the two broadest
intervals (equalling two standard deviations in levels of definition is not surprising when
each direction). The CATEGO S+ class defined considering the possible variability in incidence
schizophrenia showed quite similar rates in each of various schizophrenia related disorders with
centre varying around 1-0 per 10000. The more or less symptomatological resemblance or
combined male and female rates varied from 0-7 of borderline or dubious character. The variation
in Aarhus to 1-4 in Nottingham. For males the is lower for the most restrictive definition by the
variation was between 0-8 (Chandigarh, urban S+ CATEGO class which is an indicator of
area) and 1-7 (Nottingham) and for females symptomatologically more severe and florid
between 0-5 (Aarhus) and 1-4 (Moscow). The schizophrenia. Although the numbers of patients
variation overall was at most two-fold and the so defined are smaller the lower variation
differences nowhere reached significant levels. observed is not merely explained by decreasing
Inclusion of the CATEGO classes S?, P+ and the statistical power of analysis. When the mean
P? and O + raised the incidence rates to around rate also decreases even small fluctuations due
1-5 but also increased the variation of female to chance, for instance by missing or erroneous
rates among centres now varying between 0-7 inclusion of a few cases will tend to cause a
(Aarhus) and 2-3 (Chandigarh, rural area, and higher variation. This however did not occur,
Moscow) whereas the variation for males and indicating that the S + CATEGO class presents
for the combined sexes still differed within a a robust syndromatical representation of schizo-
two-fold non-significant variation. The female phrenic conditions that may be detected reliably
variation is significant but not merely explain- in various study areas and seems to occur with
able by inclusion of patients with acute onset or about the same frequency.
of reactive character. The highest female rates
appeared both in a developing and a developed Age- and sex-specific rates
centre and apart from the first rank symptoms The age- and sex-specific rates for the
the patients included by the CATEGO classes CATEGO-SPO - definition of schizophrenia
S?, P + , P? and O+ show a symptomatic and for the CATEGO S + class definitions, are
profile similar to the S+ patients although of presented graphically in Fig. 3.2. The sex-specific
less severity. rates are presented in tabular form in Table 3.1.
48 A. Jablensky and others

Aarhus SPO, FRC Aarhus S+

25-29 35-39 15-19 25-29 35-39 45-49

Chandigarh/Rural SPO, FRC Chandigarh/Rural S+

25-29 35-39 45-49 15-19 25-29 35-39 45-^9

Chandigarh/Urban SPO, FRC Chandigarh/Urban S+

25-29 35-39 45-49 15-19 25-29 35-39 45-49

FIG. 3.2. For caption see page 50.

For the narrow S+ definition rates show
marked peaks for males in five centres, one in
age group 15-19 and four in age group 20-24;
for females in four centres, one in age group
30-34 and two in age group 45-49. For the
broad SPO definition male rates show marked
peaks in the same age groups in the same five
centres and in one further centre a peak appeared
in age group 15-19. Female rates lost their peak
patterns except for the two centres with peaks in
age group 20-24. The pattern of age- and sex-
specific rates thus showed a tendency to earlier
onset in males than in females, a tendency which
is further illustrated by Fig. 3.3 which shows the
cumulative curves of onset in males and females
in developed countries and in developing
 Schizophrenia: A World Health Organization Ten-Country Study 49

Dublin SPO, FRC Dublin S+

25-29 35-39 45^9

Honolulu SPO, FRC

25-29 35-39 45^9 15-19 25-29 35-39 45^9

Moscow SPO, FRC Moscow S+

25-29 35-39 45^9 15-19 25-29 35-39 45^9

Fio. 3.2. For caption see page 50.

countries respectively. It suggests that women
'consume' their risk of developing a schizo-
phrenic illness at a slower and more even rate
than men, and that this pattern holds for both
types of setting.
Considering the sensitivity of age- and sex-
specific rates to small changes in sample com-
position, the emergence of a consistent pattern
showing a sex difference, i.e. a later manifestation
of schizophrenia in females in culturally and
demographically different areas of the world
represents an important finding. Apart from its
significance in relation to genetic and clinical
research in schizophrenia, it supports the validity
of the diagnostic definition of the disorders
included in this study by demonstrating their
50 A. Jablensky and others
Nagasaki SPO, FRC
15-19 25-29 35-39 45^9
Nottingham SPO, FRC
0
15-19 25-29 35-39 45-49
FIG. 3.2. Age- and sex-specific incidence rates per 10000 population for a 'broad' definition (CATEGO S, P, O cases) and for a
'restrictive' definition (CATEGO S + cases) of schizophrenia. • , Males; + , females.
cross-cultural similarity on such key para-
meters.
Since the age- and sex-specific manifestation
rates of schizophrenic disorders are considered
to be of particular importance from a geneticist's
point of view (Gottesman & Shields, 1982), the
relevant findings for each study area are sum-
marized below.
Aarhus
The 'broad' category of schizophrenia and
related disorders shows a marked preponderance
of males in the age groups 20-29, and a slight
excess of females after the age of 30. The
incidence in males shows a clear peak in the age
group 20-24; a less prominent peak of incidence
is observed in females of the same age group. In
the CATEGO S + category, the pattern is almost
the same, except that no excess of female cases
aged 30+ is observed. The shape of the curve
of age-specific incidence of the S+ cases is
Nagasaki S+
0-
15-19 25-29 35-39 45^(9
Nottingham S+
15-19 25-29 35-39 45^49
very similar to that of the 'broad' diagnostic
group.
Chandigarh {rural area)
In the 'broad' diagnostic category there is no
clear-cut gender-related pattern, and no age-
related peak of incidence can be identified. In
the S + group, the number of males exceeds that
of females in all age groups up to 34. There is,
however, a clustering of female cases in the age
group 40-49. The absence of a consistent age-
and sex-related pattern may be due, in part, to
the relatively small number of cases in this rural
area, with relatively low numbers of population
in the older age groups.
Chandigarh (urban area)
There is an excess of males in age groups 20-24
and of females in age group 35-44 in the ' broad'
diagnostic group. There is a marked peak of
incidence in the age group 20-24 in males but
 Schizophrenia: A World Health Organization Ten-Country Study 51

group 45-49 where females predominate). In the

S + category, males exceed females in the age
groups 15-29 and 35-39, while females exceed
males in the age 45-49. In females, a peak can be
observed in the age range 45-49, for S+ cases
only.

Honolulu
There is a marked excess of males in age group
20-24, and an excess of females after the age of
29, for both diagnostic categories. The peak of
incidence is in the age 20-24 for males; and no
clear-cut peak is found in the female population.

Moscow
15 20 25 30 35 40 45 50 55 At the level of the 'broad' diagnostic definition,
Age at onset there is an excess of males in the age group
(b)
18-19 only; the rates are similar for the two
100 -: sexes in ages 20-39, and females are markedly
overrepresented after the age of 40. There is a
peak of incidence in males in the age 18-19, for
80 - females in the age group 30-34.

§,60-3
40 -
20 -
0 -
15 20 25 30 35 40 45 50 55
Age at onset
FIG. 3.3. Cumulative percentages of male (
subjects having the onset of a schizophrenic disorder ('broad'
definition) by a specific age. (a) Developing countries; (b) developed
countries.
not in females. In the S+ group, males
predominate in the age range 20-24 but have
somewhat lower rates than females in the other
age groups. The peak of incidence in males is in
the 20-24 age group.
Dublin
In the 'broad' diagnostic category, there is an
excess of males in the age groups 25-29 and
35-39 (but not in the age group 20-24 where the
two sexes are equally represented, or in age
Nagasaki
In both the 'broad' diagnostic category and the
S + category, there is an excess of males in age
groups 15-19 and 25-29; the two sexes show
similar rates in other age groups. There is a peak
of incidence in males aged 15-19 and in females
aged 20-24. The shapes of the curves of the age-
specific incidence of the 'broad' definition group
and of the S + group are very similar.
Nottingham
In both diagnostic categories there is a marked
) and female ( )
excess of males in the age groups 20-29, and
some excess of females in the age groups 40-49.
There is a peak of incidence in males in the age
group 20-24 but no comparable peak in females.
The shapes of the two distributions are very
similar.
The overall impression is that the centres
show similar patterns except for the Chandigarh
rural and Moscow centres. In most centres the
rates tended to be higher in the younger age
groups particularly for males, whereas the
Chandigarh rural and Moscow centres had high
rates in the older age groups and also showed
high variations among age groups for SPO
defined rates.
52 A. Jablensky and others

u. Morbid risk
s oooooo oo
The expectancies of developing a disorder, for
males and females passing through the entire
en period of risk between 15 and 54 years of age,
O
O LL.
oooooooo
are given in Table 3.2. For the' broad' diagnostic
U
category, the probability of developing the
'enia
CAT

disorder varies from 050% in Honolulu to

• «
m ^ t ( N — Is- Oso
1-72% in Chandigarh (rural area), and for
a, S
ON
the CATEGO S+ category from 0-26% in
.§ OOOOOOOO
Honolulu to 0-54 % in Nottingham.
ion f o r ,

u.
n » \c ^ fi r~ ^ * COMPARISONS OF THE RESULTS WITH
S o o o o o o o <
PREVIOUS DATA
f

S O Having established that the incidence rates and

OH"
en"
morbid risk estimates reported in this study can
U-
be accepted as valid with a fair amount of
'ase

O
O
confidence, it is important to examine the extent
:ATI

to which they confirm or contradict previously

reported epidemiological data.
1
-Si
S oooooooo In Aarhus, both the rate for the 'broad'
group of schizophrenia and related disorders
and the rate for CATEGO S + cases, appear to
;t be higher than the Danish averagefirstadmission
s rate for schizophrenia (0-51 per 10000) for the
3 0 period 1970-9, calculated on the basis of the
0." national psychiatric case register (Joensen &
O
Wang, 1983). Nielsen (1976) calculated the 'first-
0 ^ - 0 0 0 0 0 0
0 time referral' rate for schizophrenia (excluding
CAT:

reactive psychoses) on the island of Samso near

or age 15-5

Aarhus to be 20 per 10000. A direct comparison,

s oooooooo
however, could be misleading because the
Outcome study category of 'broadly' defined
schizophrenia includes a proportion of cases
r

OH" U-
with centre diagnosis of reactive paranoid
O'lC^OfiOmtNO
- " • « OC <T) — r- «D psychosis. In the national statistics of Denmark,
'isk (

s
GOS,

< ^H o O •—" O o
all reactive psychoses (ICD 298, which includes
13
UJ
H
also those with predominantly affective or
^S 5^ confusional features) are reported separately,
f ^ o r r ^ ( N
and show a rate higher than schizophrenia (3-2
1
sis or

b ^ O l N O O ' t - ^P <>
OfN-— O O - — O O
0
per 10000). A recent analysis of the national
a case register data over the period 1970-84
(Munk-J0rgensen, 1986) indicated a trend of
Clirlical

o ^H A o O — O c decreasingfirst-admissionrates for the diagnosis

H of schizophrenia accompanied by a significant
increase in the frequency of diagnoses such as
paranoid and unspecified psychoses, or 'bor-
U derline states'. Compared with other Scandi-
c navian studies, the Aarhus rates for 'broad'
schizophrenia, as defined by the Outcome study
si 5.
criteria, do not appear to be very different.
<6UQISZZ
 Schizophrenia: A World Health Organization Ten-Country Study 53

Table 3.3. Frequency of schizophrenic disorders in geographically defined populations

Prevalence Incidence Morbid risk
Author/Year Country (per 1000) (per 1000) (per 100) Remarks

Birth cohort studies

Klemperer (1933) Germany 100* — 1-40 •Estimated with correction for
cohort attrition
Fremming (1947) Denmark — — 0-90
Helgason (1964) Iceland — — 0-57-0-69 (M) Up to age 61
0-90-102 (F)
Census studies
Brugger (1931) Germany 2-4* (1-9**) — 0-38 •Per 1000, aged 10 +
••Per 1000, all ages
Brugger (1933) Germany 2-2* — 0-41 •Per 1000, all ages
Brugger(1938) Germany 2-3* (1-8)** — 0-36 •Per 1000, aged 10 +
••Per 1000, all ages
Stromgren (1938) Denmark — — 0-58
Sjogren (1948) and Larsson Sweden 4-6 — 1-60* •Equal for M and F
& Sjogren (1954)
Hollingshead & Redlich USA — 0-30 — •Age 15 +
(1958)
Book (1953) Sweden 9-5 — 2-66* •Age 15-50
Book et al. (1978) Same population* 17-0 — 2-68 (M) •Genetic isolate
2-27 (F)
Essen-Moller et al. (1956) Sweden 6-7 (3-9*) — — •Psychotic on census date
Hagnell (1966) Same population 4-5 — —
Lin et al. (1969) Taiwan 1-4* — — •Lifetime, all ages
Crocetti et al. (1971) Yugoslavia 5-9 — — Area known for high rate of
psychosis
Rotstein (1977) Sample USSR 3-8 — —
Wijesinghe et al. (1978) Sri Lanka 5-6 — — Age 15 +
Service Contact Studies
Odegaard (1946) Norway — 0-24* 1-87 •Age 10 +
Norris (1959) United Kingdom — 0-17
Adelstein et al. (1968) United Kingdom — 0-35 (M) — Age 15 +
2-26 (F)
Walsh (1969) Ireland — 0-57 (M) — Age 10 +
0-46 (F)
Hafner & Reimann (1970) Federal Republic of — 0-54 —
Germany
Lieberman (1974) USSR — 0-20 (M) — Patients with onset in 1950-
0-19 (F) 1964 personally investigated
by author
Temkov et al. (1975) Bulgaria 2-8 — —
Nielsen (1976) Denmark 2-7* 0-20 — •Census on 1 January 1964
Ouspenskaya (1978) USSR 5-3 — — •Lifetime prevalence, per 1000,
age 14 +
Helgason, L. (1977) Iceland — 0-27 0-43 (M)
0-54 (F)
Shen Yu-cun et al. (1981) China — 011 0-46
Krupinski & Alexander (1983) Australia — 018 —
Munk-j0rgensen (1986 a, b) Denmark — 0-15(M) — All ages
0-90 (F) —

Most Scandinavian studies have produced
estimates of morbid risk, using different methods
but taking advantage of the demographic stab-
ility of the population and of the exhaustive
census or case register data. Using the census
method, Stromgren (1938) estimated the risk for
schizophrenia (both sexes) at 0-47% for the
island of Bornholm, a figure very close to the
estimate 0-56% for 'broad' schizophrenia in the
Aarhus area in this study. Other Scandinavian
data are quoted in Table 3.3. In this connection,
it should be noted that morbid risk figures for
schizophrenia in European populations appear
to be not only similar but also remarkably stable
over time. On the basis of data from 18 German,
Swiss, and Scandinavian studies conducted in
the 1920s and 1930s, Stromgren (1950) derived a
lifetime risk estimate of 0-72% for men and
54 A. Jablensky and others
women combined. The risk estimates for the
European centres in this study (0-59 % in Aarhus,
0-83 % in Dublin, 113 % in Moscow, and 0-80%
in Nottingham) are of the same order of
magnitude.
The rates in Chandigarh cannot be compared
with analogous Indian data, because practically
all previous epidemiological studies of schizo-
phrenia on the Indian subcontinent (reviewed by
Wig, 1982) have been prevalence surveys. These
surveys have produced prevalence rates for
psychoses of the order 5-10 per 1000 population,
which is comparable to prevalence rates in
European and North American populations.
First admission data on schizophrenia are,
however, available for the Indian population in
Mauritius (Murphy & Raman, 1971; Raman &
Murphy, 1972). For the year 1956, the total rate
in age groups 15-44 was 1-4 per 10000 (Hindu
Indians) and 0-9 per 10000 (Moslem Indians).
Much higher rates, especially in women aged
35-44, were found in Indians living in Singapore
(Kadri, 1963; Murphy, 1968). In spite of the
differences in methods and setting, these data do
not conflict with the Chandigarhfindings,which
also suggest a rather high rate in females after
the age of 35.
The Dublin data are of special interest, in
view of previous findings of unusually high
prevalence (Walsh et al. 1980; Terry et al. 1984)
and first admission rates for schizophrenia (5-7
per 10000 in males and 4-6 per 10000 in females
-Walsh, 1969). These survey findings are in
accordance with national first admission data
(45 per 10000 for the two sexes and all ages in
1978 - O'Hare & Walsh, 1980) and at variance
with the data for Northern Ireland (Murphy &
Vega, 1982) where the first admission rates are
of the order of 1-0 to 3-7 per 10000, i.e. very
similar to the first-contact rate for Dublin
reported here. Although the allegedly high Irish
incidence rates have led to interesting speculation
about possible underlying factors (Torrey, 1980),
it seems that a large part of the explanation may
be found in the overestimation of the number of
first admissions by hospitals, a practice common
to many reporting systems. Since the estab-
lishment of three case registers in Ireland (Blake
et al. 1984), which ensured the accurate enu-
meration of first-contact patients at the psy-
chiatric services, the rates for schizophrenia no
longer appear to be higher than elsewhere. The
incidence rates found in a survey in three Irish
counties in which the Present State Examination
was used as the principal clinical instrument (ni
Nuallain et al. 1984) agree quite well with the
rates established in the present study, although
the possibility cannot be ruled out that in some
rural areas in the western part of Ireland true
pockets of high rate of schizophrenia may exist.
No incidence or first admission rates, pro-
duced by earlier research, are available for
Honolulu but psychiatric case register data
(Weiner & Marvit, 1977) suggest that the
prevalence of schizophrenia among Caucasians
in Hawaii is the same as in mainland US (it is
somewhat higher in other ethnic groups in
Hawaii). Most of the US epidemiological studies
(reviewed by Babigian, 1980) have reported
treated incidence rates for the population aged
15+ in the range of 3-0 to 12-0 per 10000. The
Monroe county case register (Kramer, 1980) has
produced the figure 6-8 per 10000 but only one
half of these first admissions were actually first-
in-lifetime hospitalizations and, in addition,
some of the patients had had earlier spells of
out-patient treatment. The older US rates have
also been affected by the very broad definition of
schizophrenia prior to the introduction of DSM-
III. According to Babigian, the application of
DSM-III criteria to the Monroe county data
would result in a further reduction of the rate by
20-30%, and would bring it down to about 2-0
per 10000, which would be close to the Honolulu
rate in this study.
Previous epidemiological studies in Moscow
(Liebermann, 1974; Shmaonova, 1983) indicate
an incidence rate of schizophrenia of 1-91 per
10000 (1-98 in males and 1-85 in females). The
figures are based on a census of all patients
registered by the psychiatric dispensary and a
retrospective dating of onsets. These rates are
somewhat lower than the first-contact rates
found in the present study. Closer to the latter
are the first admission rates for schizophrenia
(2-3 per 10000) reported by Krasik & Semin
(1980) for the city of Tomsk in Western Siberia.
All previously reported Soviet rates show peaks
of incidence in young males (up to age 19) and
in middle-aged females.
The Nagasaki rates in this study cannot be
compared with other Japanese data because
previous surveys in that country have been
almost exclusively prevalence-orientated (Kato,
 Schizophrenia: A World Health Organization Ten-Country Study
1969; Torrey, 1980). The latter have shown a
fairly stable average national prevalence of 2-3
per 1000, although considerable regional vari-
ations have been described. Morbid risk for
schizophrenia has been estimated on the basis of
at least 15 prevalence surveys carried out
between 1940 (Nakane et al. in press). The risk
estimates have varied from 0-48 to 2-47%, with
a median value of 0-82%.
The Nottingham data should be compared
with what comes closest to the case definition in
the present study - namely, the 'first-ever con-
tact' rate for patients entering the psychiatric
registers in Camberwell (London) and Salford
(Wing & Fryers, 1976). These rates were of
similar size in Camberwell (total rate 1-4 per
10000; 1-4 for males and 1-5 for females) and
somewhat lower in Salford (total rate 11 per
10000; 1-2 for males and 1-0 for females).
The above overview shows that the first-
contact incidence rates and morbid rates, par-
ticularly for the 'broad' definition of schizo-
phrenia and related disorders established in the
55
present study, correspond well to previous
findings or estimates of the frequency of these
conditions in the individual study areas or
countries. By applying a uniform case definition
and case finding methodology, as well as
standardized tools for the clinical description of
the patient populations in the different settings,
the study has shown that no large difference is
seen in the manifestation rate of schizophrenic
disorders across cultures and geographical areas
that are as wide apart as Denmark and India, or
Japan. This, of course, does not exclude the
possibility that pockets of unusually high or
unusually low incidence may be found in
different populations (e.g. the morbid risk of
2-66% calculated by Book, 1953, and Book
et al. 1978, in a population isolate in northern
Sweden). The reasons for such deviations are far
from being well understood, but it is increasingly
clear that they represent exceptions rather than
the rule in the global epidemiological pattern of
schizophrenia.
 Chapter 4 Two-year course and outcome
The medium-term course and outcome of the Table 4.1. Characteristics of the patients who
disorders manifested by the original 1379 sub- completed the follow-up and of those who did not
jects who met the inclusion criteria of the project
and had been assessed at the initial examination Not
were evaluated by means of two follow-up Followed up followed up
Variable (AT =1078) (TV = 301) Difference
examinations, scheduled at one year and at two
years from the date of the first assessment (the Mean age (years) 27-9 26-6 NS
date of the initial PSE was taken as the reference Sex (M/F) 1-1 1-3 NS
Percentage single 61-6 62-8 NS
point). Percentage acute 39-2 41-3 NS
In each research centre, the patients and, in onset
Percentage using 14-2 20-9 P < 001
most instances, also key informants, were invited drugs
for a follow-up interview; if no response to the Percentage 55-1 44-2 P < 0001
letter of invitation resulted, the patients were CATEGO S +
Percentage ICD 28-8 28-9 NS
visited at their homes. Every attempt was made 295.31
to trace subjects who had changed their place of Percentage ICD 241 22-8 NS
residence, and to collect at least a minimum of 295.4*
information on those who could not be re- 1
Paranoid; 2
Acute schizophrenic episode.
interviewed. The latter represented a minority
(301 out of 1379 study subjects, or an overall
1200 -r
'drop-out' rate of 21-8%) of the original patient
series. The analysis of follow-up data reported
in this chapter is, therefore, based on a total of
1078 cases (the totals in the tables which follow 1000 -
may not add up to this figure because of missing
data on some patients in specific tabulations).
The sociodemographic and diagnostic charac-
teristics of the patients who were not re-assessed
did not deviate in any systematic manner from
those of the patients who were available for
follow-up. The principal characteristics of the
patients who dropped out and were not re-
assessed are shown in Table 4.1. There were no
significant differences between patients re-
assessed and patients not re-assessed on variables
such as age, gender, marital status, and type of
onset. Patients with reported use of street drugs
were over-represented among the 'drop-outs'
and the difference was significant at the 0-01 0
level. Considering diagnostic classification, there 18 19 20 21 22 23 24 25 26 27 28 29 30
was no difference at the level of the 3-digit ICD- Months of follow-up
9 diagnosis, but patients falling into CATEGO FIG. 4.1. Distribution of cases by number of months of follow-up
classes other than S+ were more likely to be within the range 18-30 months.
lost to the follow-up than class S + cases
(P < 0-001). field research centres: Aarhus 19-2, Agra 6-4,
The 'drop-out' rate (%) showed highly Cali 9-7, Chandigarh (rural) 5-6, (urban) 30-9,
significant differences (P < 0-001) among the Dublin 14-9, Honolulu 57-4, Ibadan 310,
56
 Schizophrenia: A World Health Organization Ten-Country Study
Moscow 16-8, Nagasaki 35-2, Prague 18-7,
Rochester 43-6.
The differences in the proportions of patients
who were followed up were unrelated to the
developing/developed country dichotomy.
The 1078 cases with a complete follow-up
assessment (78-2 % of the original series) pro-
vided sufficient data to enable the evaluation of
the main variables describing the course and
outcome of schizophrenic disorders over a period
of an average length of two years following the
initial examination. The actual range of the
follow-up was between 18 and 30 months (i.e. it
allowed for a deviation of up to 6 months either
way from the target date for completion of the
follow-up which had been set at 24 months after
the first assessment). The distribution of cases
by the completed number of months of follow-
up within the permissible range of 18 to 30
months is shown on Fig. 4.1.
METHODS AND INSTRUMENTS USED
ON FOLLOW-UP EXAMINATIONS
Every patient, available for a follow-up as-
sessment, had a PSE interview. Both patient and
informant provided information for the Follow-
Up Psychiatric and Personal History Schedule
(FU-PPHS); in many instances this information
was supplemented with data from hospital or
clinic notes. Apart from an updating review of
the main demographic and social data about the
patient, the FU-PPHS contains a month-by-
month chart of symptomatology, treatment,
and life events, which was designed to enable a
reconstruction of the course of the condition
over the preceding 12 months. Upon completion
of the PSE and the FU-PPHS, the investigators
were required to record their overall impressions
and conclusions in the Follow-Up Diagnostic
and Prognostic Schedule (FU-DPS), and to
write a narrative summary of the patient's
progress. An additional instrument, the WHO
Disability Assessment Schedule (WHO-DAS)
was also rated at follow-up examinations, and
the results of the analysis of the data obtained
with it will be reported in subsequent publi-
cations.
The extensive data collected on follow-up The categories used to classify the course of the
examinations were processed and tabulated at disorder were as follows.
WHO Headquarters, and reviewed at a meeting
of the collaborating investigators. An agreement
57
was reached on how to aggregate the large
number of variables that had been followed up,
and each centre produced its own summary
chart of the main course and outcome charac-
teristics on every patient. These summary charts
were coded and double-checked for consistency
against the original dataset at WHO Head-
quarters, any discrepancies between the centres
and WHO Headquarters were resolved through
correspondence or discussion. The information
used in the analyses presented below has,
therefore, been subjected to multiple checks.
GENERAL DESCRIPTION OF THE TWO-
YEAR COURSE AND OUTCOME
The following variables were assessed with a
view to describing the general features of the 2-
year course and outcome of the study patients:
(1) pattern of course (a composite rating of the
number of discrete psychotic and non-psychotic
episodes observed over the follow-up period,
and of the number and clinical quality of the
remissions, if any); (2) proportion of the total
length of the follow-up period during which the
patient was in psychotic episodes; (3) proportion
of the follow-up period during which the patient
was in a complete remission (symptom-free); (4)
proportion of the follow-up period during which
the patient was on anti-psychotic medication;
(5) proportion of the follow-up period during
which the patient was in psychiatric hospital;
and (6) proportion of the follow-up period
during which the social functioning of the patient
was unimpaired. Each of these variables had an
operational definition, and the ratings provided
by the centres were checked at Headquarters.
The results described below apply to all
patients who met the original inclusion criteria
and completed the follow-up, i.e. to the patients
falling into the 'broad' diagnostic category of
schizophrenia, which was based on the presence
of either an eligible clinical (centre) diagnosis in
ICD-9 terms, or a CATEGO class S, P, or O on
initial examination.
Pattern of course
1, single psychotic episode followed by a
complete remission;
58 A. Jablensky and others

