Women, poverty and common mental disorders in four

restructuring societies

Vikram Patela,b,*, Ricardo Arayac, Mauricio de Limad, Ana Ludermire,Charles Toddf

aSangath Centre, 841/1 Alto Porvorim, Goa 403521, IndiabInstitute of Psychiatry, London, UK

cUniversity of Santiago, Santiago, ChiledUniversidade Federal de Pelotas, Pelotas, RS-BrazileSecretaria Estadual de Saude de Pernambuco, Brazil

fUniversity of Zimbabwe Medical School, Harare, Zimbabwe

Abstract

Background: Poverty and female gender have been found to be associated with depression and anxiety in

developed countries. The rationale behind this paper was to bring together ®ve epidemiological data sets from four

low to middle income countries to examine whether key economic and development indicators such as income and

poor education, and female gender, were associated with common mental disorders.

Method: The paper is based on ®ve datasets: three based on primary care attenders in Goa, India; Harare,

Zimbabwe and Santiago, Chile; and two based on community samples in Pelotas, Brazil and Olinda, Brazil. All ®ve

studies estimated prevalence of common mental disorders along with variables to measure economic deprivation and

education.

Findings: In all ®ve studies, female gender, low education and poverty were strongly associated with common

mental disorders. When income was divided into tertiles, with the lowest tertile as a reference value, there was a

signi®cant trend for reduced morbidity for the lower two tertiles.

Discussion: These ®ndings have considerable implications since the rapid economic changes in all four societies

have been associated with rising income disparity and economic inequality. Examples of population based

prevention strategies based on increasing the proportion of those who complete schooling and on high-risk

strategies such as providing loan facilities to the impoverished are potential outcomes of these ®ndings.

Development agencies who focus on women as a priority group have failed to recognize their unique vulnerability

to common mental disorders and need to reorient their priorities accordingly. # 1999 Published by Elsevier Science

Ltd. All rights reserved.

Keywords: Mental health; Women; Poverty; Cross-cultural; India; Zimbabwe; Chile; Brazil

Introduction

Common mental disorders (CMD) was a term

coined by Goldberg and Huxley (1992, pp. 7±8) to

Social Science & Medicine 49 (1999) 1461±1471

0277-9536/99/$ - see front matter # 1999 Published by Elsevier Science Ltd. All rights reserved.

PII: S0277-9536(99 )00208-7

www.elsevier.com/locate/socscimed

* Corresponding author. Fax: +91-832-415-244.

E-mail address: vpatel@vsnl.com (V. Patel)

describe ``disorders which are commonly encountered

in community settings, and whose occurrence signals a

breakdown in normal functioning''. CMD, also

referred to as non-psychotic mental disorders or neuro-

tic disorders, manifest with a mixture of somatic,

anxiety and depressive symptoms. CMD are the third

most frequent causes of morbidity in adults (prevalence

rates) worldwide (World Health Organization, 1995).

They are an important cause of disability and pose a

signi®cant public health problem (Ormel et al., 1994).

The recent WHO report ``Investing in Health Research

and Development'' predicts that depression will be the

single most important cause of disability by the year

2020 in the developing world (World Health

Organization, 1995). The warning of a mounting crisis

of unmet needs for the countless millions with such

disorders have been building up over the past 20 years.

Evidence of a high prevalence of CMD has been gener-

ated from a range of settings in low and middle

income countries such as rural Lesotho, primary health

clinics in Santiago and the urban general practices of

India (Shamasundar et al., 1986; Holli®eld et al., 1990;

Araya et al., 1994). These studies reveal prevalence

®gures of CMD exceeding 30% in community samples

and approaching 50% in primary care samples.

The WHO Multinational study of the prevalence,

nature and determinants of CMD in general medical

care settings was conducted in 14 countries (Ustun et

al., 1995b). The startling ®nding of this study was that,

despite the use of standardized methods in all centres,

there were enormous variations in most variables.

Indeed, the only similarities across centres were the

general observations of the ubiquity of CMD, the

comorbidity of anxiety and depression, and the associ-

ation of CMD and disability even after adjustment for

physical disease severity. On the other hand, speci®c

variables showed substantial variations; thus the preva-

lence rates of CMD ranged from 7 to 52% of primary

care attenders; physician recognition of CMD varied

from 5% to nearly 60% and the association of key

variables such as gender, physical ill-health and edu-

cation with CMD were in opposite directions in di�er-

ent centres. These ®ndings demonstrate the need for

locally relevant studies with locally validated method-

ologies whose aim is to identify local needs and inform

local health services (Patel and Winston, 1994).

Female gender, social, economic and interpersonal

factors remain the most consistently demonstrated risk

factors for CMD in industrialised societies. There is

growing evidence of an association between socioeco-

nomic deprivation as represented by unemployment

(Warr, 1987; Bartley, 1994; Gunnell et al., 1995), low

income (Eaton and Ritter, 1988; Power et al., 1991)

and lower social class (Brown and Harris, 1978;

Meltzer et al., 1995), with suicide rates (Platt and

Kreitman, 1990) and psychological disorder. There is

consistent evidence of an association between economic

deprivation as measured by social class, income and

employment, in developed countries and CMD (e.g.

Bartley, 1994). A recent household survey from the

United Kingdom demonstrated a strong association

between CMD and low household income and not sav-

ing from income (Weich et al., 1997). Similarly, many

studies from these settings have demonstrated a greater

risk for women to su�er CMD (Jenkins, 1985).

