Culture and Common Mental Disorders in Sub-Saharan Africa 9781134837342, 1134837348, 9781317840916, 1317840917, 9781315826455

The influence of culture on mental illness has been the subject of considerable academic investigation and debate in rec

354 75 3MB

English Pages 128 [145] Year 1998

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Culture and Common Mental Disorders in Sub-Saharan Africa
 9781134837342, 1134837348, 9781317840916, 1317840917, 9781315826455

Citation preview

MAUDSLEY M ONOGRAPHS 41

Culture and C om m on M ental D isorders in Sub-Saharan Africa VIKRAM PATEL

Published by Psychology Press on behalf o f T UI . M A UD SLKY

THE MAUDSLEY Maudsley Monographs

MAUDSLEV M O N O G RA PH S HENRY MAUDSLEY, from whom the series of monographs takes its name, was the founder of The Maudsley Hospital and the most prominent English psychiatrist of his generation. The Maudsley Hospital was united with the Bethlem Royal Hospital in 1948 and its medical school, renamed the Institute of Psychiatry at the same time, became a constituent part of the British Postgraduate Medical Federation. It is now associated with King's College, London, and entrusted with the duty of advancing psychiatry by teaching and research. The Bethlem-Maudsley N H S Trust, together with the Institute of Psychiatry, are jointly known as The Maudsley. The monograph series reports work carried out at The Maudsley. Some of the monographs are directly concerned with clinical problems; others, less obviously relevant, are in scientific fields that are cultivated for the furtherance of psychiatry. Editor

Professor Sir David Goldberg M A D M M Sc FRCP FRCPsych D P M Assistant Editors

Professor A S David MPhil M Sc FRCP MRCPsych M D Dr T Wykes BSc PhD MPhil Previous Editors

1955-1962 1962-1966 1966-1970 1970-1979 1979-1981

1981-1983 1983-1989

1989-1993

Professor Sir Aubrey Lewis LLD DSc M D FRCP and Professor C W Harris M A M D DSc FRS Professor Sir Aubrey Lewis LLD DSc M D FRCP Professor Sir Denis Hill M B FRCP FRCPsych D P M and Professor J T Eayrs PhD DSc Professor Sir Denis Hill M B FRCP FRCPsych D P M and Professor C S Brindley M D FRCP FRS Professor C S Brindley M D FRCP FRS and Professor C F M Russell M D FRCP FRC(ED) FRCPsych Professor G F M Russell M D FRCP FRCP(ED) FRCPsych Professor G F M Russell M D FRCP FRCP(ED) FRCPsych and Professor E Marley M A M D DSc FRCP FRCPsych DPM Professor G F M Russell M D FRCP FRCP(ED) FRCPsych and Professor B H Anderton BSc PhD

Maudsley Monographs number forty-one

Culture and Common Mental Disorders in Sub-Saharan Africa Vikram Patel Clinical Research Fellow (Lecturer), Section of Epidemiology and General Practice, Institute of Psychiatry, DeCrespigny Park, London SE5 8AF, UK (formerly Visiting Lecturer, Department of Psychiatry, University of Zimbabwe Medical School, Harare, Zimbabwe)

V p Psychology Press A

Taylor & Francis Group HOVE A N D NEW YORK

First published 1998 by Psychology Press 27 Church Road, Hove, East Sussex, B N 3 2 FA Simultaneously published in the U S A and Canada by Psychology Press 270 Madison Ave, New York, N Y 10 0 16 www. psy press.com Transferred to Digital Printing 2009

Psychology Press is part o f the. Taylor & l-'rancis Group, an informa business © 1998 by Psychology Press Ltd All rights reserved. N o part o f this book may be reprinted or reproduced or utilized in any form or by any clcctronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers.

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library IS B N 978-0-86377-530-7 IS S N 0076-5465

Typeset by Quorum Technical Services Ltd, Cheltenham

Publisher’s Note The publisher has gone to great lengths to ensure the quality o f this reprint but points out that some imperfections in the original may be apparent.

To Bharati and Gauri

This page intentionally left blank

Contents

Glossary of abbreviations and Shona terms Acknowledgements Abstract

ix

xi

xiii

1. Introduction 1 The “ new” cross-cultural psychiatry 1 Comm on mental disorders (CM D) 4 Culture and common mental disorders 5 Assessment o f mental disorders across cultures 8 Epidemiology o f mental illness in Sub-Saharan Africa 10 Explanatory models o f mental illness in Sub-Saharan Africa The study setting 17 Medical pluralism in Zimbabwe 19 Summary 22 2. The studies 23 Objectives 23 The Ethnographic Study 25 The Phenomenology Study 27 The Shona Symptom Questionnaire Study The Case -Control Study 37

32

Viii

CONTENTS

3. Results of the studies 41 C oncepts o f m ental illness o f prim ary care providers 41 Sym ptom s an d explanatory models o f C M D 48 D evelopm ent o f the Shona Symptom Q uestionnaire 51 R elationship between biomedical and indigenous models o f illness Prevalence, associations, and risk factors o f C M D 66 4. Discussion 73 Lim itations o f the research m ethodology 73 Presentation and assessment o f C M D 76 Relevance o f indigenous models o f m ental illness Epidem iology o f com m on m ental disorders 90 D irections for future research 94 5. Conclusions

101

References

105

Appendix 1 Appendix 2 Appendix 3 Author index Subject index

117 119 121 123 127

83

Glossary of abbreviations and Shona terms

ABBREVIATIONS BDQ: Brief Disability Questionnaire CISR: Revised Clinical Interview Schedule CM D: C om m on mental disorders CPN: Com m unity psychiatric nurse EM I: Explanatory model interview FGD: Focus group discussions G P: General practitioner PH C : Prim ary health care clinics SRQ: Self-reporting Questionnaire SSQ : Shona Symptom Q uestionnaire T M P: T raditional medical practitioner VHW: Village health worker

FREQUENTLY USED SHONA TERMS kufungisisa: Thinking too much mamhepo: “ Bad airs” mudzimu: Ancestral spirits muroyi: W itchcraft n’anga: T raditional healer profita: Faith healer ix

This page intentionally left blank

Acknowledgements

The work described in this m onograph was supported by tw o principal grants. The Beit M edical Fellowships provided personal su p p o rt and allowed m e to stay in Zim babw e for two years (A ugust 1993-A ugust 1995) conducting the field w ork and coordinating the project. The International D evelopm ent Research C entre (C anada) provided a generous two year grant which supported all o ther project expenses and extended it well into 1996 and facilitated the com pletion o f an outcom e study and the dissem ination o f findings to health care providers and policy m akers in Zim babw e. Two oth er bodies provided small b u t im p o rtan t contributions: the G erm an Technical C o operation (G T Z, H arare) gave a grant to enable the field w ork to begin within m onths o f my arriving in H arare; and Blair R esearch L aboratories (Zim babw e M inistry o f H ealth) funded tw o sm aller projects, one o f w hich helped to establish a sam pling fram e o f T M P in the study areas. 1 am indebted to my field research colleagues w ithout w hom none o f the d ata collection would have been possible. In particu lar Fungisai G w anzura and Essie Sim unyu becam e indispensable colleagues and friends. O thers who helped included Tecla B utau, P at M aram ba, and T arisai M usara. The task o f translation involved all these persons. In addition, Sekai N hiw atiw a, Prim rose M anyika, E dw ard G ushiri, C harles Singende, Jane M utam birw a, and G ertrude K hum alo-S akatukw a p ro ­ vided assistance. I am grateful to D r O. M bengeranw a, D irector o f the City o f H arare H ealth D epartm ent and Professor G . C havunduka, xi

Xii

ACKNOWLEDGEMENTS

Secretary-General of Zim babwe N ational T raditional Healers Association for allowing us access to their respective health care facilities. M ost o f all, I am indebted to the nurses and TM Ps who enthusiastically participated in every stage o f the project and provided some o f the insights which m ade my work in Zim babwe a fascinating voyage o f learning. I am grateful to my colleagues in the University o f Zim babwe who provided support, advice, and encouragement. In particular, M ark W inston and Charles Todd were good friends and sources o f constant intellectual stim ulation. Professor Wilson Acuda provided supervision; Jane M utam birw a and Alfred Chingono advised regarding the eth n o ­ graphic com ponents o f the research; Professor Lincoln Moses, Simba R usakaniko, and S. Siziya gave statistical advice; and Paul Linde, Laurie Schultz, Sunanda Ray, Farai M adzim bam uto, Bob Kitchin, Lawrence H ipshm an, and Val T horpe bccame sparring partners and friends. In L ondon, Glyn Lewis, M artin Prince, Paulo Menezes, and G raham D unn advised me on study design and statistics. M ost o f all, I am grateful to two people. A nthony M ann provided unstinting support to my ideas, yet consistently sm oothed my rough edges as my interest in a culturally sensitive epidemiology has evolved. Second, I owe an immeasurable am ount o f gratitude to my wife, G auri, who has travelled with me while letting her own career drift. She has been my strongest critic and my closest friend through the years and, for us, the two years in Zimbabwe are already am ongst the best times o f our lives. I wish to acknowledge the permission o f the editors of journals in which some o f the material in this m onograph was published: Psychological Medicine (for the Ethnographic and Phenomenology Studies), Acta Psychiatrica Scandinavica, Central African Journal o f Medicine, and Social Psychiatry and Psychiatric Epidemiology (material from the SSQ study) and the British Journal o f Psychiatry (the Case-Control Study).

Abstract

C om m on M ental D isorders (C M D ) are am ongst the m ost frequent disorders in prim ary care attenders. They are characterised by the clinical presentation o f som atic sym ptom s, anxiety, and depression and are associated with significant disability. In A frica, as in m any o ther lowincome regions, epidem iological research has alm ost entirely consisted o f cross-sectional surveys o f prevalence rates using m ethodologies and diagnostic models developed by E uropean and A m erican researchers. T raditional medical practitioners (TM P), w ho are am ongst the key prim ary health care providers in A frica, have rarely been included in epidem iological studies. D espite these lim itations, research has dem on­ strated a high prevalence o f psychiatric m orbidity in com m unity and prim ary care settings. In contrast, ethnographic research has usually focused on descriptive studies with small num bers o f T M Ps. Such research has revealed a range o f causal beliefs and types o f m ental illnesses in A frica. T he principal lim itation o f the ethnographic approach has been the lack o f a link with practical issues related to m ental health services. The investigation o f psychological disorders across cultures needs to blend both epidem iological and ethnographic approaches to ensure th a t research is n ot only sensitive to the local culture but is also relevant to m ental health service developm ent and is com m unicable across cultures. This m onograph begins by reviewing the key concepts and literature on cross-cultural psychiatric epidem iology, and the epidem iological and ethnographic literature on C M D in Sub-Saharan Africa. The overall xiii

xiv

ABSTRACT

objectives of the research studies described in the m onograph were to obtain a complete picture o f C M D in prim ary care attenders in H arare, Zim babwe, using the existing literature and contem porary theory as a guide for developing its methodology. The research adopted a sequential design involving four distinct field research stages. Thus, this m onograph departs from the trend o f many before it in that it focuses not on one specific research study but on a series o f four consecutive distinct studies linked with each other as the objectives evolved from ethnographic descriptions to epidemiological measurements. The aims o f the E thno­ graphic Study were to describe the concepts o f mental illness o f a diverse range of prim ary health care providers and to elicit constructs o f illness whose relationship to biomedical concepts o f C M D could be examined. N ext, in the Phenomenology Study, screening criteria elicited in the Ethnographic Study and the clinical assessment o f prim ary health care providers were used to identify prim ary care attenders with conspicuous psychiatric morbidity. This study elicited the idioms o f distress and explanatory models o f these primary care attenders. The Shona Symptom Q uestionnaire Study involved the development and evaluation o f the psychom etric properties o f a measure o f C M D derived from the idioms of distress elicited in the previous study. This measure was then used as a case finding instrum ent in the final study, the C ase-C ontrol Study, which investigated the risk factors and associations of C M D . TM Ps and prim ary health clinics were involved in all four studies while general practitioners were also involved in the C ase-C ontrol Study, reflecting the pluralistic nature o f prim ary health care in Africa. The outcom e of these studies includes a description of the concepts of mental illness held by a diverse group o f health care providers and the description o f the explanatory models o f illness experience o f persons with conspicuous psychiatric morbidity. The Shona Symptom Questionnaire, a 14-item questionnaire for the detection and measurem ent o f C M D in the Shona language, was developed and its reliability and validity dem on­ strated. The studies dem onstrated that the prevalence o f C M D was greater am ongst attenders at T M P as com pared to PH C and that two indigenous models of illness (thinking too much or kufungisisa and the model o f supernatural causation) were closely related to the biomedical construct of CM D . The relationship between different criteria for determining the presence o f a psychological disorder, for example those based on biomedical models and those based on local care provider concepts, were examined. The C ase-C ontrol Study dem onstrated that female gender, economic impoverishment, infertility, and disability were strongly asso­ ciated with CM D ; this study replicated the association between the two indigenous constructs o f illness and C M D . In the discussion, material from all four studies is brought together to explore the contribution o f culture

ABSTRACT

XV

to the presentation, assessment, classification, and risk factors for C M D in primary care in H arare. In the final analysis, this book attem pts to present an innovative and pragm atic approach to the investigation o f psycholo­ gical disorders across cultures. The relationship o f culture, political structures, and ethnicity are a confusing and complex area. When one refers to the dom inant paradigm s in psychiatry, it is obvious that the m ajor influences have originated from Europe and the USA. In attem pting to use a phrase which denotes this dom inant influence, I have had to consider a range o f words: “ Western nations” was considered inappropriate due to the complex ethnic mixture o f these societies; and “ industrialised societies” was dropped due to the rapid industrialisation affecting many hitherto less developed societies. Two terms are used in the book when referring to two concepts with some overlap: “ biomedical” is used when referring to the medical model o f psychiatry, such as current international classifications. However, it is recognised that this term does n ot exclude the fact that the movement for a more culturally sensitive epidemiology has substantial biomedical roots. “ Euro-American cultures” is used when referring to European and American systems o f thought about mental illness which form the dom inant paradigm behind current biomedical classifications. The lim ita­ tions o f this term are recognised in th at much o f cross-cultural psychiatry has been driven by researchers from Euro-A m crican cultures.