2, single psychotic episode followed by an

incomplete remission;
3, single psychotic episode followed by one or
more non-psychotic episodes, with complete -15
remissions between all or most of the episodes;
4, single psychotic episode followed by one or
more non-psychotic episodes, with incomplete
remissions between all or most of the episodes;
5, two or more psychotic episodes, with
complete remissions between all or most of
the episodes; J
6, two or more psychotic episodes, with
incomplete remissions between all or most of 3 ii
u%
the episodes;
7, continuous psychotic illness (no remission);
psychotic symptoms present most of the
time;
8, continuous non-psychotic illness (no re-
mission) ; psychotic symptoms may be present
for some time but non-psychotic symptoms
predominate throughout; ov •* f

9, information inadequate for rating the

pattern of course.
The distribution of the patients over these
different patterns of 2-year course is shown in
Table 4.2. Considering the entire series of cases
with completed follow-up, the majority of the
patients (50-3 %) had a single psychotic episode,
i.e. fell into one of the patterns 1-4. A substantial
proportion (33-1 %) had two or more psychotic -I
episodes, i.e. pattern 5 or 6, and only a minority
(14-6%) of the patients had an unremitting,
continuous psychotic illness (pattern 7). How-
ever, there was significant variation among the
centres. For example, the percentages of cases
with single psychotic episodes (patterns 1-3) in
the course of the follow-up ranged from 27-5 in
Aarhus to 75-0 in Chandigarh (rural area); those
patients with two or more psychotic episodes
(patterns 5 and 6) were in the range between 19-2
(Chandigarh, rural area) and 52-5 (Aarhus); and
those subjects with continuous psychotic illness
were in the range between 2-0 (Ibadan) and 32-9
(Nagasaki).
X!
The individual patterns can be combined in
different ways to obtain more global descriptors
of the course of the disorder. A summation of .1 8
the cases of patterns 1, 3 and 5 indicates that the o 5i u C u
•as v, j=
-.2 & E S.
proportion of remitting schizophrenic illnesses s £e '§ B s
•S - S s-S
E o Ka t
with complete remission is high and amounts to frs »•* S--
mil
no less than 48-1 % of all cases. The proportion
ft a ft P. C
E &s S- = B = -a j
.SiE+ 8 +
of patients with incomplete remissions is 35-3 %;
 Schizophrenia: A World Health Organization Ten-Country Study 59

Table 4.3. Distribution of cases by percentage of follow-up spent in psychotic episodes

Percentage of time in psychotic episodes
No of
Centre patients 1-5 6-15 16-45 46-75 76-100 Total

Aar 80 26-3 17-5 27-5 7-5 21-3 1001

Dub 57 140 50-9 19-3 1-8 14-0 1000
Hon 29 34-5 20-7 13-8 — 310 1000
Mos 164 17-7 31-7 24-4 7-3 18-9 1000
Nag 69 4-4 18-8 30-4 11-6 34-8 100-0
Not 86 24-4 25-6 19-8 7-0 23-3 1001
Pra 87 17-2 47-1 25-3 2-3 8-1 1000
Roc 31 32-3 35-5 22-6 — 9-7 1001
Agr 76 23-7 34-2 17-1 40 21-2 100-2
Cal 139 6-5 26-6 22-3 17-3 27-3 1000
Cha/R 49 27-5 35-3 27-5 3-9 5-9 1001
Cha/U 107 23-4 37-4 21-5 6-5 11-2 1000
Iba 96 20-8 531 19-8 4-2 21 100-0
All 1070 18-8 33-6 22-7 7-0 17-9 1000

and that of cases with unremitting psychotic

symptoms is 14-5%.
Proportion of the follow-up period spent in
psychotic episodes
The proportions of the cases which fall into the
different percentiles of the total follow-up time
spent in psychotic episodes (obtained by sum-
ming up the duration of all discrete episodes) are
presented in Table 4.3. Nearly identical propor-
tions (18-8 % and 17-9 %) of patients fall into the
extremes of very short (up to 5 % of the length
of the follow-up period) and very long (76-100 %
of the period) total duration of the psychotic
episodes. Within these two extreme categories,
there is marked variation in the share of each
field research centre. Thus, the proportions of
patients who spent in psychotic episodes less
than 5% of the follow-up period vary from
4-4 % in Nagasaki to 34-5 % in Honolulu. Higher
proportions (over 20%) were observed in all of
the centres in developing countries except Cali,
and also in three of the centres in developed
countries (Aarhus, Honolulu and Nottingham).
As regards the subjects who spent 76-100% of
the time in psychotic episodes, their proportions
range from 2-1% in Ibadan to 34-8% in
Nagasaki; these proportions generally tend to
be higher in the centres in developed countries
(except for Dublin and Prague) but they are
similarly high in two of the centres in developing
countries (Agra and Cali).
Proportion of the follow-up period in complete
remission
The percentage of time during which patients
are symptom-free is not simply the reciprocal
value of the percentage of time spent in psychotic
episodes because a certain number of subjects
had non-psychotic episodes or incomplete remis-
sions, in addition to having been psychotic for
some of the time. However, there is a fair
correspondence between the distributions of
cases over' time psychotic' and' time in complete
remission' (Table 4.4).
Overall, 29-4% of the patients were symptom-
free (complete remission for 76-100% of the
time; on the other hand, 42-9 % never attained a
complete remission during the follow-up. The
proportion of cases in complete remission over
46-100% of the follow-up period is 44-6%.
The extremes of the distributions by centre
are illustrated by Nagasaki and Ibadan where
7-3 % and 73-1 % respectively of the patients fell
within the range of 76-100% symptom-free
time, and by Ibadan and Moscow, with 7-5%
and 77-4% respectively of the patients not
having had any symptom-free interval during
the follow-up.
Proportion of time on antipsychotic medication
This measure of the course of psychotic disorders
is based on a month-by-month review of the
treatment chart contained in the FU-PPHS in
which every prescribed medication was
recorded; the study design did not envisage
PSM
60 A. Jablensky and others

Table 4.4. Distribution of cases by percentage of the follow-up period spent in complete remission
Percentage of time spent in complete remission
No. of
Centre patients 0 1-5 6-15 16-45 46-75 76-100 Total

Aar 80 700 _ _ 1-3 11-3 17-5 1001

Dub 56 55-4 10-7 12-5 21-4 1000
Hon 28 57-1 3-6 3-6 7-1 14-3 14-3 1000
Mos 164 77-4 1-2 1-2 1-2 4-3 14-6 99-9
Nag 69 65-2 — 2-9 101 15-5 7-3 1000
Not 86 30-3 — 3-5 10-5 16-3 39-5 1001
Pra 87 29-9 9-2 21-8 391 100-0
Roc 31 54-8 — — 12-9 3-2 29-0 99-9
Agr 76 21-1 1-3 2-6 1-3 10-5 63-2 1000
Cal 138 37-0 0-7 5-8 24-6 21-7 101 99-9
Cha/R 50 280 2-0 80 32-0 30-0 1000
Cha/U 108 23-2 0-9 6-5 14-8 25-0 29-6 1000
Iba 93 7-5 2-2 — 6-5 10-8 731 1001
All 1066 42-9 0-8 2-5 9-4 15-2 29-4 100-2

Table 4.5. Distribution of cases by percentage time of the follow-up during which the patients were
prescribed antipsychotic medication
Percentage of time on psychotic medication
No of
Centre patients 0 1-5 6-15 16-^5 46-75 76-100 Total

Aar 80 3-8 6-3 16-3 23-8 500 100-2

Dub 56 5-4 1-8 8-9 8-9 19-6 55-4 1000
Hon 29 6-9 10-3 20-7 241 3-5 34-5 1000
Mos 164 — 0-6 3-7 3-7 4-3 87-8 1001
Nag 70 2-9 4-3 100 17-1 65-7 1000
Not 84 3-6 9-5 10-7 250 13-1 36-9 99-8
Pra 86 11 2-3 80 14-8 21-6 52-3 1001
Roc 31 6-5 6-5 22-6 12-9 25-8 25-8 1001
Agr 76 40 32-9 32-9 22-4 5-3 2-6 1001
Cal 139 3-6 9-4 180 360 23-0 101 1001
Cha/R 49 8-2 14-3 28-6 26-5 18-4 4-1 1001
Cha/U 109 13-8 7-3 15-6 20-2 24-8 16-5 100-2
Iba 96 — 4-2 11-5 19-8 25-0 40-6 1001
All 1069 3-9 6-9 131 18-3 17-1 40-6 99-9

plasma level monitoring or determination of antipsychotic medication for much longer

metabolite excretion in the urine. Therefore, the
actual extent of compliance with the prescribed
medication was not known. Nonetheless, this
variable is informative as a measure of the
estimated need for pharmacological treatment
and maintenance which, in turn, reflects the
psychiatrist's perception of the severity of the
course of the illness. However, the variable also
reflects different treatment practices in different
locations.
The data (Table 4.5) show a considerable
variation among the centres in this respect.
There is a marked tendency within the centres in
developed countries to maintain patients on
periods of time, as compared to centres in
developing countries. Between 34-5% (Hono-
lulu) and 87-8 % (Moscow) of the patients in the
developed countries were prescribed neuroleptics
for 76-100% of the follow-up period. In the
developing countries, the corresponding propor-
tions were in the range between 2-6% (Agra)
and 16-5% (Chandigarh, urban area), with the
exception of Ibadan where a relatively high
proportion (40-6 %) were prescribed neuroleptic
treatment for 76-100% of the time. However,
since compliance was not monitored, and the
impression of the Ibadan investigators was that
few patients actually adhered to the treatment as
 Schizophrenia: A World Health Organization Ten-Country Study 61

Table 4.6. Distribution of cases per percentage time of the follow-up spent in a psychiatric hospital
Percentage of time in psychiatric hospital
No of
Centre patients 0 1-5 6-15 16-45 46-75 76-100 Total

Aar 80 50 28-8 28.8 250 8-8 3-8 100-2

Dub 57 15-8 21-1 43-8 140 3-5 1-8 1000
Hon 29 6-9 65-5 13-8 13-8 — — 100.0
Mos 164 1-8 13-4 47-6 34-8 1-8 0-6 1000
Nag 70 28-6 4-3 22-8 25-7 7-1 11-4 99-9
Not 86 10-5 27-7 38-4 20-9 3-5 — 1000
Pra 87 5-8 40-2 47-1 6-9 1000
Roc 31 — 32-2 54-8 6-5 3-2 3-2 99-9
Agr 76 73-7 11-8 5-3 5-3 2-6 1-3 1000
Cal 139 23-7 46-8 27-3 2-2 — — 1000
Cha/R 45 911 6-7 2-2 — — — 1000
Cha/U 109 80-7 110 6-4 ' 1-8 — — 99-9
Iba 97 691 10-3 17-5 31 — — 1000
All 1070 310 20-2 27-9 16-8 2-7 1-4 1000

prescribed, it is highly unlikely that the good there were only 1-4% who spent between 76%
outcome of the majority of the cases in that and 100% of the follow-up period in hospital,
centre was in any way related to a high and there was no centre, except Nagasaki, where
medication rate. this percentage exceeded 3-8. Not a single case in
On the other hand, very few patients in any the developing countries had been continuously
centre had been considered in no need of in hospital throughout the follow-up period.
neuroleptic treatment (3-9 % of the total study Although 69% of the study patients were
population). The percentage ranged from 0 % in admitted at some point to hospital, 48-1
Moscow to 13-8% in Chandigarh (rural area). remained there for less than 15 % of the follow-
All in all, 40-6 % of the subjects in the study were up period (20-2 % were hospitalized for less than
presumed to be on anti-psychotic drug treatment 5 % of the time). It should be noted that nearly
continuously, i.e. 76-100% of the length of the one-third (31-0 %) of the patients had never been
follow-up period. admitted to hospital. Across the centres, how-
ever, this percentage varied from 0 % in Prague
Proportion of time spent in psychiatric hospital to 91-1 % in Chandigarh (rural area).
In contrast to anti-psychotic medication, the The highest percentages of patients with no
total length of time during which a patient is hospital admissions during the follow-up were,
admitted to hospital can be determined with apart from rural Chandigarh, in the urban area
accuracy. Although the probability of occur- of Chandigarh (80-7%) and in Agra (73-7%).
rence and the length of an hospital admission Higher rates of hospitalization occurred in
may be influenced by the availability of beds and several of the centres in developed countries, e.g.
by the pressure of the local caseload, none of the Nagasaki and Aarhus, which had the highest
centres participating in the study reported any proportions of patients treated in hospital for
serious difficulties in admitting project patients 46-100% of the period (18-5% and 12-6%
when necessary. It can be assumed, therefore, respectively), whereas Prague and Moscow had
that in most of the centres, in both developed the highest proportions (87-3% and 82-4%
countries and developing countries, the pro- respectively) of patients hospitalized for 6-45 %
portion of time during which patients were in of the follow-up period.
hospital was related to the severity of symptoms
and the degree of social dysfunction. Unimpaired social functioning as a proportion
The data (Table 4.6) indicate that, in the of the follow-up period
majority of the study centres, very few patients This variable was assessed on the basis of all
with a diagnosis of schizophrenia are maintained available information (recorded in the FU-
continuously in hospital. In the total sample, PPHS) from the patient, key informants, and
3-2
62 A. Jablensky and others

Table 4.7. Distribution of cases by percentage of the follow-up time during which social functioning
was unimpaired
Percentage of time of unimpaired social functioning
No. of
Centre patients 0 1-5 6-15 16-45 46-75 76-100 Total

Aar 80 350 8-8 100 15-0 31-3 1001

Dub 56 51-8 1-8 14-3 8-9 23-2 1000
Hon 28 53-6 71 71 3-6 14-3 14-3 99-9
Mos 164 58-5 0-6 31 12-2 25-6 99-8
Nag' 0 — — — — — — —
Not 86 30-2 — 3-5 5-8 186 40-7 99-8
Pra 87 17-2 — 11 11-5 25-3 44-8 99-9
Roc 31 581 — — 12-9 6-5 22-6 1001
Agr 76 211 1-3 2-6 1-3 7-9 65-8 100.0
Cal 138 12-3 2-2 8-7 28-3 26-8 21-7 100-0
Cha/R 50 9-6 — 3-9 13 5 26-9 46-2 1001
Cha/U 108 25-0 — 6-5 130 23-2 32-4 1001
Iba 96 9-4 — 21 7-3 15-6 65-6 1000
All 1000 301 0-7 3-9 109 17-7 36-7 1000

' This item was not rated in the Nagasaki Centre.

