Content uploaded by Kim Hopper
Author content
All content in this area was uploaded by Kim Hopper on Apr 07, 2014
Content may be subject to copyright.
Revisiting the Developed Versus Developing
Country Distinction in Course and Outcome in
Schizophrenia: Results From ISoS, the WHO
Collaborative Followup Project
by Kim Hopper and Joseph Wanderling
Abstract
This article examines the long-standing and provoca-
tive finding of a differential advantage in course and
outcome for persons with schizophrenia living in
"developing" countries, using results from the newly
completed World Health Organization (WHO) collab-
orative project, the International Study of
Schizophrenia (ISoS). The article addresses two ques-
tions:
Has the differential survived the 13 years since it
was last reported? If so, are the results demonstrably
not attributable to artifactual confounding? The
analysis focuses on the 809 subjects who make up the
combined incidence cohort of ISoS. These include
members of the original treated incidence cohorts of
two earlier WHO studies (the Determinants of
Outcome of Severe Mental Disorders and the
Reduction of Disability Studies) as well as subjects
drawn from two additional samples (Hong Kong and
Madras/Chennai). We first review the consistency of
the finding of a "developed versus developing" differ-
ential in course and outcome and then examine a vari-
ety of course and outcome measures for the ISoS inci-
dence cohorts. Evidence of differences in illness
trajectory in favor of the developing centers was con-
sistently found. Six potential sources of bias are then
examined: differences in followup, arbitrary grouping
of centers, diagnostic ambiguities, selective outcome
measures, gender, and age. None of these potential
confounds explains away the differential in course and
outcome. We conclude with suggestions for further
research, with particular attention to the need for close
documentation of everyday practices in the local
moral worlds that "culture" refers to.
Keywords: Schizophrenia, culture, developing
versus developed, recovery, WHO.
Schizophrenia Bulletin, 26(4):835-846, 2000.
Considerable interest and no little skepticism have long
been aroused by reports of better outcome for schizophre-
nia in less industrialized, more "traditional" societies.
Methodological hazards aside, the claim appears to fly in
the face of clinical reason and experience. Yet the long-
term outcome of schizophrenia in societies of relatively
comparable "development" status remains stubbornly
diverse, a fact documented in studies of varied prove-
nance (e.g., Bleuler 1978; Warner 1985; Harding 1988;
Strauss 1994; Davidson and McGlashan 1997). Few seem
eager to concede that course of illness may be "hard-
wired" by prognostic factors that are essentially fixed by
the time of first onset—or at any rate, by the time the suf-
ferer is first treated. Yet the alternative suggestion—that
broad cultural factors, to some extent independent of ini-
tial illness severity or premorbid functioning, not only
influence course but may substantially set differential
probabilities of recovery for what seem to be clinically
similar entities—still seems a difficult proposition. The
well-earned methodological suspicions aside, considera-
tions of therapeutic competence and armamentaria, con-
sistency of treatment and follow-through, the depreda-
tions of poverty, and the uncertainties of informal support
are the usual shoals on which provisional acceptance of
such a proposition founders.
Before potential reasons for the difference bear
examining, however, the thing to be explained—that dif-
ferential rates of recovery persist in defying the usual crit-
icisms advanced for discounting them—should be
soundly established. The most recent of the WHO fol-
lowup studies of schizophrenia offers an opportunity for
revisiting this much-contested finding.
Statement of the Problem
In 1967, WHO initiated a set of studies investigating the
manifestation, consequences, and course of schizophrenia
and related disorders. Since then, nearly 30 research sites
in 19 countries have participated. These studies—specifi-
Send reprint requests to Dr. K. Hopper, Nathan Kline Institute, 140 Old
Orangeburg Rd., Orangeburg, NY 10962; e-mail: hopper@nki.rfrnh.org.
835
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Schizophrenia
Bulletin,
Vol.
26,
No.
4, 2000K. Hopper
and
J.
Wanderling
cally, the International Pilot Study of Schizophrenia
(IPSS,
1967) and the Determinants of Outcome of Severe
Mental Disorders (DOSMeD, 1978), with initial followup
periods ranging from 2 to 5 years—have consistently
found persons clinically diagnosed with schizophrenia
and related disorders in the industrialized West (chiefly
Europe and the United States) to have less favorable out-
comes than their counterparts in "developing" countries
(countries in Africa, Asia, and Latin America) (WHO
1973;
Jablensky et al. 1992). Although the number of dis-
tinctive "cultures" was small1 and there were a few anom-
alies,2
the durability of this finding, extensively docu-
mented and assessed with increasingly sophisticated
instruments, is quite remarkable—arguably the more so
for being built on so anthropologically rickety a founda-
tion. By the late 1980s, the documentation that persons
diagnosed as suffering from schizophrenia consistently do
better in the long run in non-Western settings was being
hailed as possibly "the single most important finding of
cultural differences in cross-cultural research on mental
illness" (Lin and Kleinman 1988, p. 563).
But it was far from clear whether the pronounced dif-
ferences seen in short-term followups would hold up over
time.
Questions have been raised as well about the con-
ceptual adequacy of such labels as "developed" and
"developing" (Hopper 1991; Edgerton and Cohen 1994),
a point implicitly illustrated by the anomalous refusals
(cited in footnote 2) of a few centers to group with their
assigned class. Diagnostic ambiguities invariably cloud
the picture when so many different investigators, some
hailing from distinctive psychiatric traditions, are
included; the ambiguities are compounded when as much
as a quarter-century has elapsed since the initial assess-
ment (Gureje 1996, p. 128). Most relevant here, what
accounts for the apparent "benefits" of underdevelopment
was not at all apparent (Kleinman 1988). Speculation
ranged widely. Cultural signposts certifying the expecta-
tion of recovery, self-exempting modes of illness attribu-
tion, the therapeutic benefits of accommodating work,
kin-based stores of supportive social capital, the relative
anonymity of life in the industrialized world—all of these
have been proposed as explanatory mechanisms (Cooper
and Sartorius 1977; WHO 1979; Warner 1985).
