Background Traditional healers provide a popular and accessible service across the African continent. Little is known of the characteristics or mental health status of those using these services.
Aims To determine and compare the prevalence of common mental disorder among, and the characteristics of, those attending primary health care clinics (PHCs) and traditional healer centres (THCs) in Dar-es-Salaam.
Method The Clinical Interview Schedule – Revised was used to determine the prevalence of mental disorders in 178 patients from PHCs and 176 from THCs, aged 16–65 years.
Results The prevalence of common mental disorders among THC patients (48%) was double that of PHC patients (24%). Being older, Christian, better educated, and divorced, separated or widowed were independently associated with THC attendance. None of these factors explained the excess of mental disorder amongTHC attenders.
Conclusions The high prevalence of mental disorders among THC attenders may reflect the failure of primary health care services adequately to detect and treat these common and disabling disorders. Traditional healers should be involved in planning comprehensive mental health care.
The categories of mental disorders recognised in Europe and America are also found in Africa ( Giel & Van Luijk, 1969; Woeber, 1975; Ndetei & Muhangi, 1979; Odejide, 1979; Dhadphale et al, 1983; Ben-Tovim, 1985a; Jacobson, 1985; Westermeyer, 1985; Bondestam et al, 1990). Stock ( 1995) identified 14 studies from sub-Saharan Africa of mental disorders in primary care or general population settings. The neuroses were the most common disorders, depression and anxiety the most common diagnoses. By analogy with the West, one might expect these disorders to present often in primary care. In most African countries primary care is provided by traditional healers as well as by biomedical practitioners, the former often being more accessible ( Ben-Tovim, 1985b; Swantz, 1990; Patel et al, 1995). Knowledge of traditional healing practice and of the use of these services by people with common mental disorders is limited and impressionistic. This paper describes a study in Tanzania comparing the prevalence of common mental disorder among attenders at primary health clinics and traditional healer centres in Dar-es-Salaam. Our hypothesis was that the prevalence would be higher in the traditional healer centres than in the clinics, given that the only previous study similar to our own, in Zimbabwe, had recorded a particularly high prevalence of mental disorders among users of traditonal healers’ services ( Patel et al, 1997). In an exploratory analysis we sought to explain any observed difference in prevalence between the two settings by the characteristics of users of the two services.
This was a cross-sectional study of patients in two health care settings. Ethical approval was obtained from the research department of the Tanzanian Ministry of Health.
The study was conducted in Ilala district of Dar-es-Salaam, in Tanzania. Two types of health care providers were involved: primary health clinics and traditional healer centres.
Primary health clinics
Mnazi Mmoja health centre and Amana district hospital were selected. These were the only local biomedical providers, where medical assistants and nurses manage patients. Patients can choose freely which of the two to attend.
Traditional healer centres
In traditional healer centres patients are managed by herbalists, diviners, herbalist-ritualists and faith healers. Traditional healers in modern Tanzania are known as fundi (engineers or technicians), but were formerly known as waganga wa kienyeji or waganga wa jadi (traditional practitioners or doctors). They can be grouped into four categories, but with some overlap.
Diviners: these include traditional diagnosticians (wapiga ramli), diviners (ramli) and spiritualists (a mashetani, midzimu); these healers consult with spirits who may identify the type and cause of the illness. Diviners may treat accordingly or refer on to herbalists. They will differentiate between normal health problems (magonjwa ya kawaida) such as cancer, diabetes or acquired immune deficiency syndrome (AIDS) and traditional health problems (magonjwa ya kienyeji) that involve the control or removal of spirits.
Herbalists: these use plants and roots as medicines typically applied through scarification, steam baths, and mineral and animal extracts.
Herbalists-ritualists: these use both rituals and herbal medicines to diagnose and treat, in addition healing the specific spirits deemed responsible for a patient’s problem.
Faith healers: these use Koranic phrases (for example, kombe – a phrase from the Koran written on a piece of paper and given to a patient for treatment or protection from specific and non-specific misfortunes), or recite texts from the Bible for healing purposes. This group sometimes uses herbal medicines as well.
Traditional healers in Tanzania are typically inducted through one of four routes ( Gessler et al, 1995): inheritance within a family kinship; ancestor-spirits (midzimu) contacted through dreams; the experience of having an illness cured by traditional medicine; a personal decision, followed by a period of apprenticeship.
Normally there is no special procedure to see traditional healers. Patients refer themselves. Some are referred by one healer to another, or transfer themselves. Tanzanians commonly seek biomedical and traditional treatments simultaneously or alternately, when one or the other fails to produce quick results.