any other relevant sources. Social functioning
was considered to be unimpaired if, in the
judgement of the rater, the patient's overall
performance of social and occupational roles
was commensurate with that expected of an
' average' person of the same age, sex, social and
educational background, and culture.
Of all patients who completed the follow-up,
36-7 % were rated as unimpaired in their social
functioning for 76-100% of the entire period; a
nearly identical proportion (30-1 %) were in
some degree impaired throughout the follow-up
period (Table 4.7).
The proportions of patients who were un-
impaired for 76-100% of the time ranged from
14-3% in Honolulu to 65-8% in Agra. At the
other extreme, the percentages of subjects who
were socially impaired during the entire period
of the follow-up, varied between 9-4 in Ibadan
and 58-5 in Moscow.
Sex differences in course and outcome
There were surprisingly few sex-related dif-
ferences in course and outcome when data from
centres in developed and developing countries
were aggregated. When considered separately
for centres in developed and developing coun-
tries, some suggestions of gender differences
appeared, e.g. as regards the pattern of course,
there was an excess of female subjects falling
into pattern 1 (single psychotic episode, followed
by a complete remission), in both developed and
developing countries (Table 4.8). In the devel-
oping countries, there was a relative predomi-
nance of males in pattern of course 6 (two or
more psychotic episodes with incomplete remis-
sions between most of them), while in the
developed countries males were over-represented
in pattern 7 (continuous psychotic illness). On
the remaining variables, such as percentage
of the follow-up period spent in psychotic
episodes, percentage time in complete remission,
percentage time on antipsychotic medication,
percentage time in hospital treatment, and
percentage time of unimpaired social function-
ing, there were virtually no differences between
the sexes (Table 4.9) when location was not
considered.
Thus, it could be said that there was an over-
representation of females in the most favourable
pattern of course, and an over-representation of
males in the two least favourable patterns, but
that on the whole the course of schizophrenia,
when analysed from the point of view of
individual variables (i.e. not controlling for
location), exhibited few consistent associations
with the gender of the patient. This conclusion,
however, needs to be qualified in the light of
findings from the multivariate analysis of pre-
dictors of course and outcome, reported below.
Sex did emerge as a predictor, but the magnitude
of the effect was not of an order that would
justify regarding it as a key prognostic factor.
 Table 4.8 A. Pattern of course by sex {percentage distribution): centres in developed countries
Aar Dub Hon Mos Nag Not Pra Roc All developed countries

Both
Pattern of M F M F M F M F M F M F M F M F M F M+ F
course 1 A1 = 47 N = 33 A7 = 30 N = 27 AT = 2 1 A1 = 8 N = 61 A7 = 1 0 3 AT = 3 6 A7 = 34 A7 = 56 A7 = 3 0 A7 = 2 8 A7 = 5 9 A7 = 1 6 N = 15 A7 = 295 A7 = 309 A7 = 604

1 6-4 21-2 13-3 14 8 4-8 9-8 6-8 8-3 2-9 26-8 33-3 17-9 390 12-5 26-7 13-2 18-2 15-7
2 14-9 12-1 13-3 22-2 19-0 12-5 29-5 21-4 27-8 14-7 10-7 100 143 8-5 12-5 26-7 18-6 26-2 17-4
3 21 — 6-7 111 9-5 25-0 1-6 6-8 2-8 2-9 8-9 6-7 10-7 11-9 6-3 — 5-4 7-1 6-2
4 3-3 7-4 — 8-2 16-5 — 3-6 6-8 12-5 31 7-4 5-3
5 14-9 121 200 7-4 9-5 — 3-3 6-8 13-9 26-5 250 200 321 18-6 12-5 200 15-9 13-6 14-7
6 38-3 39-4 23-3 25-9 190 250 19-7 27-2 111 23-5 10-7 10-0 10-7 8-5 250 26-7 19-7 22-7 21-2
7 23-4 15-2 16-7 7-4 28-6 250 24-6 8-7 361 29-4 179 200 7-1 6-8 18-8 — 230 12-3 171
8 — — 3-3 3-7 9-5 12-5 3-3 5-8 — — — — 3-6 — — — 2-0 2-6 2-3
a,
Total 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 a.
i
See Table 4.2 for definition of numbered patterns.

Table 4 .8 B. Pattern of course by sex {percentage distribution): centres in developing countries

Agr Cal Cha/R Cha/U Iba All developing countries
alth ( jrgamzatic

3
Pattern of M F M F M F M F M F M F Both M + F
course 1 A7 = 49 •7 N = 90 A7 = 5 0 25 A7 = 25 A7 = 60 N = 50 N = 5;i A7 = 43 A7 = 279 Af= 195 A7 = 474 §*
1 55-1 481 200 32-0 48-0 360 18 3 38-0 43-6 60-5 33-3 42-6 371
2 — — 24-4 200 80 12-0 13 3 120 5-5 2-3 12-5 10-3 11-6
3 4-1 3-7 60 40 280 8-3 140 5-5 4-7 3-9 10-3 6-5
4 — — 1-1 40 80 80 50 — — 2-3 2-1 2-6 2-3
5 18-4 22-2 18-9 180 40 120 16-7 200 25-5 25-6 18-3 200 190 Co
6 61 — 14-4 80 200 18-3 8-0 16-4 2-3 14-7 4-6 106 K
7 14-3 25-9 200 120 40 40 11-7 80 1-8 2-3 12-2 9-7 11-2
8 — 8-3 — 1-8 — 11
9 — — 11 — 40 — — — 1-8 — 11 — 0-6
Total 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 100-0 1000

1
See Table 4.2 for definition of numbered patterns.
64 A. Jablensky and others

Table 4.9. Distribution of selected course variables (percentages) by sex: all patients with a
follow-up, all centres
Percentage of the follow-up period

Course variables 0 1-5 6-15 16-45 46-75 76-100 Total

Percentage time in psychotic episodes M(N = 610) — 20-3 32-1 210 7-9 18-7 1000
F(JV = 514) 18 9 36-4 220 5-4 17-3 1000
M + F(JV= 1124) 19-7 341 21-4 6-8 181 1000
Percentage time in complete remission M(N = 609) 44-2 0-3 3-4 100 13-8 28-2 1000
F(AT = 511) 40-7 1-2 10 9-8 17-8 29-5 1000
M + F(iV= 1120) 42-6 0-7 2-8 9-9 15 6 28-8 1000
Percentage time on antipsychotic medication M(/V = 610) 41 7-0 131 16-8 19-9 39-2 1000
F(Af = 514) 61 7-4 12-7 20-4 14-9 38-6 1000
M + F(JV = 1124) 50 7-2 12-9 18-4 17-6 38-9 1000
Percentage time in hospital treatment M(AT = 610) 30-3 210 28-5 151 3-3 1-8 1000
F(/V = 513) 34-6 19-8 25-2 17-3 2-7 0-4 1000
M + F(/V = 1125) 32-3 20-4 27-0 161 3-0 1-2 1000
Percentage time of unimpaired social M(Af = 571) 30-5 0-4 5-8 11-7 16-8 34-8 1000
functioning F(JV = 475) 28-8 1-0 1-2 11-4 18-9 38-5 1000
M + F(JV= 1046) 29-7 0-7 3-7 11-6 17-8 36-5 1000

Differences between developing countries and Table 4.10. Percentage of patients in the
developed countries developing countries and in the developed
The examination of the follow-up results for countries falling into selected categories of course
such differences was an important task, con- and outcome variables
sidering the findings of the IPSS which indicated,
Developing Developed
for the first time on a large scale and with the use Course and outcome category countries countries
of standardized methods, that the course and
outcome of disorders diagnosed as schizophrenic 1 Remitting course with full remission 62-7 36-8
(1 + 3 + 5)
were more favourable in the developing coun- Continuous or episodic psychotic 35-7 60-9
tries than in the developed countries. In view of illness, without full remission
(2 + 4 + 6 + 7)
the importance of replicating these findings, this 2 In psychotic episodes 1-5 % of FU 18-4 18-7
issue was addressed in the Outcome Study on a period
larger and more representative series of patients, In psychotic episodes 76-100% of FU 15-1 20-2
period
and with more refined methods. The principal 3 In complete remission 0 % of FU 24-1 57-2
results in this respect, from the point of view of period
simple univariate analysis, are presented on In complete remission 76-100% of FU 38-3 22-3
period
Table 4.10 which shows proportions of patients 4 No antipsychotic medication 5-9 2-5
who had met the inclusion criteria and during throughout FU
the follow-up fell into the extreme ends of the On antipsychotic medication 15-9 60-8
76-100% of FU period
distributions of the six major variables describing 5 Never hospitalized 55-5 8-1
course and outcome. Hospitalized for 76-100% of FU 0-3 2-3
period
As regards the 'best possible' outcomes, in 6 Impaired social functioning 15-7 41-6
five out of six comparisons, the proportions of throughout FU
Unimpaired social functioning for 42-9 31-6
patients in the centres in developing countries 76-100% of FU period
falling into these categories are considerably
higher than the proportions of patients in the
centres in developed countries. For example, the who were symptom-free (in complete remissions)
percentage of patients in the developing coun- for over three-quarters of the length of the
tries who exhibited a remitting course over the follow-up period was 38-3 in the developing
two years of the follow-up (i.e. patterns 1, 3 and countries and 22-3 in the developed countries.
5), was 62-8, as compared with 36-9 in the Similarly, the percentage of patients in devel-
developed countries. The percentage of patients oping countries who functioned without social
 Schizophrenia: A World Health Organization Ten-Country Study 65

Table 4.11. Pattern of course by initial diagnostic classification of the cases

{percentage distribution)
Pattern of course
Diagnostic classification
on initial examination 1 2 3 4 5 6 7 8 9 Total N

CATEGO class S + 24-0 16-9 6-4 4-0 150 161 15-7 1-7 0-2 1000 626
CATEGO classes S, P, O + 24-3 16-2 6-8 4-2 15-4 15-9 15-0 2-1 01 1000 859
CATEGO classes S, P, O 24-9 15-4 7-1 4-3 15-6 15-8 14-5 2-2 0-2 1000 968
Clinical ICD-9 diagnosis or 25-4 15-3 7-1 4-0 15-8 160 14-1 2-0 0-3 1000 1134
CATEGO classes S, P, O

1, Single psychotic episode, complete remission; 2, Single psychotic episode, incomplete remission; 3, Single psychotic episode one or more
non-psychotic episodes, complete remissions; 4, Single psychotic episode, one or more non-psychotic episodes, incomplete remission; 5, 2 +
psychotic episodes, complete remissions; 6, 2 + psychotic episodes, incomplete remissions; 7, Continuous psychotic illness; 8, Continuous
non-psychotic illness; 9, Missing data.

impairment for 76-100% of the time was 42-9, schizophrenia identified different two-year pat-
compared with 31-6 in the developed countries. terns of course and outcome. The four levels of
The only category for which no difference was diagnostic definition were: (i) clinical ICD-9
found between developing and developed coun- diagnosis of schizophrenia or of a specified
tries was that of the proportion of cases with related disorder, or CATEGO class S, P, or O;
relatively brief psychotic illnesses, i.e. with a (ii) CATEGO classes S, P, or O; (iii) CATEGO
total length of time in psychotic episodes less classes S, P, or O + ; and (iv) CATEGO class
than 5% of the follow-up period (18-4 in the S +. It has been shown that each one of these
developing countries and 18-7 % in the developed four alternative diagnostic definitions, and es-
countries). pecially (i) and (iv), was related to a different
In the range of categories characterizing the level of severity of the florid or 'positive'
'worst possible' outcome, the proportions of psychotic symptoms of schizophrenia. If
patients in the centres in developed countries diagnosis-related differences in the course and
were consistently higher than the corresponding outcome of the disorders could likewise be
proportions of patients in the centres in devel- demonstrated, the hypothesis that the diagnostic
oping countries: 38-3% compared to 21-6% as classification of schizophrenia possesses predic-
regards cases with continuous or episodic psy- tive validity would receive considerable support.
chotic illness without complete remission; 41-6 %
compared to 15-7% as regards presence of Diagnostic inclusion criteria and course and
impaired social functioning throughout the outcome
follow-up; and 20-2% compared to 15-1% as Table 4.11 provides a clear answer to this
regards being in psychotic episodes for 76-100 % question: there are virtually no differences
of the length of the follow-up. between the percentage distributions over the
Comparisons on a centre by centre basis, as different categories of the variable pattern of
regards various course and outcome variables, course between the patients series meeting each
are generally consistent with the overall trend one of the four sets of inclusion criteria of
outlined above. Such comparisons, however, are 'caseness'. The 'restrictive' definition of schizo-
phrenia based on CATEGO S+ on initial
best considered in the context of other issues and
the relevant data are presented in the remaining examination does not select cases that are in any
sections of this chapter. way different, as regards pattern of course, from
the cases identified by the 'broad' diagnostic
category based on either a clinical ICD-9
DIAGNOSIS AND SUBSEQUENT diagnosis or on a CATEGO class S, P, or O. Put
COURSE AND OUTCOME in a different way, this finding suggests that the
An important question concerns the extent to pattern of course is unrelated to the degree of
which ' caseness' for schizophrenia as denned by symptomatological specificity of the inclusion
each one of the four diagnostic definitions of criteria for schizophrenia adopted in this study.
66 A. Jablensky and others

Paranoid schizophrenia (295.3) Acute schizophrenia episode (295.4)

Remitting Remitting, Unremitting Remitting Remitting,

residual residual
symptoms symptoms

Paranoid (297), psychogenic and unspecified

All other schizophrenia sub-types (298.3-9) psychoses

Remitting Remitting, Unremitting Remitting Remitting, Unremitting

residual residual
symptoms symptoms
FIG. 4.2.1. Pattern of course (all patients with a follow-up) by clinical diagnosis made atfieldresearch centre on initial examination.
Remitting: patterns I, 3 and 5; Remitting, residual symptoms: patterns 2, 4 and 6; Unremitting: pattern 7. In Figs 4.2.1-4.2.6,
• indicates 'developed countries', S 'developing countries'.

psychiatrists in the field research centres on the

Course and outcome according to clinical b a s i s o f t h e i n i t i a i examination PSE, previous
diagnostic subtype history, and any other data, bear any prognostic
The next question to be considered is whether implications. For the purposes of this analysis
the different clinical diagnoses, made by the all diagnostic assessments made on initial exam-
 Schizophrenia: A World Health Organization Ten-Country Study
ination (resulting in the assignment of a main
clinical diagnosis) were grouped into five cate-
gories, with a view to ensuring a sufficient
number of cases within each group. The five
categories are:
(1) schizophrenia, paranoid type (ICD 295.3)
- 2 6 1 patients;
(2) acute schizophrenic episode (ICD 295.4)
- 2 1 8 patients;
(3) all other types of schizophrenia listed in
ICD-9 (i.e. simple, hebephrenic, catatonic,
latent, residual, schizoaffective, other, and
unspecified) - 426 patients;
(4) paranoid states (ICD 297), acute paranoid
reaction (ICD 298.3), psychogenic para-
noid psychosis (ICD 298.4), other and
unspecified reactive psychosis (ICD
298.8), unspecified psychosis (ICD 298.9)
- 82 patients;
(5) all other diagnoses (i.e. paranoid or
hallucinatory states induced by alcohol or
drugs - ICD 291.3,291.5,292.1; paranoid
or schizoid personality disorder - ICD
301.0, 301.2)-91 patients.
The percentage distributions of the patients
classified into the first four of these ICD-9
diagnostic groups are shown on Figs 4.2.1^6.
There is a clear difference in the pattern of
course distributions of cases diagnosed as para-
noid schizophrenia and cases diagnosed as acute
schizophrenic episode (Fig. 4.2.1). The seven
principal patterns of course were grouped into
fully remitting patterns (1,3, and 5); remitting
with residual symptoms (2, 4, 6); and un-
remitting (7). Patients with a diagnosis of
paranoid schizophrenia show a considerably less
favourable distribution on the pattern of course
variable than patients with a diagnosis of an
acute schizophrenic episode. The direction of
this observed difference in prognosis between
paranoid and acute schizophrenia is the same
for patients in developing countries and for
patients in developed countries. However, the
size of the difference between the two diagnostic
groups is greater within centres in developed
countries. On the other hand, the general trend
for patients in developing countries towards
better course and outcome is so pronounced
that, in fact, patients with a diagnosis of
paranoid schizophrenia in the developing coun-
tries show a more favourable pattern of course
than patients with acute schizophrenic episodes
in the developed countries.
67
As regards the residual group of all other
schizophrenia types, its breakdown by pattern
of course indicates that these cases are closer to
the paranoid type than to any other category in
the developed countries, while in the developing
countries they are more similar to the cases of
acute schizophrenic episodes. The most favour-
able pattern of course is seen in the 4th group of
paranoid states, psychogenic paranoid and
unspecified psychoses.
Paranoid Acute schizo- All other Paranoid (297),
schizophreniE phrenic epi- schizophrenia psychogenic and
(295.3) sode (295.4) subtypes unspecified psy-
chosis (298.3-9)
FIG. 4.2.2. Percentage of FU period in psychotic episodes (mean
and s.D.) by clinical diagnosis on initial examination. (See
Fig. 4.2.1.)
The distributions of the mean percentages of
follow-up time during which patients were in
psychotic episodes (Fig. 4.2.2) similarly indicate
a favourable trend for acute psychotic episodes
and an unfavourable trend for paranoid schizo-
phrenia. In congruence with this finding, the
mean percentage time in complete remission
(Fig. 4.2.3) is greater for patients with acute
schizophrenic episodes than for patients with
paranoid schizophrenia. Both patients with
paranoid schizophrenia and patients with acute
schizophrenic episodes spent relatively brief
mean periods of time (as percentage of the
follow-up) in hospital (Fig. 4.2.4), although on
this measure, too, the trend is less favourable for
paranoid schizophrenia. There is a marked
difference, as regards this variable, between
PSM
68 A. Jablensky and others

100

80.5
(31.0)
80 71.3
(34.1)
63.5
(34.5)
60

40 34.3
(28.0)

20

Paranoid Acute schizo- All other Paranoid (297), Paranoid Acute schizo- All other Paranoid (297),
schizophrenia phrenic epi- schizophrenia psychogenic and schizophrenia phrenic epi- schizophrenia psychogenic and
(295.3) sode (295.4) subtypes unspecified psy- (295.3) sode (295.4) subtypes unspecified psy-
chosis (298.3-9) chosis (298.3-9)
FIG. 4.2.3. Percentage of FU period in complete remission (mean FiG. 4.2.5. Percentage of FU period on anti-psychotic medication
and s.D.) by clinical diagnosis on initial examination. (See (mean and S.D.) by clinical diagnosis on initial examination. (See
Fig. 4.2.1.) Fig. 4.2.1.)