Most epidemiological studies of CMD in low-income

countries have concentrated on prevalence rate esti-

mations, rarely examining the role of risk factors. The

aim of this paper is to bring together data collected by

the authors in ®ve separate studies conducted in four

low and middle income countries in di�erent stages of

economic development to examine two hypotheses:

®rst, that female gender is associated with CMD and,

second, that poverty is associated with CMD. The

rationale for these hypotheses was that these risk fac-

tors had been demonstrated in some studies from in-

dustrialised societies. If similar associations were

demonstrated in low and middle income countries, the

implications would be of great importance since they

would re¯ect not only the universal nature of these

risk factors, but also have a bearing on the impact of

the dramatic economic changes in these countries on

the increased morbidity of CMD. These studies were

conducted by the authors in their own countries, using

methodologies that were sensitive and valid for the

local setting. Thus, this paper is not based on a multi-

Table 1

Methodology of studiesa

Setting Olinda Pelotas Harare Goa Santiago

Investigator AL ML VP,CT VP RA

Population community community PHC PHC PHC

Design survey survey case-control survey survey

Sample 621 1277 199 cases,197 controls 303 4200

Sampling random multistage random systematic random systematic

Case criteria SRQ SRQ SSQ CISR GHQ/CISR

a GHQ=General Health Questionnaire (Goldberg, 1978); SRQ=Self Reporting Questionnaire (Harding et al., 1980);

SSQ=Shona Symptom Questionnaire (Patel et al., 1997a); CISR=Revised Clinical Interview Schedule (Lewis et al., 1992).

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±14711462

national study in the traditional sense of having its

emphasis on uniformity. Instead, this is a model of

what the authors propose as locally sensitive research

whose ®ndings can be collated to examine themes aris-

ing out of diverse cultural settings.

Method

The overall methodology of the 5 studies is summar-

ised in Table 1. Brief information on key aspects of

the methodology are described below. Details for each

study can be obtained from the original publication

describing the studies as referenced.

Harare, Zimbabwe (Patel et al., 1997b)

Harare is the administrative and ®nancial capital of

Zimbabwe. The study was set in primary care in two

high-density suburbs of the city. The high-density sub-

urbs are the poorer, more crowded areas of Harare;

they were, during the colonial era, assigned for the

African population of the city. The study design was a

case-control investigation of the risk factors of CMD.

The primary care sites were 2 Primary Health Clinics

(PHC), 4 GP surgeries and 14 Traditional Medical

Practitioners (TMP) in these suburbs. Recruitment was

by systematic sampling at GP and PHC while all eli-

gible TMP attenders were recruited. The measure of

CMD was the Shona Symptom Questionnaire, an indi-

genously developed instrument for the measurement

and detection of CMD in primary care attenders with

an established validity and reliability (Patel et al.,

1997a). 199 cases and 197 non-cases of CMD were

recruited from the 3 care provider sites.

Goa, India (Patel et al., 1998)

Goa is the smallest state of India. 40% of its popu-

lation is urbanized. Its main economy is based on

mining, agriculture, ®shery and tourism. Patients were

recruited from 2 Primary Health Clinics, one in a rural

and one in an periurban district of Goa. The study de-

sign was a cross-sectional survey. PHC attenders were

recruited by random sampling of consecutive attenders

(n= 302). The measure of CMD was the Konkani ver-

sion of the Revised Clinical Interview Schedule (CISR)

(Lewis et al., 1992), a structured interview for the

measurement and diagnoses of CMD in primary care

and community settings.

Olinda, Brazil (Ludermir, 1998)

Olinda is part of the Recife Metropolitan Area in

the state of Pernambuco in north-east Brazil. Leisure,

tourism, commercial sales and services represent the

main economic activities. More than a third of the

population live in shanty towns. Subjects were

recruited from a community survey in Olinda, which is

a high-density urban population area. A sample of 621

adults were recruited by random sampling of all enum-

erated households. The measure of CMD was the

Portuguese version of the 20 item Self-Reporting

Questionnaire (SRQ) which is one of the most widely

used screening measures for identi®cation of CMD in

developing countries (Harding et al., 1980).

Pelotas, Brazil (Lima et al., 1996)

Pelotas is a city located in the extreme south of

Brazil, near the Uruguayan border. It is the commer-

cial and services centre of an active agricultural, sub-

tropical area. The local economy is based on

agriculture (mainly rice, soybeans and cattle) and on

the food industry, this being one of the most developed

areas of Brazil. Although most of the urban popu-

lation is middle class, there are a few shanty towns

located near to the central urban area. The study set-

ting was 30 randomly selected census tracts (out of 259

tracts comprising approximately 300 households each)

in the central urban area. Using multi-stage sampling,

a total of 1277 adults from 600 households were

recruited. The measure of CMD was, as with the

Olinda study, the Portuguese version of the SRQ.

Santiago, Chile (Araya, 1994; Araya et al., 1994)

Santiago is the administrative and commercial capi-

tal of Chile. Chile has been undergoing dramatic socio-

economic changes in the past 5 years as a result of a

sustained economic growth. The study setting was

Primary Health Clinics in high-density suburbs (i.e.

poorer areas) of Santiago. The study design was a

cross-sectional survey of adult attenders in 24 PHC. A

sample of 4200 subjects was recruited by systematic

sampling. The measures of CMD were the Spanish ver-

sions of the 12 item General Health Questionnaire

(Goldberg, 1978) and the Revised Clinical Interview

Schedule in a two-stage case identi®cation method.