1 The "developing" world was represented by samples from Ibadan
(Nigeria), Cali (Colombia), and Agra (India) in IPSS; by Ibadan, Cali,
Agra, and Chandigarh (India) in DOSMeD; and by Chandigarh, Madras
(India),
Beijing (China), and Hong Kong in ISoS. If one combines the
three studies, the category takes in a single Latin American and African
example each, two from Asia, and three from the Indian subcontinent.
2 Early on, for example, it was apparent that the short-term pattern of
course for IPSS subjects in Cali "approximated centres in developed
countries" (WHO 1979, p. 369n). Conversely, in a recent recursive parti-
tioning analysis of short-term outcome in DOSMeD (Craig et al. 1997),
Prague (Czechoslovakia) and Nottingham (United Kingdom) tended to
align with the developing centers.
Hence the timeliness of the recently completed ISoS,
the latest of the WHO Collaborative Projects. In early
1997,
investigators completed data collection in followup
interviews of both the original IPSS prevalence cohort (26
years after the episode of inclusion) and the DOSMeD
cohort (13-16 years after initial episode), as well as two
other groups of subjects—an incidence cohort from each
of three centers of the WHO Reduction and Assessment of
Psychiatric Disability Study (RAPyD, 1978) and a mixed
set of subjects (two treated incidence cohorts, one preva-
lence) from three additional invited centers (see Hopper et
al.,
in press).
This article has a modest aim: to examine as closely
as the available data permit the durability and soundness
of that provocative finding of a differential advantage in
course and outcome for the developing countries. Has the
differential outcome survived the 13 years since last
reported for (some of) these same subjects? If so, are the
results demonstrably not attributable to artifactual con-
founding?
We focus here on course and outcome for the com-
bined incidence cohort of ISoS—that is, for the 809 sub-
jects followed since "first (treated) episode" of psychosis
(table 1), only some of whom appeared in earlier WHO
analyses of DOSMeD cohorts. We first review the consis-
tency of the finding of a "developed versus developing"
differential in course and outcome in three WHO studies.
Next, we examine a variety of course and outcome mea-
sures for the ISoS incidence cohorts that bear upon differ-
ences in illness trajectory for the two groups. We analyze
five potential sources of bias and assess their likely
impact on these reported differences. We conclude with
some directions for further analyses.
Methods
With 13 research centers, spread across 11 countries, and
a training, data gathering, and analysis period that
spanned nearly a decade, questions of standardization of
methods inevitably arise. Detailed descriptions of the
ISoS research protocol and instruments, reliability exer-
cises,
and analysis of cohort bias are available elsewhere
(Sartorius et al. 1996; Drake et al., in press; Siegel et al.,
in press). Suffice it to note here that differential attrition
seems not to have substantially biased the followup
cohort, although there was a nonsignificant trend of sub-
jects with poorer prognostic traits (male subjects, those
with slow illness onset) being more likely lost to fol-
lowup.
Entry diagnoses were assigned by local clinicians,
most of whom had undergone a training regimen in the
use of a common psychopathology assessment tool (the
Present State Examination [PSE-9], supplemented in
836
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Developed vs. DevelopingSchizophrenia Bulletin, Vol. 26, No. 4, 2000
Table 1. WHO ISoS—treated
Location
"Developing" Centers
incidence cohorts
Original n Followup n1Lost to followupDeaths
Chandigarh (D)2—urban
Chandigarh (D)—rural
Hong Kong (I)
Madras (I)
Total
155
55
100
100
410
67
83
71
70
72
115
99
118
58
60
813
80
38
70
77
265 (65%)
37
63
26
56
52
57
86
79
33
55
544 (67%)
61
7
19
14
101 (25%)
22
11
41
7
10
51
4
28
25
3
202 (25%)
14
10
11
9
44(11%)
8
9
4
7
10
7
9
11
0
2
67 (8%)
"Developed" Centers
Dublin (D)
Groningen (R)
Honolulu (D)
Mannheim (R)
Moscow (D)
Nagasaki (D)
Nottingham (D)
Prague (D)
Rochester (D)
Sofia (R)
Total
Note.—ISoS
= International Study of Schizophrenia; WHO = World Health Organization.
1 Followup time ranged from
13
to
17
years.
2 Original study in parentheses. (D): DOSMeD. (R): RAPyD. (I): Invited. Note that of the original developing centers of DOSMeD,
Agra,
Cali,
and Ibadan did not participate in ISoS; neither
did
Aarhus,
a developed center.
most cases by the Psychiatric and Personal History
Schedule). The PSE was translated into indigenous lan-
guages as needed. For over half of the ISoS sites (those
made up of the DOSMeD centers), these entry diagnoses
were reviewed by a WHO-convened group of experts.
Both before and during the study, investigators were
required to demonstrate reliability (within centers and
across centers) by rating videotaped clinical interviews of
subjects. There is little question that these diagnoses rep-
resent local psychiatric "ethnographic reality" as it
appeared 15 years ago. The official WHO analyses
(Hopper et al., in press) use the nomenclature of the most
recent ICD-10 standard, obtained by using an algorithmic
conversion from the original ICD-9 diagnoses (WHO
1994).3
In order to minimize the risk of classification arti-
fact, in the analyses below we present findings by several
different diagnostic conventions.