20 |— 80

15 - 60
51.9
(35.2)

10 - 40 -

20

Paranoid Acute schizo- All other Paranoid (297), Paranoid Acute schizo- All other Paranoid (297),
schizophrenia phrenic epi- schizophrenia psychogenic and schizophrenia phrenic epi- schizophrenia psychogenic and
(295.3) sode (295.4) subtypes unspecified psy- (295.3) sode (295.4) subtypes unspecified psy-
chosis (298.3-9) chosis (298.3-9)
FIG. 4.2.4. Percentage of FU period in hospital admissions (mean
and S.D.) by clinical diagnosis on initial examination. (See FIG. 4.2.6. Percentage of FU period in unimpaired social functioning
Fig. 4.2.1.) (mean and s.D.) by clinical diagnosis on initial examination. (See
Fig. 4.2.1.)

patients in developed countries and patients in latter having even shorter hospital admissions
developing countries falling into the category of than acute schizophrenics. The distributions
other types of schizophrenia, the former being according to mean percentage follow-up time
similar to the paranoid schizophrenics, and the on antipsychotic medication (Fig. 4.2.5) follow
 Schizophrenia: A World Health Organization Ten-Country Study 69

the same general pattern. Finally, the mean Table 4.12. Occurrence of CATEGO class S +
proportions of time during the follow-up, in and classes other than S+ at successive PSE
which the patients' social functioning was examinations during the follow up of patients (a)
unimpaired (Fig. 4.2.6), show again, from a initially categorized as S+ and (b) initially
different perspective, the superior outcome of categorized as not S+
acute schizophrenic episodes as compared to
paranoid schizophrenia. Initial 1-year 2-year
The consistent pattern of the above findings examination follow-up follow-up N %
suggests that the clinical subtyping of schizo-
phrenia according to the ' conventional' clinical S+ S+ S+ 30 8-2
diagnostic criteria has clear predictive impli- s+ NotS + S+ 17 4-6
s+ S+ NotS + 40 10-9
cations, especially as regards the prognostic s+ NotS + NotS + 279 76-2
distinction between the two diagnoses which Total 366 1000
were most frequently made in the present study, (A)
paranoid schizophrenia and acute schizophrenic NotS + S+ S+ 6 2-2
episode. Patients meeting the inclusion criteria NotS + NotS + S+ 11 4-1
NotS + S+ NotS + 7 2-6
for schizophrenia and related disorders in NotS + NotS + NotS + 244 910
developing countries indeed have a more favour- Total 268 1000
able course and outcome on most measures than
patients in developed countries, but the trends
concerning differences in severity between the while only 8-9 % of the patients who, on initial
two diagnostic entities of paranoid schizo- examination, had a CATEGO class other than
phrenia and acute schizophrenic episode are S +, were classified as S + on any one of the
very similar in the two kinds of setting. The follow-up assessments. This indicates that the
results are less clear cut as regards the other symptomatological characteristics defining S +
diagnostic groupings. ('the central schizophrenic syndrome') tend to
recur, regardless of the fact that the occurrence
The CATEGO classification and the course of of Schneiderian first-rank symptoms on initial
schizophrenia examination did not predict a particular pattern
The CATEGO classification which reflects dif- of course of the disorder. This suggests that,
ferences in symptom profiles of patients has while the occurrence of the constellation of
important implications for the interpretation of symptoms defining S+ on a single occasion
data on the first-contact incidence of schizo- carries little prognostic weight, cases in which
phrenia and related disorders in the different first-rank symptoms tend to be present on two
catchment areas of the study. Therefore, the or more consecutive cross-sections of the mental
relationship between CATEGO classes and the state may also differ in some of their other
variables describing course and outcome needs characteristics from cases in which S+ is
an examination. followed or preceded by other CATEGO classes.
As pointed out above, the allocation of From this follows the more general question
the patients included in the study to classifi- whether particular CATEGO sequences or
cation categories based on different subsets of ' strings' composed of the classes assigned on the
CATEGO classes on initial examination did not three consecutive cross-sections of the mental
identify subgroups of schizophrenic patients state of the patients might be significantly
that are different from one another as regards associated with course and outcome.
two-year pattern of course. However, the initial To explore this proposition, the actually
CATEGO classification appeared to be related occurring combinations of CATEGO classes
to another aspect of prognosis. As shown in during the follow-up were listed and ordered on
Table 4.12, 23-7% of the patients who were the basis of purely clinical assumptions (Table
assigned to CATEGO class S + on the basis of 4.13). For example, it was hypothesized that
the initial PSE had the same S + CATEGO class patients who either had one of the three
on at least one of the subsequent two PSE CATEGO classes S, P, or O on each of the three
examinations (at one year and at two years), follow-up cross-sections, or had an S, P, or O
70 A. Jablensky and others

Table 4.13. Types of' strings' of CATEGO classes occurring in the follow-up study
Code
number Combination of CATEGO classes observed on the three examinations N %

10 S, P or O on each of the three occasions, or 365 35-2

S, P or O on any two occasions and missing PSE on remaining occasion
9 S, P or O on any two occasions; on the remaining one occasion: A, B, X, NO 119 11-5
81 S, P or O on one occasion; on the remaining two occasions: either missing PSE data, or A, B, X, 140 13 5
or NO and missing PSE data
8 S, P or O on one occasion; on the remaining two occasions: either twice A or B, or twice X or 168 16-2
NO, or A, B and X, NO
7 S, P or O on one occasion and M on another occasion, with missing PSE data on the remaining 28 2-7
one occasion; or
S, P or O on two occasions and M on the remaining one occasion
6 S, P or O on one occasion and D, R or N on another occasion, with missing PSE data on the 89 8-6
remaining one occasion; or
S, P or O on two occasions and D, R or N on the remaining one occasion
5 S, P or O on one occasion and M on both remaining occasions; or 28 2-7
S, P or O on one occasion, M on another occasion, and either D, R, N or A, B, X, NO on the
remaining one occasion
4 S, P or O on one occasion, D, R, or N on another occasion, and either D, R, N or A, B, X, NO 43 4-2
on the remaining one occasion
3 M on one occasion, either M or D, R, N or A, B, X, NO on another occasion, and missing PSE 22 21
data on the remaining one occasion; or
M on at least one occasion, and any combination of M, D, R, N, A, B, X, NO on one or two
occasions
2 D, R or N on one occasion, missing PSE data on the remaining two occasions; or 33 3-2
D, R, or N on one occasion and either D, R, N or A, B, X, NO on another occasion, with
missing PSE data on the remaining one occasion, or
D, R, or N on at least one occasion, and any combination of D, R, N, A, B, X, NO on one or
two occasions
1 A or B on one occasion, A, B or X, NO on another occasion, with missing PSE data on the 2 01
remaining one occasion; or
A or B on at least one occasion, and any combination of A, B, X, NO on one or two occasions
1037 1000

S, Schizophrenic psychosis; P, Paranoid psychosis; O, Borderline and doubtful psychosis; M, Manic and mixed affective psychosis; D,
Depressive psychosis; R, Retarded depression; N, Neurotic depression; A, Anxiety state; B, Obsessional neurosis; H, Hysterical condition;
X, Other; NO, No abnormality.

class on any two occasions and missing data on 7, 6 Schizoaffective disorder

the other one, would be different, as regards
course and outcome, from patients who had an
S, P, or O class on one occasion only, and a non-
psychotic CATEGO class on the remaining two
occasions. The rest of the syndrome list presented
in Table 4.13 was constructed in a similar way,
and descriptive clinical labels were assigned to
the different 'strings' before examining the
course and outcome of the patients with those
'strings'. The clinical labels were chosen as a
matter of convenience only, and have no
terminological implications outside this context,
The different combinations of CATEGO classes
were grouped according to the following clinical
concepts.
CATEGO 'string' Corresponding clinical con-
cept
10, 9 Schizophrenia
8.1, 8 Schizophrenia-like disorder
5, 4 Atypical affective disorder
3 Bipolar affective disorder
2 Unipolar affective disorder
1 Neurotic disorder
The numbers of patients and the percentages
given in Table 4.13 indicate that with such use of
the CATEGO classification (i.e. considering only
the serial or consecutive PSE data and ignoring
other diagnostically relevant information), not
more than 5-5 % of all included patients remain
outside those sequences of CATEGO classes in
which there is at least one S, P, or O. When the
frequency with which each of the CATEGO
' strings' occurred in the course of the study in
the individual catchment areas is examined it
can be seen that 'string' 10 ('schizophrenia')
was clearly predominant in all the centres in
developed countries. In the centres in the
developing countries 'string' 8 ('schizophrenia-
 Schizophrenia: A World Health Organization Ten-Country Study
Prague
Nottingham
Nagasaki
Moscow
Honolulu
Dublin
Aarhus
FIG. 4.3. Percentage distribution of specified sequences of CATEGO classes on initial and on two follow-up examinations
by centre.
8.1 9 10
FIG. 4.4. Type of onset by sequence of CATEGO classes. •
Acute; g|, subacute; M, gradual; H , other or not rated.
like psychosis') was modal in Agra, Chandigarh
(rural and urban), and Ibadan, but not in Cali;
in all of these centres, however, considerable
percentages of the cases fell into categories 10
and 9. If anything, these data may indicate that
the symptomatological manifestations of psy-
chosis in the centres in developed countries tend
to be more consistent over time than psychotic
disorders in patients in the developing countries;
further confirmation of thisfindingwill however
be necessary before it can be accepted with
certainty.
The pooling of the data on patients in all of
the centres shows that the CATEGO 'strings'
are strongly associated with the mode of onset
of the disorder (Fig. 4.4). For example, 50 or
more per cent of the patients classified into types
8 (' schizophrenia-like'), 2 (' unipolar affective'),
71
6 7 8 8.1 9 10
„ ^.1 9 10
FIG. 4.5. Pattern of course by sequence of CATEGO classes. • .
Mild, patterns 1-3; ^ , intermediate, patterns 4, 5; U, poor,
patterns 6, 7.
and 1 (' neurotic') had an acute onset, in contrast
to types 10 ('schizophrenia'), and 7 and 6
(' schizoaffective'), in which over 50% of the
patients had gradual onset.
No less important is the observed association
between CATEGO ' strings' and the pattern of
course. Fig. 4.5 indicates that 'mild' patterns of
course (i.e. patterns 1, 2, and 3) characterize
predominantly the disorders falling into the
categories 1 ('neurotic'), 3 ('bipolar affective'),
4 ('atypical affective'), and 8 ('schizophrenia-
like'), in contrast to the 'poor' patterns of
course occurring most frequently in patients
classified into categories 7 ('schizoaffective')
and 10 ('schizophrenia').
72 A. Jablensky and others

Table 4.14. Variables used in the analysis of predictors of course and outcome and number of
patients for whom data were available
Source of No. of
Variables Categories information patients

Outcome variables
Pattern of course Good - patterns 1-3 Synopsis table 500
Intermediate - patterns 4-5 222
Poor - patterns 6-7 334
Percentage follow-up time psychotic < 15% Synopsis table 527
16-44% 266
> 45 % 276
Percentage follow-up time in complete < 15% Synopsis table 491
remission 16-44% 100
>45% 475
Percentage follow-up time in incomplete < 15% Synopsis table 571
remission 16-44% 168
> 45 % 325
Percentage follow-up time when social < 15% Synopsis table 347
functioning was unimpaired 16-44% 109
> 45 % 544
Percentage follow-up time in hospital <5% Synopsis table 556
6-15% 298
> 15% 224
Percentage follow-up time on antipsychotic < 15% Synopsis table 265
medication 16-44% 196
> 45 % 617
Explanatory variables
Gender Male PPHS 574
Female 504
Age < 25 years PPHS 509
> 25 years 569
Marital status Single, widowed, divorced, married 655
Common-law marriage, separated 408
Type of household One person, unrelated persons PPHS 116
Nuclear family 694
Extended family, joint family 231
Setting (level of industrialization) Developing country PPHS 604
Developed country 474
Frequency of contact with relatives None PPHS 426
Rare 328
Frequent 273
Frequency of contact with close friends None PPHS 598
Rare 251
Frequent 171
Frequency of contact with casual friends None PPHS 458
Rare 371
Frequent 171
Avoidance of patient by family members None PPHS 705
Some 138
Marked 52
Avoidance of patient by relatives None PPHS 634
Some 74
Marked 26
Increased contact 46
Avoidance of patient by close friends None PPHS 457
Some 74
Marked 36
Increased contact 16
Avoidance of patient by casual friends None PPHS 556
Some 91
Marked 57
Increased contact 14
Affective relationship to spouse (among Never had close relationship PPHS 112
married, common law, separated patients) (No interest shown)
Never close but showed interest 66
Only casual contact 62
Relationship before onset only 164
Relationship after onset 43
 Schizophrenia: A World Health Organization Ten-Country Study 73

Table 4.14. {com.)

Source of No. of
Variables Categories information patients

Overall adjustment in childhood Good adjustment PPHS 690

Transient problems 166
Persistent problems 32
Overall adjustment in adolescence Good adjustment PPHS 593
Transient problems 200
Persistent problems 74
Street drug use No use, none suspected PPHS 878
Sporadic use known or suspected 63
Five or more instances known or suspected 82
Number of months since onset of disorder Continuous variable PPHS 986
CATEGO class S+ PSE +CATEGO program 594
Not S + 484
Main diagnosis
Group 1 295.3 DPS 307
Group 2 295.1, 295.6 111
Group 3 296.7, 295.4, 295.2 328
Group 4 Other schizophrenia 159
Group 5 Non-schizophrenia (other diagnosis) 173
Type of onset Acute Synopsis table 405
Sub-acute 216
Slow 402

The conclusion which can be drawn from
these data is that, while the CATEGO clas-
sification of PSE symptoms at the point of the
initial examination does not predict the sub-
sequent pattern of course, the sequential pattern
of CATEGO classes determined on two or three
follow-up examinations is predictively associ-
ated with an independently derived measure of
the course of the disorder. Classes S, P, and O,
if and when they recur, are clinical markers of a
relatively poor prognosis, while the appearance
of affective or neurotic CATEGO classes at any
point in time, with or without an S, P, or O class
on a single occasion, is associated with a more
favourable evolution of the disorder.
It may be argued that in a general sense the
CATEGO classes and the different patterns of
course are not entirely unrelated (e.g. a
CATEGO class representing psychosis could be
a priori expected to be associated with a less
favourable course than a class representing a
neurotic illness). However, apart from the
general criteria defining a psychotic and a non-
psychotic episode, no specific symptomatology
data were used in the operational definition of
the pattern of course; the assessment of the
latter was based only on a review of the
longitudinal characteristics of the disorder. In
contrast, the PSE/CATEGO assessment, as used
in this study, presupposes the sampling of
symptoms on a short-term (one month) basis,
without any reference to the history of symptoms
outside that period. Thefindingof an association
between the pattern emerging from consecutive
'point' assessments by CATEGO, and the
pattern emerging from longitudinal data should
qualify the statement made above about the lack
of predictive power for the CATEGO classifi-
cation. It should also caution against the
conclusion that symptomatology is not pre-
dictive of course and outcome; such a conclusion
could only be made if a single cross-section of
the mental state were considered. The data
reported here advise strongly against prognostic
judgements based on isolated cross-sections of
the disorder.
PREDICTORS OF TWO-YEAR COURSE
AND OUTCOME
The principal method used to identify predictors
of two-year course and outcome was that of log-
linear analysis. Log-linear modelling is a general
term applying to a set of statistical techniques
developed for analysing multidimensional cross-
classifications. The particular technique used
here is known as polytomous logit, or a log-odds
model. It predicts the log-odds of a certain
outcome measured by a discrete variable by
using a linear combination of effects due to the
explanatory variables.
The log-odds model makes no distributional
 Table 4.15. Effect of type of onset and setting on seven outcome variables
Effect on good outcome Effect on poor outcome

Estimated Estimated
partial partial
Outcome variable Explanatory variable N Coefficient 2 derivative Coefficient 2 derivative
x x
Pattern of course Type of onset
Acute 403 + 0-445*** 23-88 + 0111 -0-543*** 24-92 -0116
Sub-acute 216 + 0052 0-25 + 0-013 -0197 2-74 -004
Gradual* 389 -0-497 -0-124 + 0-740 + 0158
Setting &.
Developing country* 449 + 0-227 + 0057 -0-290 -0062
Developed country 559 -0-227*** 11-43 -0057 + 0-290*** 14-63 + 0062 i?
Percentage follow-up time in psychotic episode Type of onset a:
Acute 404 + 0-540*** 34-27 + 0-135 -0-573*** 22-77 -0107 I
Sub-acute 215 + 0163 2-47 + 0-041 -0-314* 5-50 -0059
Gradual* 398 -0-703- -0-175 + 0-887 + 0166 &
Setting R.
Developing country* 450 -0048 -0010 + 0149 + 0-028
Developed country 567 + 0048 0-50 + 0010 -0149 3-50 -0028 3<^>-
Percentage follow-up time in complete Type of onset 3
remission Acute 400 + 0-495*** 28-32 + 0123 -0-458*** 21-96 -0113
Sub-acute 210 + 0-240* 509 + 0-059 -0-330** 8-71 -0-082
Gradual* 405 -0-735 -0-182 + 0-788 + 0-195
Setting
Developing country* 571 + 0-355 + 0-880 -0-524 -0130
Developed country 444 -0-355*** 26-80 -0088 + 0-524*** 54-28 + 0130
Percentage follow-up time in incomplete Type of onset
remission (and non-psychotic episode) Acute 400 -0020 005 -0005 -0-047 0-21 -0010
Sub-acute 210 -0068 0-42 -0017 -0-009 001 -0002
Gradual* 404 + 0088 + 0-022 + 0-056 + 0-012
Setting
Developing country* 596 + 0-306 + 0076 -0-542 -0114
Developed country 445 -0-306*** 20-72 -0076 + 0-542*** 48-23 + 0114
Percentage follow-up time in hospital Type of onset
Acute 402 + 0162 2-29 + 0040 -0054 018 + 0008
Sub-acute 213 -0019 0-02 -0005 -0128 0-81 + 0021
Gradual* 411 -0143 -0036 + 0-182 -0-030
Setting
Developing country* 574 + 1-243 + 0-310 -1-423 -0-235
Developed country 452 -1-243*** 245-49 -0-310 + 1-423*** 91-30 + 0-235
 Schizophrenia: A World Health Organization Ten-Country Study 75

assumptions and can handle interactions be-

tween variables easily. It differs from the more
O\ t m
conventional methods of analysing the relation-
tt (N
© o
N
©
o ©
ships between categorical variables, such as the
© © © © © 6 © ©
I + + I + I I + I + chi-square tests, in the manner in which the
dependent variable is represented in the model.
While the traditional methods represent the
dependent variables in terms of proportions of
subjects falling into specified categories, the log-
odds model uses the natural logarithm of the
ratio of frequencies for any two categories of the
ON ON dependent (outcome) variable. The log-odds
r- r-
O ©
I + +
© o ©
©
I
© ©
I +
© o
I +
model, therefore, explores how the odds (expres-
I +
S sed in a log form) that a subject will appear in
o one outcome category rather than in another are
linked to the explanatory, or predictor, variables.
W> ( N
909
rl

6 ©
—
©
©
©
—
©
©
©
©
©
By comparing the natural logarithms of the
© ©
+ I I
©
+ I + + i +i odds, rather than the proportions of cases in
each category, the log-odds model has the effect
that the distance between the proportions is
' stretched' as they approach the values of zero
and one. For example, in a log-odds model a
distance between 0-1 and 0-2 will be about twice
the distance between 0-5 and 0-6 (Kritzer, 1979).
-
m
©
^-
©
CN
The log-odds model for polytomous data,
©
+
© ©
I I
©
+ i
© © ©
when the dependent variable has n categories,
can be written as a set of n — 1 equations:

ln =M
T3

where j represents categories of the dependent

variable, P} is the frequency of responses in
St
category j , u} is a consonant, X(s are the
explanatory variables, and fifjs are the coef-
ficients to be estimated. The model was imple-
mented using maximum likelihood techniques
)SDQ and the SAS version 5 CATMOD procedure.
The results presented below include both the
estimated coefficients fii} as well as the estimated
partial derivative evaluated at the sample mean
which can be used to approximate the percentage
point change in the dependent variable per unit
change in the explanatory variable. (For cate-
gorical variables the formula for the estimated
partial derivative is an approximation - Peterson
& Kronmal, 1985.)
S.S Table 4.14 lists all variables used in the
II predictor analysis and their categorization.
Seven measures of course and outcome were
used, namely pattern of course, percentage of
the follow-up period during which the subject
76 A. Jablensky and others
was in psychotic episodes, percentage of the
follow-up period in complete remissions, per-
centage of follow-up period in incomplete
remissions, percentage of the follow-up period
in hospital admissions, percentage of the follow-
up period on anti-psychotic medication, and
percentage of the follow-up period in unimpaired
social functioning.
The main reason for using seven different
outcome variables was that one variable, by
itself, could not capture the multi-faceted nature
of outcome. It was felt that examining a profile
of outcome variables will give a much better
understanding of the course and outcome of the
disorder. In the analysis, not only was statistical
significance considered, but also the consistency
of the relationship with seven outcome measures.
Thus, if an explanatory variable had a con-
sistently significant relationship with the ma-
jority of outcome variables, this was considered
strong evidence for believing that the variable in
question was indeed an important predictor of
the outcome of the disorder. If the relationship
was inconsistent, the evidence was considered to
be weaker.
The analysis was carried out in two steps.
First, the relationship between outcome and
type of onset (acute, sub-acute, gradual) and the
setting (developed/developing country) was
examined. As the second step, other explanatory
variables were tested, one at a time, controlling
for the type of setting (developed/developing)
and type of onset.
EFFECTS OF THE PRINCIPAL
EXPLANATORY VARIABLES
The results of the first step of the analysis in
which type of onset and type of setting
(developed/developing country) were used as
explanatory variables are presented in Table
4.15. The first response function in these tables
concerns predictors of good outcome, and the
second response function concerns predictors of
poor outcome. The coefficients for the log-linear
model are presented, as well as the estimated
partial derivatives which measure the estimated
percentage point change in the probability of
having either a good outcome or poor outcome
for a unit change in the explanatory variable.
The partial derivatives are obtained by multi-
plying the coefficient by P(\ — P), where P is
the mean probability of good outcome or poor
outcome depending on the response function.
The type of setting, (i.e. centres in developed
countries or centres in developing countries) was
shown to be a highly significant predictor of all
outcome measures with the exception of the
percent time spent in psychotic episodes. Patients
living in developing countries were significantly
more likely to have a more favourable outcome
than patients living in developed countries, on
six of these outcome measures. Because of the
correlation between the variable describing the
setting (developed/developing country) and
many of the other explanatory variables it was
necessary to control for the variable 'setting'
when examining other predictors.
Three categories of the type of onset were
considered: acute (development of florid psy-
chotic state within a week); sub-acute (de-
velopment of a psychotic state within a month);
and gradual (slow, incremental development of
psychotic symptoms over periods exceeding one
month). Type of onset stood out as a significant
predictor of five out of the seven outcome
variables, namely the pattern of course, per-
centage of time spent in psychotic episodes,
time (%) spent in complete remissions, time (%)
spent in unimpaired social functioning, and time
(%) on anti-psychotic medication. Patients with
acute onsets had the most favourable outcomes
on these five measures, while patients with
gradual onsets had the least favourable out-
comes. Those with sub-acute onsets were in the
middle.
Since mode of onset appeared to be a highly
significant predictor, and the proportion of
patients with acute, sub-acute and gradual onsets
varied between centres, it was decided to control
for the type of onset in all subsequent analysis.
It should be mentioned that the interactions
between type of onset and ' setting' (developed
or developing country) were examined and found
to be insignificant for two of the most important
outcome variables, pattern of course and per-
centage of time psychotic. (They were not
examined for the other outcome variables.)
ADDITIONAL EXPLANATORY
VARIABLES
The results of the second step of the predictor
analysis are presented in Tables 4.16 and 4.17.
 Schizophrenia: A World Health Organization Ten-Country Study 77

Table 4.16. Summary of predictors and cause-and-outcome variables: predictors of good outcomes
Course and outcome variables

% of FU
% of FU in % of FU in % of FU in % of FU % of FU unimpaired
Pattern psychotic complete incomplete in antipsychotic social
Predictors of course episodes remission remission hospital medication functioning"

Age NS NS NS NS • •
NS *
Sex * * NS NS NS *
Marital status ** *•* *** ***
Type of household NS NS NS NS NS NS s
Setting (developed v. developing country) *** NS ** *** *** *•• * It*
Type of onset *** *** ** NS NS ** * 1*
N months since onset NS NS NS NS NS * fc*
Diagnostic group (ICD-9) * • * ** **• NS • *• * :*
^ATEGO class on initial examination NS NS s NS * NS S
Adjustment in childhood NS NS NS NS NS • 1"*
••• X*
Adjustment in adolescence NS ** NS NS *
Drug use * •
** NS NS • * :*
Heterosexual relationships NS NS NS * NS NS
Affective relationships with spouse NS NS s NS NS NS NS
Contact with relatives NS NS NS NS NS *
Contact with close friends * * * NS *• • * #**
Contact with casual friends * • • NS NS **• *
Avoidance by family members NS NS ** NS NS NS
Avoidance by close friends NS NS NS NS NS NS NS
Avoidance by casual friends NS NS NS NS NS NS NS

* Significant at P ^ 0-05; ** significant at P < 0 0 1 ; *** significant at P < 0-001; NS not significant.
a
Not ascertained for Nagasaki.