Indicators of poverty

Emphasis was placed on using indicators that had

been found to be sensitive and valid for the local set-

ting. In four studies, income was considered the most

reliable indicator of economic status. However, the

type of income measured varied. Thus, in the Brazilian

and Chilean studies, household income data were col-

lected because investigators were con®dent of eliciting

accurate data from the index respondent and, in the

case of the Brazilian studies, the community setting of

the studies. Thus, in both studies, interviews took

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±1471 1463

place at home. In Harare, household data was con-

sidered unreliable because of the variable de®nitions of

a household in the high-density suburbs. Thus, house-

holds could consist of unrelated persons living in the

same home (who would not share information on their

income), persons who had informal occupations and

undisclosed incomes, partners/spouses who had

families and children in other homes as well and so on.

Thus, personal income of the index patient was used

as the indicator of poverty. Other indicators of poverty

in Harare were indebtedness, hunger in the previous

month due to lack of money to buy food, and having

cash savings. In Goa, however, income, was considered

unreliable since the majority of PHC attenders were

women who were often not directly involved in income

generation; thus, their responses to a question on

income may not re¯ect the real economic situation of

the household. Many households involved more than

one earning member and accurate estimates of house-

hold income were not possible on the basis of inter-

viewing the index patient. Finally, much income in

India is `undeclared' and there is discomfort in sharing

sensitive information on income with researchers.

Instead, economic indicators were discussed amongst

the research team and two indicators were chosen:

whether the subject was in debt and whether the sub-

ject had been unable to buy food due to lack of money

in the previous month. The same two `proxy' indi-

cators were also used in the Harare study.

Analysis

First, each investigator examined the associations of

gender, age and educational status with CMD. Next,

each investigator identi®ed indicators of economic

deprivation from their study data and analysed their

association with CMD. Income data were grouped

into tertiles and the odds ratios for CMD were com-

puted using the highest tertile as the reference value.

Statistical strength of associations was examined using

odds ratios (adjusted for age and sex) and Chi-square

test for trends. Data collection was not standardised

since the studies were conducted at di�erent times;

thus, some data such as years of education were con-

tinuous in some studies and categorical in others. T-

tests were used for comparisons of continuous data.

All signi®cance tests are two-tailed.

Results

Sociodemographic data and prevalence of CMD

These are summarised in Table 2.

Association of socio-demographic variables with CMD

Tables 3 and 4 show that there is a consistent associ-

ation between CMD and female gender, older age and

lower education. While there is a trend for those who

were previously married (i.e. widowed, separated/

divorced) and those who were unemployed, this was

not consistent after adjustment for age and sex.

Association of economic indicators with CMD

For the four studies where income data was elicited,

the samples were categorized on the basis of income

tertiles (Tables 5 and 6). Using the lowest income ter-

tile as a reference value, there were signi®cant trends

for lower morbidity as income levels were higher. In

the two studies in which proxy variables were used to

measure poverty, strong associations were found

between morbidity and economic problems.

Discussion

The aim of this paper was to collate data from 5

data-sets elicited in 4 countries which are currently in

the midst of radical economic reforms, to examine the

associations between female gender and economic indi-

cators with common mental disorders. In contrast to

the stress laid on uniformity in multi-national study

designs, all 5 studies collated in this paper were con-

ducted independently. All the authors of the studies

described in this paper were resident in the areas of

their study and this paper is a post-hoc analysis of

Table 2

Sociodemographic characteristics and prevalence of CMD of study populationsa

Setting Goa Olinda Pelotas Santiago

Female % 69 57 55.5 69.5

Age (mean years) 44.6 (S.D. 14.3) 36.1 (S.D. 16.0) 41.4 (S.D. 18) 37.5 (S.D. 14.5)

Education: % not completed secondary school 78 56 58 58

Prevalence of CMD (%) 46 35 23 52

a Harare data is not applicable for this table due to the case-control design.

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±14711464

their datasets. Thus, there is no standardisation of the

criteria of psychiatric caseness or for the choice of

measure of economic deprivation. While this may be

argued to be a limitation of this paper, the authors are

of the view that in this diversity of methodologies lies

the strength and unique nature of the collation of data

presented in this paper. Thus, each study employed

variables, instruments and case criteria which were sen-

sitive and valid for the local culture; yet, the data

yielded, though not identical, were comparable and, as

this paper shows, can be used to derive themes and as-

sociations across the study settings. Thus, this paper is

evidence that in order to achieve cross-cultural com-

parability in multinational studies of mental illness, it

is not necessary to use the same instruments or

measures in all settings. What is crucial is that the

method used is valid for the local setting.

Another potential limitation is the lack of data on

individual sub-categories of CMD such as depression,

anxiety and so on. Although a categorical classi®cation

of CMD has been devised for use in primary health

care (Ustun et al., 1995a), there is substantial evidence

that in community and primary care settings, a dimen-

sional approach to morbidity is more valid. This dis-

crepancy is accounted for by the fact that

classi®cations have tended to re¯ect the results of psy-

chiatric assessments at tertiary care level. In most pri-

mary care patients, symptoms of anxiety and

depression co-exist to such an extent that their categor-

isation in either group is di�cult. For example, the

WHO multinational study of CMD found that for all

speci®c psychiatric disorders (excluding alcohol depen-

dence), comorbidity rates (with other psychiatric dis-

orders) exceeded 50% (Ustun et al., 1995b). This

suggests that one of the basic criteria of a successful

classi®cation, i.e. the mutual exclusiveness of di�erent

categories, was not achieved. Indeed, Goldberg and

Huxley (1992) state that ``it is becoming clear that the

idea that CMD should be thought of as discrete dis-

ease entities with distinct causes, course and treatment

is probably untenable''. This problem of validity is

arguably even greater in cultures di�erent from those

which dominate the ICD10 classi®cation; for example,

there is evidence that sub-categories of CMD lack con-

ceptual validity in many African settings and that even

their categorisation as a psychiatric disorder may be

partly to blame for the low recognition rates in pri-

mary care (Patel, 1996).