Findings
Consistency of the Developed versus Developing
Differential in Course and Outcome. As table 2 illus-
3 See Craig et al., in press, for a discussion of diagnostic stability over
time in ISoS.
trates,
the finding of a consistent outcome differential
favoring the "developing" centers is remarkably robust. It
extends across all three WHO collaborative projects. It
holds for followup periods ranging from 2, to 5, to 15
years.
It applies when various diagnostic groupings are
used (for ISoS: ICD-9 schizophrenia, converted ICD-10
schizophrenia, and all psychoses). It holds when country
groupings shift (note the changes in table 1 in the makeup
of developing and developed groupings from DOSMeD to
ISoS).
It even appears to be relatively constant, as indi-
cated by the odds ratios for recovery calculated in the far
right column of table 2.
Other Course and Outcome Indicators for ISoS. The
late course differential in table 2 (percentage showing no
psychotic episodes vs. percentage with continuous illness
in the most recent 2 years of followup) holds for other
outcome indicators as well. As table 3 illustrates, the dif-
ferential favoring developing centers applies to general
clinical state (the Bleuler scale; Bleuler 1978), symptoma-
tology (scores on the Global Assessment of
Functioning-Symptoms scale; APA 1987), disability
(scores on the Global Assessment of Functioning—
Disability scale [APA 1987] and Disability Assessment
Schedule [WHO 1988]), and social functioning (at least
for paid work or housework). It holds up, too, whether a
narrow (ICD-10) or broad ("spectrum") classification of
837
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Schizophrenia Bulletin, Vol. 26, No.
4,
2000K. Hopper and
J.
Wanderling
Table
Study
2.
WHOoutcomestudies—a synopsis
Percentages
Developed2
"Best" vs. "Worst"
Developing
Outcomes1
OddsRatio3
IPSS(1967-)
2-yr followup
5-yr followup
DOSMeD(1976-;
All subjects
ICD-9
SZ
35
vs.
33
23 vs.
24
33 vs.
17
32 vs.
19
52 vs.
19
38
vs.
14
49 vs.
11
49
vs.
13
ISoS 15-yr followup (incidence only),
%
never psychotic
in
last
2
yrs vs. continuously psychotic
ICD-9
SZ
40 vs.
33
58 vs.
23
ICD-10SZ
37
vs.
38
53 vs.
27
All psychoses
45
vs.
30
58 vs.
22
2.01
2.05
1.95
2.04
2.07
1.92
1.69
Note.—DOSMed
=
Determinants
of
Outcome
of
Severe Mental Disorders study; IPSS
=
International Pilot Study
of
Schizophrenia;
ISoS = International Study
of
Schizophrenia; SZ = schizophrenia; WHO = World Health Organization.
1 Various measures
of
patterns
of
course were used in the individual studies.
2 Assignment of centers to categories
of
developed
vs.
developing
is
as per individual studies.
3 These are the odds
of
good outcome in developing centers
vs.
the odds of good outcome in developed centers.
Table 3. ISoS results—"developed" vs. "developing" centers
Outcome
measures
Bleuler scale
% recovered
% severe
GAF-S
%
> 60
GAF-D
%
> 60
Global
DAS
% excellent
or
good
Last
2-yr
course
% never psychotic
% continuously psychotic
Working2
most
of
last
2 yrs (%)
ICD-10SZ
Developed
n = 319
44
12
43
41
24
37
38
46
Developing
n = 183
55
9
70
65
53
53
27
73
Diagnostic
SZ + SA,
Developed
n = 410
49
10
48
44
28
40
33
49
Grouping
SZ-like1
Developing
n
=
230
60
7
73
69
57
58
23
77
All psychoses
Developed
n
=
516
55
9
53
47
32
45
30
51
Developing
n
=
260
59
7
73
69
58
58
22
79
Note.—DAS
=
Disability Assessment Schedule; GAF-S
=
Global Assessment
of
Functioning-Symptoms; GAF-D
=
Global
Assessment
of
Functioning-Disability and Disability Assessment; ISoS
=
International Study on Schizophrenia; SA = schizoaffective;
SZ = schizophrenia; SZ-like = acute schizophrenialike psychoses.
1 This approximates the ICD-9 schizophrenia category.
2 This indicates
paid
job or housework.
schizophrenia is used, or if the diagnostic net is expanded
to take in all psychoses. Table 4 (A and B) shows that the
odds ratios in favor of the developing centers range from
1.57 to 3.51 for ICD-10 diagnosis of schizophrenia, and
from 1.59 to 3.61 for the broad spectrum diagnosis (schizo-
phrenia, schizoaffective disorder, and acute schizophrenia-
like psychoses) when Hong Kong is included as a develop-
ing center. The same table shows how the odds ratios shift
somewhat when Hong Kong is classified as a developed
center but still consistently favor the developing group.
838
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Developed vs. DevelopingSchizophrenia Bulletin, Vol. 26, No. 4,2000
Table 4. Odds ratios and various outcome measures, "developed" vs. "developing"
A. ICD-10 schizophrenia
Outcome measure
Developed
319(n = 388)1
Developing
n=183(n=114)1
Odds Ratio2 (Confidence Interval)
Hong Kong as Hong Kong as
"developing" "developed"
Bleuler scale, % recovered 44 (46)
GAF-S,
% > 60 43 (49)
GAF-D,
% > 60 41 (46)
Global DAS, % excellent or good 24 (28)
Last 2-yr course, % never psychotic 37 (38)
Working most of last 2 yrs, % 46 (49)
55 (56) 1.57 (1.09-2.27) 1.52 (1.00-2.32)
70(69) 3.15(2.12-4.70) 2.32(1.48-3.65)
65 (66) 2.64 (1.79-3.90) 2.25 (1.45-3.51)
53(49) 3.51(2.27-5.41) 2.47(1.57-3.89)
53 (59) 1.97 (1.36-2.86) 2.33 (1.52-3.57)
73(82) 3.13(2.09-4.70) 4.71(2.78-8.00)
Note.—DAS = Disability Assessment Schedule; GAF-S = Global Assessment of Functioning-Symptoms; GAF-D
Assessment of Functioning-Disability.