Each additional variable was entered in a

separate model in which setting and type of
onset were controlled for. When the additional
variables were added to the model, there were
very few changes in the significance of the
coefficients for mode of onset or for setting. The
percentage of time spent in hospital had par-
ticularly large centre specific variation. There-
fore for this outcome measure, there were two
runs: one run controlling for individual centres,
and one run controlling for developed or
developing country setting. Since there were
very few differences between these two runs,
only the runs controlling for setting are reported.
Age
Age was significantly related to only two of the
outcome measures, percentage of follow-up time
in which social functioning was unimpaired and
percentage of follow-up time spent in hospital.
Patients at least 25 years-of-age at the initial
examination were more likely to spend more
than 4 5 % of the follow-up period with un-
impaired social functioning, and were more
likely to have spent very little time ( < 5 % of
follow-up period) in hospital than younger
patients. The number of months since the onset of the
Gender
Gender was significantly related to pattern of
course, percentage of time psychotic, percentage
of time in complete remission, percentage of
time of unimpaired social functioning, and
percentage of time in hospital. Female subjects
tended to have more favourable outcomes than
male subjects on those five outcome variables.
Marital status
Marital status was significantly related to all
seven measures of outcome. Patients who were
married at the point of initial examination
(including married, common law-married, and
married but separated) had the better outcomes
on all measures, and patients not married at the
time of initial examination (i.e. never married,
widowed, or divorced) had the worst outcomes.
Significance was generally high. The consistency
of the relationship and the high levels of
significance indicate that marital status is an
important predictor of outcome.
Number of months between onset and initial
examination
78 A. Jablensky and others

Table 4.17. Summary of predictors and course-and-outcome variables: predictors of poor outcome
Course and outcome variable

% of FU
% of FU in % of FU in % of FU in % of FU % of FU unimpaired
Pattern psychotic complete incomplete in antipsychotic social
Predictors of course episodes remission remission hospital medication functioning"

Age NS NS NS NS * NS NS
Sex ** ** ** NS * NS ***
Marital status *** *** ** NS **• NS • ••

Type of household NS NS NS NS NS NS NS
Setting (developed v. developing country) •*• NS • ** **• *** *•* • •
Type of onset ••* *** *•* NS NS * ***
N months since onset NS * • ** NS NS NS ***
Diagnostic group (ICD-9) • * *** ** NS *** ***

CATEGO class on initial examination NS NS NS * NS NS NS

Adjustment in childhood NS NS ** NS NS NS *
Adjustment in adolescence • ** **• *** NS NS NS *•

Drug use ** ** ** NS NS *> * **

Heterosexual relationships NS NS ** NS NS NS
Affective relationships with spouse NS NS NS NS NS NS NS
Contact with relatives NS NS NS NS NS ** * **
Contact with close friends * *** *** NS NS * • **
Contact with casual friends * **• * NS NS * *
Avoidance by family members NS ** *** NS NS NS ***

Avoidance by close friends NS NS NS NS NS NS NS

Avoidance by casual friends NS NS NS NS NS NS NS

• Significant at 005; ** significant at 0 0 1 ; **• significant at 0-001; NS not significant.

* Not ascertained for Nagasaki.

disorder is particularly difficult to estimate in the ICD 295.7, 295.2 and 295.4 (including the
case of patients with gradual onsets. Therefore, Moscow diagnosis of periodic or recurrent
several consistency checks were applied to the schizophrenia);
data to ascertain the accuracy of the information.(4) other schizophrenia subtypes;
The results indicate that the length of illness (5) other diagnoses.
before initial examination was associated with Diagnosis was a significant predictor of all
the percentage of time in psychotic episodes outcome variables except for percentage of
during the follow-up, percentage of time in follow-up time in hospitals. In general, patients
complete remission, and percentage of time that in group 5 tended to have the best outcomes,
social functioning was unimpaired. The longer and patients in group 2 had the worst outcomes.
the time since onset, the more likely the patient (An exception to this was the percentage of time
was to have spent a longer period of time in a spent in incomplete remission.) Group 3 had the
less favourable state. However, this variable was second best outcome (or a very close third) in
not significantly related to the remaining three terms of pattern of course, percentage of time
outcome variables. psychotic, percentage of follow-up time spent in
unimpaired social functioning, and percentage
Main clinical diagnosis on initial examination of time on anti-psychotic medication. However,
In the predictor analysis, the patients were although group 3 spent relatively little time in
divided into five diagnostic groups: psychotic episode, this group spent the most
(1) paranoid schizophrenia ICD 295.3 (the time in incomplete remission, and relatively little
diagnosis of shift-like progressive type time in complete remission.
according to the Moscow criteria was also
included here); CATEGO class
(2) hebephrenic and residual schizophrenia Patients classified as S + were significantly more
ICD 295.1 and 295.6 (including the con- likely to have spent more than 45% of the
tinuous progressive type according to the follow-up period in incomplete remission and
Moscow criteria); were less likely to have spent < 5 % of follow-up
(3) schizoaffective, catatonic, and acute types time in hospital, than those in other CATEGO
 Schizophrenia: A World Health Organization Ten-Country Study
classes. These were the only statistically signifi-
cant relationships found between the seven
outcome measures, and the classification S + .
Overall adjustment in childhood
The overall adjustment in childhood was signifi-
cantly related to only two of the seven outcome
variables, namely percentage follow-up time in
complete remission, and in unimpaired social
functioning. Those with good adjustments in
childhood had the best outcomes on these two
measures, those with persistent problems had
the worst outcomes, and those with transient
problems were in the middle.
Adjustment in adolescence
Three categories were considered: those with
good adjustment in adolescence, those with
transient problems, and those with consistent
problems. There was a significant relationship
between adjustment in adolescence and five of
the seven outcome variables (the exceptions
were percentage follow-up time on anti-psy-
chotic indication, and percentage follow-up time
in hospital). Those with good adjustment con-
sistently had more favourable outcomes (on
these five variables) than those with persistent
problems. Those with transient problems were
usually in the middle, but not consistently so.
Use of'street' drugs
The use of' street' drugs was significantly related
to five of the seven outcome variables (exceptions
were percentage of follow-up time in incomplete
remission, and percentage of follow-up time in
hospital). For four of the outcome measures the
use of street drugs was less likely to be associated
with a favourable outcome. However, the per-
centage of time on anti-psychotic medication
was shorter for those who used 'street' drugs
than for those who did not.
Heterosexual relationship
For those patients who were single, divorced, or
widowed, information was asked about the
extent of their relationship with the opposite
sex, both before and after the illness. This
variable was not significantly related to four of
the seven outcome variables. However there was
a significant relationship with percentage of
follow-up time in complete remission, percentage
of follow-up time in hospital, and percentage of
time on anti-psychotic medication. Those who
79
were able to have a heterosexual relationship
even after falling ill had the most favourable
outcomes.
Social isolation
There were two ways in which the social isolation
of the patients was considered. The first was
whether other persons in the ' social field' of the
patient (relatives, friends, household members)
were avoiding him or her. There were no
significant relationships found between any of
the outcome variables and avoidance of the
patient by either close or casual friends. There
were three outcome measures that were signifi-
cantly related to avoidance of the patient by
family members; however, these relationships
were not consistent with one another.
The frequency of contact with close friends,
casual friends and relatives was used as another
indicator of social isolation. Patients with
frequent contact with close friends had signifi-
cantly more favourable outcomes on six outcome
measures (the exception being percentage of
follow-up time in incomplete remission). Those
with no contact with close friends had the least
favourable outcomes, and those with rare
contact with close friends were in the middle.
For contact with casual friends the general
pattern was similar, but the levels of significance
were lower, and the relationships were significant
for only five of the outcome measures. (Pattern
of course, percentage follow-up time psychotic,
percentage of time in complete remissions,
percentage of follow-up time that social func-
tioning was unimpaired, and percentage of
time on anti-psychotic medication.) There was a
significant association between frequency of
contact with relatives for only three of the
outcome variables, (percentage time that social
functioning was unimpaired, percentage follow-
up time in complete remission, and percentage
of time on anti-psychotic medication). Here
again, the more frequent the contact, the better
the outcome.
Variables unrelated to any outcome measure
The following variables did not appear to be
significantly related to any of the seven outcome
measures used in the analysis:
(1) type of household;
(2) affective relationship to spouse;
(3) avoidance of the patient by either close or
casual friends.
 Chapter 5 Discussion of the findings and
conclusions
In spite of nearly a century of research, the
epidemiology of schizophrenia still contains
many uncharted or poorly explored areas. First,
the possible existence of differential rates of
occurrence of the disorder in various age groups
and sexes in populations geographically and
culturally apart, has been a matter of con-
siderable interest (Murphy, 1978; Torrey, 1980).
However, until the present time only fragmen-
tary empirical data have been available on this
issue, and the different views expressed resort
more often to speculation rather than to an
epidemiological data base.
Secondly, few clinical studies have been based
on representative samples of cases in early stages
of the illness, before its symptoms and course
have been modified by treatment and, very
likely, by social attitudes. This has hampered
research on the 'natural history' of schizo-
phrenia, which continues to be viewed by many
as a process invariably leading to some degree of
deterioration and disablement. Prognostic indi-
cators that could be used early in the course of
the illness have usually been sought on relatively
small and unrepresentative patient samples, and
often without appropriate data collection instru-
ments and statistical models.
Thirdly, widely diverging views continue to be
held on the scope and boundaries of the
diagnostic concept, as well as on the extent to
which a diagnosis of schizophrenia made in one
setting, or by one ' school' of psychiatry, can be
meaningfully replicated in another setting, or by
another 'school'.
These issues are basic to the very notion of
schizophrenia as a specific psychopathological
condition. It has been pointed out (Shepherd,
1982) that progress in the search for aetiological
factors, pathogenic mechanisms, and disease
markers can be seriously impeded unless the
essential parameters of its incidence, cross-
culturally constant clinical manifestations, and
pattern of course are laid down with sufficient
clarity.
80
THE STRATEGY OF MULTI-CENTRE
COLLABORATIVE RESEARCH
In the instance of a condition of a relatively low
population incidence, such as schizophrenia, the
multi-centre collaborative research has advan-
tages over single centre studies and is more likely
to significantly increase epidemiological and
clinical knowledge. Large numbers of cases can
be accumulated in less time, and the individual
patient samples, identified in different areas,
provide opportunities for multiple replications
of the search for ' robust' characteristics of the
disorder that show constancy and repeatability
in the face of cultural variation and other,
demographic, ecological and biological dif-
ferences between populations.
Such multi-centre collaborative research has
been the hallmark of the mental health pro-
gramme of the World Health Organization in
the past two decades, and the present project on
Determinants of Outcome of Severe Mental
Disorders is part of a series of cross-cultural
investigations which up to date have involved
teams of investigators in over 40 countries.
Together with the International Pilot Study of
Schizophrenia - IPSS (1202 patients in nine
countries - WHO, 1973, 1979); the Study on
Psychiatric Impairments and Disabilities (520
patients in seven countries - Jablensky et al.
1980); the Study on Depressive Disorders in
Different Cultures (570 patients in four countries
- Sartorius et al. 1983); and a recent multi-
centre investigation of acute transient psychoses
(to be published), the Outcome study is a
component of a collaborative programme which
possesses several features that are rarely encoun-
tered together. These features are as follows:
(i) multi-centre and multi-cultural setting in
which field work is carried out by research
workers belonging to the same culture as the
subjects;
(ii) use of uniform and standardized research
instruments and techniques which make possible
 Schizophrenia: A World Health Organization Ten-Country Study
comparisons between patient samples in dif-
ferent settings;
(iii) prospective observation, including mul-
tiple follow-up examinations of patients and key
informants;
(iv) assessment of patients by highly skilled
clinicians, rather than by lay interviewers or
research assistants.
In addition to these features, the Outcome
study employed an epidemiological approach to
case finding based on symptoms, which repre-
sents a considerable extension of, and improve-
ment over, the first admission-by-diagnosis
method, used previously in investigations of the
incidence of schizophrenia.
Conceptually, the Outcome project reflects
aspects of the 'state of the art' in schizophrenia
research in the mid-1970s, when new trends like
the experimentation with 'operational' diag-
nostic criteria, and their application to studies of
the major mental disorders, were gaining ground.
A number of studies, initiated around the same
time, have taken different methodological
approaches. While none of these studies (the
Outcome project included) have yet provided
conclusive answers to the unresolved questions
about the nature of schizophrenia, it will be
important, as results are being published, to
assess carefully whether the understanding of
schizophrenia can be significantly furthered by
epidemiological and clinical means, or a major
breakthrough ought to be first awaited in the
domain of biological research before any new
large-scale epidemiological study is undertaken.
Having reviewed the principal methods and
results of the 'core' study of the Outcome
project in the preceding chapters, it is now
important to examine their significance and
possible interpretation in relation to several
questions resulting from the statement made in
the opening paragraphs above.
1. In what sense can it be said that the patient
population assessed in the present study is
' representative' of the universe of disorders that
would broadly be classified as schizophrenic
according to internationally accepted criteria?
In this context, it would be especially important
to compare the Outcome study patient samples
with the IPSS series of patients.
2. How effective are the psychopathological
and diagnostic criteria employed in the present
study as tools for dividing the disorders broadly
81
grouped as schizophrenic into groups and
subtypes with distinctive symptomatological and
other associated features which could provide
useful leads for further, aetiologically orientated
research?
3. What new knowledge has been collected in
the present study on the course and outcome of
schizophrenic disorders and on their deter-
minants?
4. What are the implications of the findings
reported here on the incidence of schizophrenia
in different geographical areas?
A COMPARISON BETWEEN THE
OUTCOME STUDY AND THE IPSS: THE
ISSUE OF REPRESENTATIVENESS
A review of epidemiological, including follow-
up, studies on schizophrenia in the 1960s and
early 1970s (WHO, 1979) concluded that
methodological problems, especially lack of
standardization of diagnostic criteria and out-
come criteria, limited seriously the comparability
of results reported by different research workers
and in different cultures. Nevertheless, it was
pointed out that' despite improvements in social,
psychopharmacological and other treatments of
schizophrenia, no matter how narrowly or
broadly schizophrenia is defined, this disorder
still has the potential to develop into a chronic
disorder'.
This conclusion, based on a selective review of
about a dozen major studies published between
1939 and 1972, had to be modified in the light of
the results of the two-year follow-up of IPSS
patients, which demonstrated that, 'at least as
far as short-term course is concerned, sympto-
matologically similar schizophrenic patients may
differ greatly with regard to course and outcome'
(WHO, 1979). One of the strikingfindingsof the
IPSS was the marked tendency for schizophrenic
patients in developing countries to have less
severe course and outcome than patients with
comparable initial clinical pictures in the de-
veloped countries.
The IPSS findings were underscored by the
results of several other studies, published sub-
sequently. Although based on methods and
design often quite different from those of the
WHO study, they reported high proportions of
favourable outcomes not only for patients
diagnosed as schizophrenic in traditional cul-
82 A. Jablensky and others
tures (Waxier, 1979), but also for European
patients, followed up over years, and, in some of
these studies, over decades (Bleuler, 1972;
Ciompi, 1980; Gross et al. 1981). In the report
on a 5-year prospective investigation of 121
patients, Watt et al. (1983) pointed out that little
attention had been paid in previous studies to
the representativeness of patient groups, and
described the manner in which their own patients
were selected with a view to obtaining a
representative group of schizophrenics. This
study used assessment methods and criteria
comparable to those of the IPSS, and its findings
showed a good outcome in nearly 50 % of the
cases. However, only one-third of the patients
were first admissions, and although the sample
could be considered a representative cross-
section of the schizophrenic population under
treatment, it was not an incidence cohort.
The problem of representativeness is central
to research into the clinical spectrum and
' natural history' of the conditions diagnosed as
schizophrenic. It implies the identification of all
cases (or a representative sample of them) who
meet specified criteria for the disorder and are
members of a defined population. If selection
criteria restrict admission to the study only to
patients with clearly established symptoms and
signs of schizophrenia as stipulated by many of
the current ' operational' criteria (Fenton et al.
1981), it is likely that the sample would consist
predominantly of patients with severe illnesses,
many of them chronic, and with a history of
previous treatment. Such a sample would be
perfectly adequate for assessing the effects of a
given treatment, but it is unlikely to advance
research into risk factors and mechanisms
underlying the course of the disorder. It would
not include, for example, the less severe and less
' typical' forms which are a natural part of the
clinical spectrum of any disorder (not only
schizophrenia), and it would miss the early
stages of illness, which may be important for the
prediction of its subsequent evolution.
The present study was designed with a view to
making some advance over the IPSS in regard of
these aspects of representativeness of the patient
population. As pointed out in chapter 1, the aim
of the case-finding strategy chosen was to collect
a series of patients at a point as close as possible
to the onset of their illnesses, and in a stage
when their clinical presentation and social
functioning would not have been affected by
treatment or by the institutionalized roles that
are frequently imposed on schizophrenics and
their families by the environment. The findings
suggest that, by and large, this aim has been
achieved and that the study has been successful,
probably for the first time on such a scale, in
collecting a sample of practically untreated
patients of recent onset.
A comparison with the IPSS shows that 40 %
of the patients with a diagnosis of schizophrenia
in that study had had previous attacks and
treatment; all the patients in the present study
were included during the first episode of psy-
chotic illness, and 86 % entered the study within
less than a year after the onset of symptoms.
While in the IPSS the episode of inclusion had
lasted six or more months before the initial
assessment in 34 % of the cases, the correspond-
ing figure for the Outcome study is 20%. The
proportion of IPSS schizophrenic patients with
sudden onset of the episode was 12%; using
somewhat different criteria, this proportion was
found to be 36% in the present study.
All these findings show that, in accordance
with the research aims, the Outcome study
sample is characterized by less chronicity, shorter
previous history of illness, and less exposure to
psychiatric services. At the same time, a number
of the sociodemographic characteristics of the
two samples are quite similar. Although the
IPSS included patients within the age groups 15
to 44, the male to female ratios of the schizo-
phrenic patients in the two samples are com-
parable (see Table 5.1).
Table 5.1. Male to female ratios of schizophrenic
patients in the IPSS and Outcome study groups
Outcome
Age group IPSS study
15-24 1-3 1-6
25-34 1-2 1-2
0-6 0-7
All ages 10 1-2
The percentage of single, divorced or separ-
ated patients in the Outcome study is slightly
higher (63%) than in the IPSS (59%) but the
proportions of patients living alone is the same
(9 %). Ten per cent of the IPSS patients, and
12% of the Outcome study patients, were
 Schizophrenia: A World Health Organization Ten-Country Study
illiterate and never went to school. The pro-
portion of schizophrenic patients who had never
been employed was 21 % in the IPSS and 24%
in the Outcome study. Exactly the same propor-
tions (26%) of the IPSS schizophrenics and of
the Outcome study patients (26 %) had at least
one relative with mental illness, although the
method of data collection in the latter study was
considerably more elaborate.
Another aspect of the problem of representa-
tiveness is related to the question whether the
sample of patients in the Outcome study is
comprehensive, as regards inclusion of all the
varieties of the schizophrenic syndrome, and
specific, in the sense of not being diluted with
conditions which on closer scrutiny would not
satisfy even broad criteria for a diagnosis of
schizophrenia.
In the absence of external validating criteria,
schizophrenia remains a clinical concept, and
the sampling of cases cannot be guided by
anything better than a carefully evaluated
knowledge base, shared by the greatest possible
number of investigators. This was the assump-
tion which led to the adoption of a dual criterion
for inclusion in the present study: the sim-
ultaneous use of the clinical diagnosis of the
investigators and of the CATEGO computer
classification of psychopathological syndromes.
The' outer' boundary of the disorders qualifying
as schizophrenia or schizophrenia-related was
denned by either: (i) the clinical diagnosis of
schizophrenia or one of the several ICD-9
categories phenomenologically close to it; or (ii)
the S, P and O CATEGO classes, in instances
where the clinical diagnosis might prove too
restrictive. The CATEGO S + class, denned by
the presence of Schneider's first-rank symptoms,
was used as an index defining a ' core' of cases
less likely to evoke disagreement as to their
nosological status than the broader definition.
Compared to the IPSS, this procedure was
intended to be both more flexible, by giving
more room to clinical judgement in doubtful
cases, and more specific, by providing oper-
ationally defined levels of diagnostic classifi-
cation.
Although it is difficult to draw strict parallels
between the IPSS and the Outcome study with
regard to the symptomatology and diagnostic
class distribution of the patient samples (because
of the use of different editions of the PSE and
83
slightly different CATEGO versions), some gross
differences and similarities stand out. At the
level of diagnosis assigned by the field research
centres, the ratios of schizophrenia, paranoid
states, and reactive paranoid psychoses were
very similar in the two study samples - 91:3:6
in the IPSS and 92:3:5 in the Outcome study.
However, there were differences in the dis-
tribution on individual diagnostic subtypes.
Paranoid schizophrenia (ICD 295.3) was more
frequent in the IPSS sample (40 % of all cases)
than in the present study (29 %). Schizoaffective
disorder accounted for 13 % of all diagnoses of
schizophrenic patients in the IPSS, but for less
than 6% of the Outcome study diagnoses. The
largest difference concerns the frequency of the
diagnosis of acute schizophrenic episode - over
2 3 % in the present study and less than 10% in
the IPSS. The difference is probably a reflection
of the case finding strategy of the Outcome
study because it correlates with the recency of
onset in a substantial proportion of the patients.
As regards symptomatology, comparison be-
tween the two studies can be made on the basis
of the major CATEGO classes assigned to
different subgroups of patients. This, of course,
is meaningful only for the centres which partici-
pated in both studies: Aarhus, Agra, Cali,
Ibadan, Moscow, and Prague. Table 5.2 shows
the percentages of IPSS and Outcome study
patients with centre diagnosis of schizophrenia,
paranoid states, or psychogenic paranoid psy-
chosis, who were assigned to CATEGO classes
S, P and O. While in four of these centres the
rates of concordance in the two studies are very
similar, there was some decrease of the con-
cordance rate in Aarhus, and an increase in
Moscow. Changes in the diagnostic habits may
have taken place in the centres over the years
that separate the two studies, but it is more
likely that the differences, as regards Aarhus are
closely related to the nature of the sample, and
in the case of Moscow, to the earlier timing of
the PSE assessment after inclusion of the patients
in the present study. Many of the Aarhus IPSS
schizophrenics had a longer history of previous
illness than the patients in the other IPSS
centres. This was not the case in the present
study, and it is therefore not surprising that the
rate of CATEGO S, P, O would be lower. The
greater concordance between clinical diagnosis
and CATEGO class in the Outcome study
84
Table 5.2. Comparison of diagnostic concordance rates between the International Pilot Study of
Schizophrenia (IPSS) and the present study (only centres which took part in both investigations)
Percentage patients with centre clinical diagnosis ICD 295, 297
or 298.3-9 who have been assigned CATEGO classes S, P or O
sample in Moscow may be due to the fact that in
the present study the patients were administered
the PSE early after intake, i.e. at the height of
florid symptoms, while in the IPSS such assess-
ments often took place somewhat later, often
after a period of treatment, when the intensity of
symptoms had abated. These comparisons
strengthen retrospectively the case for consider-
ing the IPSS series of schizophrenic patients also
to be a representative selection of cases, in spite
of the less rigorous nature of the inclusion
procedure from an epidemiological point of
view.
VALIDITY OF THE DIAGNOSTIC
CLASSIFICATION OF THE CASES
Assuming that a representative sample, reflecting
a wide clinical horizon of non-affective psychotic
disorders, has been obtained in the Outcome
study, the first question that arises is: in what
sense can the disorders investigated in the
Outcome study be regarded as 'schizophrenic'?
It is well known that' schools' and traditions
in psychiatry, across and within countries, differ
with regard to the conceptual definition of
schizophrenia and the criteria thought to be of
importance in establishing the diagnosis (Berner
et al. 1983). Different theoretical positions on
these issues are also represented within the
network of research centres which took part in
the present study. However, a critical review
of the existing variation in the approaches to
the taxonomy of schizophrenia (WHO, 1981)
nevertheless concluded that 'schizophrenia is a
valid and useful concept, defining a disorder or
groups of disorders of world-wide occurrence
for which a predisposition is genetically trans-
mitted'.
The extent to which diagnostic variation can
actually invalidate the assessment of epidemi-
ological data on schizophrenia is not precisely
A. Jablensky and others
Aar Agr Cal Iba Mos Pra
IPSS 95 91 88 96 63 83
Outcome 78 92 94 95 79 92
study
known. Considering the remarkable stability of
prevalence rates of schizophrenia, reported from
different parts of Europe where various psy-
chiatric 'schools' have been competing for
decades, one is inclined to think that the effects
of such diagnostic variation need not be exag-
gerated. The differences between the various
'schools' and approaches usually concern the
margins of the diagnostic concept, but not its
core; therefore they are likely to affect the
estimates of the upper limits of the range of
morbidity indices, but would have less effect on
its lower limits.
The study presented here took an explicitly
empirical and pragmatic approach to the di-
agnostic definition of 'caseness'. Instead of
taking as its point of departure any one
particular concept of schizophrenia, the first
step of case definition at the screening stage
consisted in the ascertainment of symptoms and
signs which most psychiatrists would describe as
either psychotic or strongly suggestive of psy-
chosis, and which were unlikely to be a mani-
festation of an underlying affective disturbance.
Since gross brain damage was also excluded at
that stage, these criteria restrict the universe of
'caseness' to an area that might be subdivided
and labelled differently, but is certain to contain
most of the symptoms and syndromes which
would be regarded as schizophrenic by any
'school'. The results of the second stage, at
which diagnostic classification was carried out
after a standardized clinical assessment, confirm
the advantage of starting the case finding with
an enquiry about symptoms and signs rather
than about diagnoses: in the eight areas for
which incidence rates are reported, a total of 80
cases would not have entered the study if case
finding had been based only on a clinical
diagnosis. All of these 80 cases, which met the
screening criteria but were given centre diagnoses
other than schizophrenia and schizophrenia-
 Schizophrenia: A World Health Organization Ten-Country Study 85