The data presented in this paper provide compelling

evidence of an association between CMD and female

gender, older age, low education and economic depri-

vation. While it may be possible that these associations

were not representative of the community in the stu-

dies in primary health care settings, the associations

were also demonstrated in both of the community

based studies.Table3

Associationofgender,ageandeducationwithcommonmentaldisorders:categoricalvariables(oddsratiosareadjustedforageandsex;95%

con®denceintervals)

Harare

Goa

Olinda

Pelotas

Santiago

Femalegender

1.5(1,2.3)

3.7(2.1,6.5)

3,1(2.2,4.4)

1.6(1.2,2.1)

1.9(1.7,2.2)

Widowedorseparated

1.4(0.8,2.3)

0.9(0.5,1.7)

1.4(0.8,2.5)

1.2(0.8,1.8)

1.2(1.05,1.4)

Notinformalemployment

1.4(0.8,2.3)

1.3(0.7,2,2)

1.5(1,2.2)

notavailable

1.4(1.2,1.7)

Categoricaleducation

continuous(seebelow)

notpassedOlevels2.4(1.2,4.7)

<

8years2.9(2,4.3)

continuous(seebelow)

<

8years1.4(1.7,1.2)

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±1471 1465

The association of female gender with CMD

The ®ndings of the studies in this paper corroborate

the ®ndings of several earlier community-based and

studies of treatment seekers which demonstrate that

women are disproportionately a�ected by mental

health problems (e.g. Orley and Wing, 1979; Desjarlais

et al., 1995; Pearson, 1995). Women's mental health

cannot be considered in isolation from social, political

and economic issues. When women's position in so-

ciety is examined, it is clear that there are su�cient

causes in current social arrangements to account for

the surfeit of depression and anxiety experienced by

women. The multiple roles played by women such as

child-bearing and child-rearing, running the family

home, caring for sick relatives and, in an increasing

proportion of families, earning income are likely to

lead to considerable stress. Di�culties for women are

encountered in a number of di�erent areas such as

their social position, aspirations and domestic pro-

blems. The reproductive roles of women, such as their

expected role of bearing children, the consequences of

infertility and the failure to produce a male child and

postnatal depression, are examples of mechanisms

which make women vulnerable to CMD.

Cox studied women attending an antenatal clinic at

a health centre in Uganda and found that 30% had

psychiatric morbidity in the antenatal period, and

among those followed up into puerperium, 10% suf-

fered from postnatal depression (Cox, 1979, 1983). A

community study from Zimbabwe showed that 18% of

mothers in the eighth month of pregnancy had a sig-

ni®cant emotional disorder; 16% had postnatal de-

pression (Nhiwatiwa et al., 1998). A study of mothers

living in a squatter settlement in Brazil reported that

36% had emotional problems (Reichenheim and

Harpham, 1991).

Violence against women is emerging as a pervasive

global issue and contributes signi®cantly to preventable

morbidity and mortality for women across diverse cul-

tures. Violence has serious psychological and emotion-

al consequences including depression, anxiety, Post

Traumatic Stress Disorder (PTSD), dissociation dis-

orders, somatization, sexual dysfunction and self-harm

behaviour (Fischbach and Herbert, 1997).

Table 4

Association of gender, age and education with common mental disorders: continuous variables (the ®gures quoted are for CMD

subjects versus those without CMD)

Age (mean, S.D.; t-test, p ) Years of education (mean, S.D.; t-test, p )

Harare 34.5 (12.8) versus 31.8 (11.5), t= 2.2, p= 0.02 7.8 (3.5) versus 9 (3.3), t= 3.4, p< 0.001

Goa 46.8 (12.3) versus 42.6 (15.5), t= 2.5; p= 0.01 categorical (see above)

Olinda 39.5 (16.4) versus 34.2 (15.5), t= 4; <0.0001 categorical (see above)

Pelotas 44.1 (18.8) versus 40.4 (17.6), t= 2.9 ; p= 0.004 5.1 (3.8) versus 7.3 (4.5), t= 8.4; <0.0001

Santiago 38.7 (14.1) versus 36.2 (14.8), t= 5.4; <0.0001 categorical (see above)

Table 5

Association of indicators of poverty with common mental disorders: association between income (categorized in tertiles) and

CMDa

Harare

(income: personal)

Olinda

(income: household per capita)

Pelotas

(income: household per capita)

Santiago

(income: household total)

Tertile1 1 1 1 1

Tertile2 0.59 (0.3, 1.1) 0.66 (0.4, 1.1) 0.75 (0.5, 1) 0.7 (0.6, 0.9)

Tertile3 0.46 (0.2, 0.9) 0.47 (0.3, 0.8) 0.54 (0.4, 0.8) 0.5 (0.4, 0.7)

Chi square test for trend 5.6, p= 0.05 11.4, p= 0.003 15.8, p< 0.001 60.2, p< 0.001

a All odds ratios adjusted for age and sex with 95% con®dence intervals. Tertile 1 is the lowest income tertile and is the reference

value.

Table 6

Association of indicators of poverty with common mental dis-

orders: associations of proxy indicators of poverty with

CMDa

Goa Harare

Debt 2.8 (1.7, 4.6) 1.1 (0.7, 1.7)

Hunger 3.2 (1.9, 5.2) 2.1 (1.3, 3.2)

a All odds ratios adjusted for age and sex with 95% con®-

dence intervals.