1 Data in parentheses include Hong Kong in the "developed" group.
2 These are the odds of good outcome in developing centers
vs.
the odds of good outcome in developed centers.
B. Schizophrenia spectrum1
Global
Outcome measure n =
Bleuler scale, % recovered
GAF-S,
% > 60
GAF-D,
% > 60
Global DAS, % excellent or good
Last 2-yr course, % never psychotic
Working most of last 2 yrs, %
Developed
410(n = 480)2
49 (49)
48 (52)
44 (47)
28(31)
40 (40)
49 (50)
Developing
n = 230(n = 160)2
60 (63)
73 (74)
69(71)
57 (56)
58 (64)
77 (85)
Odds Ratio3 (Confidence Interval)
Hong Kong as
"developing"
1.59(1.15-2.21)
3.05(2.13-4.38)
2.74(1.93-3.89)
3.48 (2.38-5.08)
2.07(1.49-2.88)
3.61 (2.48-5.25)
Hong Kong as
"developed"
1.77(1.23-2.56)
2.71 (1.81-4.05)
2.73(1.84-4.04)
2.87(1.94-4.26)
2.62(1.80-3.79)
5.70 (3.52-9.23)
Note.—DAS = Disability Assessment Schedule; GAF-S = Global Assessment of Functioning-Symptoms; GAF-D = Global
Assessment of Functioning-Disability.
1 This is ICD-10 schizophrenia, plus schizoatfective disorder and acute schizophrenialike psychoses.
2 Data in parentheses include Hong Kong in the "developed" group.
3 These are the odds of good outcome in developing centers vs. the odds of good outcome in developed centers.
Potential Sources of Bias. Such findings could be artifac-
tual if it were the case that underlying, nonrandom differ-
ences between the two groups were operating in a way
that favored a finding of relative benefit in the developing
group. Several candidates are plausible.
Ascertainment. Even if the original groups were
comparable diagnostically, if systematic differences crept
in during the followup endeavor, this could skew the pic-
ture of comparative outcome. As table 1 shows, however,
lost-to-followup rates are comparable for the two groups
(25%),
and differences owing to mortality are small (11%
vs.
8%). But suppose the difference entered in who was
lost to followup: if it were the case, for example, that
"developed" center subjects who recovered were more
mobile, less easily located through clinical records, and
less likely to be interested in participating once relocated
(all of which was suggested, anecdotally, in reports from
the Rochester, NY, field team), then the "analyzable"
group would be artificially weighted in the direction of
poorer outcome (having "lost" more of those who were
better off). This turns out not to be the case. For both nar-
row and broad diagnostic classifications, the chances of
839
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Schizophrenia
Bulletin,
Vol. 26,
No.
4, 2000K. Hopper and J. Wanderling
showing up in the long-term followup are better for those
subjects in the developed world with a favorable early (2
years) course of illness (table 5).4
Arbitrary groupings. The developed versus devel-
oping grouping has been rightly criticized as more
bureaucratic convenience than analytic distinction. If any-
thing, the passage of 15 years has compounded the diffi-
culties, especially with respect to Hong Kong.5 With
Hong Kong assigned to the developing group, the effect
should be to subvert the detection of difference; in fact,
when it is shifted to the developed group, the odds ratios
tend to shrink with respect to symptoms and widen some-
what with respect to functioning. That the finding holds
up across time, study design, composition of groups, and
measures of outcome suggests that, however anthropolog-
ically incoherent the classification may be, it is marking
something real. Whether that something is best
"explained" by the crude taxonomy used to capture it is
another question.
Diagnostic ambiguities. In anthropological circles, it
is commonly argued that the term "schizophrenia" is part
of a powerful discursive practice that not only authorita-
tively names but also materially shapes the objects of its
attention. But this does not radically distinguish it from
other "disease entities" in the psychiatric (or biomedical)
taxonomy (Baruch and Teacher 1978; Sedgwick 1982;
Good 1994). Like race, another "dynamic [if disputed]
cultural category" (Barrett 1996, p. 305), schizophrenia
has long served as a marker of disorder and dysfunction.
Until a more disaggregated, phenomenologically based
4 A brief word on selection bias may be in order here. Murphy (1982)
has argued that irregularities in sampling in IPSS meant that more
patients with longer histories of schizophrenia (even with the 5-year
limit on onset) were included in the developed centers. (What his analy-
sis actually shows is that acute onset is more common among the devel-
oping centers and prior psychiatric contact less common.) Given huge
differences in local resources for treatment and custodial care—easily
accessible clinics and state-supported institutions chiefly—the converse
bias seems equally plausible. At the time of the IPSS, for example, Agra
had one psychiatric facility of 718 beds for a catchment area of 17 mil-
lion people (WHO 1973, pp. 54-55). To the extent that ease of access
affects the threshold at which families, significant others, or the police
resort to psychiatric facilities, there should be profound differences in
the mix of cases brought to clinical attention. Where thresholds of acces-
sibility are high (owing to transportation difficulties, costs, suspicion of
unfamiliar clinics, etc.) and local alternatives exist, "therapy-managing
groups" (Janzen 1978) may decide it makes practical sense to triage
informally all but the most recalcitrant cases.