This page intentionally left blank

CHAPTER ONE

Introduction

THE "NEW " CROSS-CULTURAL PSYCHIATRY Epidemiology is o f particular im portance in psychiatry for several reasons: first, mental health resources are insufficient and epidemiological research can help establish the scale o f the need by describing the frequency of psychiatric disorders and the burden they pose on individuals, families, and health care systems; second, most psychiatric disorders are poorly understood and epidemiological research can help unravel the m ulti­ factorial aetiology o f these disorders; third, epidemiological research can help improve our understanding o f the relationship between physical and psychological illness (Tansella, de G irolam o, & Sartorius, 1992). Historically, cross-cultural studies in psychiatric epidemiology have suffered several problems. First, case identification techniques varied from site to site and m ethods were not standardised (C om pton et al., 1991). These inconsistencies led to a movement to standardise the process of psychiatric measurem ent and diagnosis. This process o f standardisation was driven by psychiatric classification systems originating in EuroAmerican societies. Standardised interviews which mimicked the clinical psychiatric evaluation were developed and becamc the criteria for determining “caseness” in epidemiological investigations (Williams, Tarnopolsky, & H and, 1980). After standardisation in Euro-American cultures, the interviews were subsequently used in other cultures. M ost of these subsequent cross-cultural psychiatric investigations relied on implicit, 1

2

CULTURE A ND CM D IN SU B-SAH ARAN AFRICA

largely untested assumptions: (1) the universality o f m ental illnesses, implying th at regardless o f cultural variations, disorders as described in Euro-A m erican classifications occur everywhere; (2) invariance, implying that the core features o f psychiatric syndromes are invariant; and (3) validity, implying th at although refinement is possible, the diagnostic categories o f current classifications are valid clinical constructs (Beiser, Cargo, & W oodbury, 1994). This approach, termed as the “ etic” or universalist approach, became the most popular m ethod for epidem iolo­ gical investigations o f mental illness across cultures. The etic approach offered the perspective that, since mental illness was similar throughout the world, psychiatric taxonom ies, their measuring instrum ents, and models o f health care were also globally applicable. There are two dom inant systems o f psychiatric classification, the ICD-10 Classification o f M ental and Behavioural Disorders (W orld Health O rganisation, 1992) and the D iagnostic and Statistical M anual (American Psychiatric Association, 1985), which reflect the psychiatric nosologies o f Euro-American medicine. D iagnostic criteria o f syndromes can and do change over time as is well dem onstrated by the regular revisions o f international psychiatric classifications; these revisions are considerably influenced by attitudinal, political, and historical factors (Westermeyer, 1985). Some cross-cultural researchers and psychiatric researchers in low-incomc countries have argued for the effectiveness and universal applicability o f current classification systems (e.g. C orin & M urphy, 1979; K erson & Jones, 1988; Odejide, 1979; Sen & M ari, 1987). Problems with the etic approach arose when cross-cultural researchers pointed o u t th at there was risk o f confounding culturally distinctive behaviour with psychopathology on the basis o f superficial similarities o f behaviour patterns or phenom ena in different cultures (D raguns, 1984). It was argued that classification o f psychiatric disorders largely reflected American and E uropean concepts o f psychopathology based on implicit cultural concepts o f norm ality and deviance (Baskin, 1984; K irm ayer, 1989). Some argued that cross-cultural psychiatry should examine the influence of culture on mental illness in Euro-American society itself, rather than assume th at these illnesses were “ natu ral” and free o f any cultural bias (M urphy, 1977). Critics accused the etic approach o f contributing to a world view which “ privileges biology over culture” (Eisenbruch, 1991) and ignoring the cultural and social contexts o f psychiatric disorders. T he field o f medical anthropology has exerted a growing influence on health research, particularly in low-income countries. This influence has seen a shift in paradigm s in public health and epidemiology from its unifocal and positivist “ scientific” approach to the recognition th at illness is the result o f a “ web o f causation” which includes the individual’s

1. INTRODUCTION

3

sociocultural environm ent (Heggenhougen & D raper, 1990). Medical anthropology has been one o f the key factors which fuelled the development o f the “emic” approach in cross-cultural psychiatry. At a general level this approach argued that the culture-bound aspects of biomedicine, such as its emphasis on medical disease entities, limited its universal applicability (Helman, 1991). M ore specifically, this approach argued that culture played such an influential role in the presentation o f psychiatric disorders th at it was wrong to presume a priori that Euro-A m erican psychiatric categories were appropriate throughout the world (Littlewood, 1990). P art o f this argum ent was based on the lack o f specific pathophysiological changes which could be identified in psychiatric disorders, which effectively made the diagnostic categories “ illnesses” as com pared to “diseases” (Helm an, 1981; Littlewood, 1991). The emic approach proposed to evaluate phenom ena from within a culture and its context, aiming to understand its significance and relationship with other intracultural elements. Purely emic studies have also draw n their share o f criticism, the most fundam ental one being that they are unable to provide data which can be compared across cultures (M ari, Sen, & Cheng, 1989). These studies are usually small in scale and are unable to resolve questions o f the long-term course and treatm ent outcom e of illness episodes (Kirmayer, 1989). The reliability o f emic studies is in doubt due to the lack o f standardisation o f research m ethods and the biased findings based on the interpretations of individual researchers. The emic approach has been criticised for not suggesting plausible alternatives, such as a set o f principles which would help ensure cultural sensitivity, or models upon which to fashion culturally sensitive nosologies (Beiser et al., 1994). It is argued that culture is not a static concept; all cultures are constantly evolving and changing and with the increasing influence o f Euro-American values and urbanisation in many low-income societies, “ traditional” beliefs may not be as rigidly held as is supposed. Furtherm ore, any individual may hold a multiplicity o f ideas regarding his illness and any or all o f these ideas may change with time (Eisenbruch, 1990). Despite m ajor strides in the international classification o f mental disorders and in the ethnographic approach to studying mental illness, a truly international psychiatry does n ot exist (W estermeyer, 1989). Thus, there are strengths and weaknesses o f both the etic and emic approaches in cross-cultural psychiatry. It is increasingly accepted that the integration o f their methodological strengths is essential for the development of the “ new cross-cultural psychiatry” or a culturally sensitive psychiatry (Kleinman, 1987; Littlewood, 1990). Value must be given to both folk beliefs about mental illness as well as to the biomedical system of psychiatry (Leff, 1990). It is im portant to investigate the patients’ “explanatory models” , i.e. how patients understand their problems, their nature, origin,

4

CULTURE AND C M D IN SU B-SAHARAN AFRICA

consequences, and remedies since these can radically assist patien t-d o cto r negotiations over appropriate treatm ent (Kleinman, 1980). Similarly, researchers should examine the psychiatric symptoms o f persons who are considered to be mentally ill by the local population and to interview the T M Ps and other prim ary carers to ascertain the diagnostic systems used. In essence, the central aim o f the “ new” cross-cultural psychiatry would be to describe mental illness in different cultures using m ethods which are sensitive and valid for the local culture and resulting in data which are com parable across cultures. In order to tackle this difficult task, psychiatric research needs to blend both ethnographic and epidemiological m ethods and emphasise the unique contribution of both approaches to the understanding o f m ental illness across cultures.

CO M M ON MENTAL DISORDERS (CMD) This is a term coined by G oldberg and Huxley (1992) to describe “ disorders which are commonly encountered in com m unity settings, and whose occurrence signals a breakdown in norm al functioning.” C M D , also referred to as non-psychotic mental disorders, encompass a broad group o f distress states which manifest with a mixture o f anxiety and depressive symptoms. C M D are the contem porary equivalent of the neuroses, a descriptive category which has become increasingly unpopular because o f its vague meaning and stigma. CM D have been classified in ICD-10 in two m ain categories: “ neurotic, stress-related and som atoform disorders” with a num ber o f subcategories and “ mood disorders” (with specific reference to unipolar depression). A simpler classification o f C M D has been devised for use in prim ary health care (Ustun et al., 1995b; see Table 1.1). In practice, the subcategories of CM D are n ot w ithout their conceptual problems: for example, obsessive-compulsive disorders are n ot com m on in com m unity settings, whereas phobic disorders may not be counted as T A B L E 1.1 C lassification of co m m on mental disorders for primary health care in IC D - 1 0

F32 F40 F4I.0 F41.1 F41.2 F43 F44 F45 F48 F51

Depression Phobic disorder Panic disorder Generalised anxiety Mixed anxiety and depression Adjustment disorder Dissociative disorder Unexplained somatic symptoms Neurasthenia Sleep problems

1. INTRODUCTION

5

mental illnesses by some investigators. Part o f this problem may be accounted for by the fact that classifications have tended to reflect the results o f psychiatric assessments at tertiary care level. In most prim ary care patients, symptoms o f anxiety and depression coexist to such an extent that their categorisation in either group is difficult. The W HO m ultinational study o f C M D found that for all specific psychiatric disorders (excluding alcohol dependence), com orbidity rates (with other psychiatric disorders) exceeded 50% (Ormel et al., 1994) suggesting that one o f the basic criteria o f a successful classification, i.e. the m utual exclusiveness o f different categories, was not achieved. Indeed, G oldberg and Huxley (1992) state that “ it is becoming clear that the idea th at C M D should be thought o f as discrete disease entities with distinct causes, course and treatm ent is probably untenable.” The W orld H ealth R eport based on 1993 statistics shows that neurotic, stress-related and som atoform disorders arc the third m ost frequent causes o f m orbidity (prevalence rates) worldwide (W HO, 1995). C M D are an im portant cause o f disability and have been identified as a significant public health problem (Blue & H arpham , 1994; W orld Bank, 1993). The m ultinational study o f the prevalence, nature and determ inants o f C M D was conducted in 14 countries, including Nigeria in Africa (U stun & Sartorius, 1995). The startling finding o f this study was that, despite the use of standardised methods in all centres, there were enorm ous variations in m ost variables. Indeed, the only similarities across centres were the general observations o f the ubiquity o f C M D and the com orbidity o f anxiety and depression and the association o f C M D and disability even after adjustm ent for physical disease severity. On the other hand, specific variables showed substantial variations; thus the prevalence rates o f C M D ranged from 7% to 52% of primary care attenders; physician recognition o f C M D varied from 5% to nearly 60% ; and the association o f key variables such as gender, physical ill health, and education with CM D were opposite in different centres (Goldberg & Lccrubier, 1995). For example, the relative risk o f having a C M D for females varied from 1.9 in Bangalore to 0.3 in Ibadan. Similarly, the relative risk for depressive disorder for those with two or more children varied from 3.3 in Athens to 0.19 in Ibadan. These marked variations suggest the need for regional studies with local health service-driven priorities to complement m ulti­ national studies with their emphasis on uniformity and universality (Patel & W inston, 1994).

CULTURE AND CO M M ON MENTAL DISORDERS Study of the influence o f culture on mental illness is im portant for several reasons: for example, it enables us to understand how patients from

6

CULTURE AND C M D IN SU B-SAH ARAN AFRICA

different ethnic groups experience and express mental distress and, further, by shedding light on aetiological factors it plays an im portant role in the development o f psychiatric theory by illum inating the diverse influences on mental distress posed by culture, society, and biology (Beiser, 1985; Sartorius, 1986). In recent years there has been increasing concern regarding the validity o f cross-cultural psychiatric studies which have mainly utilised the ctic approach. The uncertainty arises from the diagnosis o f mental illness which relies alm ost entirely on clinical presentations, there being no “ gold standard” validating pathological or diagnostic tests. It is generally accepted th at culture plays a profound role in the expression of idioms o f distress and th at in psychiatry language is the very m eans o f conveying symptoms. Yet,the validity o f the descriptive categories o f current classifications for other cultures has rarely been evaluated. The description o f illnesses in Euro-A m erican classifications are autom atically assumed to be cross-culturally valid, prom pting accusations o f “psychiatric imperialism” (Fernando, 1991). Concerns about validity should be greatest for syndromes o f depression and anxiety, where the boundaries between “ norm al sadness” and “clinical depression” are blurred (Kirmayer, 1989). F o r example, while the experience o f norm al sadness may be similar across cultures, the clinical significance o f depression as a unique illness category may vary considerably (Beiser, 1985). On the other hand, it has been argued that dysphoria is a univeral hum an experience and that depression can be recognised in many cultures. A lthough there may be indigenous categories o f mental illness, this does n ot necessarily invalidate the application o f international psychiatric categories for epidemiological purposes (Bebbington, 1993). The first problem one encounters in examining the cross-cultural validity o f the clinical category of depression is that many languages have no conceptually equivalent term for depression (Chaturvedi, 1993; Ihezue, 1989; M anson, Shore, & Bloom, 1985; Swartz, Ben-Aric, & Teggin, 1985). Conceptualisation o f depression in cross-cultural research is m ade especially difficult by the widely varying idioms o f distress expressed by patients and the varying contextual significance o f such idioms (Angst, 1973; Lutz, 1985). It cannot be assumed that even “ core” features o f depression in one culture have the same meaning in another. F o r example, Obeyesekere (1985) argues that hopelessness, a core cognitive feature o f the biomedical model o f depression, is perceived to be a positive feature o f m ental state for Buddhists. A lthough some African studies have reported that the “core” sym ptom s o f depression were the same in their patients, these researchers sampled patients who were attending psychiatric facilities or who had already been diagnosed as having a depressive disorder by psychiatrists (Keegstra, 1986; M ajodina & A ttah Johnson, 1983; M akan-

1. INTRODUCTION

7

juola & Olaifa, 1987); it is unclear whether these patients were representative o f depression in the community. Somatic symptoms have often been percieved to be a com m on mode o f presentation o f depression in low-income countries. Recent studies have shown that, contrary to popular belief, som atic presentations o f depression were also comm on in Euro-American societies (Bridges & G oldberg, 1985). Thus, som atic sym ptom s o f depression appeared to be universal to many cultures, though this did not imply that the appropriate name for the disorder was “depression” , but merely th at in Euro-A m erican cultures the everyday experience o f sadness came to the fore to the point that it became the most characteristic feature o f “depression” . It has been suggested that som atisation is “ an expression o f personal and social distress in an idiom o f bodily com plaints and medical help seeking” nonspecific to particular diagnoses (Kleinman & Kleinman, 1985). For example, Cheng’s study in Taiw an (1989) suggests that for a substantial num ber o f psychiatric patients in prim ary care, som atisation was a form o f illness behaviour manifested in neurotic patients from a wide diagnostic spectrum including anxiety and depression. C an depression be diagnosed in patients who do not experience the cognitive features o f the illness? While some authors have assumed that depression is the “ true” illness in patients presenting with nonspecific somatic sym ptom s (Ndetei & M uhangi, 1979), others have expressed reservations in diagnosing depression when the central cognitive features o f the illness are absent (Venkoba Rao, 1994). Illness patterns with their own characteristic clinical and epidemiological features, but with no E uroAmerican equivalent, are seen to be “ masked” presentations o f an “ underlying” depression. In these situations the fact that some patients complained o f low m ood or achieved “ cut-off” scores on depression rating scales was taken as evidence for the assum ption that the “ true” diagnosis was depression, even though it is well recognised that emotional responses such as low m ood and apprehension comm only occur as a reaction to a num ber o f medical and psychiatric conditions. Thus, patients presenting with a prim ary com plaint o f “ loss o f semen” may also be depressed (Cheng, 1989); is depression com orbid with the semen loss syndrome (dhat syndrom e in India), or is the latter a “ masked” or “ som atised” form of depression? Angst (1973) argues th at the current concept o f depression is so rooted in European culture that it is strongly influenced by a cultural bias; in his view, then, the concept o f “ masked depression” is as representative o f a depressive syndrome as the classic descriptions. Questions regarding the cross-cultural validity o f the clinical category o f depression remain unresolved, n o t least because “ the universality o f the category o f depression (and other categories o f neurotic disorder) is assumed, eliminating the need to establish validity, and the tautological

8

CULTURE AND C M D IN SU B-SAH ARAN AFRICA

circle is completed when the symptoms th at serve as criteria for the diagnosis, because they are believed to reflect specific psychophysiological and horm onal states, are assumed to be universal” (G ood, G ood, & M oradi, 1985). Theoretic assumptions underlying the etic and emic approaches have influenced the choices researchers m ake o f the m ethod o f assessment o f psychiatric disorders across cultures. These m ethods will now be described.