related disorders, were classified by the 1959) were from patients assigned to CATEGO
CATEGO program into one of the 'schizo- classes S?, P, and O. The results reported in the
phrenic' classes S, P, or O, and were sympto- preceding chapters show clearly that CATEGO
matologically similar to the patients who were S + patients constituted either the majority, or a
assigned to the same CATEGO classes but had substantial minority, of the patients meeting the
a centre diagnosis of schizophrenia. inclusion criteria in all the centres. The compari-
The high rate of concordance between the ICD sons between S + and non-S -I- cases in the total
diagnostic inclusion criteria and a CATEGO study population can be summarized as follows.
classification as S, P, or O (no less than 80 % in
seven of the study areas and 64 % in one area) is Variables on which S+ and non-S + cases
an indication of the valid inclusion of the large differ
majority of cases. Both in the IPSS (WHO, (i) Percentage of cases with clinical diag-
1973) and in other studies (Scharfetter et al. nosis of schizophrenia (87 % of S + and
1976; Brockington et al. 1978; Lewine et al. 81% of non-S+ ).
1982), membership in the 'broad' S, P, O group (ii) Percentage of cases with any one of the
has been shown to correlate highly with clinical clinical diagnoses required for inclusion
consensus on a diagnosis of schizophrenia. Since, (95% of S + and 89% of non-S+ ).
however, the reverse is not necessarily true (i.e. (iii) Percentage scores on all PSE psychotic
not all CATEGO-discrepant cases with a clinical symptoms (uniformly higher in S + than
diagnosis of schizophrenia in one setting would in non-S+ ).
be diagnosed as schizophrenia in another set- (iv) History of convulsions (2-3 times more
ting), it was considered important to calculate likely in S+ than in non-S+ ).
not one but several rates of incidence for each (v) History of emotional or conduct prob-
area, and to include the marginal cases in the lems in childhood (19 % of S + and 14 %
'broad' definition, but not in the other three of non-S+ ).
definitions, which require S, P, O membership. (vi) History of emotional or conduct prob-
Assuming that a representative sample, re- lems in adolescence (27% of S+ and
flecting a broad clinical agreement on the scope 20% of non-S -I-).
of schizophrenia and related conditions has (vii) History of psychotic illness in the mother
been obtained in the Outcome study, the next (in 50 out of the 108 S+ patients with at
question concerns the further diagnostic sub- least one parent mentally ill, compared
division and classification of the patients who to 24 out of 67 non-S + patients).
have met the inclusion criteria for schizophrenia
or a related disorder. If schizophrenia is, Variables on which S+ and non-S+ do not
genetically and aetiologically, a heterogeneous differ
group of disorders - an idea which goes back to (i) Percentage scores on PSE affective
Bleuler (1911) and Kraepelin (1920) and has symptoms.
been resuscitated by modern biological research (ii) Mode of onset.
(McGuffin, 1988)-it is important to know if (iii) Pattern of course.
any particular diagnostic subdivision of the The evidence summarized above does not
sample would provide' points of rarity' (Kendell lend sufficient support to a notion of a discrete
& Brockington, 1980) that may help to identify entity of 'nuclear' schizophrenia, qualitatively
real discontinuities or well defined syndromes. different from the other syndromes exhibited by
the patients in this study. As previously shown
Subdivision by CATEGO class S + (first-rank (p. 37) the S+ patients have more florid
symptoms) 'positive' psychotic symptoms. Such patients
In the exploratory analyses, carried out in the may represent one extreme of a continuum of
context of the present study, a special attempt psychopathology spanning all the distance be-
was made to establish how different patients tween oligosymptomatic cases, manifesting pre-
assigned to CATEGO class S + , which by dominantly deficits and 'negative' symptoms,
definition identifies patients with highly charac- and cases with the full-blown picture of schizo-
teristic symptoms of schizophrenia (Schneider, phrenic psychosis. In this sense, the presence of
86 A. Jablensky and others

Schneider's first-rank symptoms can be regarded Table 5.4. Age group distribution of (a)
as an index of severity of 'positive' psychotic schizophrenic psychoses (excluding acute) and (b)
disturbances in schizophrenic patients. The acute schizophrenic episode, by developed
uniform occurrence of such symptoms in pa- countries and developing countries
tients belonging to different cultural environ-
ments provides a justification for their use as Schizophrenic Acute
epidemiological tools. psychoses schizophrenic
(excluding acute) episode
Subdivision by acute/non-acute clinical subtype Age group Developed Developing Developed Developing
Mode of onset, i.e. the time elapsed since the 15-24 44-2 500 600 72-6
first appearance of an unequivocally psychotic 25-34 33-1 34-8 21-4 23-5
manifestation and the point at which a recog- 35-44 15-1 12-9 11-4 3-4
45-54 7-6 2-3 7-1 0-5
nizable clinical syndrome or symptom complex
is present, emerged as an important variable in
a number of the analyses described in previous
chapters. This raises the question of a possible Table 5.5. Most common symptoms at initial
existence in the study population of a subgroup examination in patients with centre diagnosis
of cases which is not merely a variant of the of acute schizophrenic episode, by developed
manifestation of the same basic disorder, i.e. countries and developing countries
schizophrenia but a psychotic condition sui
Most common symptoms (occurring in > 30 % of patients)
generis which may be distinct from schizophrenia Developed countries % Developing countries %
in several aspects. If this were the case, this
group of acute illnesses should be classified 1 Systematization of 62-9 1 Voice speaking to 51-5
delusions subject
separately from 'mainstream' schizophrenia. 2 Depressed mood 55-7 2 Systematization of 47-5
In order to test this proposition, the age and delusions
3 Delusions of reference 51-4 3 Acting out delusions 47-1
sex distribution of the disorders diagnosed in the 4 Acting out delusions 51-4 4 Delusions of reference 43-1
centres as acute schizophrenic episodes was 5 Delusional 44-3 5 Delusions of 42-6
misinterpretation and persecution
examined in the developing and the developed misidentification
countries, and was compared with the dis- 4 Delusions of 44-3 6 Visual hallucinations 35-8
tribution for patients with all other schizophrenic persecution
7 Evasiveness 44-3
diagnoses. The results (Tables 5.3 and 5.4) 8 Preoccupation with 38-6
indicate that: (i) in both developed and devel- delusions and
oping countries the male/female ratio in the hallucinations
9 Delusional mood 35-7
group of patients with acute schizophrenic 10 Morning depression 32-9
episodes is lower than in the group of patients
with other schizophrenic diagnoses (0-9 and 1-2
respectively for acute, 1-2 and 2-0 respectively
for all other schizophrenia), i.e. the proportion
Table 5.3. Sex ratio in (a) schizophrenic of women is higher in the acute group; (ii) there
psychoses (excluding acute) and (b) acute is a relative excess of patients in younger age
schizophrenic episode, by developed countries and groups (15-24) among the cases diagnosed as
developing countries acute schizophrenic episodes in both developing
Schizophrenic Acute
and developed countries.
psychosis schizophrenic Further, the most common symptoms (those
(excluding acute) episode occurring in 30% or more of the patients)
Setting M F M F were tabulated for acute schizophrenic episodes
in the developed and the developing countries
Developed countries 54-4 45-6 48-6 51-4 (Table 5.5). The symptoms which appear among
(245) (205) (34) (36)
M:F 1-2 0-9 the most common ones in one type of setting but
Developing countries 66-2 33-8 53-9 461 not in the other are in italics. Out of 10
(200) (102) (110) (94)
M:F 2-0 1-2
symptoms which appear in over 30% of the
patients in the developed countries, 4 occur with
 Schizophrenia: A World Health Organization Ten-Country Study 87

the same frequency in the developing countries; important conclusions reached in that previous
and out of the 6 common symptoms in the study and, on the other hand, contribute new
patients in developing countries, 4 are also knowledge about certain aspects of the prognosis
common in patients in the developed countries. of schizophrenia.
Thus, considerable symptomatological simi-
larity exists between the disorders diagnosed as Diversity of patterns of course
acute schizophrenic episodes in the two types of First, it has again been clearly demonstrated
setting. that the course of conditions meeting clinical
Next, the most common symptoms (occurring criteria for schizophrenia is far from being
in 30 % or more of the patients) were tabulated uniform and does not conform to a single
for acute schizophrenic episodes and for all pattern. In a substantial proportion (over 50 %)
other schizophrenia subtypes, for all patients of the cases the psychotic disturbance is limited
regardless of setting. Seven out of the 8 common to a single episode of a varying duration (several
symptoms in the acute group were identical with weeks to several months) which may be followed
the most common symptoms (their total number either by a complete remission in which the
also was 7) in the group of other schizophrenia patient is practically symptom-free, or by a
subtypes (Table 5.6). The results of this analysis, symptomatic but non-psychotic state charac-
therefore, do not support the proposition that terized by affective and neurotic disturbances or
the acute schizophrenic episodes included in the by mild aberrations or changes of personality
present study represent a type of disorder that is and behaviour. Undoubtedly, the relatively high
nosologically different from schizophrenia. frequency of the complete recovery after a
psychotic episode qualifying for a diagnosis of
schizophrenia (approximately every fourth pa-
Table 5.6. Most common symptoms at initial tient in the developing countries, and every
examination in patients with centre diagnosis ofseventh in the developed countries) is one of the
schizophrenic psychosis (excluding acute) and most important findings of this study. Although
acute schizophrenic episode (all centres) this pattern of course occurs more frequently in
Most common symptoms (occurring in S 30 % of patients)
the developing countries, it is not uncommon in
Schizophrenic psychoses Acute the developed countries.
(excluding acute) % schizophrenic episode %
In a proportion of the cases (about 31 %) the
1 Systematization of 53-7 1 Systematization of 51-5 pattern of course is episodic, with two or more
delusions delusions psychotic attacks, each followed by a remission
2 Acting out delusions 50-0 2 Acting out delusions 48-2
3 Delusions of reference 49-1 3 Voice speaking to 45-6 which may be complete or incomplete, in the
subject sense referred to above. This pattern of course in
4 Delusions of 44-1 4 Delusions of reference 45-3
persecution
schizophrenia bears a similarity to what is
5 Depressed mood 33-9 5 Delusions of 431 commonly accepted to be the characteristic
persecution course of the affective disorders but the symp-
6 Voice speaking to 32-7 6 Visual hallucinations 38-1
subject
tomatology of these cases is not necessarily of
7 Delusional 32-6 7 Depressed mood 30-3 the schizoaffective type.
misinterpretation and In yet another percentage of the patients
misidentification
8 Evasiveness 32-6 8 Evasiveness 30-3 (about 15 %) the course of the psychotic disorder
is continuous and unremitting, leading to severe
impairment. These cases are more frequent in
the developed countries than in the developing
HAS NEW KNOWLEDGE BEEN ADDED countries, but it is more important to recognize
TO THE UNDERSTANDING OF THE that they occur in both types of setting and that
COURSE AND OUTCOME OF their clinical characteristics are very similar.
SCHIZOPHRENIA? Thus, the Outcome study did not identify any
Being separated by nearly a decade from the particular pattern of the course and outcome of
follow-up data collection phase of the IPSS, the schizophrenic illnesses which could be regarded
Outcome study has provided data which, on as specific to a given area or culture. The
the one hand, confirm and refine some of the descriptive categories used to classify the di-
 A. Jablensky and others