5 Hong Kong hardly fits the mold of a developing country today—and
did not even at the time of the baseline interviews, a fact suggested
emphatically by the availability of detailed demographic data for Hong
Kong (but not for the developing centers of ISoS) in the 1980 United
Nations Demographic Yearbook. We include it here to make the compar-
ison of developing and developed the most conservative. An alternative
would be to exclude it from the analysis altogether—as was done with
Taipei in the original IPSS analysis, because of the per capita physician
rate,
availability of medical facilities, and leading causes of death (WHO
1979,
p. 148n).
set of categories proves its worth,6 even culturally dis-
puted terms are useful as signifiers of difference. The
detection of consistent patterns across time and place sug-
gests that durable differences are at stake, no matter how
problematic the surface markers.
A more pointed challenge is posed, however, by the
hypothetical notion of "non-affective acute remitting psy-
chosis" (NARP, Susser and Wanderling 1994; cf. Stevens
1987;
Desjarlais et al. 1995): a psychotic disorder misdiag-
nosed as schizophrenia with a markedly better prognosis
that may explain the "developing" advantage. (Note that
because course is part of
the
definition of
NARP,
that advan-
tage is built into the diagnosis.) To test this, we calculated
recovery rates within NARP and non-NARP groups (the rel-
evant results are displayed in table 6). Briefly, while NARP
turns out to be more common among cases diagnosed as
schizophrenia in the developing world, that "selection"
advantage is countered by an interaction effect: the differ-
ence NARP makes in enhancing the chances of recovery is
more profound in the developed world. The two effects
operate in opposing directions. The same table also high-
lights the contrary contextual (or interaction) effect in cen-
ters "corrected" for the unwarranted boost to recovery given
by NARP: recovery rates for non-NARP subjects in the
developing centers are 52 percent, as compared with 38 per-
cent in the developed world. Similar results, 55 percent ver-
sus 42 percent, are obtained for the broad spectrum schizo-
phrenia diagnosis as well.7
Selective outcome measures. It could further be the
case that the measures of outcome systematically privilege
contextual features of recovery that have little to do with
actual social function. Hospitalization, for example, is more
of an "administrative outcome" reflecting policy and
resource availability than an elastic indicator of need met
(Harrison et al. 1994). We have avoided most of these insti-
tutional effects in the outcome indicators shown here. Then
too,
the rather generous range of capacities tapped (from
symptom control to interference with function) and the con-
sistency of the differential across them argue otherwise.
Still, problems remain. Where welfare states offer reason-
able stipends (in the form of disability payments) to persons
certified as "disabled," the spur of necessity is blunted and
motivation to work may suffer. It might be argued, however,
that certain features of local necessity may well be central to
6 Say, for example, people begin using a symptom- or syndrome-based
classification system that takes explicit account of the clinical picture of the
80 percent of the world's population that is not part of North America or
Europe, continents that have supplied the cases on which the current
knowledge base of psychiatry is built (Kleinman and Cohen 1997).
7 Subtracting NARP-mediated recoveries and recomputing the rates
among developed and developing centers has little effect on the relative
odds of recovery; for Bleuler ratings of recovery, for example, the odds
ratio rises slightly from 1.3 to 1.4.
840
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Developed
vs.
DevelopingSchizophrenia Bulletin, Vol. 26, No. 4, 2000
Table 5. Ascertainment bias? Odds ratio of appearing in the long-term followup
ICD-10 SZ
Percent found
Percent lost
Odds ratio1
SZ spectrum2
Percent found
Percent lost
Odds ratio1
Developing
Remitting
course
84
160.85
84
161.00
Centers
Unremitting
course
86
14
84
16
Developed Centers
Remitting
course
87
13
. 83
17
Unremitting
course
81
19
1.56
80
20
1.22
Note.—SZ =
schizophrenia.
1 These are the odds of inclusion in study given remitting course
vs.
the odds of inclusion given unremitting course.
2
This
is ICD-10 schizophrenia, plus schizoaffective disorder and acute schizophrenialike psychoses.
Table 6. The effect of NARP
Center grouping
ICD-10 SZ; onset <
1
week1
Developed
Developing
SZ spectrum; onset <
1
week2
Developed
Developing
NARP
79
71
83
75
% Recovered
Non-NARP
38
52
42
55
"Effect"
(recovery ratio:
NARP/non-NARP)
2.1
1.4
2.0
1.4
Note.—NARP = non-affective remitting psychosis; SZ = schizophrenia.
1 The percent NARP in developed centers is 9.8; in developing centers, it is 15.5.
2 This is ICD-10 schizophrenia, plus schizoaffective disorder and acute schizophrenialike psychoses; the percent NARP in developed
centers is 12.5; in developing centers, it is 27.2.
the beneficial effect observed rather than just an artifact of it
(Warner 1985; Wikan 1996). Specifically, where nonmarket
work roles allow for great variation in the tasks and compe-
tencies socially valued, lingering disability may be less of a
barrier to useful employment. Conversely, useful work may
slow or arrest the evolution of disability. Far from being
mere incidental cultural music, tightly strapped circum-
stances and flexible means of addressing them may provide
therapeutic benefits forgone under circumstances of
enforced supported dependency.8
8 In the treated incidence group, the percentage of ICD-10 schizo-
phrenia subjects working for most of the past two years in the develop-
ing centers was 73 percent, vs. 46 percent for the developed centers; for
broad spectrum schizophrenia, the figures were 77 percent versus 49 per-
cent, respectively. Intriguingly, when looking only at subjects who were
rated as having substantial symptoms, significant disability, or both, the
developed centers actually reported slightly more (ICD-10 schizophre-
nia) subjects working
(21.3%
vs. 15.4%) or doing housework
(39.5%
vs.