A SS E S S M E N T OF MENTAL DISORDERS A C R O SS CULTURES The quantitative assessment o f mental disorders, such as the measurem ent o f psychiatric m orbidity and determ ination o f prevalence rates, requires standardised questionnaires. There are two m ethods o f using question­ naires across cultures, viz., using preexisting measures developed in other cultures or developing measures de novo. M ost cross-cultural studies use instrum ents developed in one culture (to date, always a Euro-Amcrican culture), translate them, and apply them to another culture. Given the central im portance o f language in expressing symptoms, the translation o f the instrum ent is perhaps the single m ost im portant step in etic studies. The translated version should be evaluated on a num ber o f different param eters such as its content, technical, conceptual, and criterion equivalence (Flaherty et al., 1988; Krause, 1990; Sartorius, 1993). A part from an emphasis on translation, it is im portant to evaluate the validity o f instrum ents when they are to be used as case finding instrum ents. F o r example, though the Self-reporting Questionnaire (SRQ) is used by many authors with a standard cut-off score o f 6-7, investigators who have evaluated its validity in Africa have shown th at higher cut-off scores are more sensitive and specific for case identification (K ortm ann & T en-H orn, 1988; Patel & T odd, 1996). A lthough m ost studies using etic instrum ents deal superficially with the issue of translation and validity, there is a growing literature on research using etic instrum ents suitably translated and validated for use in different cultural settings. This research has shown that if careful attention is placed on the issues o f validity and translation, etic instrum ents can be used with confidence across cultures (e.g. Bravo, C anino, Rubio-Stipec, & W oodbury-Farina, 1991; M anson et al., 1985; M um ford ct al., 1991b). There have been a few successful attem pts at developing new m ethods o f assessment for C M D which integrate emic and etic approaches both in Africa and elsewhere. The following are four examples o f such studies from non-A frican settings. Kinzie et al. (1982) described the development o f a Vietnamese language rating scale for depression. Items were derived from the Beck Depression Inventory, Vietnamese terms elicited from a

1. INTRODUCTION

9

lexicon generated by bilingual mental health workers, som atic symptoms frequently presented by Vietnamese patients, and items designed to tap the behavioural and somatic symptoms o f depression. O f the 15 items which discriminated for depression 10 were unrelated to either lowered m ood or the Western concept o f depression. The Bradford Somatic Inventory (BSI) developed in the U K for use with patients from the Indian subcontinent (M um ford et al., 1991a) emphasised the im portant role o f som atic symptoms in the expression o f emotional distress. The BSI consisted of somatic symptoms recorded in the case notes o f patients in the U K and Pakistan. These items were checked against the case notes o f patients in India and N epal and over 90% coverage o f all somatic sym ptom s was achieved. The U rdu and English versions were then administered to bilingual students in Pakistan to determine linguistic equivalence. Conceptual equivalence was determined by studying the factor analysis o f responses by patients with functional disorders in Britain and Pakistan. The seven-item Prim ary Care Psychiatric Questionnaire (PPQ) was developed in India (Srinivasan & Suresh, 1990; Suresh, Suresh K um ar, Bashyam, & Srinivasan, 1993) on the rationale th at patients preferred to express their distress in som atic terms and prim ary health care staff were more com fortable in discussing such symptoms. Therefore, a screening measure consisting o f som atic symptoms would be more appropriate for the Indian setting. Eleven symptoms frequently presented by patients with neurotic disorders were administered to a random sample o f new prim ary care attenders. Seven symptoms which occurred more often in cases (as judged by a psychiatric interview) formed the PPQ. The Chinese Health Questionnaire (CHQ) was developed with the 30-item G eneral Health Questionnaire as its starting point (Cheng & Williams, 1986; G oldberg, 1978). An additional 30 items based on the Chinese concepts o f illness, such as the concern about the heat and coldness o f food, were added. The resulting 60-item questionnaire was validated against a standardised psychiatric interview. The final questionnaire consisted o f 12 items which discriminated cases best; half originated from the G H Q and the rem ainder were emic items. In Africa, there are three published examples o f attem pts to develop culturally sensitive psychiatric instrum ents, all from W est Africa. Beiser and colleagues’ (1972, 1976) studies with the Serer people o f Senegal began with ethnographic w ork setting the stage for eliciting a local taxonom y o f mental illness and lexicon o f Serer illness terms. A group o f patients with “ illnesses o f the spirit” were interviewed showing that these illnesses were closely related to psychiatric concepts o f mental illness. The research group further developed an interview schedule based on a preexisting ques­ tionnaire and the lexicon o f illness terms. The Somatic Screening Instrum ent designed by Ebigbo (1982) in Nigeria was developed by listing

10

CULTURE A N D C M D IN SU B-SAH ARAN AFRICA

com plaints o f patients diagnosed as suffering from anxiety neurosis. This questionnaire was then tested on a patient and control group to identify items with discrim inant validity. O f the original 65 complaints, 46 were found to distinguish male cases from norm als while 30 items distinguished female cases from normals. The N igerian version o f the SRQ was developed because many o f the symptoms represented in the item content were found not to be presented spontaneously and were subject to considerable “ yea-saying” bias. Ten “culture-specific” items were added to the SRQ and the 30-item version used with samples o f cases o f C M D and their relatives. The 20 items which discriminated best between these two groups went on to form the Nigerian SRQ-20 which contained nine “culture-specific” items (M artyns-Yellowe, 1995). All the above m ethods were innovative in developing assessment measures by creating lists o f items derived either entirely from idioms o f distress o r by adding these idioms to a preexisting questionnaire and then evaluating its validity using psychiatric diagnosis as elicited by clinical or standardised interview as the criterion.

EPIDEMIOLOGY OF MENTAL ILLNESS IN SU B-SA H A R A N AFRICA Perhaps the earliest attem pt at a psychiatric epidemiology in Africa was the w ork o f C arothers (1953) who reported that depression virtually never occurred in Africa and th at this disorder was 13 times as comm on in the U K . His w ork was largely discredited since it was based on hospital populations and had racist overtones in his implication that the lack o f depression was evidence o f the underdevelopm ent o f the African brain. Similar racist ideologies were evident from some researchers in South Africa; for example, Le Roux (1973) argued that the apparent lack o f specific terms equivalent to E uropean psychopathology suggested an “ obvious ignorance” o f the nature o f such derangem ents and th at this “ primitiveness o f the subconscious” could emerge as “ physical assault” . Studies in the late 1960s to the mid-1970s which used operationalised definitions o f psychiatric disorder, structured questionnaires, and com m u­ nity or prim ary care samples o f patients showed that psychiatric disorders were at least as comm on in Africa as in Euro-A m erican societies (Binitie, 1981; Giel & Van Luijk, 1969; Gillis, Lewis, & Slabbert, 1968; Leighton et al., 1963). Pyschiatric epidemiology in Africa has been reviewed by several authors in the recent past (G erm an, 1987; Odejide, Oyewumi, & Ohaeri, 1989; Parry, 1996; Reeler, 1986). The objective o f the review in this m onograph is to provide a brief overview o f research in prim ary care and com m unity settings with a specific aim o f evaluating the cultural sensitivity o f the

1. INTRODUCTION

11

study methodology. O f the studies published since the mid-1970s, 14 were identified through bibliographic searches representing 8 Sub-Saharan A frican countries. Sub-Saharan Africa was selected as the geographical region for review (for both this and the following section) because its nations share num erous cultural and historical features distinct from those of predom inantly A rab N o rth Africa (Stock, 1995). A review o f the study settings, methods, and key findings is presented in Table 1.2. The prevalence o f mental disorder was reported to be greater in both rural comm unity studies as compared to similar studies in E uropean or American settings. The rural U gandan study in which researchers used m ethods identical to those used in a study in London reported that hypom anic disorders were more than five times as frequent, depressive disorders twice as frequent, and anxiety states three to four times as frequent am ongst female respondents. In rural Lesotho, depression was four times as comm on and panic disorder was five times as comm on in both sexes when com pared to data from the USA. On average, in both studies, more than one in four rural inhabitants were diagnosed as suffering from a current mental illness. Before accepting these findings, one must consider their validity and meaning to the community. To do so, one would need to elicit the patients’ and care providers’ views about the nature o f the illness. Only the U gandan study explicitly asked the patients’ views about whether they thought they had a mental illness or any illness in the past involving the symptoms discussed during interview with the Present State Exam ination (PSE; Wing, C ooper, & Sartorius, 1974). The authors rem arked that “ it was fairly rare to get a positive answer, and negative answers were considered unreliable” (Orley & Wing, 1979). The relationship o f psychiatric disorder and explanatory models in the Lesotho study was only presented for patients with a panic disorder. M ore than tw o-thirds of patients (68%) said they did not know the cause o f their problem; 23% attributed psychological causes and only 3% mentioned spiritual causes. N o explanation was provided for w hat a “ psychological” cause or “spiritual” cause meant. Nearly three-quarters o f patients (72%) did not know what the treatm ent would be and, perhaps surprisingly, not one patient said that a TM P might be helpful. This finding was at odds with the finding that nearly a quarter (23% ) had sought traditional medical help for panic symptoms, casting some doubt on the reliability o f the findings. Prevalence rates in the PH C studies ranged from 10% to 69% with a median prevalence o f 22.5% and a mean prevalence (excluding the one extreme prevalence figure o f 69% ) of 21.8%. In Zimbabwe, widely varying findings were reported ranging from an annual prevalence rate o f mental disorder of 26% over a four-year study period (Hall & Williams, 1987) to a prevalence o f 10% over a one-m onth study period (Reeler, Williams, &

T A B L E 1.2 Epidem iological studies of mental disorder in primary care an d com m unity settings in S u b -S a h a r a n Africa after 1978

Authors/ Country 1. Ndetei & M uhangi, 1979; Kenya 2. Orley & Wing. 1979; Uganda 3. H arding et al., 1980; Sudan 4. Diop et al., 1982; Senegal 5. Dhadphale ct al., 1983; Kenya 6. Oduowlc & Ogunyemi, 1984, Nigeria 7. de Jong ct al., 1986; Guinea-Bissau 8. Hall & Williams, 1987; Zimbabwe 9. Dhadphale et al., 1989; Kenya 10. Abiodun, 1989; Nigeria 11. Hollifield et al., 1990; Lesotho 12. Jegede et al., 1990, Nigeria 13. Gureje & Obikoya, 1992, Nigeria 14. Reeler et al., 1993; Zimbabwe

Setting

Sample size

Emic element! translation

Stages

Interviews

Findings

U PH C

140

N o information

1

R COM

221

1

R PHC

360

2

SRQ, PSE

Overall: 25% DEP: M 14.3%, F 22.6%, ANX: M 3.1% , F 4.3% Overall: 10.6%

R PHC

933

Translation/back translation Brief d ata on translation No inform ation

Clinical examination PSE

1

SRQ

Overall: 16.2%

SU /R OPD

388

2

SRQ, SPI

1

GHQ-30

Overall 29% D EP 9.3%, ANX 8.5% , M D I 4.9% , SCZ 1.5% Overall: 69%

2

SRQ, PSE

Overall 12%

2

SRQ. PSE

Overall > 10% M = F

SRQ, SPI, H D RS PSE

Overall 25% D EP 9%

Short DIS, SCL-90 PSE, Clinical interview, CES-D GHQ-12 and C ID I SRQ

Overall: 22.8% D EP 12.4%; PD 11%, Overall (PSE) 39.7% Overall (CES-D) 42% Overall (Clinical) 47.7% Overall: 35.1% D EP 8.8% , G A D 9.1% Som atoform dis 16.7% Overall: 26%

U OPD

80

R PHC

251

R OPD

448

Modified SPI item on bewitchment T ranslation/back translation Portuguese version of SRQ from Brazil N o inform ation

R OPD

881

N o inform ation

2

R PHC

214

No inform ation

1

R /U COM

356

2

U OPD

104

U PH C

787

Translation/back translation Translation/back translation Translation/back translation N o inform ation

R/U PH C/ OPD

1236

1 2 1

Overall: 20% M 19%; F 22%

Overall: 22.5% D EP 14%,

Key: R = rural; SU = sem i-urban; U - u r b a n ; C O M = com m unity; O P D = hospital outp atien t departm ent; PH C = prim ary health clinic; D IS = D iagnostic Interview Schedule; H D R S = H am ilton D epression R ating Scale; PSE = Present State Exam ination; SCL-90 = Sym ptom CheckList; SPI = Standard Psychiatric Interview; SRQ = Self R ating Q uestionnaire; G H Q -3 0 = 30-item G eneral H ealth Q uestionnaire; C E S-D = C entre for Epidemiological Studies o f D epression Scale; C lD I —C om posite International D iagnostic Interview; A N X = anxiety; D E P = depression; DYS = dysthym ia; G A D = generalised anxiety disorder; M D I = m anic depressive illness; PD = panic disorder; SCZ = schizophrenia.

1. INTRODUCTION

13

Todd, 1993). Diagnostic breakdown was given in half the studies and showed that neurotic disorders were the comm onest conditions in prim ary care and, o f these, depression and anxiety were the comm onest diagnoses. Phobic disorders, obsessive-compulsive disorders, and disassociative disorders were rarely, if ever, identified. The most comm on presenting complaints in PH C attendcrs were somatic symptoms such as fever, headache, epigastric discomfort, abdom inal and chest pains, cough, gcnito-urinary symptoms, and constipation; patients very rarely complained o f “ psychological sym p­ tom s” . The authors o f one study diagnosed depression and anxiety in patients who displayed none o f the subjective cognitive sym ptom s o f either (Ndetei & M uhangi, 1979). These authors suggested that in African culture “ concepts like sadness and anxiety do not carry medical implications and so on a cultural level alone it is unlikely th at patients would complain of these states.” Three studies attem pted to investigate the coexistence o f comm on medical diseases or nutritional problems which are im portant causes o f somatic morbidity in Sub-Saharan Africa (Gureje & Obikoya, 1992) but only one reported on com orbidity o f physical illness, showing that a fifth of patients with a psychiatric disorder also suffered a physical illness (Abiodun, 1989). Case recognition by health clinic staff was reported in five studies and was always found to be low. For example, in a Nigerian study, only 14.6% o f m orbidity was recognised by care providers (A biodun, 1989), while in a Zim babwean study, the figure was even lower at 4.2% (Hall & Williams, 1987). Patients with psychotic and suicidal symptoms were most likely to be recognised as having a mental illness, whereas those with somatic “equivalents” were least likely. All the studies which elicited the diagnostic assessment of the health care provider were doing so to examine how many “cases” were detected by them, rather than to determine their views as clinicians with much experience in prim ary care. Even though psychiatry in Africa is principally concerned with the “psychiatry o f psychoses” (Asuni, 1991), the hospital-based psychiatric instrum ents designed in a foreign culture were the gold standards for diagnosis in comm unity and prim ary care settings. Thus, even when neither the patient nor health clinic staff considered a mental illness to be present both were considered to be of secondary im portance to, and less “ reliable” than, the instrum ent. Baasher (1982) noted the unique difficulties in conducting epidem iolo­ gical investigations in low-income countries, such as organisational deficiencies, shortage o f trained personnel and lack o f reliable medical recording. Despite these problems, researchers in a num ber o f African countries have successfully conducted epidemiological investigations and dem onstrated that psychiatric morbidity is common in com m unity and prim ary care populations. Some studies have also dem onstrated that such

14

CULTURE AND C M D IN SU B-SAH ARAN AFRICA

disorders are often not recognised by prim ary care staff and others have noted that this may lead to inappropriate medication, unnecessary investigations, chronicity o f morbidity, and dissatisfaction with health care services (Freem an, 1991). However, from the perspective o f a culturally sensitive epidemiology, none o f the 14 studies reviewed employed a significant emic element. All the intrum ents used were etic; the m ajority were developed in Britain. Only one study provided detailed inform ation on translation and this study identified many linguistic and conceptual problem s with the instrum ent used (PSE). Validity was rarely evaluated for the instruments. The SRQ was the most frequently used instrum ent. In the W H O study, which led to its developm ent and described its use in four low-income nations, the cut-off score varied from 3-4 in one centre to 10-11 in another (H arding ct al., 1980). Yet, all the African studies used a standard cut-off score o f 6-7. Recent research has shown that higher cut-off scores, such as 9-10, are m ore appropriate for the SRQ in African settings (K ortm ann & Ten H orn, 1988; Patel & T odd, 1996). K ortm ann and T en-H orn (1988) examined the validity o f the SRQ in E thiopia and found that the instrum ent lacked criterion validity, that m any items lacked concept validity and th at scores often reflected helpseeking behaviour rather than mental illness. Similarly, Martyns-Yellowe (1995) found that, in Nigeria, many o f the SRQ items were not presented spontaneously and were subject to a yea-saying bias. A lthough the PSE was designed for use in hospital-based populations, the instrum ent was used in a num ber o f studies w ithout any mention o f issues pertaining to its validity in com m unity or primary care settings (Parry, 1996). In conclusion, epidemiological studies in Africa have used the etic approach and none can be said to apply the principles o f the new crosscultural psychiatry. Despite the frequent observation th at patients with C M D tend to consult T M P (O latawura, 1982), none o f the studies included T M P attenders. Forty years o f psychiatric epidemiology in Africa have alm ost entirely focused on cross-sectional estimates o f prevalence using im ported methodologies with little attention to indigenous categories o f m ental illness, local sym ptom profiles, care provider diagnostic concepts and patient explanatory models.