versity manifest in the evolution of schizophrenic or tested at the present time, in the absence of
disorders were equally applicable to the patient established genetic markers, indicators of aeti-
series in the developing and in the developed ology or other underlying mechanisms of dis-
countries. ease. It is, however, possible to reject another
hypothesis which can be formulated in clinical
Higher frequency of good outcome in the and descriptive, rather than biological terms.
developing countries This is the conjecture that the patient sample in
The Outcome study replicated in a clear and, the developing countries might contain an
possibly, conclusive way the major finding of the excessive number of cases of so-called acute
IPSS, that of the existence of consistent and transient psychoses, for which some evidence
marked differences in the prognosis of schizo- exists now that they are both clinically and
phrenia between the centres in developed coun- aetiologically distinct from schizophrenia. The
tries and the centres in developing countries. On evidence reviewed in a preceding section of this
five out of six of the measures and dimensions of chapter is sufficient to reject the hypothesis that
two-year course and outcome which have been an inclusion of atypical transient psychotic
used in the analyses reported here (pattern of illnesses among the schizophrenic cases in the
course, proportion of the follow-up period in developing countries could explain the better
complete remission, proportion of the time course and outcome in these areas. Moreover, it
during which the patient was on anti-psychotic can be shown that the difference in the course
medication, proportion of the follow-up period and outcome of schizophrenia between the two
spent in psychiatric hospital, and proportion of groups of centres clearly persists if the com-
the follow-up during which the social func- parison is limited only to cases of schizophrenia
tioning of the patient was unimpaired), patients with a gradual or insidious onset. Table 5.7
in the developing countries show a more shows that while less than 30 % of the gradual
favourable evolution than their counterparts in onset cases in developed countries had 'mild'
the developed countries (the only dimension patterns of course, this figure was over 40 % for
showing no difference was the percentage of the patients with the same type of onset in the
follow-up period spent in psychotic episodes). developing countries. On the other hand, 53 %
As demonstrated by the multivariate statistical of the gradual onset patients in developed
analysis, these differences between patients in countries, compared with 4 3 % in developing
the two types of setting cannot be explained by countries, had a ' severe' pattern of course.
other variables and remain highly significant
when such possible influences are controlled for. Table 5.7. Pattern of course (2 year-follow-up)
It can now be said with a fair amount of by type of onset and setting {percentages)
confidence that they are not the result of differing
sample composition in the two groups of centres, Pattern of course
in the sense of a selection bias in favour of more Type of
Setting onset Mild Intermediate Severe
pre-inclusion chronicity in the developed coun-
tries and more recent onsets in the developing Developed countries Acute 52-1 251 22-6
X> = 40-3 Subacute 41-3 23-9 34-7
countries. In this study, the average length of the P < 0001 Gradual 29-8 17-5 52-6
illness prior to inclusion into the study did not All types 38-9 21-1 39-8
differ significantly between the developing and Developing countries* Acute 620 21-0 16-9
the developed countries. X* = 26-4 Subacute 58-7 23-8 17-4
P < 0001 Gradual 40-2 16-3 43-4
A more complex issue is the possibility that All types 55-7 20-2 240
the clinical conditions meeting the inclusion
* Ibadan excluded.
criteria of the study in the two types of setting
may be heterogeneous and include varying
proportions of aetiologically and genetically Having excluded, for lack of support by the
different disorders which may be distinguishable data described in this report, the explanation of
from one another at the level of the phenotype, the observed difference between the prognosis of
i.e. the symptoms and syndromes. This possi- schizophrenia in developing and in developed
bility exists but it cannot be properly examined countries as an artefact, a strong case can be
 Schizophrenia: A World Health Organization Ten-Country Study
made for a real pervasive influence of a powerful
factor which can be referred to as 'culture'.
Unfortunately, neither the IPSS nor the Out-
come study could penetrate in sufficient depth
below the surface on which the impact of this
unknown factor was established - tentatively in
the IPSS and definitively in the present study.
Although other components of the Outcome
study (e.g. the investigation on the 'expressed
emotion' in families of schizophrenic patients in
Aarhus and Chandigarh, Wig et al. 1987; Leff
et al. 1990) demonstrated important differences
at the level of day-to-day social interaction of
patients and key figures in their environment in
the two types of setting, it is unlikely that the
variation in course and outcome between devel-
oping countries and developed countries could
be reduced to a single variable. The contribution
of the present study, therefore, is not in providing
the answer but in clearly demonstrating the
existence of the question.
Predictors of course and outcome
By using a log-linear model in the analysis of the
relationship between a number of variables
which had been assessed on initial examination
and variables characterizing the two-year course
of the disorder, it has been possible in the
Outcome study to refine the data on the
prediction of course and outcome. Overall, there
is good agreement between the conclusions
concerning predictors which were reached in the
IPSS and the conclusions about prognostic
indicators in the present study. Although the
instruments for collection of previous psychiatric
and social history data were different in the two
studies, some of the potentially predictive vari-
ables, such as type of onset, marital status, or
length of previous illness were assessed in a
similar manner (but more systematically and in
greater detail in the Outcome study). There is
further a high level of comparability for the
dependent variables in the analysis of predictors
in the two studies, as pattern of course and per-
centage of the follow-up in psychotic episodes
were defined and assessed in a similar way.
However, while stepwise multiple regression was
the main tool of predictor data analysis in the
IPSS, a log-odds (log-linear) model was selected
for the Outcome study. The latter is better suited
to the handling of psychiatric and social data
(which are often of a categorial nature) because
89
it makes no distributional assumptions. Besides,
upon examining the estimated partial derivatives
in the log-linear model, the contribution of
individual predictors to the course and outcome
variables can be understood in terms of per-
centage point differences, which is a simpler and
perhaps clearer 'mental representation' of the
mathematical function.
The results of the log-linear analysis highlight
the key significance of two predictor variables:
the type of onset of the disorder and the type of
setting (developed or developing country). Be-
cause of the strong association of these two
predictors with the pattern of course, their
effects had to be controlled for in the estimation
of the contribution of each one among the
remaining explanatory variables. While it is true
that both the type of onset and the developing/
developed country dichotomy had already been
shown to be predictors of course and outcome in
the IPSS, their overriding importance has been
put into a much sharper focus by the analysis
reported in chapter 4 of this report. It should be
emphasized again that their contribution to the
prediction of course and outcome is indepen-
dent; i.e. the better prognosis of schizophrenic
disorders in the developing countries is not
reducible to the relative excess of acute onsets in
such settings, nor to any other of the predictor
variables tested in the model. Type of onset,
therefore, appears in the light of the findings of
the present study, as one of the critical variables
in schizophrenia research.
The Outcome study findings confirm the
relatively modest but still quite definite prog-
nostic significance of the diagnostic classification
of schizophrenic disorders according to ICD-9
subtypes. The direction in which the individual
subtypes predict the pattern of course is con-
sistent with previous knowledge and data. While
the hebephrenic subtype is associated with the
worst prognosis, the subtypes characterized by
acuteness of onset, presence of affective symp-
toms, and catatonic disturbances (i.e. 295.4,
295.7 and 295.2) are linked with the best chances
of recovery.
Of the remaining predictors, variables such as
sex, marital status, and persistent adjustment
problems in adolescence also seem to confirm
previous knowledge. What is new is the finding
that frequency of contacts outside the family
(close and casual friends) is just as useful a
90 A. Jablensky and others