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±14711466

Physical health problems and women's roles as

carers can also place considerable burdens on their

mental health. For example, in Harare, women had to

cope with a high level of death and morbidity related

to HIV disease. A recent survey showed that 30% of

pregnant women attending antenatal PHC in Harare

were HIV-positive (Mbizvo et al., 1996); thus, women

have to cope not only with illness in their male part-

ners and children but with their own failing health as

well. It is also possible that biological factors may play

some role in explaining the female risk for CMD.

Gender dynamics and power relations which lead to

an unequal status for women in a variety of situations

are likely to make their lives more stressful. Indeed, ``it

is not surprising that the health of so many women is

compromised from time to time. Rather, what is more

surprising is that stress related health problems do not

a�ect more women'' (WHO, 1993). In addition to

being more likely to su�er CMD, the precarious status

of women in their husband's family comes to the fore

when they su�er mental illness (Shiva, 1992; Raghavan

et al., 1995). Because of the di�erent expectations and

evaluations of men's and women's behaviour, mental

illness in women attracts a greater amount of shame

and dishonour and has a greater impact on family life

due to the womans role in running the domestic activi-

ties of the household (Skultans, 1991; Malik, 1993).

In view of the enormous social, physical and econ-

omic stresses facing women, the association of female

gender with CMD is, arguably, not surprising. Its im-

plications for health research and development activity

are, however, immense. The bulk of health research

and development in low-income countries currently

focuses on `maternal and child health' or the health of

young women. Paradoxically, the approach taken by

funding and development agencies is to focus almost

entirely on reproductive health, as if this area of health

was somehow discrete and separate from psychological

health. Indeed, the remarkable paucity of data on

postnatal depression and the relationship between

reproductive symptoms and infertility with psychoso-

cial health is a marker of the myopic view of health

research in low and middle income countries. Thus, it

is essential to conduct research on women's health that

encompasses the holistic nature of health, viz., incor-

porating psychological, reproductive and social view-

points. Examples of potential research themes are: the

ethnographic description of the experience of CMD by

women and their interpretation of the symptoms and

causes; examination of the social and cognitive matrix

which underlies the strong association between female

gender and CMD; the association of CMD with vio-

lence against women, childbearing, infertility, child-

birth and reproductive symptoms; the impact of CMD

on the daily lives of women and the nature of disabil-

ity that it imposes; the management of CMD using

locally available personnel and resources and the devel-

opment of locally sensitive intervention strategies.

The association of age, low income and education with

CMD

The ®ve studies show clear associations between low

education, low income and older age with CMD.

These studies are consistent with ®ndings in other low

and middle income countries (e.g. Bahar et al., 1992;

Desjarlais et al., 1995; Kishore et al., 1996; Mumford

et al., 1997). In public health terms, the most powerful

implication of these ®ndings is that low education is a

potentially preventable risk factor. It is perhaps im-

portant to recognize that the key factor may not be

whether 100% of children are in primary school, but

rather the proportion of children who fail to complete

the minimum years needed to obtain a secondary

school certi®cate (10±12 years in most countries). This

is a far more signi®cant landmark in society, for with-

out it the number of years of schooling is irrelevant to

prospective higher educational institutions or employ-

ers. Thus, even though there are impressive gains in

increasing school enrollment, there may need to be

further emphasis on reducing school drop-out rates; in

India and Zimbabwe, for example, less than half the

children who are in primary school go on to complete

their 10 years of secondary education. In Brazil, only

one in seven children completes elementary school.

One of the reasons for this high dropout rate is the

need to earn money very early in life (Iacoponi et al.,

1991). Education re¯ects socioeconomic circumstances

of the family at early life (Rutter and Madge, 1976)

and is an important cause of perpetuating inequalities

in Brazil (Urani, 1995), given its role in sorting individ-

uals into occupations. Education permits greater

choices in life decisions and in¯uences aspirations, self-

image (Brown et al., 1986) and opportunities to

acquire knowledge, which may motivate attitudes and

behaviour toward lifestyle and health status (BMA,

1987).

Older age is consistently associated with CMD; this

has implications for all 4 societies studied where aver-

age life spans are gradually increasing even as birth

rates are falling, thus increasing the proportion of the

elderly. It is essential for health services to respond to

the needs of the older subjects, and for social services

to recognise that as rural societies change, the fabric of

support for the elderly is also changing. Increasing

numbers of elderly persons are living out their twilight

years in loneliness as their family structures change. In

a similar way, elderly people in many low and middle

income countries do not have enough ®nancial support

from the government and su�er considerable economic

di�culties. Research on the needs of older women is

even rarer in many countries, as if the health of

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±1471 1467

women is less relevant once their reproductive years

are completed.

Low income groups are more vulnerable to su�er

CMD, irrespective of the overall state of development

of the society they live in. Thus, it would appear that

it is relative poverty, i.e. low income, which is a risk

factor for CMD. This ®nding is consistent with those

from developed countries (Weich and Lewis, 1998).

This ®nding would predict that even if developing

countries reduce the proportion of those who live in

absolute poverty, if income inequality grows, the

impact on reducing the burden of mental illness may

not be as remarkable. If this was the case, then the po-

tential implications for the prevalence of CMD are

ominous, since across all four nations included in this

paper, there are radical economic structural changes

being implemented. Despite raising average income,

these changes are also widening socioeconomic

inequality, characterized by marked inequalities in the

distribution of income and access to education and

another basic needs. For example, in Brazil the pro-

portion of the GNP by the poorest 50% of the popu-

lation dropped from 17.4 to 12.6% from 1960 to 1989,

the equivalent ®gures for the richest 10% rose from

39.6 to 51.3% (Pereira, 1988). In Zimbabwe, a 1995

poverty assessment survey found that 45% of house-

holds were living below the food poverty line and 61%

below the total consumption line (Government of

Zimbabwe, 1996). Thus, economic inequality is being

enhanced which, in turn, may lead to higher levels of

CMD amongst low income groups.