EXPLANATORY MODELS OF MENTAL ILLNESS IN SU B-SA H A R A N AFRICA Explanatory models (EMs) is a term coined by Klcinm an (1980) to denote the “ notions about an episode o f sickness and its treatm ent th at are employed by all those engaged in the clinical process.” EMs are formed from a variable cluster o f cultural symbols, experiences and expectations associated with a particular category o f illness. EM s reveal sickness

1. INTRODUCTION

15

labelling and cultural idiom s for expressing the experiences o f illness. E xplanatory m odels o f illness influence health seeking behaviour and health service utilization (F osu, 1981). A review by the a u th o r covered w ork from 11 Sub-Saharan nations, viz., N igeria, Senegal, U ganda, Zim babw e, Botsw ana, E thiopia, G h an a, Sw aziland, South A frica, G uineaBissau, and K enya. T he detailed review has been published elsewhere (Patel, 1995a) and only an overview o f the m ain findings are presented here. V irtually all the cultures reviewed differentiated between the m ind and the body, an d the concepts relating to the m ind were shared to som e extent by the different cultures. T he existence o f T M P s w ho specialised in the m anagem ent o f m ental illness (G elfand, M avi, D rum m ond, & N dem era, 1985; G ood, 1987; O dejide, O lataw ura, Sanda, & Oyenye, 1977; S taugard, 1985) was further evidence o f a m in d -b o d y differentiation. T here were som e similarities to E uro-A m erican concepts, evidenced by reports o f a “ goodness-of-fit” between the tw o system s’ identification o f m ental illness, an d the attem pts by som e au th ors to com pare traditional concepts ab o u t the p arts o f m an to E uro-A m erican concepts, such as F reudian theories ab o u t the personality (M utam birw a, 1989). W hat differed were the sem antics used to describe the m ind an d concepts relating to its function an d localisation. Illnesses o f the “ sp irit” and o f the “ soul” were probably analogous to m ental illness. A lthough there was som e variation a b o u t th e som atic positioning o f the m ind, it was mostly localised in the head, chest, and abdom inal regions. As E bigbo (1986) states, this localisation o f the m ind to som atic structures m ay explain the phenom enon o f som atic presentations o f nonpsychotic m ental illnesses. T hus, w hat was perceived by E uropean p atients as being p alpitations associated w ith anxiety due to “ stress” and w orries in the m ind, m ay be perceived by an A frican patien t as p alpitations due to im balance in the function o f the heart. A num ber o f com m on threads could be identified in the diversity o f taxonom ies and aetiological models. First, although there was a com m on assum ption th a t classification was solely on aetiological grounds, there was evidence th at phenom enological classifications were also im p o rtan t. T hus, the type and severity o f behavioural disturbance was often used to classify broad categories o f m ental illness; it was w ithin these categories th a t aetiological m odels were used for further classification. Second, the classifications used were flexible and p atient dependent; thus, even though phenom enology was used by a healer to u nderstand the n atu re o f the illness, an aetiological model was alm ost always provided since it gave the illness experience m eaning for the patient. T hird, there was a general classification o f illness into the two categories o f “ n a tu ra l” and “ u n n atu ral” illness; these have clear im plications on health service

16

CULTURE AND C M D IN SU B-SAHARAN AFRICA

utilisation, since the form er category was perceived as being related to physical or environm ental causes and could be equally well treated by both biomedical and traditional methods, whereas the latter was seen to be related to traditional beliefs o f misfortune and illness and was more likely to be brought to TM Ps. F ourth, the role of supernatural factors was prom inent in the causation o f illness. Though witchcraft is outlawed in m ost o f Sub-Saharan Africa, this category continued to be used as an im portant way of explaining misfortune. Finally, there was growing evidence that with the influence o f colonisation and urbanisation, views about illness were also changing. F or example, a study with psychiatric outpatients in Nigeria found that less than half the patients held a supernatural cause as the source o f their problem; neurotic illness was m ore likely to be ascribed to supernatural causes than psychotic illness (Ilechukwu, 1988). Similarly, the diminishing im portance o f supernatural causal models in urban settings was noted by G ood in Kenya (1987). The review suggested that, m ost commonly, EMs o f mental illnesses in Africa equated them with the biomedical construct o f psychotic disorders. Thus, in many studies, when healers or key inform ants were asked open questions to describe mental illness, they most often described behavioural features related to psychotic illness (Ugorji & Ofem, 1976). It was only when there was more detailed inquiry or the presentation o f case vignettes that the extended ramifications o f the concepts and classification o f mental illness which included neurotic-like disorders were elicited. K o rtm an n ’s (1987) finding that, although there existed a general and “ n eutral” A m haric term to denote mental illness in Ethiopia this excluded many less severe forms o f psychiatric disturbance which were not considered as being illnesses, has echocs in studies from other African cultures. Similarly, when case vignettes o f different presentations o f mental disorder were presented to com m unity samples or primary health care staff, neurotic disorders were rarely viewed as being psychiatric problems (A biodun, 1991; Erinosho & A yonrinde, 1978). Indeed, the view th at TM Ps may not be able to distinguish neurotic and psychotic disorders (G elfand, 1967) may be largely due to the fact th at they do n ot recognise the form er as being related to the latter. There was a striking similarity in the behavioural sym ptom s o f acute psychotic disorder across cultures, with some behaviours such as aggression being particularly common. A lthough such patients would be identified as suffering from a psychotic illness by etic criteria, there was much less emphasis on cognitive features such as delusions which are central to psychiatric diagnosis. N eurotic disorders, although often not perceived to be mental disorders, were still recognised by local com m u­ nities and TM Ps as being sources o f illness and misfortune. Somatic features predom inated, and there were some sym ptom s th at were reported

1. INTRODUCTION

17

to be particularly com m on, such as the sensation o f craw ling under the skin and uncom fortable sensations originating in th e region o f the heart and abdom en. A num ber o f cognitive features were also identified, though they were often less em phasised. These included fearful feelings, thinking too m uch, and the m ind going blank. Finally, there were several phenom ena th a t bore close sim ilarity to anxiety and depression, such as insom nia, palpitations, an d headache. Even though, in general, no distinctions were m ade between the various categories o f neurotic disorder, som e studies involving detailed questioning o f T M P s uncovered su b ­ categories which bore similarities to dissociative and p anic disorders. O bsessive-com pulsive disorders and p h obias were rarely recognised.

THE STUDY SETTING T he studies described in this m onograph were conducted in H arare, the capital city o f Zim babwe. Zim babw e is a landlocked n ation in south central A frica. She gained independence from colonial rule after a bitter liberation w ar in 1980. In the 1992 census, Z im babw e’s populatio n was recorded to be ju st over 10 million (C entral Statistical Office, 1995). Just under 70% o f the popu latio n lived in the rural areas. T he populatio n was relatively young w ith 45% aged below 15 years and only 3% aged over 65 years. O f the ad u lt popu latio n (i.e. over 15 years o f age), 56% were m arried. O f the ad u lt p opulation, 38% were either students, “ hom e­ m akers” , o r were unable to w ork due to sickness o r old age. O f the rem ainder, 22% were unem ployed. O f those in em ploym ent, agriculture accounted for 43% o f jobs, m ost o f which were in com m unal farm ing. G iven the highly seasonal natu re o f agriculture in Zim babw e an d the fact th a t m ost farm ing is subsistence, this occupation does n o t offer long-term stability o r security to m ost farmers. The m ajority o f Z im babw e’s people are o f A frican origin (98% ). A lthough the 1992 census does n o t report the ethnic com position o f the A frican population, it is widely accepted th a t the Shona tribe are by far the m ore num erous, particularly in the H arare region. T his g ro u p is com posed o f num erous subclans (e.g. Z ezuru, M anyika, K aranga), b u t all speak S hona (with variations in dialects for each subclan). T he o th er m ajor group are the N debele who are concentrated in the southw estern regions o f the country. T heir language has ro o ts in the Z ulu language and is linguistically distinct from Shona. A n um ber o f other A frican nationalities are represented in Zim babw e, particularly in H arare. M any o f these im m igrants are also conversant in Shona. T hus, S hona is the m ost com m only spoken language in H arare and w as the language used for the studies described in this book. In the non-A frican p o p u latio n are the whites and A sians. B oth com m unities, in particu lar the form er, are

18

CULTURE AND C M D IN SU B-SAH ARAN AFRICA

highly influential in terms o f their control o f private industry and farming land. H arare is the adm inistrative and commercial capital o f Zim babwe. It was founded about 100 years ago by British settlers, and until independence in 1980 was called Salisbury. It has a population o f approxim ately 1 120000 inhabitants (Central Statistical Office, 1995). The city’s 42 suburbs can be classified into two types: low-density suburbs which, for historical reasons, house the professional and educated classes (now multiracial), and high-density suburbs, which house m ost o f the city’s black population. Household size varies from 2.9 in the low-density suburb o f Borrowdale to 4.9 in the high-density suburbs o f W orkington and M ufakose (City o f H arare H ealth D epartm ent, 1994). H ealth service use differs between these two types o f suburb; residents o f low-density suburbs use private health facilities, while those o f high-density suburbs m ake greater use o f public health facilities and TM P. Private general practitioners (GPs) are also playing an increasing role in prim ary health care in high-density suburbs. Zim babwe is a society in transition. It is witnessing an increasing rural to urban m igration and a change in the traditional extended family and clan kinship networks (Romm e, 1987). Increasing exposure to E uropean and American culture through films and music is evident in the towns and cities and this is identified as a key influence in the breakdow n o f traditional kinship ties (Bourdillon, 1987). Zim babwe m ade impressive progress in health care in the years following independence. The increased funds available through taxes and foreign aid and the governm ent’s policy o f supporting prim ary health care led to impressive im provem ents in health indices; for example, infant m ortality rates reduced from 110 per 1000 live births in 1960 to 73 per 1000 in 1986 (Roemer, 1991). These achievements prom pted a recent U N IC E F report to describe Zim babwe as a “ beacon for progress tow ards child survival and development in SubSaharan A frica” (Lennock, 1994). Since 1990, however, the achievements of post-independence Zimbabwe are under serious threat due to a com bination o f reasons. The public resources available for investm ent in the social sector are being reduced due to slow economic grow th in the 1980s, overdependence on overseas financial aid, drought, and, since 1991, the budgetary constraints imposed as a result o f the im plem entation o f the W orld Bank Economic Structural A djustm ent Program m e (ESAP). One key plank o f ESAP has been the reduction in governm ent expenditure which led to the introduction o f “ user fees” for public health services. This policy has been directly blamed for a fall in prim ary health clinics attendances despite a growing population (Logie & W oodroffe, 1993). It is anecdotally reported th at there are increasing num bers o f patients now consulting TM Ps. A recent comm unity survey has reported an association

1. INTRODUCTION

19

between relative poverty, nonconsultation for illness, and p o o rer health outcom es affecting a sizeable p ro p o rtio n o f the sam ple (W inston & Patel, 1995).

M EDICA L PLU RA LISM IN Z IM B A B W E H ealth care in Zim babw e, like m ost A frican countries, is provided by both biom edical and traditional health care providers (Ben-Tovim , 1985). Patients are faced with a varied choice o f health care practitioners and may consult different practitioners sim ultaneously o r consecutively. Biomedicine was introduced in Zim babw e by the m issionaries who preceded the colonial settlers. Follow ing colonisation cam e the doctors w ho catered largely to the white settler p opulation. T his was the first exposure o f biom edicine to the c o u n try ’s black people. Biomedical services were dichotom ised in public an d private health services. Public health care was provided throug h hospitals and clinics, and were funded either by the missions o r the governm ent. Facilities run by the governm ent tended to be segregated on racial grounds while mission hospitals dealt largely with blacks in the rural areas. Private m edical practitioners were concentrated in the urban areas an d catered alm ost exclusively to the white population. M any were econom ically protected th rough m edical insurance schemes (R oem er, 1991). As the governm ent service grew, an increasing need for district- and rural-based doctors was ap p aren t and, by the 1950s, m any such posts becam e unfilled vacancies. S tandards o f peripheral care dropped further as the central hospitals sucked in m ore o f the health budget in funding high technology medicine. The new medical school, which graduated its first doctors in 1968, had little im pact on the shortage o f medical m anpow er since m ost new graduates stayed in the u rb an areas o r left the country (M ossop & S tratfo rd , 1986). Sadly, even after independence this trend has n o t changed with m any medical an d nursing graduates m igrating to work in neighbouring Botsw ana o r South A frica for the better wages offered there. Follow ing independence, the govern­ m ent em barked on an am bitious program m e to redevelop the peripheral health services and created the new position o f the village health w orker with a preventive health role. D istrict hospital staffing levels im proved w ith m any expatriate doctors, though the situation rem ained critical w ith over a third o f posts reported to be vacant in 1995. O f Z im babw e’s 1600odd m edical practitioners, over 1000 are concentrated in the two m ain cities o f H arare and Bulawayo. T hus the average d o cto r patien t ratio o f 6500 is unrepresentative o f the situation in p o o r u rban areas and rural areas. T his shortage o f m anpow er is even m ore acute in the field o f psychiatry; thus, the situation a t independence with seven psychiatrists, all