34.6%) (see Hopper et al., in press).
Gender and age. Because a number of long-term fol-
lowup studies (though not all) have found female gender
to predict better outcome, we examined gender differ-
ences in the assessed cohort, in recovery rates and in sub-
jects lost to followup in developed and developing
groups. None showed evidence of a gender bias.
Similarly, NARP's effect on recovery is nearly the same
for men and women, in both developed and developing
groups.9
With respect to age, in both developed and develop-
ing countries, older subjects (41+ at the time of followup)
had better prospects of recovery. The developing centers'
9 In the developing world, however, non-NARP women are slightly
more likely to recover. Since NARP is more common among men in the
developing (15.1%) than in the developed world (6.6%), we adjusted to
equalize the two rates. This produced only an additional five cases of
recovery in the developed world.
841
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Schizophrenia
Bulletin,
Vol.
26,
No.
4, 2000K. Hopper and J. Wanderling
subjects were disproportionately younger, however. If
anything, then, the age distributions should give the
developed centers an advantage.
This brief digression into the potential effect of
NARP suggested another line of inquiry as well. Suppose
one were to divide the putative effect of context into an
early and later stage. The early stage would focus on ini-
tial course—specifically, remitting versus nonremitting or
continuous illness. The later stage would focus on the rel-
ative prospects of recovery in both those initial course
groupings. Table 7 shows this two-stage effect and the net
impact it has on the odds ratios of recovery for developing
and developed centers. In the first two years of illness tra-
jectory, the effect of context is registered primarily as
more favorable initial course; nearly half of the subjects
in the developing centers have remitting illness course, as
compared with less than a third in the developed centers.
Over the next 13 years, subjects with early remitting
course in both groups do quite well—with about two-
thirds showing recovery in the long run. But a delayed
secondary advantage is seen in the developing centers
with respect to subjects whose early course was unfavor-
able:
42 percent of them (compared with 33 percent in the
developed centers) go on to recover. If ISoS may be said
to yield any evidence for the "slow, uphill returns to
health" that Harding and colleagues (1992, p.34) have
urged us to look for, it may be found here: in the 22 per-
cent of subjects in the developing centers, and the 23 per-
cent of their counterparts in the developed world, who go
on to recover despite the poor prognosis suggested by an
unremitting early course of illness.
A final observation on diagnostic bias may be in
order. It is sometimes argued that "acute brief psychoses"
followed by full recovery are incompatible with a bona
fide schizophrenia diagnosis and may unfairly handicap
the picture of illness course in the developing world (e.g.,
Stevens 1987). Table 8 shows the effect on recovery rates
if all subjects with single episode psychoses are excluded
from the analysis of the broad spectrum schizophrenia
group. Rates drop substantially in both the developing and
developed groups, to 49 percent and 40 percent, respec-
tively, but still preserve the differential.
Discussion
That the developed versus developing differential has
proven so robust is generally taken as prima facie evi-
dence for the relevance of "culture" in influencing course
Table 7. A two-stage
Early course
(0-2 yrs)
Good
Poor
effect (ICD-10 schizophrenia)
Difference
(developing vs.
developed)
47%
vs.
31%
53%
vs.
69%
Odds ratio1
1.97
N.A.
Recovered at 15
yrs (developing
vs.
developed)
70%
vs. 65%
42%
vs. 33%
Odds ratio1
1.26
1.48
Delayed improvement: poor early course x late recovery:
Developing
Developed
22%
23%
1 These are the odds of good outcome in developing centers vs. the odds of good outcome in developed centers.
Table 8. The effect of single-episode psychosis on recovery rates1
Recovery Rate
Developing centersDeveloped centers
Early remitting patients with
single-episode psychosis, % (n))
Other recovering patients (early
remitting patients with single-
episode psychosis removed
from analysis
2), % (n)
78(96/123)
49(63/129)
59(80/136)
40(118/296)
1 This is for schizophrenia spectrum (ICD-10 schizophrenia, plus schizoaffective disorder and acute schizophrenialike psychoses)
patients.
2 Recovery rates (good outcome at 15 years) for patients with single-episode psychoses are 75% in developing centers, 78% in
developed.
842
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Developed vs. DevelopingSchizophrenia Bulletin, Vol. 26, No. 4, 2000
and outcome in schizophrenia (e.g., Jablensky et al. 1994;
Davidson and McGlashan 1997; Malla et al. 1999). But if
the WHO followup studies have shown anything, it is
how difficult it will be to tease out the specific patterns
and timing of cultural influence using conventional instru-
ments of psychiatric epidemiology (Hopper, in press). At
the same time, the durability of this old puzzle suggests
that further anthropological exploration is well warranted.
Close, on-the-ground documentation (of the sort necessar-
ily forgone in large-scale epidemiological studies) of the
local contingencies of illness trajectories (clinical and oth-
erwise) is badly needed. In that regard, the staid anthropo-
logical staples long hypothesized as likely factors promot-
ing recovery—supportive kin, auspicious or alternative
beliefs, flexibly configured work, forgiving domestic
space, more socially integrated subjectivities (Warner
1985;
Hopper 1991)—are both relevant and in need of
refinement.