In Harare and Goa, local communities had impli-

cated indebtedness as being a consistent source of

stress and worry. This was especially so because the

poorest who needed loans to cover short-term ®nancial

shortfalls were not considered credit-worthy by banks

and, in desperation, relied on loan-sharks. The latter

charged exorbitant interest rates and it was not

uncommon for the children of a family to spend their

lives toiling to repay the interest of relatively small

loans taken out by their parents. We tested the hy-

pothesis that being in debt was associated with CMD

and found this to be the case in both settings. It is

clear that here lies another potential preventive strat-

egy in that local banks could step in and review their

process of assessing credit-worthiness for persons who

belong to the poorest sectors of society. Radical com-

munity banks and loan facilities such as those run by

SEWA in various parts of India could be involved in

setting up such loan facilities in areas where they do

not exist. Further evidence of this relationship is

fuelled by a recent spate of suicides by farmers in the

Indian state of Andhra Pradesh where the seasonal

crop failed and farmers were deeply in debt to local

money-lenders1. A potential research theme would be

evaluating CMD and suicidal behaviour in populations

who have access to such facilities and comparing them

to populations with similar economic and social cir-

cumstances but without loan facility access.

What is unclear is the direction of the association

between poverty, low education and psychological dis-

order. As with any cross-sectional study design, the

®ndings of the ®ve studies collated in this paper cannot

be used to provide a de®nite indication of causal direc-

tion. However, as Wilkinson (1996, pp. 81±82) points

out, there are strong arguments against `reverse causal-

ity' when considering the relationship between ill-

health and income distribution. He states that ``if the

arrow of causality pointed in that direction, we would

be obliged to say that health is one of the most im-

portant determinants of income distribution. Not only

Fig. 1. A model to explain the relationship between poverty

and common mental disorders.

1 The Hindu newspaper of January 1, 1998, reported that

the Indian Government had set up a panel to investigate why

farmers in Andhra Pradesh tend to commit suicide following

the crop failure. The article mentioned that the farmers had

to take loans from local money lenders to help their families

through the ®nancial crises.

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±14711468

does this run counter to economic theory, but it ¯ies in

the face of commonsense notions of the in¯uence on

income distribution of employment and unemploy-

ment, pro®ts, tax and bene®ts'' Psychologists have

conceptualized poverty as a situation with a number of

attendant conditions which individually or collectively

in¯uence the development of the individual, rendering

him or her less capable of overcoming poverty by per-

sonal e�ort (Sinha, 1997).

Fig. 1 presents a model which provides a basis for

explaining the association and mediating factors

between poverty and psychological disorder. It is clear

that this is not a unidimensional one-way relationship

but is an interactive and complexly dynamic one.

Thus, for example, while poor nutrition may be a

stressor which triggers depression in a woman already

facing other deprivation related problems, depression

in turn robs the woman of the necessary coping skills

and energy to overcome her problems. Further, the po-

tential stresses imposed by absolute poverty may be

considerably di�erent from those of relative poverty. It

is suggested that the psychological impact of relative

poverty is the result of both the indirect (e.g. increased

exposure to behavioural risk factors due to psychoso-

cial stress) and direct (e.g. physiological e�ects of

chronic mental and emotional stress) e�ects of psycho-

social circumstances associated with social position.

The mechanism involved is one of `cognitive compari-

son' whereby people are made aware of the vast di�er-

ences in socioeconomic status that prevail, and the

knowledge of how the richer ``other half live'' a�ects

psychosocial well-being and thus, overall health status

(Wilkinson, 1997).

Conclusions

The key ®ndings of this paper are that female gen-

der, low income and other measures of poverty, older

age and low education are associated with CMD in all

®ve studies from 4 low to middle income countries.

The analyses reported in this paper were conducted

post-hoc and, in this sense, the ®ndings of this paper

may be considered as preliminary, needing replication

through prospective studies with the stated aims of

examining these associations and controlling for all

possible confounders, such as employment status,

social class, family size, etc. However, the pervasive-

ness of the key associations across the study settings,

despite the use of varying methodologies and the con-

sistency of this association with developed country stu-

dies, lead us to suggest that, even if there are

unmeasured confounders, the associations are valid.

The vulnerability of women needs to be more widely

acknowleged by those who fund and work in the area

of women's health so that psychosocial factors are

properly evaluated and included in their research and

service agendas. If the associations of income inequal-

ity are replicated, this would be of serious concern to

all four study countries where the formula for econ-

omic development is leading to a reduction in public

health expenditure, a rising inequality between the rich

and poor, increased migration to urban areas with its

attendant rise in urban squalor and rapid culture

change as the great urban centres take on an inter-

national cosmopolitan ¯avour. Research which is sensi-

tive to regional factors, such as the ®ve studies

reported in this paper, is needed to inform local health

planners and policy makers on the potential mental

health needs of the community as these economic

changes profoundly alter the fabric of their societies.

Acknowledgements

The study in Harare was funded by the

International Development Research Centre (Canada);

in Goa by the Wellcome Trust (UK); in Olinda by

FACEPE and CNPq (Brazil); in Santiago by Fondo

Nacional de Ciencia y TecnologõÂ a (FONDECYT,

Chile) and in Pelotas by CAPES (Ministry of Health,

Brazil). All authors are indebted to the numerous indi-

viduals and researchers who participated in each of the

studies and are acknowledged in detail in the main

publications referenced for each study.