20

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

in u rb an areas (H ollander, 1986), had barely im proved to a b o u t a dozen psychiatrists in 1995. T he M inistry o f H ealth and the City o f H arare H ealth D ep artm en t provide biom edical health care services in H arare. T he two m ain central teaching hospitals are adm inistered by the M inistry o f H ealth and serve as tertiary referral centres for no rth ern Zim babw e; the local hospitals an d prim ary health care centres are adm inistered by th e City o f H a rare H ealth D epartm ent. Psychiatric services in the public sector are largely concen­ trated in the central hospitals, each o f which houses an acute unit. A t the tim e o f the studies described in this b ook (1993-1995), these units were staffed m ainly by expatriate consultant psychiatrists, local psychiatric trainees and interns, nurses, and occupational therapists. T he vast m ajority o f p atients suffered from a psychotic illness, often w ith disturbed behaviour. T he em phasis o f inpatient care was to obtain a psychiatric diagnosis and provide pharm acological treatm ents; psychotherapeutic ap proaches were seldom used (R om m e, 1987). A study with 67 referrals to a H arare psychiatric unit revealed th a t m any p atients had consulted a T M P earlier and m ost h ad bypassed biom edical prim ary care services (Reeler, 1992). The m ajority (74% ) were acutely psychotic with disturbed behaviour. D elays in seeking psychiatric care were not evident for those consulting T M Ps. P H C are run by the City H ealth D epartm ent and are staffed by general nurses. Eleven com m unity psychiatric nurses were posted in different districts. T heir w ork was concentrated on com m unity care o f the chronically m entally ill and they w ould only occasionally see p atients w ith C M D . M edical officers visited clinics twice or thrice weekly to hold clinics. H IV -related disease was the single m ost im p o rtan t cause o f death, accounting fo r a third o f all deaths in the 25-44 year age group. Suicide was the fourth m ost com m on cause o f d eath in the 15-24 year age group accounting for 5% o f all deaths (City o f H arare H ealth D ep artm en t, 1994). T raditional medicine is an im p o rtan t source o f health care in m odern Zim babw e. T M P are grouped into tw o broad categories: the n ’angas include spirit m edium s, diviners and herbalists; and the profitas, who belong to one o f the m any A frican C hristian churches which flourish in the country and use m ethods which syncretise traditional treatm ents and C hristian beliefs. The latter belong mainly to the A postolic churches and divine the causes o f illness th rough the the H oly Spirit (as opposed to the ancestral spirits in the case o f n ’anga). T he m ost obvious role o f these different T M P s as depicted in much o f the cross-cultural literature is th at o f the “ native m edicine-m an” in which health is defined along biom edical concepts o f disease entities. In addition to health care, however, T M P s arc also religious consultants, legal advisors, social w orkers, m arriage counsellors, health educators, and family therapists (C avender, 1991; N gw enya, 1992; N yam w aya, 1992; S taugard, 1985). T M P s are relatively

1. INTRODUCTION

21

well organised; shortly after independence in 1980, the Zim babw e N atio n al A ssociation o f T raditional H ealers (Z IN A T H A ) was form ed by am alga­ m ating eight other small organisations, with aim s n o t dissim ilar to professional bodies representing biom edical practitioners. W hile Z IN A T H A plays an im p o rtan t role in registering healers and publishes occasional papers, its role in research and training is limited due to a shortage o f trained research staff, financial constraints, and political problem s (C havunduka, 1986). A lthough m aintaining registers o f T M P s is an im portant function o f these organisations, it is generally th o u g h t th at less than h alf o f the 40 000 T M P in Zim babw e are registered with the Z IN A T H A (C havunduka, 1994). A recent com m unity survey o f T M P s was conducted in tw o suburbs o f H arare which were to be the focal points for m uch o f the research described in this book (W inston, Patel, M usonza, & N yathi, 1995). By m eans o f a m ultistage tracing procedure, 189 T M P s were identified in the tw o highdensity suburbs o f D zivarasekw a and M ufakose. By contrast, the suburbs were served by 58 prim ary care nurses and 5 private doctors, which represented a ratio o f 3 T M P s per biom cdical practitioner. A stratified sam ple o f 110 T M Ps were invited for a detailed interview. It was noted th a t 46% o f the final census o f T M Ps were not registered with a T M P association. N one o f the profita were registered, suggesting th a t the attem pts by Z IN A T H A to include them in their official registers had not m et with much success. The registration bias was relevant since profitas and n ’angas constituted different groups o f T M P. Profitas were younger and better educated, entered practice at a younger age, and had fewer years experience in practice. V irtually all were affiliated to an A frican church, m ost com m only the A postolic C hurch. The greater age o f n 'anga an d their clients (W inston & Patel, 1995) may reflect a change in health care-seeking patterns, from those based on a traditional w orld view favoured by an older generation o f T M P who are preferred by older com m unity m em bers to those rooted within contem porary syncretic C hristianity an d favoured by younger m em bers o f the com m unity. Both groups o f T M P shared com m on characteristics in their practices. The m ajority were full tim e an d often saw clients in the evenings an d at weekends; in doing so, they appeared to offer a practical and convenient service available at times when clients were better able to access them . T he average T M P reported seeing only three to fo u r patients daily. Profita tended to charge lower fees and reported themselves as busier than n ’anga, two findings which m ay be linked. M any T M Ps referred patients to clinics. T he predom inant reason given for referral was physical illness. M any o f those interviewed wished for closer relationships with the biom edical sector, either through easier and m ore personal referral systems or through opportunities to share o r transfer care between the traditional and biom edical sectors. The survey concluded th a t T M P s were

22

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

num erous, accessible a t hours convenient to clients, fam iliar w ith a range o f problem s, aw are o f cu rren t health concerns, an d favourably disposed tow ards biom edical care (W inston et al., 1995).

SU M M A RY T he universalist and culturally relativist approaches to cross-cultural psychiatry need to be integrated to produce research which is n o t only com parable across cultures b u t also sensitive to the local culture. Such research needs to be linked to practical clinical goals. Integrating ethnographic and epidem iological m ethods offers the poten tial o f achieving the goal o f a culturally sensitive psychiatry. C om m on m ental disorders (C M D ) are am ongst the m ost frequent an d disabling o f hum an afflictions. C M D include a wide range o f distress states, often presenting w ith nonspecific som atic com plaints an d having p rom inent features o f anxiety an d depression. Epidem iological studies o f m ental illness in prim ary care and com m unity settings in S ub-S aharan A frica have show n th a t m ental illness can be identified with psychiatric interviews in a b o u t a q u a rte r o f attenders at prim ary care clinics. E thnographic studies o f m ental illness have tended to show th at, although m ental illness is recognised as a distinct entity from physical illness, the concept o f m ental illness is m ostly related to acute an d severe behavioural disturbance an d often excludes the distress states subsum ed u n d er C M D . S up ern atu ral forces such as w itchcraft are one o f the com m onest causal m odels for CMD. Z im babw e is one o f the youngest independent n ations in Africa. Its people consult b o th biom edical p ractitioners an d T M P for health problem s. Psychiatric services c an n o t even cope w ith the dem ands placed by those w ith severe m ental disorders; thus, C M D are alm ost entirely seen an d m anaged by p rim ary health w orkers who are m ainly nurses in prim ary care clinics and T M Ps. Research is needed to describe the concepts o f C M D , the indigenous m odels used to describe an d explain C M D , the sym ptom s which best identify C M D , and the prevalence and risk factors o f C M D in biom edical an d traditional medical attenders. T h ro u g h the use o f culturally sensitive m ethods, such research m ay achieve the tw in goals o f com m unicating n o t only with academ ic researchers b u t with those w ho are in the frontline o f prim ary health care in Zim babwe.

CHAPTER TWO

The studies

OBJECTIVES The overall objectives o f the studies in this book were to obtain a description o f C M D in prim ary care attenders using m ethods which integrate ethnographic and epidemiological techniques in the study o f psychological disorders. The studies aimed to explore the contribution o f culture to the presentation, assessment, classification, and risk factors for C M D in prim ary care in H arare. The specific objectives were: (1) to describe the concepts o f mental illness o f carers in comm unity and prim ary care based settings; (2) to describe the sym ptom s and explanatory models of patients with conspicuous psychiatric m orbidity (i.e. prim ary care attenders whom their care providers feel have a C M D ); (3) to develop an indigenous measure o f C M D which could be used as a case finding instrum ent for future epidemiological investigations; (4) to examine the relationship between indigenous and biomedical models o f mental illness; (5) to determine the prevalence, associations, and risk factors o f C M D in primary care attenders. 23

FIG. 2.1 A n overview of the studies 24

2. THE STU D IES

25

THE ETHNOGRAPHIC STU DY1 Aims T he aims o f this study were to describe the concepts o f m ental illness held by a diverse range o f prim ary care providers in H ara re and to define screening criteria for distress states which approxim ated C M D fo r use by health care providers to identify cases for the next study.

Sample F o u r groups o f care providers were recruited: 22 T M P s recruited through T M P key inform ants; 9 o u t o f 11 com m unity psychiatric nurses (C PN ) w orking in the C ity o f H arare PH Cs; 30 o f the 36 village health w orkers (V HW ) from the periurban settlem ent o f Epw orth; and 15 relatives (R E L ) o f patients attending psychiatric clinics a t the Parirenyatw a H ospital in H arare.

Data collection F ocus group discussions (F G D ) were selected as the m ethod o f d a ta collection. A focus g roup questionnaire was designed based on clinical experience and previous ethnographic research suggesting that, although m ental illness was recognised as a category o f illness by the com m unity and T M Ps, it was often equated with severe behavioural disturbances akin to psychotic disorders. Since we wished to generate emic d a ta to com pare w ith biom edical concepts o f C M D , the F G D were introduced by inform ing the care providers th a t the F G D were n o t referring to severe m adness (kupenga) alone b u t any o f the illnesses o f the spirit/soul (mweya) and m ind/thinking (pfungw a) and to the problem s o f those patients w ho presented with illnesses which were n o t related to a physical illness. The key concepts which were covered in the F G D s were: the m eaning o f the term m ental illness; the site o f the “ m ind” in the body; the functions o f the “ m ind” ; the types and causes o f m ental illness; the effect o f m ental illness on the sufferer; the im pact o f m ental illness; and the types o f health care which were appropriate. A fter covering these concepts, three case vignettes describing typical cases o f C M D were presented. A 40-year-old woman with depression and suicidal ideas. F o r a few m onths, a 40-year-old w om an has been looking very sad, m iserable, and

' See Patel, M usara, M aram ba, and B utau (1995a).

26

CULTURE AND C M D IN SU B-SAHARAN AFRICA

unable to look after her home and children, slow in speech and movements. She says that life is not w orth living. N othing seems capable o f cheering her up. She does n ot eat or sleep well and lies on a bed for days w ithout doing anything. Once she even tried to take her own life. A 34-year-old man with panic attacks and agoraphobia. A 34-year-old m an has been unable to use public transport for the past m onth because he feels unwell and frightened to do so. He used to go shopping with his wife b u t now feels uncom fortable in markets. Crowds make him break out in a sweat and he feels tense and panicky. When this happens, he feels like som ething terrible is going to happen and now he spends m uch time indoors. A 38-year-old woman with multiple unexplained somatic symptoms. A 38-year-old woman has been complaining o f body aches and pains, especially headaches, crawling sensations in her skin, chest discomfort, tiredness, and backache. Physical tests and exam inations do not show any sign of a physical illness. G roups were asked to consider for each vignette the following: w hat, if anything, was the problem in this case; w hat were the causes o f this problem ; w hat should the person do about it?

Focus group procedure A total o f nine F G D were held with an average o f seven to nine participants. The com position o f each F G D was determined by the type o f carer concerned; thus, groups consisted o f T M P or R EL o r C PN o r VHW . The three F G D with VHW were conducted outdoors on a large m onolithic rock in Epw orth which was the site o f their regular meetings. Two o f the three F G D with T M P were conducted in the homes o f T M P while the third was in the medical school. The remaining F G D (one with CPN and two with REL) were conducted in the medical school. All F G D were conducted by persons familiar to the carer group; thus, the F G D with T M P were conducted by a TM P; the F G D with the VHW were conducted by a nurse who was involved in coordinating their activities as p art o f the medical school Family H ealth Program m e; the F G D with C PN and R EL was conducted by a psychiatric nurse. All F G D were in Shona.

Data analysis The author was present at all FG D and recorded the proceedings on tapes; these were later transcribed and translated to English for further analysis. D ata analysis was done separately for the four carer groups. Analysis

2. THE STUDIES

27

involved exam ining the d a ta for them es an d categories which were related to the questions posed. T hem es which recurred across F G D s were o f special interest because they were potentially m ore representative. A fter ordering d ata according to the questions posed, d a ta were categorised based on responses w ith sim ilar characteristics and possible associations between d ata were explored (Varkevisser, P ath m an ath an , & Brownlee, 1991).

Link to next study A series o f guidelines which could serve as screening criteria for C M D were draw n up using m aterial from the F G D . T hus, persons w ith a C M D could include: 1. Patients com plaining directly o f m ental illness categories, such as m adness o r kupenga and thinking to o m uch o r kufungisisa. 2. T hose with illnesses which may be related to m ental illness, such as m uroyi o r w itchcraft. 3. T hose w ith specific causes o f m ental illness such as mbanje or cannabis and alcohol abuse and head injuries. 4. T hose with relationship problem s such as m arital an d sexual problem s. 5. T hose with social problem s an d life events such as unem ploym ent an d bereavem ent. 6. T hose with illnesses th a t were perceived to arise either in the head (musoro) o r the heart (moyo). 7. T hose with typical psychiatric presentations including: expressions o f sadness, tearfulness o r suicidal ideas; fear, apprehension o r co n stan t worrying; unexplained m ultiple physical sym ptom s; acute attacks o f fear with physical sym ptom s such as a rapid heart beat.

THE PHENOMENOLOGY STU D Y 2 Aims T he aim o f this study was to record the sym ptom s and explanatory m odels o f prim ary care attenders who were considered to suffer from a m ental illness by care providers (“ conspicuous psychiatric m orbidity”). An additional objective was to translate and field test the Revised Clinical Interview Schedule, a standardised psychiatric interview

2 See Patel, G w anzura, Simunyu, Lloyd, and M ann (!995d).

28

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

Study design C ross-sectional survey psychiatric m orbidity.

o f prim ary

care

attenders

with

conspicuous

Site T hree PH C s in three suburbs and fo u r T M P s selected on recom m endation from key inform ants (including the district T M P organisation secretary) on the basis th a t these T M P s were locally well know n and had sufficient patients to allow the research to be com pleted within the limited time available. T he fo u r T M P s included three n ’angas an d one profita.

Sample T he care providers were asked to select patients from consecutive o u tp atien t attenders who had consulted them for illnesses which they thought were presentations o f a m ental illness. T he care providers used b o th their clinical judgm ent an d the screening guidelines from the E thnographic Study in the selection o f patients.