At least for the ISoS cohorts, cultural integrity is no
longer quite the phantom promise it once seemed. The
notion that "developing" might serve as a crude surrogate
for "traditional" has long irritated critics of the WHO
studies, and for good reason. But if one subtracts Hong
Kong from the developing group in ISoS, the remaining
members of the "developing" group are all Indian—
Madras/Chennai, and the two Chandigarh centers. While
this surely restricts the generalizability of the finding of
differential advantage, it also simplifies the cultural ques-
tion considerably. One might legitimately inquire into
salient cultural aspects of the Indian subcontinent in ways
that would be foreclosed were members of the group
spread all over the globe. Indeed, the extraordinary
engagement of Indian families in the course of treat-
ment—from the initial decision to seek help, to attending
to basic needs and medication adherence during hospital-
ization, to support afterward, including monitoring med-
ications and functioning—is surely one of the signature
features of psychiatry in that country (Nunley 1998).
Styles of family interaction over issues of illness, for
which preliminary work in Chandigarh was done as a
substudy of DOSMeD (Wig et al. 1987a, 19876), are thus
an obvious candidate for close longitudinal research. As
the measurement problems posed by "expressed emo-
tion" in those studies illustrate,10 useful constructs will
have to be embedded in everyday practice and observed
over time if their meaning as "variables" is to be inter-
pretable.
10 Methodological difficulties include the validity of assessing "emo-
tion" from verbal material only (Kleinman 1988), and scaling problems
that arise from so ratcheting up the local threshold for rating "overin-
volvement" in this most familially engaged culture that virtually no
Indian household qualifies as "high" (Nunley 1998).
The same is true of other fields of inquiry not ordi-
narily seen as part of the "cultural" portfolio. Two in par-
ticular are worth noting. First, duration of ufitreated psy-
chosis could not be reliably measured in ISoS but shows
promise (at least in the West) as a predictor of poor out-
come (Loebel et al. 1992; Davidson and McGlashan
1997).
But since part of the influence of duration of
untreated psychosis may be due to its degrading effect on
functioning in the pranorbid phase, this too may be sub-
ject to cultural coloring. If onset is slow and uneven, a
setting flexible enough to accommodate fluctuating levels
of capacity, motivation, and attentiveness might continue
to sustain, even cultivate, the not-yet-impaired aspects of
self longer than those that practice early and enduring
assignment to the sick role. Were this the case, the clinical
price paid for late formal treatment would be muted.
Second, anthropologists are increasingly prone to empha-
size the particularities of the "local" and the intercon-
nected nature of the "contextual" in studies of culture.
Sustained documentary efforts of the sort undertaken in
traditional ethnography commonly reveal beliefs and
practices to be highly variable, often internally inconsis-
tent, situation dependent, and riven by class, gender, eth-
nic,
religious, and other structural divisions (Ortner 1995).
. This suggests that culture is pointedly not the sort of thing
to be assessed by structured questionnaires inquiring
bluntly, say, into habits of illness attribution. Not that
indirect evidence of the importance of local area is lack-
ing: despite stubborn measurement hurdles, "neighbor-
hood effects" seem well documented for a variety of indi-
vidual outcomes in the United States (Ellen and Turner
1997),
although little attention has been devoted to the
status of resident disabled members. Even if we assume
its relevance, the workings of solidarity in ethnically
diverse, politically charged neighborhoods are bound to
be complex, changeable, and difficult to trace (e.g.,
Sanjek 1999). Still, it seems prudent to seek out distinc-
tive ways in which the workings of solidarity shape the
"life worlds" of persons with schizophrenia (Corin 1988).
The ISoS findings offer other lessons as well.
Culture and the Dilemma of Context. An instructive
suggestion may be drawn from the interaction effect of
NARP and center seen here, one further highlighted by
the two-stage model of effect proposed. That is, while it
may be folly to attempt to deconstruct culture into a bun-
dle of discrete, measurable variables, it may make sense
nonetheless to think that distinctive types of influence
may be at work, that these may well be keyed to phases or
contingencies of recovery, and that the relevant cultural
factors (or better, complexes) may well differ from place
to place. With respect to the early phase, for example, in
addition to prognostic differences of subtypes of the dis-
843
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Schizophrenia
Bulletin,
Vol.
26, No. 4, 2000K. Hopper and J. Wanderling
order, host resiliency factors, and competent clinical
attention, there may be something about the overall
response to breakdown (its reception and the cultivation
of residual capacity) that increases the likelihood of it
being short-lived. Over the longer term, should the disor-
der prove persisting, other cultural factors—say, at the
microlevels of household, neighborhood, work, and
everyday social interaction—may be engaged that, over
time,
more effectively coax "a functional sense of self
(Davidson and Strauss 1992, p. 131) out of the disloca-
tions and suffering occasioned by the disorder. In short,
we may well end up investigating not only native "idioms
of distress" (Nichter 1981), but also—to some extent at
least—native idioms of recovery and the enabling prac-
tices that make them vital and effective.
Rehabilitation. The "work of recovery" can be uneven
and prolonged. Clinical appreciation for the reconstitutive
function of prolonged troughs in the recovery trajectory
(e.g., Strauss et al.'s [1985] notion of "woodshedding")
has been hard won but, along with anthropological obser-
vations on the value of "positive withdrawal" and negoti-
ated terms of social interaction (Corin 1988), has usefully
enhanced our understanding of the many fronts and vary-
ing paces at which the work of recovery proceeds
(Davidson and Strauss 1995). This suggests a more
nuanced, developmentally flexible model of rehabilita-
tion—one that, again, directs our attention back to a
closer analysis of the microcontexts of support in the lives
of those patients with poor early illness course who go on
to recover. (Whether formal contexts of care could be
designed to mimic such homespun dynamics is another
question.)