References

Araya, R., 1994. Emotional Disorders in Primary Health

Care. Ministerio de Salud and Universidad de Chile.

Araya, R., Wynn, R., Leonard, R., Lewis, G., 1994.

Psychiatric morbidity in primary health care in Santiago,

Chile. Preliminary ®ndings. British Journal of Psychiatry

165, 530±532.

Bahar, E., Henderson, A.S., Mackinnon, A.J., 1992. An epi-

demiological study of mental health and socioeconomic

conditions in Sumatera, Indonesia. Acta Psychiatrica

Scandinavica 85, 257±263.

Bartley, M., 1994. Unemployment and ill health: understand-

ing the relationship. Journal of Epidemiology and

Community Health 48, 333±337.

BMA, 1987. Deprivation and Ill-Health. Board of Science

and Education, British Medical Association.

Brown, G.W., Harris, T.O., 1978. Social Origins of

Depression: a Study of Psychiatric Disorder in Women.

Tavistock, London.

Brown, G.W., Harris, T.O., Bifulco, A., 1986. Long-term

e�ects of early loss of parent. In: Rutter, M.L., Lizard,

C.E., Read, P.B. (Eds.), Depression in Young People:

Developmental and Clinical Perspectives. Guilford Press,

New York, pp. 251±296.

Cox, J.L., 1983. Postnatal depression: a comparison of

African and Scottish women. Social Psychiatry 18, 25±28.

Cox, J.L., 1979. Psychiatric morbidity and pregnancy: a con-

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±1471 1469

trolled study of 263 semirural Ugandan women. British

Journal of Psychiatry 134, 401±405.

Desjarlais, R., Eisenberg, L., Good, B., Kleinman, A., 1995.

World Mental Health: Problems and Priorities in Low-

Income Countries. Oxford University Press, New York.

Eaton, W.W., Ritter, C., 1988. Distinguishing anxiety and de-

pression with ®eld survey data. Psychological Medicine 18,

155±166.

Fischbach, R.L., Herbert, B., 1997. Domestic violence and

mental health: correlates and conundrums within and

across cultures. Social Science and Medicine 45, 1161±

1170.

Goldberg, D., 1978. Manual of the General Health

Questionnaire. NFER Publishing Company, Windsor.

Goldberg, D., Huxley, P., 1992. Common Mental Disorders:

a Biosocial Model. Tavistock/Routledge, London.

Government of Zimbabwe, 1996. Poverty Assessment Study

Survey 1995. Ministry of Public Service, Labour and

Social Welfare, Harare.

Gunnell, D.J., Peters, T.J., Kammerling, R.M., Brooks, J.,

1995. Relation between parasuicide, suicide, psychiatric

admissions, and socioeconomic deprivation. British

Medical Journal 311, 226±230.

Jenkins, R., 1985. Sex di�erences in minor psychiatric mor-

bidity. In: Psychological Medicine. Cambrige University

Press, London (Monograph Suppl. No 7).

Harding, T.W., De Arango, M.V., Baltazar, J., Climent, C.E.,

Ibrahim, H.H.A., Ladrigo-Ignacio, L., Srinivasa Murthy,

R., Wig, N.N., 1980. Mental disorders in primary health

care: a study of their frequency and diagnosis in four

developing countries. Psychological Medicine 10, 231±241.

Holli®eld, M., Katon, W., Spain, D., Pule, L., 1990. Anxiety

and Depression in a village in Lesotho, Africa: a compari-

son with the United States. British Journal of Psychiatry

156, 343±350.

Iacoponi, E., Laranjera, R.R., Mari, J.J., 1991. Brazil: a giant

wakes up to progress and inequality (ch. 13). In: Appleby,

L., Araya, R. (Eds.), Mental Health Services in the Global

Village.

Kishore, J., Reddaiah, V., Kapoor, V., Gill, J., 1996.

Characteristics of mental morbidity in a rural primary

health care centre in Haryana, India. Indian Journal of

Psychiatry 38, 137±142.

Lewis, G., Pelosi, A., Araya, R., Dunn, G., 1992. Measuring

psychiatric disorder in the community: a standardized

assessment for use by lay interviewers. Psychological

Medicine 22, 465±486.

Lima, M.S., Beria, J.U., Tomasi, E., Conceicao, A.T., Mari,

J.J., 1996. Stressful life events and minor psychiatric dis-

orders: an estimate of the population attibutable fraction

in a Brazilian community-based study. International

Journal of Psychiatry in Medicine 26, 213±224.

Ludermir, A.B., 1998. Socioeconomic status, employment, mi-

gration and common mental disorders in Olinda,

Northeast Brazil. Ph.D. thesis, London School of Hygiene

and Tropical Medicine.

Malik, S., 1993. Women and mental health. Indian Journal of

Psychiatry 35, 3±10.

Mbizvo, M., Mashu, A., Chipato, T., Makura, E., Bopoto,

R., 1996. and Fottrell Trends in HIV-1 and HIV-2 preva-

lence and risk factors in pregnant women in Harare,

Zimbabwe. Central African Journal of Medicine 42, 14±

21.

Meltzer, H., Gill, B., Petticrew, M., 1995. The prevalence of

psychiatric morbidity among adults aged 16±64 living in

private households in Great Britain, OPCS Surveys of

Psychiatric Morbidity in Great Britain. Report No 1.

HMSO, London.