Instruments Tw o instrum ents were used: the E xplanatory M odel Interview (E M I) and the Revised Clinical Interview Schedule (CISR). E M I. T his is a sem istructured interview developed to elicit explana­ tory m odels o f illness which incorporates qualitative and q u antitative items. T he E M I used in this study was an adapted version o f the S hort E xplanatory M odel Interview (Lloyd et al., 1996) which, in turn, is based on K leinm an’s (1980) suggested questions for eliciting explanatory models: • W hat do you call your problem? • W hat nam e does it have? • W hat do you th in k has caused your problem ? • W hy do you think it started when it did? • W hat does your sickness do to you? • H ow severe is it? W hat do you fear m ost a b o u t your sickness? • W hat are the chief difficulties your sickness has caused for you? Few problem s arose in the translation o f the E M I to Shona since m ost o f the questions contained no em otional o r psychological term s. Indeed, the open natu re o f the questions m eant th a t the first back translation from the Shona version resulted in a version conceptually sim ilar to the original interview. W ith advice from a team o f bilingual indigenous professionals

2. THE STUDIES

29

(including tw o nurses and a m edical anthropologist), som e items were adapted: for exam ple, the item a b o u t perceived origin o f illness (i.e. w hether a person felt their distress had an em otional com ponent) was adapted since the closest conceptual equivalent to “ em o tio n al” would apply to distress which affected b oth the “ m ind” (pfungwa) as well as the “ soul” (mweya) (M utam birw a, 1989). A further seven closed items regarding causal m odels were added after piloting because it was ap paren t th at m any patients gave “ d o n ’t know ” responses to open questions a b o u t perceived causes. T his was, in p art, due to em barrassm ent caused by stating supernatural o r nonm edical causes to an interviewer who was from a medical background. Seven causal m odels were used fo r the closed questions, which were to be asked after the open question. These causes were based on those elicited in earlier studies on indigenous views o f causes o f m ental disorder (C havunduka, 1978; G elfand, 1967): (1) mudzimu, ngozi, mashave (i.e. ancestral spirits, aggrieved spirits, an d alien spirits, respectively); (2) m uroyi (witchcraft); (3) mamhepo (“ bad airs” , occasionally related to w itchcraft); (4) mbanje (cannabis); (5) alcohol; (6) nhaka (refers to heredity, but m ay imply a spiritual inheritance); (7) kufungisisa o r thinking to o much. The Revised Clinical Interview Schedule (C 1 S R ). T his is a revised version o f a standardised interview developed for clinical evaluation o f m ental state by G oldberg and colleagues (1970). The C IS R is sem i­ structured interview th at can be used by lay interviewers in com m unity and prim ary care studies o f C M D (Lewis, Pelosi, A raya, & D unn, 1992). The interview is com posed o f 14 key areas corresponding to m ajor com ponents o f nonpsychotic m ental disorders as recognised by biom edical psychiatry: som atic sym ptom s, fatigue, concentration, sleep problem s, w orry ab o u t physical health, irritability, w orry, depression, depressive ideas, anxiety, panic, phobias, obsessions, and com pulsions. Each key area contains one o r two m an d ato ry questions which determ ine w hether the particular sym ptom group was experienced in the previous week; if the patien t responds positively to the m an dato ry question, a series o f scoring questions rates the severity o f the sym ptom s in the previous week leading to a key area score. The total o f 14 key area scores com prises the total score (range 0-57) which is a m easure o f the severity o f nonpsychotic psychiatric m orbidity. A cut-off score o f 11-12 has been validated to determ ine casencss (Lewis et al., 1992). T his was the first time the C ISR was used in Zim babwe. The Shona version o f the C ISR was prepared th rough a series o f stages beginning with translation into Shona by two bilingual m ental health professionals (a psychiatric nurse and a psychiatrist). T he tw o Shona versions were presented to a third bilingual m ental health professional (a

30

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

psychiatric nurse) for back translation. Based on this back translation, item s from either version w ith the closest sem antic equivalence to the original text were selected. F o r som e items, neither translatio n was correct and the third tran sla to r provided a new version fo r th a t item. F o r som e items, b o th the versions were close to the original an d were b o th retained in the interview as alternatives. T he Shona version was now presented to five S hona health professionals (a medical an thropologist, tw o psychiatric nurses, one sociologist, and one T M P ). Sim ultaneously, it w as also presented to a Shona linguist with no previous m ental health experience fo r back translation. Em phasis was now laid on co ntent and technical equivalence. As a result o f the opinion o f the experts, som e item s were adapted. F o r exam ple, the item “ phobias while using public tra n sp o rt” was rew orded to inquire ab o u t fear whilst using “ emergency taxis” (a com m on form o f public transport).

Interview procedure T w o bilingual researchers w ho were psychology graduates conducted interviews under supervision by the au th o r. Both underw ent training including role play and interviewing psychiatric patients, an d special em phasis was laid on achieving com petence with conducting qualitative interviews and the use o f open-ended probes. A fter being selected by the care provider and providing consent, the patien t was interviewed by a research w orker. T he E M I was adm inistered first so th a t the closed questions o f the C IS R would n o t bias the reporting o f sym ptom s.

Data analysis Shona d a ta was translated to English for analyses. T he q u alitative d a ta generated from the E M I were converted into num erical codes following the m ethod o f stepwise reduction o f d ata to discrete categories (V arkevisser et al., 1991): 1. T he first 30 interviews were reviewed by the research team . Q ualitative d a ta for each item were collated and b ro ad them es identified; for exam ple, for the item “ reasons for co n su ltatio n ” , them es were “ aches and p ains” , “ o th er som atic co m p lain t” , “ supernatural com plaint” etc.; 2. These them es were allocated num erical codes and the 30 cases coded independently by tw o researchers. T he in terrater reliability o f the rating codes was estim ated. F o r items with k ap p as < 0.7, rating codes were refined and altered (see Table 2.1 for exam ples o f final codes); 3. Em ic sym ptom s were elicited by open-ended probing with the key question “ w hat docs your illness d o to you? F or exam ple, w hat does

2. THE STU D IES

31

TABLE 2.1 R a tin g c o d e s for the S h o n a explan atory m od el interview

Reason fo r consultation* 1. Aches and pains (e.g. headache, stom ach ache) 2. O ther som atic com plaints (e.g. difficulty walking, tiredness) 3. Specific som atic diagnoses (e.g. high BP, diabetes) 4. A utonom ic com plaints (e.g. palpitations) 5. Behavioural com plaints (e.g. sleep and appetite, aggression) 6. Psychological com plaints (e.g. thinking loo much) 7. M arital/fam ily problems (e.g. beaten by spouse) 8. Socioeconomic problem s (e.g. homelessness, hunger) 9. Supernatural/spiritual problem s (e.g. mudzimu, being bewitched) 10. N o t applicable/missing Worries and difficulties caused by illness 1. Ability to care for children/family 2. Ability to m aintain o r get employment 3. M arital difficulties (e.g. worry that illness may prevent m arriage) 4. Difficulties in other relationships (e.g. with in-laws, neighbours) 5. Econom ic difficulties (e.g. homeless, hungry) 6. W orry about sym ptom s/outcom e o f illness 7. Supernatural worry (e.g. bewitched) 8. O ther difficulties 9. N o t particularly w orried/no specific difficulties 10. M issing/not applicable * Similar facets were used for items on nam e o f illness, cause o f illness, and reason for timing.

it do to your body and m ind?” . Item s were enum erated and those which occurred in at least five cases were allocated independent codes. These items were coded dichotom ously (not reported/ reported) for the entire sample.

Links to the next study The Phenom enology Study was linked to the next stage by the idiom s elicited from patients which went on to form the prelim inary version o f an indigenous m easure o f C M D , the Shona Sym ptom Q uestionnaire (SSQ). Tw o causal m odels and labels for C M D were elicited frequently from patients in this study, viz., kufungisisa and sup ern atu ral problem s, particularly in relation to w itchcraft. The relationship o f b o th these causal m odels and C M D were to be exam ined in subsequent studies. The C ISR was found to have reasonable criterion validity in this setting. N early 75% o f patients with conspicuous psychiatric m orbidity were rated as cases on the interview. Some o f the C IS R items were rew orded following

32

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

the experiences o f the interviewers w ho reported problem s with the p atien ts’ understanding o f these items. The item s needed to be ad ap ted o r rew orded fo r different reasons. F o r exam ple, the m andatory question on obsessions was often confused with “ thinking to o m uch” ; in o rd er to distinguish the tw o, stress was laid on the distinction between th e tw o concepts by giving exam ples o f obsessional ideas. The Shona translation o f som e w ords needed to be adapted; for exam ple the literal translation o f “ co n cen tratio n ” could n ot be und erstood by m any patients while the sam e m eaning was adequately conveyed by a S hona w ord whose back translation to English was “com prehension” . Finally, som e questions needed two versions since cither could convey the sam e m eaning in case o f difficulties in u n d er­ standing one translation, for exam ple a question on hopelessness. In terrater reliability o f the Shona C IS R was estim ated by the m ethod o f observer co­ ratings, one research w orker interviewing the p atient and scoring the C IS R , the o th er research w orker independently scoring the C ISR . In this way 46 p atients in the sam ple were co-rated. K ap p a values for individual items ranged from 0.58 to 1.00 (average 0.79). T he lower kappas were for the last three key areas o f the interview, viz., phobias, obsessions, and com pulsions which were recorded infrequently.

THE SH O N A SYM PTO M QUESTIONNAIRE ST U D Y 3 Aims T he overall objective o f this study was to develop the Q uestionnaire, to estim ate the prevalence o f C M D in attenders and to exam ine the association between constructs, viz. kufungisisa an d supernatural causation constructs o f C M D .

S hona Sym ptom T M P an d P H C tw o indigenous with biom edical

Study design A cross-sectional survey o f prim ary care attenders was undertaken.

Site T he study site was tw o suburbs in western H arare, D zivarasekw a and M ufakose. These suburbs were allocated to the study by the City o f H arare H ealth D epartm ent. E ach suburb was served by a single P H C m anaged by the City H ealth D ep artm en t and staffed by nurses (26 in D zivarasekw a and 32 in M ufakose). All the nurses w orked prim arily as

5 See Patel et al. (1997b).

2. THE STUDIES

33

general medical nurses, though each clinic also had one C PN . D zivarasekw a had tw o private general practitioners and M ufakose three; neither suburb had a pharm acist. A cross-sectional com m unity survey in two suburbs (carried out by a team including the au th o r) identified a total o f 189 T M Ps (W inston et al., 1995) o f whom 35 were profita. Ten T M P s (five n ’anga and five profita) were random ly selected from a subsam ple o f all the T M Ps in the tw o suburbs w ho were consulted by at least five p atients a day (n = 21).

Sample size estimation The sam ple size was estim ated by pow er calculations (using the EPI6 softw are) based on the hypothesis th at the causal model o f kufungisisa was associated with the presence o f C M D in this p opulation. T his hypothesis was based on studies b oth by the a u th o r an d o ther researchers (A bas, B roadhead, M bape, & K hum alo-S akatukw a, 1994) which had suggested that this causal explanation was closely related to C M D . The phenom en­ ology study show ed th a t ab o u t 80% o f p atients with conspicuous psychiatric m orbidity stated th at their problem had been caused by kufungisisa. The hypothesis was th at this causal model was m ore associated with cases o f C M D as com pared to noncases in the attender p opulation. It was hypothesised th at the prevalence o f this causal explanation was 80% in the cases as com pared to 60% in noncases; pow er calculations estim ated th at the m inim um sam ple sizes needed to dem onstrate this hypothesis were 89 cases and 178 noncases (95% confidence and 90% pow er assum ing th a t the ratio o f noncases to cases was 2:1).

Sampling PH C patients were recruited by system atic sam pling o f consecutive attenders to the clinic (ratio 1:4). The exclusion criteria were: (1) patients attending the an ten atal clinic; (2) patients under the age o f 16 or over 65 years; (3) patients suffering from an acute medical illness. If a patient was excluded for any o f the above reasons, the next p atient in the queue was approached. Sam pling in the P H C to o k place on consecutive weekdays over a period o f a m onth. O n average, seven or eight patients were recruited daily. As there were fewer daily attenders at individual T M P clinics, all consecutive attenders becam e potential recruits. N one o f the T M P attenders were consulting for anten atal purposes n or did patients with acute medical emergencies consult T M Ps. T hus the only exclusion criterion that applied to T M P attenders was being under the age o f 16 or over 65 years. Sam pling periods at individual T M P s varied on the basis of num bers o f patients recruitcd from each T M P and the times at which the

34

CULTURE AND C M D IN SU B-SAH ARAN AFRICA

individual T M P was available for consultation. In all, sampling lasted six weeks and included weekends and evenings (in lieu o f two weekdays) when attendances were highest. On average, five patients were recruited daily.

Instruments The preliminary version o f the Shona Sym ptom Questionnaire (S S Q ). Shona idioms o f distress elicited in the Phenomenology Study were collated. A ltogether, 41 phenom ena were reported by at least five patients in the sample. The actual idiom for each type o f com plaint was identified from the original interview data. In the case o f six phenom ena, two idioms were comm only used to describe them, so both were included in the questionnaire. These six phenomena were: gait problems/difficulty walking, abdom inal ache/pain in the navel region, side achc/mabayo (a specific type o f pricking side pain), being easily startled/feeling panicky, losing interest in speaking to others/losing interest in things one cared about or enjoyed, and feelings o f hopelessness/suicidal ideas. The questionnaire was designed to be administered by a lay interviewer to patients with any level of literacy. Emphasis was placed on simple and clear wording. Questions asked about the presence o f symptoms during the previous week: patients were required to respond yes (score 1) or no (score 0). This form o f scoring is comm on in psychiatric screening questionnaires, for example the SRQ. The sum o f the question scores was hypothesised to reflect the severity o f psychiatric morbidity. The SSQ was then presented to five bilingual mental health and prim ary health care professionals to assess the clarity o f the wording and to check the face validity o f the items. Specifically, they were asked if the instrum ent appeared to be measuring mental illness and w hether it covered all the relevant domains. All items were seen to be relevant in the context that this was a preliminary questionnaire which was to be more thoroughly evaluated in the SSQ Study. The questionnaire was then piloted with 95 participants including community residents, PH C , and T M P attenders. The wording was found to be clear. As a result o f piloting, an item on seizures was dropped due to its rarity and the one item on gait problem s was dropped due to its similarity to another item. M any researchers who have developed culturally sensitive measures o f psychiatric disorder have recognised the im portance o f established psychiatric measures in developing new instrum ents and combined such items with those derived with emic techniques. Cheng and Williams (1986) state in relation to the development o f the Chinese H ealth Questionnaire, “ it is foolish to ignore a large body o f developmental w ork.” Following the same principles, items from the SRQ (H arding et al., 1980) were added to the 45-item prelim inary SSQ. The SRQ is the most comm only used case

2. THE STU D IES

35

finding instrum ent in A frica and the Shona version had been previously used in Zim babw e (H all & W illiams, 1987; Reeler et al., 1993). T he SR Q items were adapted to fit in with the form at o f the prelim inary SSQ, i.e. they were rew orded to ask ab o u t sym ptom s in the previous week. It was noted th at 9 o f the 20 items o f the SR Q were conceptually and sem antically sim ilar to item s on the prelim inary SSQ and were assum ed to be the sam e. T hus, an additional 11 items were added. F o u r items on positive m ental health derived from the Shona version o f the W H O Q uality o f Life Interview (K uyken, O rley, H udelson, & Sartorius, 1994) were random ly distributed in the prelim inary SSQ. These were included to generate a score o f positive m ental health to examine its relationship with SSQ scores and to detect any yea-saying bias. T hus the version o f the SSQ used in this study consisted o f 60 item s from three sources (see A ppendix 1). The Explanatory M odel Interview (E M I). A n abridged version o f this sem istructured interview was used. The following item s were included: (1) reasons for consultation; (2) the nam e o r label the patien t gave to their illness experience; (3) onset o f illness; (4) the p atien t’s view o f the source o f his/her illness, i.e. w hether the illness was purely som atic (arising from the muviri o r body) o r also involved the m ind (pfungwa) or soul (m weya); (5) the p atien ts’ view on w hether kufungisisa had caused their illness; (6) the p atients’ view on w hether su p ernatural factors such as w itchcraft had caused their illness. The last tw o questions were closed with a dichotom ous (yes/no o r d o n ’t know ) response form at. The rating codes for the first two items were reduced from those o f the original version described in T able 2.1. T he new rating codes were: som atic (including codes 1-4 in Table 2.1); psychosocial (including codes 5-8 in Table 2.1); spiritual/supernatural; n o t applicable/m issing. The care provider judgm ent. F o r each p atient recruitcd in this study, the care provider (PH C nurse o r T M P ) w ho had been consulted was interviewed with a question adapted from the health staff ratin g schedule used in W H O studies in prim ary carc (H arding ct al., 1980), viz., w hether the patien t’s presenting illness was mainly a physical illness, mainly an em otional or m ental illness, o r a m ixture o f both.