Narrative Evidence. A similar observation might be
made with respect to those intriguing accounts of how
mental illness unfolds, and interacts with aspects of a non-
ill
self,
over time in ordinary lives (e.g., Strauss 1989;
Estroff et al. 1991). Such accounts, which bring the "I" as
agent back into prominence on the cultural stage, are
especially revealing with respect to what people do to
help themselves restore the sense of efficacy eclipsed by
recurring psychosis (Strauss 1994). Given their prove-
nance, the ISoS illness narratives (now in construction)
are likely to be heavily clinically inflected. Even so, our
examination of initial drafts yielded tantalizing glimpses
of circumstances (e.g., direct and lasting responsibility for
child care; late-breaking moves for independence from
parents) that appear to boost resolve to weather the strains
of episodic disorder but are not typically found in the
usual roster of recovery-related factors.
No matter the fresh complexities to be pursued, an
unmistakable message of optimism may be read across
the board in the ISoS findings. For a substantial portion of
subjects followed, the guarded hope first voiced by the
17th-century Cervantes—that Sancho Panza's hero
remained, perplexingly, "mad in patches, full of lucid
intervals" (as cited in Kerr and Snaith 1986)—seems war-
ranted. The trick may lie in how the intervals out from
under unreason's rule might be lengthened.
References
American Psychiatric Association. DSM-III-R:
Diagnostic and Statistical Manual of Mental Disorders.
3rd ed. Washington, DC: APA, 1980.
Barrett, R. The Psychiatric Team and the Social Definition
of Schizophrenia. New York, NY: Cambridge, 1996.
Baruch, G., and Teacher, A. Psychiatry Observed. Boston,
MA: Routledge and Kegan Paul, 1978.
Bleuler, M. The Schizophrenic Disorders. New Haven,
CT:
Yale University Press, 1978.
Boyle, M. Schizophrenia: A Scientific Delusion? New
York, NY: Routledge, 1990.
Cooper, J., and Sartorius, N. Cultural and temporal varia-
tions in schizophrenia: A speculation on the importance of
industrialization. British Journal of Psychiatry,
130:50-55,
1977.
Corin, E.E. Facts and meaning in psychiatry: An anthro-
pological approach to the life-world of schizophrenics.
Culture, Medicine and Psychiatry 14:153-188, 1988.
Craig, T; Siegel, C; Hopper, K.; Lin, S.; and Sartorius,
N.
Outcome in schizophrenia and related disorders com-
pared between developed and developing countries: A
recursive partitioning reanalysis of the WHO DOSMD
fate. British Journal of Psychiatry, 170:229-233, 1997.
Craig, T.J.; Wanderling, J.; and Janca, A. Long term diag-
nostic stability in international cohorts of persons with
schizophrenia and related psychoses. In: Hopper, K.;
Harrison, G.; Janca, A.; and Sartorius, N., eds. Prospects
for Recovery From Schizophrenia—An International
Investigation: Report From the WHO-Collaborative
Project, The International Study of Schizophrenia.
Westport, CT: Psychosocial Press, in press.
Davidson, L., and McGlashan, T.H. The varied outcomes
of schizophrenia. Canadian Journal of Psychiatry,
42:34-43,
1997.
Davidson, L., and Strauss, J.S. Sense of self in recovery
from severe mental illness. British Journal of Medical
Psychology, 65:131-145,1992.
Davidson, L., and Strauss, J.S. Beyond the biopsychoso-
cial model: Integrating disorder, health, and recovery.
Psychiatry, 58:44-55, 1995.
844
by guest on July 6, 2011schizophreniabulletin.oxfordjournals.orgDownloaded from
Developed vs. DevelopingSchizophrenia Bulletin, Vol. 26, No. 4, 2000
Desjarlais, R.; Eisenberg, L.; Good, B.; and Kleinman, A.
World
Mental Health. New York, NY: Oxford, 1995.
Drake, C; Levine, R.; and Laska, E. Identifying prognos-
tic factors that predict recovery in the presence of loss to
follow-up. In: Hopper, K.; Harrison, G.; Janca, A.; and
Sartorius, N., eds. Prospects for Recovery From
Schizophrenia—An International Investigation: Report
From the WHO-Collaborative Project, The International
Study of Schizophrenia. Westport, CT: Psychosocial Press,
in press.
Edgerton, R., and Cohen, A. Culture and schizophrenia:
The DOSMD challenge. British Journal of Psychiatry,
164:222-231,
1994.
Ellen, I.G., and Turner, M.A. Does neighborhood matter?
Assessing recent evidence. Housing Policy Debate,
8:833-866, 1997.
Estroff,
S.E.; Lachicotte, W.S.; Illingworth, L.C.; and
Johnston, A. Everybody's got a little mental illness:
Accounts of illness and self among persons with severe,
persistent mental illness. Medical Anthropology
Quarterly, (new series) 5:331-369,1991.
Good, B. Medicine, Rationality and Experience. New
York, NY: Cambridge, 1994.
Gureje, O. Schizophrenia. In: Tantam, D.; Appleby, L.;
and Duncan, A., eds. Psychiatry in the Developing
World.
London, U.K.: Gaskell, 1996. pp.
111-131.
Harding, CM. Course types in schizophrenia: An analysis
of European and American studies. Schizophrenia
Bulletin, 14(4):633-643, 1988.
Harding, CM.; Zubin, J.; and Strauss, J.S. Chronicity in
schizophrenia: revisited. British Journal of Psychiatry,
Supplement. 18:27-37,1992.
Harrison, G.; Mason, P.; Glazebrook, C; Medley, I.;
Croudace, T.; and Docherty, S. Residence of incident
cohort of psychotic patients after 13 years of follow up.
British Medical Journal, 308:813-816, 1994.
Hopper, K. Some old questions for the new cross-cultural
psychiatry. Medical Anthropology Quarterly, (new series)
5:299