predictor as the frequency of contacts with pecially for the most restricted definition by
family members, while avoidance of the patient CATEGO S+ class.
by others is not (the avoidance of the patient by Apart from two centres with differing patterns
family members on the other hand is a useful of age- and sex-specific rates, the rates even for
predictor). Being based on a large patient series the broadest defined level varied among the six
in different cultures, this finding may be of other centres at most by a factor of 2-2-5. By
considerable potential interest for the under- inclusion of schizophrenia related disorders by
standing of the nature and effects of social clinical diagnosis or by CATEGO class O?
support networks in schizophrenia. It certainly considerable variations among centres would be
raises the important question about the in- expected because of varying incidence of acute
terpretation of other research evidence, collectedor reactive disorders or drug induced disorders.
in the present study, on the role of ' expressed Whether the actual variation particularly in the
emotion' in the family in the determination of Chandigarh and Moscow centres is explained by
the pattern of course. the inclusion of such cases is difficult to
The predictor analysis demonstrated further determine. For the Moscow centre the variation
that the proportion of the follow-up period is mainly caused by high female rates in higher
during which patients are in incomplete remis- age groups which hardly would be caused by
reactive or drug induced cases, and the possible
sions is that aspect of the course of schizophrenia
which is most clearly associated with the role of a different nosology of schizophrenia in
difference between the settings (developing this centre does not appear to offer an ex-
versus developed countries). Being in a developed planation either because the variation appears
country was a strong predictor of not attaining also at the SPO+ level. For the Chandigarh
a complete remission. It seems, therefore, that a centre the rural sample shows the highest rates
major part of the difference in the prognosis of causing the great differences in the variation
schizophrenia in the two kinds of setting may be which may have a number of possible ex-
reduced to the failure of many patients in the planations, such as inclusion of acute or reactive
developed countries to attain or maintain a or organic cases with SPO+ level symptoms.
complete remission of symptoms. The Chandigarh urban sample however showed
An unexpected finding which calls for further lower rates and the pattern of age and sex
exploration is the prediction of poor course and specific rates shows more resemblance to
outcome by a history of ' street' drug use (the patterns in most of the other centres.
initial expectation was that drug use may be a At the restricted S+ diagnostic level the
triggering factor in the precipitation of relatively
differences diminish and the variation is not
benign schizophreniform illnesses). The high greater than may be explained by chance
incidence of reported drug abuse among study variation. The lower number of patients by the
patients in three of the centres of the study more restrictive level does not explain the lower
(Honolulu, Rochester, and Aarhus), may be an variation which rather would be expected to
indication of an increase in the frequency of the increase by chance fluctuations. Extending the
combined occurrence of psychotic illness and casefindingperiods or doubling up the numbers
drug use. While there is no reason to regard with unchanged rates would of course increase
these cases in the Outcome study as drug- the power of the statistical evaluation and might
induced psychoses, the phenomenon merits have been able to detect significant differences
careful investigation. also at the restricted diagnostic level. However,
the actually observed rates at the various
diagnostic levels appear to differ among the
IMPLICATIONS OF THE FINDING OF centres within a modest range pointing more to
SIMILAR INCIDENCE RATES OF similarity than variability of the incidence of
SCHIZOPHRENIA IN DIFFERENT schizophrenia and schizophrenia-like disorders.
GEOGRAPHICAL AREAS
The importance of the differences depend on
A major newfindingof the study is that incidence the point of view. For health administrations the
rates for various levels of definition of schizo- differences of the broadly defined diagnostic
phrenic disorders are surprisingly similar, es- level is important for the delivery of health care
 Schizophrenia: A World Health Organization Ten-Country Study
to a number of patients which may vary with the
factor of two or three among centres. For
aetiological studies exploring the cause of schizo-
phrenic disorders, the similarity of incidence
rates at the S+ CATEGO class levels is an
important finding.
The absence of marked variation in the
incidence of schizophrenia will not easily lend
itself to an interpretation, unless the nature of
the relationship of the schizophrenic phenotype
to the underlying causes and pathophysiology of
the disorder is clarified. Linked to this is the
question whether schizophrenia is a single
disease entity or a heterogeneous group of
conditions which exhibit a 'common final
pathway'.
In a lucid overview of the state of knowledge
in this respect, Bleuler (1981) classified the
different views on the subject into three
'schools'. The first, which he termed 'elemen-
tary ' school, adheres to the formula ' one disease
-one cause'. The second, or 'middle' school
regards schizophrenia as 'many diseases, each
with its own causes'. The third, or' high' school,
would see in schizophrenia neither a single
disease, nor a collection of single diseases, but
rather a manifestation of a developmental
disorder in which 'multiple influences shape
symptomatology and course'. This position goes
full circle back to the views of Kraepelin (1920)
who, at the end of his career, came to regard
schizophrenia as ' a common reaction of man-
kind to the most varied forms of noxious events'.
According to Kraepelin, 'the affective and
schizophrenic forms of mental disorder do not
represent the expression of particular patho-
logical processes but rather indicate the areas of
our personality in which these processes un-
fold ... It must remain an open question whether
hereditary factors make certain areas more sus-
ceptible and accessible to pathological stimuli'.
Current hypotheses about schizophrenia as a
neurodevelopmental disorder, which build on
91
recent advances of cerebral morphology, gen-
etics and pathophysiology (e.g. Murray & Lewis,
1988) in fact echo the conjectures referred to
above. Such a model would be in agreement
with the epidemiological data. If schizophrenia
is conceptualized not as a single disease but as a
'common final pathway' for a variety of cerebral
disorders and neurodevelopmental lesions, simi-
lar rates of its incidence in different populations
could be seen as the expression of a more or less
uniformly distributed liability for a schizo-
phrenic type of reaction to different causes. This
liability must have a genetic basis which may
be more complex than currently assumed. A
'nuclear' schizophrenic syndrome (identified by
CATEGO S + in the present study) - with its
clinical consistency and uniform occurrence - in
different cultures, may be the manifestation of a
specific segment of a complex genotype with a
much wider range of phenotypical expression.
However, the question whether the apparently
similar rates of manifestation of schizophrenic
syndromes in different populations are primarily
due to a uniformly distributed genetic liability,
or to some ubiquitous constellation of environ-
mental factors interacting with it, or to a
similarity in expression of genetically different
disorders, should be addressed in future research.
The present study has developed an extensive
database whichfillsa number of important gaps
in the descriptive phenomenology and epidemi-
ology of schizophrenia worldwide. As linkage
studies and genome mapping techniques are
now preparing the ground for a 'molecular'
epidemiology of schizophrenia, a new role may
be emerging for comparative epidemiological
research across different populations and geo-
graphical areas: that of guiding neurobiology to
more clearly defined targets, taking into account
the ultimate role of culture as the context in
which gene-environment interactions shape the
clinical picture of human disease.
 Conclusion: A synopsis of the main
findings
1 The study on Determinants of Outcome of
Severe Mental Disorder (Outcome study) is a
cross-cultural investigation, coordinated by
WHO, of schizophrenic and related disorders in
13 geographical areas in 10 countries (Colombia,
Czechoslovakia, Denmark, India, Ireland,
Japan, Nigeria, the Union of Soviet Socialist
Republics, the United Kingdom, and the United
States of America).
2 The study is based on an initial examination
and two follow-up examinations, at annual
intervals, of 1379 patients. Of the total number,
78-2% completed the follow-up and were re-
assessed two years after the initial examination.
3 The patients included in this study were
new cases, in the sense that they had contacted
a 'helping agency' for their mental health
problem for thefirsttime in their lives during the
three months preceding the initial examination,
and had practically no previous exposure to
psychiatric treatment or care. The reasons for
making a first contact with a 'helping agency'
were similar in the developing and the developed
countries (behaviour perceived as 'odd' and
feared violent behaviour towards self or others
being cited in about 90% of the cases). The
mean length of previous illness (i.e. prior to
inclusion into the study) was practically the
same for patients in developing countries and
patients in developed countries. The majority
(86%) of the patients were recruited for the
study within less than a year of the first
appearance of symptoms.
4 In 39 % of the cases the first help-seeking
contact was made with a psychiatrist; however,
especially in the developing countries, traditional
medicine is a frequent resource in the event of
mental disorder and is often utilized simul-
taneously with the services of' Western' mental
health care. It has been estimated that about 200
cases eligible for this study would have been
missed if traditional practitioners had not
cooperated in case-finding.
5 There was a considerable amount of simi-
92
larity in the early behavioural manifestations of
psychotic illnesses across the centres. In both
developing and developed countries 'negative'
behavioural disturbances (neglect of usual ac-
tivities, social withdrawal) were described more
often as the earliest perceived signs of illness
than frank psychotic manifestations such as talk
of persecution, harm or bewitchment, or be-
having as if hearing voices. Family members and
key informants in the community appeared to
be sensitive observers and served well as a case-
finding resource in the majority of the centres.
6 Of all included cases, 82 % were assigned to
CATEGO classes S, P, or O which, together
with a clinical ICD diagnosis of schizophrenia
and schizophrenia-related disorders (paranoid
psychoses, reactive paranoid and schizopheni-
form psychoses, unspecified psychoses, alcohol
and drug induced psychoses with hallucinatory
or paranoid symptoms, schizoid and paranoid
personality disorders) were considered to con-
stitute a broad group of schizophrenia and
related disorders in this study. The classification
of a patient into the broad group of schizo-
phrenic disorders required either one of the ICD
diagnoses listed above or a CATEGO class S, P,
or O. Between 60 % and 95 % of the patients in
the different centres met both criteria. It should,
however, be emphasized that the inclusion of
cases into the study through a screening process
was based on specified symptoms and behav-
iours, and not on diagnosis.
7 More than one half (56%) of the study
population had CATEGO class S+ ('nuclear'
schizophrenia), defined by the presence of one
or more of Schneider's first-rank symptoms
(Schneider, 1959). These patients were found to
have high scores on all types ' positive' psychotic
symptoms, and could be considered to be a more
severely disturbed subgroup than the rest of the
patients.
8 The PSE profiles of the patients meeting the
'broad' inclusion criteria of this study were
similar in the developed and the developing
 Schizophrenia: A World Health Organization Ten-Country Study
countries. In the latter, visual hallucinations
tended to occur more often, and in the former,
affective symptoms, especially depression, were
more common. However, these differences could
be regarded as relatively insignificant, consider-
ing the great similarity in the scores of the
remaining symptoms.
9 Schizoid traits: sensitivity, suspiciousness
and reserve, were described as manifest during
adolescence in a high proportion of the patients.
However, contrary to expectations, presence of
' positive' pre-morbid personality traits was more
frequent in patients who were classified as
CATEGO S + .
10 The annual incidence of new cases of
' broadly' defined schizophrenia was in the range
between 15 and 4-2 (both sexes) per 100000
population at risk (age 15-54). The incidence of
schizophrenia defined by CATEGO class S 't-
was in the range between 0-7 and 1-4 per 100000.
The morbid risk (expectancy) for schizophrenia,
determined on the basis of the incidence data, is
between 0-5 and 1-72% for the 'broad' di-
agnostic category, and between 0-26 and 0-54 for
CATEGO S + . The incidence of 'broadly'
defined schizophrenia was highest in India (both
the rural and the urban area of Chandigarh).
The differences between the incidence rates for
the' broad' diagnostic category of schizophrenia
in the different centres were significant and
indicate the necessity of future studies in some of
the centres, particularly the Chandigarh areas,
to further explore the nature of their high
incidence rates. In every centre, the incidence
rates tended to decrease as more specific defin-
itions of 'caseness' for schizophrenia were
applied. At the level of CATEGO S + , there
were no significant differences across the study
areas.
11 In all the study areas, the age- and sex-
specific curves of the incidence of schizophrenia
followed a similar pattern. It was demonstrated
that in developed and in developing countries
alike, the onset of schizophrenia tended to occur
at a later age in females as compared to males.
The similarity of age- and sex-related patterns of
onset of schizophrenia across the study areas is
a strong evidence that the same basic type of
disorder has been identified and investigated in
the different cultural settings of the study.
12 The majority of the patients in the study
had a remitting pattern of course over the two
93
years of follow-up: 50-3 % had a single psychotic
episode and a further 31 • 1 % had two or more
psychotic episodes followed by remissions. Only
15-7% of the patients had an unremitting,
continuous psychotic illness. The remitting pat-
terns were more common among patient popu-
lations in the developing countries.
13 On five out of six course and outcome
dimensions patients in the developing countries
had a markedly better prognosis than patients in
the developed countries. The tendency for a
more favourable course and outcome was not
limited to acute schizophrenic episodes; it was
also clearly present in the subset of cases which
had a gradual or insidious onset of schizo-
phrenia. The only variable which did not
distinguish clearly between patients in devel-
oping countries and patients in developed
countries was the proportion of the follow-up
period during which patients were in psychotic
episodes. On the other hand, the variable on
which patients in the two kinds of setting differed
most was the proportion of the follow-up period
during which patients were in incomplete remis-
sions : the mean percentage of time in such state
was considerably higher for patients in the
developed countries.
14 About one third of all the patients in the
study were never admitted to a psychiatric
hospital; of those admitted, the majority spent
only brief periods in hospital treatment. On the
other hand, 95 % of the total study sample were
prescribed neuroleptic medication for varying
lengths of time in the course of the study;
patients in the developed countries were pre-
scribed anti-psychotic drugs over longer periods
of time than patients in the developing countries.
15 The different sets of inclusion criteria
(' broad' versus ' restrictive') did not result in the
selection of patients differing according to their
prognosis. The CATEGO class assigned on
initial examination had no prognostic signifi-
cance ; however, the clinical subtyping of schizo-
phrenic disorders according to ICD-9 criteria
was associated with some significant differences
in course and outcome. The hebephrenic and
paranoid subtypes tended to have the worst
course and outcome, while the acute schizo-
phrenic episodes had the best course and
outcome.
16 Type of onset (i.e. acute, subacute, and
gradual) and setting (developing country or
94 A. Jablensky and others
developed country) were the most important
predictors of several dimensions of the two-year
course and outcome. Other significant predictors
were: clinical diagnosis on initial examination,
marital status, sex, adjustment in adolescence,
frequency of contact with friends, and history of
use of 'street' drugs. The use of a log-linear
model for the identification and assessment of
predictors has resulted in prognostic tables
which may be further tested in clinical trials.
17 The study described in this report will
raise a number of new research questions.
Schizophrenia remains an entity defined almost
exclusively by its clinical symptoms and their
characteristic evolution over time. No external
validating criteria for the diagnosis have yet
been established, in spite of a number of
suggestive biological findings, among which
genetic data appear to be the most consistent.
The study of the epidemiology of a disorder
can provide critical leads to the understanding
of causes and risk factors. In the instance of
schizophrenia, this has not occurred because
of continuing uncertainties due to the absence of
reliable diagnostic indicators and markers, and
of a sharp demarcation between its symptoms
and the symptoms of other psychiatric con-
ditions. This hampers the application of multi-
disciplinary approaches. The collaborative
studies coordinated by WHO are contributing
to the resolution of such difficulties by creating
an international, cumulative clinical and epi-
demiological database on schizophrenia and
related disorders worldwide.
REFERENCES
Adelstein, A. M., Downham, D. Y., Stein, Z. & Susser, M. (1968).
The epidemiology of schizophrenia in an English city. Social
Psychiatry 3, 47.
American Psychiatric Association (1980). Diagnosic and Statistical
Manual, 3rd edn. AMA: Washington, DC.
Babigian, H. M. (1980). Schizophrenia: epidemiology. In Com-
prehensive Textbook of Psychiatry, 3rd edn, vol. 2 (ed. H. I.
Kaplan, A. M. Freedman & B. J. Sadock). Williams & Wilkins:
Baltimore.
Berner, P., Gabriel, E., Katschnig, H., Kieffer, W., Koehler, K.,
Lenz, G. & Simhandl, Ch. (1983). Diagnostic criteria for
schizophrenic and affective psychoses. In Proceedings of World
Psychiatric Association. American Psychiatric Press: Washington,
DC.
Blake, B., Halpenny, J. V., ni Nuallain, M., O'Brien, P. F., O'Hare,
A. & Walsh, D. (1984). The incidence of mental illness in Ireland
- patients contacting psychiatric services in three Irish Counties.
Irish Journal of Psychiatry 5, 23-29.
Bleuler, E. (1911). Dementia praecox oder die Gruppe der Schizo-
phrenien. In Handbuch der Psychiatrie, 4. Abt. (ed. G. Aschaf-
fenburg), pp. 251-252. Deuticke: Leipzig.
Bleuler, M. (1972). Die schizophrenen Geistesstorungen im Lichte
langjdhrigen kranken- und Familiengeschichten. Thieme: Stutt-
gart.
Bleuler, M. (1981). Einzelkrankheiten in der Schizophrenie-Gruppe?
In Schizophrenic Stand und Entwicklungstendenzen der Forschung
(ed. G. Huber), pp. 155-166. Schattauer: Stuttgart.
Book, J. A. (1953). A genetic and neuropsychiatric investigation of a
North Swedish population (with special regard to schizophrenia
and mental disorder). Ada Genetica 4, 1-100.
Book, J. A., Wetterberg, L. & Modrzewska, K. (1978). Schizophrenia
in aNorth Swedish geographical isolate, 1900-1977. Epidemiology,
genetics and biochemistry. Clinical Genetics 14, 373-394.
Brockington, I. F., Kendell, R. E. & Leff, J. P. (1978). Definitions
of schizophrenia: concordance and prediction of outcome.
Psychological Medicine 8, 387-398.
Brown, G. W. & Birley, J. L. T. (1968). Crises and life changes and
the onset of schizophrenia. Journal of Health and Social Behaviour
9, 203-214.
Brown, G. W., Birley, J. L. T. & Wing, J. K. (1972). Influence of
family life on the course of schizophrenic disorders: a replication.
British Journal of Psychiatry 121, 241-258.
Brugger, C. (1931). Versuch einer Geisteskrankenzahlung in
Thiiringen. Zeitschrift fur die Gesamte Neurologic und Psychiatrie
133, 252-290.
Brugger, C. (1933). Psychiatrische Ergebnisse einer medizinischen,
anthropologischen und soziologischen Bevolkerungsuntersuchung.
Zeitschrift fur die Gesamte Neurologie und Psychiatrie 146,
489-524.
Brugger, C. (1938). Psychiatrische Bestandesaufnahme im Gebiet
eines medizinischanthropologischen Zensus in der Nahe von
Rosenheim. Zeitschrift fur die Gesamte Neurologie und Psychiatrie
160, 189-207.
Carpenter, W. T., Strauss, J. S. & Muleh, S. (1973). Are there
pathognomonic symptoms in schizophrenia? Archives of General
Psychiatry 28, 847-852.
Ciompi, L. (1980). The natural history of schizophrenia in the long-
term. British Journal of Psychiatry 136, 413-420.
Crocetti, G. J., Lemkau, P. V., Kulcar, Z. & Kessic, B. (1971).
Selected aspects of the epidemiology of schizophrenia in Croatia,
Yugoslavia. II. The cluster sample and the results of the pilot
survey. American Journal of Epidemiology 94, 126-134.
95
Day, R., Nielsen, J. A., Korten, A., Ernberg, G., Dube, K. C ,
Gebhart, J., Jablensky, A., Leon, C , Marsella, A., Olatawura, M.,
Sartorius, N., Stromgren, E., Takahashi, R., Wig, N. & Wynne, L.
C. (1987). Stressful life events preceding the acute onset of
schizophrenia: a cross-national study from the World Health
Organization. Culture, Medicine and Psychiatry 11, 123-205.
Essen-Moller, E., Larsson, H., Uddenberg, C. E. & White, G. (1956).
Individual traits and morbidity in a Swedish rural population.
Ada Psychiatrica Neurologica Scandinavica, Suppl. 100.
Fenton, W. S., Mosher, L. R. & Mathews, S. M. (1981). Diagnosis of
schizophrenia: a critical review of current diagnostic systems.
Schizophrenia Bulletin 7, 452-476.
Fremming, K. H. (1947). Sygdomsrisikoen for Sindslidelser og Andre
Sjaelelige Abnormtilstande i den Danske Gennemsnitsbefolkning.
Munksgaard: Copenhagen.
Gottesman, I. I. & Shields, J. (1982). Schizophrenia: The Epigenetic
Puzzle. Cambridge University Press: New York.
Gross, G., Huber, G. & Schutter, R. (1981). Uebereinstimmungen
und Unterschiede in den Ergebnissen neuerer Langzeitstudien bei
schizophrenen Kranken. In Schizophrenic. Stand und Entwick-
lungstendenzen der Forschung (ed. G. Huber), pp. 39-57.
Schattauer: Stuttgart.
Hafner, H. & Reimann, H. (1970). Spatial distribution of mental
disorders in Mannheim, 1965. In Psychiatric Epidemiology (ed. E.
H. Hare and J. K. Wing), pp. 341-354. Oxford University Press:
London.
Hagnell, O. (1966). A Prospective Study of the Incidence of Mental
Disorder. Svenska Bokforlaget: Lund.
Helgason, L. (1977). Psychiatric services and mental illness in
Iceland. Ada Psychiatrica Scandinavica, Suppl. 268.
Helgason, T. (1964). Epidemiology of mental disorders in Iceland.
Ada Psychiatrica Scandinavica, Suppl. 173.
Hoenig, J. (1984). Schneider's first rank symptoms and the tabulators.
Comprehensive Psychiatry 25, 77-87.
Hollingshead, A. B. & Redlich, F. C. (1958). Social Class and Mental
Illness. Wiley & Sons: New York.
Jablensky, A. (1987). Multicultural studies and the nature of
schizophrenia: a review. Journal of the Royal Society of Medicine
80, 16-2-167.
Jablensky, A., Schwarz, R. & Tomov, T. (1980). WHO collaborative
study on impairments and disabilities associated with schizophrenic
disorders. Ada Psychiatrica Scandinavica 62, Suppl. 285, 152-163.
Joensen, S. & Wang, A. G. (1983). First admissions for psychiatric
disorders. A comparison between the Faroe Island and Denmark.
Ada Psychiatrica Scandinavica 68, 66-71.
Kadri, Z. N. (1963). Schizophrenia in university students. Singapore
Medical Journal 4, 113-118.
Kato, M. (1969). Psychiatric epidemiological surveys in Japan: the
problem of case finding. In Mental Health in Asia and the Pacific.
East-West Center Press: Honolulu.
Katz, M. M., Sanborn, K. O., Lowery, H. A. & Ching, J. (1978).
Ethic studies in Hawaii on psychopathology and social deviance.
In The Nature of Schizophrenia: New Approaches to Research and
Treatment (ed. L. Wynne, R. Cromwell and S. Methysse). John
Wiley: New York.
Katz, M. M., Marsella, A., Dube, K. C , Olatawura, M., Takahashi,
R., Nakane, Y., Wynne, L. C , Gift, T., Brennan, J., Sartorius, N.
& Jablensky, A. (1988). On the expression of psychosis in different
cultures: schizophrenia in an Indian and in a Nigerian community:
a report from the World Health Organization Project on
Determinants of Outcome of Severe Mental Disorders. Culture,
Medicine and Psychiatry 12, 331-355.
96 A. Jablensky and others
Kendell, R. E. & Brockington, I. F. (1980). The identification of
disease entities and the relationship between schizophrenic and
affective psychoses. British Journal of Psychiatry 137, 324-331.
Klemperer, J. (1933). Zur Belastungsstatistik der Durchschnit-
tsbevolkerung. Psychosehaufigkeit unter 1000 stichprobemassig
ausgelesenen Probanden. Zeitschrift fiir die Gesamte Neurologie
und Psychiatric 146, 277-316.
Koehler, K. (1979). First rank symptoms of schizophrenia: questions
concerning clinical boundaries. British Journal of Psychiatry 134,
236-248.
Kraepelin, E. (1920). Die Erscheinungsformen des Irreseins. Zeit-
schrift fur die Gesamte Neurologie und Psychiatrie 62, 1-29.
Kramer, M. (1980). The rising pandemic of mental disorders and
associated chronic diseases and disabilities. In Epidemiological
Research as Basis for the Organization of Extramural Psychiatry
(ed. E. Stromgren, A. Dupont and J. Achton-Nielsen), pp.
382-397. Ada Psychiatrica Scandinavica, Suppl. 285.
Krasik, E. D. & Semin, I. R. (1980). Epidemiological aspects of first
admissions of schizophrenic patients. Zhurual Nevropatolgii
Psikijatrii 80, 1354-1359.
Kritzer, H. M. (1979). Analyzing contingency tables by weighted
least squares alternative to the Goodman approach. 78 Politic 5,
277-326.
Kritzer, H. M. (1979). Approaches to the analysis of complex
contigency tables: a guide for the perplexed. 79, Social Methods
Research 7, 305-329.
Krupinski, J. & Alexander, L. (1983). Patterns of psychiatric
morbidity in Victoria, Australia, in relation to changes in diagnostic
criteria 1848-1978. Social Psychiatry 18, 61-67.
Larsson, T. & Sjogren, T. (1954). A methodological, psychiatric and
statistical study of a large Swedish rural population. Ada
Psychiatrica Neurologica Scandinavica, Suppl. 89.
Leff, J., Kuipers, L., Berkowitz, R., Vaughn, C. & Sturgeon, D.
(1983). Life events, relatives' expressed emotion and maintenance
neuroleptics in schizophrenic relapse. Psychological Medicine 13,
799-806.
Leff, J., Wig, N. N., Ghosh, A., Bedi, H., Menon, D. K., Kuipers, L.,
Nielsen, J. A., Thestrup, Grethe, Korten, A., Ernberg, G., Day,
R., Sartorius, N. & Jablensky, A. (1987). Expressed emotion and
schizophrenia in North India. III. Influence of relatives' expressed
emotion on the course of schizophrenia in Chandigarh. British
Journal of Psychiatry 151, 166-173.
Leff, J., Wig, N. N., Bedi, H., Menon, D. K., Kuipers, L., Korten,
A., Ernberg, G., Day, R., Sartorius, N. & Jablensky, A. (1990).
Relatives' expressed emotion and the course of Schizophrenia in
Chandigarh: a two-year follow-up of a first-contact sample. British
Journal of Psychiatry 156, 351-356.
Leff, J., Sartorius, N., Jablensky, A., Korten, A. & Ernberg, G.
(in press) The International Pilot Study of Schizophrenia: five-year
follow-up findings. Psychological Medicine.
Lewine, R., Renders, R., Kirchhofer, M., Monsour, A. & Watt, N.
(1982). The empirical heterogeneity of first rank symptoms in
schizophrenia. British Journal of Psychiatry 140, 498-502.
Liebermann, Y. I. (1974). On the problem of incidence of schizo-
phrenia (materials from a clinical and epidemiological survey).
Zhurnal Nevropatologii Psikijatrii 74, 1224-1231.
Lin, T., Rin, H., Yeh, E., Hsu, C. & Chu, H. (1969). Mental disorders
in Taiwan, fifteen years later: A preliminary report. In Mental
Health Research in Asia and the Pacific (ed. W. Candill and T.
Lin), pp. 66-91. East-West Center Press: Honolulu.
McGuffin, P. (1988). Major genes for major affective disorders?
British Journal of Psychiatry 153, 591-596.
Marneros, A. (1984). Sichern die Symptome ersten Ranges die
Diagnose Schizophrenic? Nervenarzt 55, 365-370.
Munk-J0rgensen, P. (1986 a). Decreasing first-admission rates of
schizophrenia among males in Denmark from 1970 to 1984.
Changing diagnostic patterns? Ada Psychiatrica Scandinavica 73,
645-650.
Munk-j0rgensen, P. (1986A). Decreasing rates of first-admission
diagnoses of schizophrenia among females in Denmark from 1970
to 1984. Ada Psychiatrica Scandinavica 74, 379-383.
Murphy, H. B. M. (1968). Cultural factors in the genesis of
schizophrenia. In The Transmission of Schizophrenia (ed. D.
Rosenthal & S. S. Kety), pp. 137-153. Pergamon: Oxford.
Murphy, H. B. M. (1978). Cultural differences in incidence, course,
and treatment response. In The Nature of Schizophrenia (ed. L. C.
Wynne, R. L. Cromwell and S. Matthysse), pp. 586-594. Wiley &
Sons: New York.
Murphy, H. B. M. & Raman, A. C. (1971). The chronicity of
schizophrenia in indigenous tropical peoples. Results of a twelve-
year follow-up survey in Mauritius. British Journal of Psychiatry
118, 489^t97.
Murphy, H. B. M. & Vega, G. (1982). Schizophrenia and religious
affiliation in Northern Ireland. Psychological Medicine 12, 595-605.
Murray, R. & Lewis, S. W. (1988). Is schizophrenia a neurodevelop-
mental disorder (letter)? British Medical Journal 296, 6614-
6663.
Nakane, Y., Ohta, Y. & Radford, M. H. B. (in press). Epidemio-
logical studies of schizophrenia in Japan. Schizophrenia Bulletin.
ni Nuallain, M., Buckley, H., McHugh, B., Ottare, A. & Walsh, D.
(1984). Methodology of a study of mental illness in Ireland. Irish
Journal of Psychiatry 5, 4-9.
Nielsen, J. (1976). The Samso project from 1957 to 1974. Ada
Psychiatrica Scandinavica 54, 198-222.
Norris, V. (1959). Mental Illness in London. Maudsley Monograph.
Chapman & Hall: London.
O'Hare, A. & Walsh, D. (1980). Activities of Irish Psychiatric
Hospitals and Units, 1978. The Medico-Social Board: Dublin.
Odegaard, O. (1946). A statistical investigation of the incidence of
mental disorder in Norway. Psychiatric Quarterly 20, 381-401.
Odegaard, O. (1952). The incidence of mental diseases as measured
by census investigation versus admission statistics. Psychiatric
Quarterly 26, 212-218.
Ouspenskaya, L. Y. (1978). Problems of methodology of comparative
epidemiological studies and the characteristics of prevalence of
schizophrenia in various areas of the country. Zhurnal Nevro-
patologii Psikijatrii 78, 724-748.
Peterson, A. V. & Kronmal, R. A. (1985). Mixture method. En-
cyclopedia of Statistical Sciences 5, 579-583.
Raman, A. C. & Murphy, H. B. M. (1972). Failure of traditional
prognostic indicators in Afro-Asian psychotics: results of a long-
term follow-up survey. Journal of Nervous Mental Disorders 154,
238-247.
Rotstein, V. T. (1977). Material from a psychiatric survey of sample
groups from the adult population in several areas of the USSR.
Zhurnal Nevropatologii Psikijatrii 11, 569-574.
Riidin, E. (1916). Zur Vererbung und Neuentstehung der Dementia
Praecox. Springer: Berlin.
Sartorius, N., Jablensky, A. & Shapiro, R. (1977). Two-year follow-
up of the patients included in the WHO International Pilot Study
of Schizophrenia: Preliminary Communication. Psychological
Medicine 7, 529-541.
Sartorius, N., Davidian, H., Ernberg, G., Fenton, F. R., Fujii, I.,
Gastpar, M., Gulbinat, W., Jablensky, A., Kielholz, P., Lehmann,
H. E., Naraghi, M., Shimizu, M., Shinfuku, N. & Takahashi, R.
(1983). Depressive Disorders in Different Cultures. World Health
Organization: Geneva.
Sartorius, N., Jablensky, A., Ernberg, G., Leff, J., Korten, A. &
Gulbinat, W. (1987). Course of schizophrenia in different countries:
some results of a WHO international comparative 5-year follow-
up study. In Search for the Causes of Schizophrenia (ed. H. Hafner,
W. F. Gattaz and W. Janzarik), pp. 107-113. Springer-Verlag:
Berlin.
Scharfetter, C , Moerbt, H. & Wing, J. K. (1976). Diagnosis of
functional psychoses: comparison of clinical and computerized
classifications. Archiv fur Psychiatrischen Nervenkrankenheiten
222, 61-67.
Schneider, K. (1959). Clinical Psychopathology. Grune & Stratton:
New York.
Schwartz, C. C. & Myers, J. K. (1977). Life events and schizophrenia.
I. Comparison of schizophrenics with a community sample.
Archives of General Psychiatry 34, 1238-1241.
 Schizophrenia: A World Health Organization Ten-Country Study
Shen Yucun, Zhang Weixi, Shu Liang, Yang Xiaoling, Cui Yuhua &
Zhou Dongfeng (1981). Investigations of mental disorders in
Beijing suburban district. Chinese Medical Journal 94, 153-156.
Shepherd, M. (1957). A Study of the Major Psychoses in an English
County. Maudsley Monograph No. 3. Chapman & Hall: London.
Shepherd, M. (1982). Psychiatric epidemiology and the classification
of mental disorder. International Journal of Epidemiology 11,
312-314.
Shmaonova, L. M. (1983). Possibilities of the epidemiological method
and results of population studies on schizophrenia. Zhurnal
Nevropatologii Psikijatrii 83, 707-716.
Sjogren, T. (1948). Genetic-statistical and psychiatric investigations
of a West Swedish population. Ada Psychiatrica Neurologica
Scandinavica, Suppl. 52.
Stromgren, E. (1938). Beitrage zur psychiatrischen Erblehre, auf
Grund von Untersuchungen an einer Inselbevolkerung. Acta
Psychiatrica Neurologica Scandinavica, Suppl. 19.
Stromgren, E. (1950). Statistical and genetical population studies
within psychiatry. Methods and principal results. Proceedings of
First International Congress of Psychiatry, Paris 1950, vol. VI, pp.
155-192. Herman et Cie: Paris.
Temkov, I., Jablensky, A. & Boyadjieva, M. (1980). Use of reported
prevalence data in cross-national comparisons of psychiatric
morbidity. Social Psychiatry 3, 111-117.
Torrey, E. F. (1980). Schizophrenia and Civilization. Jason Aronson:
New York.
Torrey, E. F., McGuire, M., O'Hare, A., Walsh, D. & Spellman, M.
P. (1984). Endemic psychosis in Western Ireland. American Journal
of Psychiatry 141, 966-970.
Walsh, D. (1969). Mental illness in Dublin - first admissions. British
Journal of Psychiatry 115, 449^t56.
Walsh, D., O'Hare, A., Blake, B., Halpenny, J. V. & O'Brien, P. F.
(1980). The treated prevalence of mental illness in the Republic of
Ireland - the three county cases register study. Psychological
Medicine 10, 465-470.
Watt, D. C , Katz, K. & Shepherd, M. (1983). The natural history of
schizophrenia: a 5-year prospective follow-up of a representative
sample of schizophrenics by means of standardized clinical and
social assessment. Psychological Medicine 13, 663-670.
Waxier, N. E. (1979). Is outcome for schizophrenia better in non-
industrial societies? The case of Sri Lanka. Journal of Nervous
Mental Disorders 167, 144-158.
97
Weiner, B. P. & Marvit, R. C. (1977). Schizophrenia in Hawaii:
analysis of cohort mortality risk in a multi-ethnic population.
British Journal of Psychiatry 131, 497-503.
WHO (1973). Report of the International Pilot Study of Schizophrenia.
World Health Organization: Geneva.
WHO (1978). Mental Disorders: Glossary and Guide to their
Classification in Accordance with the Ninth Revision of the
International Classification of Diseases. World Health Organ-
ization : Geneva.
WHO (1979). Report of the International Follow-up Study of
Schizophrenia. John Wiley & Sons: New York.
WHO (1981). Current State of Diagnosis and Classification in the
Mental Health Field. Document MNH/81.11. World Health
Organization: Geneva.
Wig, N. N. (1982). Methodology of data collection in field surveys.
In Epidemiology and Mental Health Services. Principles and
Applications in Developing Countries (ed. T. A. Baasher, J. E.
Cooper, H. Davidian, A. Jablensky, N. Sartorius and E. Strom-
gren). Acta Psychiatrica Scandinavica 65, Suppl. 296, pp. 77-86.
Wig, N. N., Menon, D. K., Bedi, H., Ghosh, A., Kuipers, L., Leff, J.,
Nielsen, J. A., Thestrup, Grethe, Korten, A., Day, R., Sartorius,
N., Ernberg, G. & Jablensky, A. (1987). Expressed emotion and
schizophrenia in North India. I. Cross-Cultural transfer of ratings
of relatives' expressed emotion. British Journal of Psychiatry 151,
156-160.
Wijesinghe, C. P., Dissanayake, S. A. W. and Dassanayake, P. V. I.
N. (1978). Survey of psychiatric morbidity in a semi-urban
population in Sri Lanka. Acta Psychiatrica Scandinavica 58,
413-441.
Wing, J. K. (1976). A technique for study psychiatric morbidity in in-
patient and out-patient series and in general population samples.
Psychological Medicine 6, 665-671.
Wing, J. K. & Fryers, T. (1976). Statistics from the Camberwell and
Salford Psychiatric Registers 1964-1974. MRC Social Psychiatry
Unit, Institute of Psychiatry, London, and Dept. of Community
Medicine, University of Manchester.
Wing, J. K., Cooper, J. E. & Sartorius, N. (1974). Measurement and
Classification of Psychiatric Symptoms. Cambridge University
Press: London.
Wing, L., Wing, J. K., Hailey, A., Bahn, A. K., Smith, H. E. &
Baldwin, J. A. (1967). The use of psychiatric services in three urban
areas: an international case register study. Social Psychiatry 2, 158.