Interview procedure Tw o interviewers participated in the field work. B oth were psychology graduates w ho were bilingual in Shona and English. T heir training included role play and pilot interviews with the interview schedules. Both interviewers had earlier participated in the Phenom enology Study and were proficient in the use o f the C IS R an d E M I. A fter obtaining consent, each

36

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

patien t was seen by the respective care provider and then directed to the interviewers. O ne interviewer w ould elicit the prelim inary SSQ d ata, while the second interviewer (blind to the interviewer 1 d ata) w ould use the C IS R and interview the care provider. The two sets o f interviews were given in a ran d o m order.

Case criteria In the absence o f “ gold stan d ard s” to determ ine psychiatric casencss, the choice o f validating criteria is a difficult and controversial issue. In this study two sets o f validation criteria for the SSQ were considered: 1. Etic o r biom edical criteria: the Revised Clinical Interview Schedule (C ISR ) w ith a cut-off o f 11-12 as one which discrim inates cases best. 2. Emic criteria: The clinical ju dgm ent o f the care provider (a rating o f w hether the patien t suffered from an em otional o r m ental illness, though n o t necessarily excluding a coexisting physical illness). W hile the etic criterion offered the advantage o f stan d ard isatio n and reliability, there was a potential risk o f im posing a category fallacy. The latter lim itation was unlikely to be significant since the Phenom enology Study had show n th a t the C ISR had satisfactory criterion validity. W hile the emic criterion offered the advantage o f providing an indigenous m easure o f caseness, there was the problem o f the lack o f standardisation. Since the aim o f the SSQ was to m easure psychiatric m orbidity, it was felt th a t the criterion o f caseness had to be robust, and given the lim itations o f both the etic and emic criteria alone and the need to be able to relate emic and etic m odels o f illness, it was decided to define the “ gold sta n d a rd ” case criterion as those p articipants w ho scored 12 o r m ore on the C ISR and whom the care provider judged had a m ental illness. In addition to the above criteria, the criterion o f p atien ts’ assessm ent o f the em otional origin o f illness, as elicited from the E M I d ata , was also used as a “ self-assessed” criterion o f caseness.

Data analysis T he statistical procedure used to determ ine the items o f the prelim inary SSQ which best discrim inated for m ental disorder was discrim inant analysis with the gold stan d ard case criterion as the outcom e. Two prelim inary statistical steps were used to reduce the num ber o f item s to enter into a stepwise model for discrim ination. First, the chi-squared statistic for 56 items o f the instrum ent (excluding the 4 positive m ental health items) com paring cases and noncases were determ ined. F ro m the 56 items, 30 items with the highest chi-squared values (ranging from a value

2. THE STU D IES

37

o f 18 to 61, i > 1 and a significance o f at least 0.25 were also selected; all but two o f these items were already represented in the 30 item s selected using chi-squared tests. T hus a total o f 32 items (A ppendix 2) were selected for further analysis. These 32 items were then entered into a stepwise forw ard discrim inant analysis. T he items selected by this statistical procedure went on to form the final version o f the SSQ. T he item scores o f the SSQ items were sum m ed to generate an SSQ score for each patient. T he psychom etric properties o f the SSQ were then evaluated in the following ways: by com paring the m ean scores for cases and noncases according to different case criteria; by calculating specificity, sensitivity, positive and negative predictive values, and misclassification rates o f different cut-off scores with the gold stan d ard as the case criterion to estim ate the optim al cut-off score for caseness; by plotting receiver operating characteristics (R O C ) curves and com puting the area under the curve as an estim ate o f the overall discrim inating pow er o f the SSQ; by com paring SSQ scores with patient self-assessment o f the em otional origin o f their illness experience as a m easure o f convergent validity; by com paring SSQ scores with those o f the positive m ental health items (both individual items as well as the sum total o f the four items) as a m easure o f a divergent validity; and by com puting the intraclass correlation coefficient (C ro n b ach ’s alpha) and G u ttm a n n ’s split-half reliability as a m easure o f the internal consistency o f the questionnaire.

Link to the next study T he SSQ Study led to the developm ent o f a 14-item questionnaire with satisfactory discrim inating ability for identifying probable cases o f C M D . Its psychom etric properities revealed a high divergent and convergent validity and internal consistency. V alidity coefficients show ed optim al sensitivity and specificity for case detection a t a cut-off score o f 7-8, i.e. participants who scored 8 o r m ore were p robable cases o f C M D . T he SSQ was then used in the final study as a case detection measure.

THE C A SE -C O N T R O L ST U D Y 4 Aims T he objective o f this study was to exam ine the risk factors and associations o f C M D . The variables to be studied were sociodem ographic, 4 See Patel et al. (1997c).

38

CULTURE AND C M D IN SU B-SAHARAN AFRICA

economic, life events, alcohol use, disability, and clinical presentation including the p atient’s beliefs on kufungisisa and witchcraft and care provider diagnostic and treatm ent prescription behaviour.

Site The study took place in the same two high density suburbs as the SSQ Study, Dzivarasekwa and M ufakose. Both the PH Cs in the two suburbs, 4 o ut o f 6 GPs and 14 o ut o f 189 TM Ps in the two suburbs participated. The T M Ps were selected random ly from a sampling frame o f T M Ps in the two suburbs who reported at least five consultations a day (W inston et al., 1995).

Sample Potential participants for inclusion in the study were chosen by systematic sampling o f consecutive attenders. The sampling ratios differed at the three care provider sites due to the varying numbers o f daily attenders at each site; thus, daily attenders were highest at PH Cs where the sampling ratio was 1:4, while the sampling ratio at G P was 1:2. Since the average num ber o f daily attenders at T M P was only five, all T M P attenders were recruited. Patients were included in this study if they were aged between 16 and 65 years; they were excluded if they were tem porary visitors to H arare and thus unlikely to be available for follow-up interviews o r if they had an acute medical illness requiring immediate hospital referral. All eligible patients selected were given full inform ation about the study and only those who gave written consent for the study and follow-up were selected.

Definition of caseness All consenting patients were interviewed with the psychiatric screening measure, the Shona Symptom Q uestionnaire (SSQ). Those who scored 8 or more on the SSQ were classified as likely “cases” while those who scored 7 or less were likely “ noncases” or the controls.

Instruments A list o f variables collected from each subject is presented in Appendix 3. Clinical presentations and explanatory models. This interview elicited data on onset, presenting complaints, patient causal models and whether the patient believed that kufungisisa or witchcraft had caused their illness. Sociodemographic interview. family, and economic data.

This interview elicited sociodemographic,

2. THE STUDIES

39

Life event rating scale. This instrum ent was adapted from the Social Readjustm ent Scale used in earlier studies in Zim babwe (M yam bo, 1990). A series o f life events were inquired about in the previous 12 m onths. If the patient reported a particular event, the interviewer would probe to determine the em otional response to th at event, viz., whether the event was perceived as being stressful or not. A U D IT. The Shona version o f the Alcohol Use D isorders Identifica­ tion Test, a 10-item questionnaire developed in a m ultinational collabora­ tive study by the W H O (Babor & G rant, 1992; Babor, de la Fuente, Saunders, & G rant, 1992), was used for this study. Scoring guidelines recommend that scores o f 8 or more have the highest sensitivity for hazardous consum ption an d /o r recurrent intoxication. Disability. This interview began with an open question probing the way the illness had affected the patient’s life. Next, the Brief Disability Q uestionnaire (BDQ) (Von K orff et al., 1996), a Shona version o f which had been previously used in a W H O study on som atoform disorders, was used. This questionnaire generates a total score o f disability (range 0-22) plus two questions on the num ber o f days in the previous m onth in which the patient was unable to do his o r her regular work and was bedridden as a result o f the illness. The final com ponent was a three-point continuous measure o f the patients’ perception o f their overall quality o f life adapted from the Shona version o f the W H O Quality O f Life interview (Kuyken et al., 1994). Care provider assessment data. PH C staff and G Ps were asked to complete a clinical data sheet on each patient in the study which recorded the current diagnosis and current oral and injectable treatm ent. T M P were interviewed for each o f their patients to ascertain diagnosis and cause o f illness and their view on the role o f ancestral spirits, witchcraft, and kufungisisa in the illness.

Interview procedure A team o f four interviewers were involved in the field work. All were indigenous Zimbabweans, fluent in Shona and English, and from a social science or psychology background. Rigorous training in the use o f the interview schedule included role play and pilot interviews with psychiatric outpatients. Practice sessions continued until satisfactory interrater reliability was achieved (kappas for quantitative items >0.7). The interviewers were in the field for a period o f two m onths, one m onth in

40

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

each o f the tw o suburbs, and w orked concurrently in the different care provider sites. Screening with the SSQ took place while patients w aited to sec the care provider, while the m ain interview was random ly carried o u t either before o r after the consultation.

Data analysis Key sociodem ographic and clinical variables were com pared between the sam ples recruited from the three care provider sites. D a ta were then analysed in term s o f the study design, i.e. com paring cases with noncases. C hi-squared tests (with Y ates correction) were calculated for categorical variables. T tests and M ann -W h itn ey U tests were com puted for param etric an d no nparam etric continuous variables respectively. W here appropriate, simple odds ratios were com puted as well as odds ratios adjusted for sex, age and site o f recruitm ent.

CHAPTER THREE

Results of the studies

C O N C E P T S OF M E N T A L IL L N E S S OF P R IM A R Y C A R E P R O V ID E R S T his scction describes d ata from the E thnographic Study.

The sample A ltogether 9 focus group discussions (F G D ) were conducted involving a total o f 76 care providers. The dem ographic characteristics o f the participants were: 1. Village health w orkers (/j = 30); age m ean 46.9 years; SD 8.2 years. Sex: females = 26; males = 4; years o f experience: m ean 8.1 years; S D 1.1 years. 2. T raditional medical practitioners (n = 22); n'anga (n = 16) and profita (n = 6). Age m ean 47.4 years; S D 12.7 years. Sex: fem ales = 1 5 ; males = 7; years o f experience: m ean 17.2 years; S D 19.1 years (range 6 m onths to 71 years). 3. C om m unity psychiatric nurses (n = 9): age m ean 37.7 years; S D 4.4 years; sex: all female; years o f experience: m ean 14.3 years; S D 3.5 years. 4. Relatives (w = 15): age m ean 42.3 years; S D 7.9 years. Sex: females = 14; m ales = 1. 41

42

CULTURE AND C M D IN SU B-SAHARAN AFRICA

Psychiatric diagnoses am ong the patients: schizophrenia and manic depressive illness. All the participants spoke Shona. The m ajority subscribed to one o f the m any C hristian churches in Zimbabwe; the comm onest were M ethodist (« = 17), Salvation Army (n = 7), Apostolic ( n = l l ) , Anglican (« = 9), and R om an C atholic (« = 10). Two participants denied any religious affiliation. O ther than the nurses, only a handful of care providers were educated beyond the ‘O ’ level stage.

Basic concepts of mental illness M ental illness was recognised by all care providers as a category o f illness in which behavioural changes were characteristic. The person lost his or her ability to reason and to look after the family. The term kupenga was frequently used, literally translated as “ madness” . Such illness was similar to the biomedical concept o f acute psychosis. On probing abo u t other illnesses which affected the mind or soul, kufungisisa, mamhepo (bad airs, often associated with witchcraft) and pfukwa o r ngozi (angry spirits o f persons who were killed) were m entioned frequently. C PN s broadly defined mental illness as occurring when a person was behaving in a m anner which was unacceptable or contradictory to the norm s o f the com m unity one lived in. All the care providers agreed that the mind resided in the head region (musoro). M ost also pointed to the role o f the heart (m oyo) as being closely connected with the mind. Thus “when someone thinks a lot, the heart becomes heavy” and when someone stole “ it is the heart which decides it, and the brains work o u t how to do it.” Similarly, “ our thoughts originate from the heart and then move up to the m ind” and that “ if a person worries a lot, they can experience chest pain.” The CPNs all felt th at the mind was in the head and the brain, but also comm ented th at they were taught by the elders in the com m unity th a t the mind resided in the heart. The principal function o f the mind was to organise and plan one’s life and that o f one’s family. The mind determined right from wrong. While “ the mind is w hat makes a person do w hat they w ant to d o ,” it also imposed certain socially determined restrictions on behaviour. The mind was the “ actual person” ; a person w ithout a mind was like a corpse, because he or she could n ot think. The brain was a broader entity and had m any parts with varying functions, such as sensation and movements.