Mumford, D., Saeed, K., Ahmed, I., Latif, S., Mubasshar,

M., 1997. Stress and psychiatric disorder in rural Punjab:

a community survey. British Journal of Psychiatry 170,

473±478.

Nhiwatiwa, S., Patel, V., Acuda, S., 1998. Predicting postnatal

mental disorder with a screening questionnaire: a prospec-

tive cohort study from Zimbabwe. Journal of

Epidemiology and Community Health 52, 262±266.

Orley, J., Wing, J., 1979. Psychiatric disorder in two African

villages. Archives of General Psychiatry 36, 513±520.

Ormel, J., Von Kor�, M., Ustun, T., Pini, S., Korten, A.,

Oldehinkel, T., 1994. Common mental disorders and dis-

ability across cultures. JAMA 272, 1741±1748.

Patel, V., Winston, M., 1994. The universality of mental ill-

ness revisited: assumptions, artefacts and new directions.

British Journal of Psychiatry 165, 437±440.

Patel, V., 1996. Recognizing common mental disorders in pri-

mary care in African countries: should `mental' be

dropped? Lancet 347, 742±744.

Patel, V., Simunyu, E., Gwanzura, F., Lewis, G., Mann, A.,

1997a. The Shona Symptom Questionnaire: the develop-

ment of an indigenous measure of non-psychotic mental

disorder in Harare. Acta Psychiatrica Scandinavica 95,

469±475.

Patel, V., Todd, C.H., Winston, M., Gwanzura, F., Simunyu,

E., Acuda, S.W., Mann, A., 1997b. Common mental dis-

orders in primary care in Harare, Zimbabwe: associations

and risk factors. British Journal of Psychiatry 171, 60±64.

Patel, V., Pereira, J., Coutinho, L., Fernandes, R., Fernandes,

J., Mann, A., 1998. Poverty, psychological disorder and

disability in primary care attenders in Goa, India. British

Journal of Psychiatry 172, 533±536.

Pearson, V., 1995. Goods on which one loses: women and

mental health in China. Social Science and Medicine 41,

1159±1173.

Pereira, L.T., 1988. Economia Brasileira. Editiora Brasiliense

S.A, Sao Paulo.

Platt, S., Kreitman, N., 1990. Long term trends in parasuicide

and unemployment in Edinburgh,1968±1987. Social

Psychiatry and Psychiatric Epidemiology 25, 56±61.

Power, C., Mannor, O., Fox, J., 1991. Health and class: the

early years. Chapman and Hall, London.

Raghavan, K., Srinivasa Murthy, R., Lakshminarayana, R.,

1995. Symposium on Women and Mental Health (Report

and Recommendations). NIMHANS, Bangalore.

Reichenheim, M., Harpham, T., 1991. Maternal mental health

in a squatter settlement in Rio de Janeiro. British Journal

of Psychiatry 159, 683±690.

Rutter, M., Madge, N., 1976. Cycles of Disadvantage: a

Review of Research. Heinemann, London.

Shamasundar, C., Krishna Murthy, S., Prakash, O.,

Prabhakar, N., Subbakrishna, D., 1986. Psychiatric mor-

bidity in a general practice in an Indian city. BMJ 292,

1713±1715.

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±14711470

Shiva, M., 1992. Women and Health. In: Mukhopadhyay, A.

(Ed.), State of India's Health. Voluntary Health

Association of India, New Delhi, pp. 265±302.

Sinha, D., 1997. Psychological Concomitants of Poverty and

their implications for education. In: Atal, Y. (Ed.),

Perspectives on Educating the Poor. Abhinav Publications,

New Delhi, pp. 57±118.

Skultans, V., 1991. Women and a�iction in Maharashtra: a

hydraulic model of health and illness. Culture, Medicine

and Psychiatry 15, 321±359.

Urani, A., 1995. Crescimento e gerano de emprego e renda no

Brasil, Lua Nova. Revista de Cultura e Politica 35, 5±37.

Ustun, T., Goldberg, D., Cooper, J., Simon, G., Sartorius,

N., 1995a. New classi®cation for mental disorders with

managment guidelines for use in primary care: ICD-10

PHC Chapter 5. British Journal of General Practice 45,

211±215.

Ustun, T., Sartorius, N., Costa e Silva, J., Goldberg, D.,

Lecrubier, Y., Ormel, J., Von Kor�, M., Wittchen, H.,

1995b. Conclusions. In: Ustun, T.B., Sartorius, N. (Eds.),

Mental Illness in General Health Care: an International

Study. John Wiley and Sons, Chichester, pp. 371±375.

Warr, P., 1987. Work, Unemployment and Mental Health.

Clarendon Press, Oxford.

Weich, S., Churchill, R., Lewis, G., Mann, A., 1997. Do

socio-economic risk factors predict the incidence and

maintenance of psychiatric disorder in primary care?

Psychological Medicine 27, 73±80.

Weich, S., Lewis, G., 1998. Material standard of living, social

class and the prevalence of common mental disorders.

Journal of Epidemiology and Community Health 52, 8±14.

Wilkinson, R.G., 1996. Unhealthy Societies: the A�ictions of

Inequality. Routledge, London.

Wilkinson, R., 1997. Health inequalities: relative or absolute

material standards? BMJ 314, 591±595.

World Health Organization, 1995. The World Health Report

1995: Bridging the Gaps. World Health Organization,

Geneva.

WHO, 1993. Psychosocial and Mental Health Aspects of

Women's Health. WHO/FHE/MNH/93.1. Geneva.

V. Patel et al. / Social Science & Medicine 49 (1999) 1461±1471 1471