Types and causes of mental illness C PN s described types and causes o f mental illness similar to biomedical nosology, describing categories such as psychoses and depression. W hen asked about traditional concepts o f illness, their views resembled those o f

3. RESULTS OF THE STUDIES

43

the other care providers. T he types and causes o f m ental illness were closely related, the m ajority o f illness being caused by supern atu ral factors. S upernatural forces could cause m isfortune and illness fo r tw o reasons, either as a consequence o f acts com m itted by the p atien t o r his family or due to the actions o f o th er persons who used supern atu ral m eans to inflict misfortune. In the first group o f sup ern atu ral causes, if a person o r one o f his family broke a social tab o o , then spirits w ould be angered and cause illness. Spirits were o f m any types, the com m onest ones th o u g h t to cause severe m ental illness being ngozi and pfukw a. These were the spirits o f persons killed either by the p atient o r by one o f his ancestors. M udzim u (family ancestral spirits) an d shave (alien spirits) could also possess individuals. T he form er could cause m ental illness by aban d o n in g their role o f protecting their descendants due to their disrespectful conduct. Such possession needed to be distinguished from th a t which occurred when the spirits w anted to “ com e o u t” through a family m em ber as a sign th a t he or she was a spirit m edium ; in these cases, a n 'anga was consulted and he would clarify th a t the person was n o t really m entally ill. C om m on exam ples o f situations which could lead to illness due to angering o f ancestral spirits included: 1. Crim es com m itted either by the patien t or one o f his ancestors. These crimes varied from theft to m urder an d could have occurred several generations earlier. The continuing cycle o f illness and m isfortune could only be ended if adequate com pensation was given to the aggrieved family. 2. B reaking social rituals, e.g. mombe yeum ai in which a cow was to be given to the bride’s m other as roora (bride price) o r if a deceased person’s property was not distributed according to tradition. 3. A cquiring unnecessary charm s o r am ulets o r o th er traditional items for protection o r for acquiring wealth and the excess use o f “ love” potions o r mupfuhwira (used secretly by a person to m ake their spouse m ore affectionate and attentive) could anger the spirits. S upernatural m eans could also be used by others, such as relatives, neighbours o r friends, o r even strangers to inflict illness. T he com m onest m ethod was through w itchcraft (m u ro yi) or sorcery. W itchcraft (which was intrinsic to a person, to his or her soul o r personality) was distinguished from sorcery (which was extrinsic an d was merely a tool o r a technique employed under certain circum stances). These m ethods were com m only used as a result o f som eone’s jealousy a t the success o r prosperity o f the victim. W itchcraft could cause m ental illness in a variety o f ways, such as by the use o f evil spirits o r items such as zvivanda,

44

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

tokoloshi, and zvishiri (birds or anim als used in w itchcraft). Zvidhom a was a dead person who was raised either to bewitch som eone or to seek redress for unresolved grievances. C PN s felt th at, in som e instances, supern atu ral explanations helped p atients understand why they were suffering and w hat needed to be done to correct this. A variety o f oth er causes and types o f m ental illness were given. A lthough som e accounted for the illness as a direct consequence o f the causal factor (such as alcohol dam aging the brain), in m any instances (such as head injuries following a car accident), a supern atu ral factor was also sought to answ er the fundam ental question o f why the person had a car accident. “ Biological” causes o f m ental illness included head injuries; A ID S; epilepsy; chronic physical illness; drug abuse (especially cannabis); alcohol abuse especially abuse o f illicit brews such as kachasu or chikokiyem a; and old age. C hildren w ho suffered high fever and convulsions could grow up with m ental illness. “ Psychological” factors included: thinking to o much a b o u t problem s; bereavem ent o r sickness in close relatives; excessive reading and studying; relationship problem s, especially m arital conflict and divorce; an d loneliness. “ Socioeconom ic” factors included: poverty; infertility; and unem ploym ent. M any o f these causes could also be the result o f m ental illness. W hile it was recognised th at m ental illness ran in som e families, the cause was usually the passing on o f the spirit o f the m entally ill person. C ertain specific types o f m ental illnesses were also described (see T able 3.1).

Symptoms of mental illness A range o f sym ptom s were described (see T able 3.2). The essential features were behavioural changes and the loss o f the ability o f self-care. Som atic presentations were p artly due to the social stigm a attached to m ental illness. Sexual com plaints were difficult to elicit due to social custom s (e.g. the im propriety o f young nurses asking older p atients such personal questions) and the lack o f privacy in crow ded prim ary care clinics. In som e patients, sym ptom s w orsened during certain phases o f the lu n ar cycle.

Impact of mental illness T he m ost im portan t effects were on fam ily relationships. T he sexual relationship o f a couple could be im paired by an increased o r decreased sexual drive. W om en could becom e prom iscuous while men could force sex on their wives and become violent if they refused. In trying to determ ine reasons for the illness, family ties were often disrupted. F o r exam ple, when a w om an becam e m entally ill, it was often her family which was blam ed, leading to quarrels; som e care providers rem arked th at, in this respect, m ental illness was sim ilar to A ID S, in which blam e was passed aro u n d

3. RESU LT S OF THE ST U D IES

45

T A B L E 3.1 S h o n a c a te g o rie s o f m en tal illn e s s d e sc rib e d b y health care p ro vid e rs

C ategory o f illness Kupenga M am hepo M hengeramumba Kufungisisa Kutanda botso

Zvipotsw a Kushaya unhu Tsviyo or pfari

Biomedical equivalent

Description!comments Severe behavioural distu rb an ce w ith aggression, in ap p ro p riate a n d d isturbed behaviour Bizarre behaviour, headaches, sudden m utism , sitting m otionless for ho u rs M ild variety o f m ental illness which only m anifests in the hom e T h in k in g too m uch, excess w orry a b o u t problem s, exam s, etc. M ay ap p ear sad A cute a n d b rief behavioural disturbance follow ing breaking a m ajo r ta b o o (such as striking p arents) Person tries to h arm som eone by using w itchcraft, but fails and becom es m entally ill Person behaves in a m anner co n trad icto ry to com m unity norm s, e.g. is disrespectful Convulsive m ovem ents, often associated with behavioural problem s

Psychosis ?H ysteria 7 ?Anxiety o r depression ?Brief reactive psychosis

7 '.'Antisocial personality Epilepsy

from h u sb an d to wife an d so on. A ccusations o f w itchcraft could d isru p t fam ily ties. T he p a tie n t w as shu n n ed by o th ers a n d isolated from the com m unity an d often ended up hom eless. Since m any m en tal d isorders were incurable, there w as considerable expense in carin g fo r th e patient. T he financial burd en w as increased due to the cost o f m edication and co n su ltatio n s w ith priv ate d o cto rs an d n 'angas. T h e fam ily o f the m entally ill person w as laughed a t an d ostracised by o th ers such as neighbours. R elatives could them selves becom e m iserable as a result o f these stresses.

Sources of care for the mentally ill Both biom edical an d tra d itio n a l care p roviders played a role in the m anagem ent o f m ental illness. H o sp itals w ere especially useful for “ biological” o r “ n a tu ra l” causes such as head injuries and trea tm e n t o f physical problem s such as d e h y d ratio n . N 'angas were able to d iagnose the real cause o f th e illness, such as bew itchm ent, a n d in itiate treatm en t th a t w ould provide a definite a n d p erm an en t cure. Som e relatives felt th a t n ’angas helped the p a tie n t think m ore clcarly an d p ro tected him o r her from evil spirits. T h e trea tm e n t could include h erb al m edicines, steam bath s, scarification, o r in stru ctio n s to the p a tie n t an d fam ily to co rrect bro k en tab o o s o r crim es. F o r exam ple, if som eone h ad becom e m entally ill because o f stealing, he w as required to p lead w ith the o w ner an d com pensate him for the stolen goods. Profitas usually req u ired their

46

CULTURE AND CM D IN SUB-SAHARAN AFRICA T A B L E 3.2 Phenomena of mental disorder described by health care providers

Phenom enon B eh av io u ral

Im p a ired self-care M ood Speech

C og n itiv e

S o m atic P ercep tu al

T yp es o f disturbance a n d exam ples A ggressive: v erbal a n d physical D isin h ib ited : d isro b in g o r u rin a tin g in p ublic R isky: w an d erin g aw ay from h o m e, ru n n in g betw een c a rs, b u rn in g clo th es B izarre: e atin g o r sm earin g faeces, la u ghing in a p p ro p ria te ly , h ittin g oneself, w earin g several layers o f c lothing, h o a rd in g rub b ish A g ita tio n a n d restless b eh av io u r R e tard atio n : lying im m obile, s tarin g in to space Sleep: increased o r decreased L oss o f a p p etite a n d w eight loss Sexual: “ b a tte ry is flat” ; “ n o th in g is h a p p en in g in b e d ” N o t w ashing, e atin g d irty fo o d , w earing to m clothes, being u n a b le to lo o k a fte r fam ily Irrita b ility Sadness Increased speed Irre lev a n t nonsensical speech B ecom ing m u te T h in k in g to o m uch F o rg etfu ln ess B ecom ing suspicious, e.g. th a t fo o d is poiso n ed H ead ach es, p a lp ita tio n s, w eakness, ch est p a in s, infertility in w om en, high b lo o d p ressu re, b lack o u ts, g ait d isturbances. H e arin g voices Seeing th in g s o th e rs c a n n o t

p a tie n ts to b elo n g to th e ir c h u rc h a n d used a ra n g e o f tre a tm e n ts in clu d in g em etic p o tio n s, p ra y e r, h o ly w a te r a n d b itin g to rem o v e fo reig n bo d ies. C P N s c o m m e n te d o n th e p a rtic u la r difficulties o f w o rk in g in b u sy clinics w ith little tim e o r p riv a c y to g ain th e p a tie n t’s co n fid en ce a n d in q u ire a b o u t p e rs o n a l a n d fam ily issues w hich w ere re lev a n t to th e ir p ro b lem s. T h e y accep ted th a t m a n y fam ilies c h o o se a fo rm o f c a re b ased o n th e ir beliefs re g a rd in g th e illness a n d th a t m a n y fam ilies co n su lte d m o re th a n o n e c are p ro v id e r. F o r ex am p le, a lth o u g h th e sy m p to m s o f a sc h iz o p h ren ic p a tie n t w ere c o n tro lle d w ith an tip sy c h o tic m e d ic a tio n fro m a clinic, th e fam ily w o u ld also ta k e him o r h e r to a n ’anga to p e rfo rm necessary ritu als.

C ase 1: A 4 0 -year-old female with depression A ll th e ca re p ro v id e rs id entified su ch p e rso n s labelled th is as kufungisisa. M a n y felt th a t th is re a c tio n to severe life stresses. A s o n e R E L w o m a n is ill . . . she is suffering fro m w o rry .”

in th e ir c o m m u n ity . S o m e w as n o t a n “ illness” b u t a said “ I d o n o t th in k th is P o v e rty , m a rita l p ro b le m s

3. RESULTS OF THE STUDIES

47

such as violence by the husband, bereavement, kufungisisa, too many “ life problem s” , illness in the family, laziness, being bewitched, excess o f love potions, lack o f friends, and not sharing problems with other people were identified as possible causes. The thought o f wanting to kill herself could have been inherited from her ancestors, one o f whom may have comm itted suicide. CPNs identified this vignette as depression and comm ented that such patients often presented with sleep problem s or following an overdose, often o f antimalarials. M any VHWs felt th at she did not need medical treatm ent but would benefit by socialising in women’s clubs or church groups and sharing her problems with friends. H er husband needed education and the family could be helped with their financial problems. N ’angas could help by getting rid o f the evil spirits responsible for the wish to w ant to kill herself. Some RELs put their trust in biomedical treatm ents especially those who had suffered similar symptoms and failed to respond to traditional treatm ent. Visiting a church and talking to a cleric could also be helpful. CPNs said that many patients were taken initially to TM Ps and only those who failed to respond consulted at clinics.

Case 2: A 34-year-old male with agoraphobia and panic disorder M any care providers identified such persons in their comm unity, including some who had fought in the Liberation W ar and others who had suffered accidents (including one o f the VHWs). Road accidents were the most im portant cause; thus, the patient could imagine that another accident would happen causing him to fear getting in a car. In some cases, the patient or his ancestor may have killed someone in a car accident and did not make amends, leading to possession by the ngoii (aggrieved spirit) o f the victim. Kufungisisa, bereavement, infertility, divorce, physical illness, failures in life, and mamhepo were other causes. The patient could be possessed by ancestral spirits who were unaccustomed to cars and the smell o f petrol. He could be bewitched by people who were jealous when they saw someone working and doing well. He could have many women friends who gave him love potions. A person with this problem could go for a whole year w ithout leaving his house. Several treatm ent approaches were advocated: support and reassurance; encouragem ent to make trips in a car beginning with short distances and gradually increasing it; and being taken by friends to crowded places such as football games and being reassured that there was nothing to fear. Some TM Ps said that he should get permission from the spirits before boarding a bus or car. He could also be treated with bute (snuff) and rituals such as using a fowl to carry away the aggrieved spirits.

48

CULTURE A N D C M D IN SU B -SA H A R A N AFRICA

Case 3: A 3 8 -year-old female with unexplained somatic symptoms M any care providers not only knew such patients b u t som e adm itted th at they themselves fitted the description. Since the doctors failed to find a cause, som e R E L s felt th at the m ost likely reason for this w om an’s problem s were to do with her ancestors. Possession by mudzim u, laziness, old age affecting muscles and veins, infertility, kufungisisa, mamhepo, and being bewitched were identified as possible causes. A variety o f spiritual treatm ents used by n 'angas and profitas were ap p ro p riate for such patients. C P N s felt the m ain problem was anxiety and th a t careful questioning often revealed sym ptom s typical o f psychological distress.

S Y M P T O M S AN D EXPLANATORY M O DELS OF C M D T his section describes d a ta from the Phenom enology Study.

The sample P rim ary care providers selected 110 p atients from consecutive o u tp atien ts on the basis o f their clinical assessm ent th a t th e patien t had an em otional disorder (i.e. conspicuous psychiatric m orbidity) and 109 com pleted b oth the C IS R and E M I (1 patien t com pleted the E M I only). D ata were analysed for the sam ple as a whole, and for P H C (n = 53) and T M P cases (n = 57) separately. The m ean age o f the sam ple was 32.5 years (SD 11.7, range 15-70). Fem ales com prised 65% o f the T M P an d 85% o f the P H C sam ple; 57% were living with a p artn er, 14% were divorced o r separated and 4.5% were widowed. 85.5% had n o t com pleted school an d 82% were n o t in full-tim e em ploym ent; unskilled w ork such as selling vegetables was the m ost com m on type o f em ploym ent (7% ). O f the sam ple 80% subscribed to one o f m any C hristian churches operating in the area o f which the A postolic C hurch had the largest following (24% ). The rest o f the sam ple denied any religious affiliation.

Explanatory models Reasons fo r consultation. U p to three reasons for consultation were recorded for each patient. The m ajority o f cases presented a som atic com plaint, e.g. aches and pains (58% ) o r stated specific som atic diagnostic labels, e.g. high blood pressure (15% ) o r a nonspecific som atic com plaint, e.g. dizziness (39% ). O th er reasons stated were psychological (e.g. thinking too m uch), m arital com plaints, socioeconom ic com plaints an d super­

3. RESULTS OF THE STUDIES

49

n atural com plaints. S upernatural com plaints, e.g. “ my m udzim u has been blockcd” , were m ore com m on in the T M P g roup (see T able 3.3). Nam e o f illness. R espondents were asked w hether they had a particular nam e for their illness. Specific som atic diagnostic labels, e.g high blood pressure, were com m oner in the P H C group while su p ernatural labels, e.g. mamhepo o r being bewitched, were com m oner in the T M P group (see T able 3.3). O ther labels used were psychological labels such as kufungisisa, nonspecific som atic nam es, e.g. headache, and relationship problem s, e.g. infidelity. Onset o f illness. M ore th an h alf the sam ple had a chronic illness ( > 1 year) with 42% suffering m ore th an two years. A cutc presentations were com m oner am ongst P H C attenders while tw o-thirds o f T M P attenders had a chronic illness (Table 3.3). Parts o f body affected. R espondents were asked to nam e up to three parts o f the body which they felt were affected by their illness. T he head was quoted by over h a lf the sam ple. O th er p arts com m only cited were the TABLE 3 .3 C o m p a riso n o f exp lan ato ry m o d e ls b etw een T M P an d P H C attenders w ith c o n s p ic u o u s p sych iatric m o rb id ity

Variable

PHC (%>)

Reason for consultation Supernatural N am e o f illness Somatic Supernatural O nset o f illness > 1 year Origin o f illness Body mainly Soul mainly M ost im portant cause o f illness Somatic Supernatural Perceived causes Mamhepo (“ bad airs” ) Angry ancestral spirits R eason for tim ing o f illness Som atic Supernatural Prognostic sssessment F atal outcom e o r suicide if untreated

TMP (% )

y2 (df= 1) two-tailed

0

15

23 7.5

3.5 39

36

66

9.2;