In 1995 Tanzanian traditional healers formed an association called Chama cha Waganga na Wakunga wa Tiba Asilia Tanzania (CHAWATIATA), the Tanzania Traditional Health Practitioners Association. The work of registering healers has begun through CHAWATIATA but no comprehensive, formal register exists to date. For the purposes of this study, eight healers (two herbalists, two herbalist-ritualists, two diviners and two faith healers) were selected randomly from a sampling frame of traditional healer centres drawn up by the district CHAWATIATA chairman. This was constructed to include only those (the majority of healers) who had affordable consultation fees and a sufficient number attending daily to allow the study to be completed quickly. Fees for treatment depend on the illness or problem; for AIDS, cancer, diabetes and other illnesses that involve the control or removal of evil spirits, the fee is normally high. Njenga ( 2002) reported a growth of traditional healer practice in Dar-es-Salaam, mainly less well apprenticed healers. Conversely, traditional healers seemed to be becoming less common in rural areas. The rural phenomenon of payment in kind is not common in urban areas, as the healers operate in a cash economy and need to generate income. Patients usually pay for diagnosis and treatment separately.
Sample size was estimated to give 80% power to detect a hypothesised difference in the prevalence of common mental disorder of 15% or greater between attenders at primary health clinics (among whom we predicted a prevalence of 35%, after Gureje et al, 1992) and traditional healer centres (a predicted prevalence of 50%, although two-sided tests were used throughout). Under these assumptions, a sample of 182 participants in each setting was required. The aim was to achieve a representative sample of clients for each service, by sampling from consecutive attenders. Sampling ratios at the two settings differed owing to variation in the number of people attending each day. A sampling ratio of 1:4 was used at the primary health clinics. The daily case-load at the traditional healer centres was lighter so all those attending were invited to participate. All those aged 16–65 years, who were resident in the defined area of study and who spoke Swahili were considered eligible. Only those who suffered from an acute medical illness of a severity that rendered interview impractical were excluded.
Gender, age, religious affiliation, marital status, educational attainment, number of children, occupational status (employed or unemployed) and socio-economic status (accommodation type, whether cash saving or in debt, and ability to buy food in the past month) were recorded through completion of a questionnaire.
Explanatory model interview
The Short Explanatory Model Interview (SEMI; Lloyd et al, 1998) was used to elicit patients’ attributions regarding their presenting complaints, their previous help-seeking behaviour and their expectations regarding the index consultation. The SEMI, which combines open-ended questions with a structured coding frame, has been used successfully in a variety of countries and cultures, including southern Africa ( Patel et al, 1995).
Standardised psychiatric interview
A fully structured clinical interview, the Clinical Interview Schedule – Revised (CIS–R; Lewis et al, 1992), was used to elicit information on symptoms of common mental disorder during the week preceding interview. A previously validated cut-off score of 12 or more was used to identify those with a common mental disorder ( Lewis et al, 1992). At the end of the assessment ICD–10 diagnoses ( World Health Organization, 1992) were generated from CIS–R data using the Programmable Questionnaire System (PROQSY; Lewis, 1992). The CIS–R was designed for use in primary care in the UK and has not been used before in Tanzania. However, it has proved to be a feasible and valid instrument for the detection of common mental disorders in Zimbabwe ( Patel & Mann, 1997) and Sri Lanka ( Wickramasinghe et al, 2002), and has also been used successfully in India ( Patel et al, 1998), Taiwan ( Liu et al, 2002), Chile ( Lewis et al, 1992) and Thailand ( Silpakit, 1997). It was translated into Swahili by M.C.N.; three bilingual professionals unacquainted with the original English version completed its back-translation. On the basis of the back-translated text and the original version, M.C.N. then developed a consensus version in Swahili with advice from an independent Tanzanian psychiatrist. Content and technical equivalence were examined during the process of translation to ensure that the original content had not been significantly distorted. The interrater reliability of the Swahili CIS–R was estimated before data collection by observer co-ratings of 20 patients (M.C.N. and research assistant), one interviewer interviewing the patient and scoring the CIS–R, the other independently rating. The mean kappa value across all CIS–R items was 0.76 (range 0.59–1.00).
Data were analysed with the Statistical Package for the Social Sciences, version 10.0 (1999). We compared the prevalence of common mental disorder (CIS–R score 12 or more) between primary health clinic and traditional healer patients, and compared the characteristics of the two groups. Pearson’s chi-squared test was used to compare categorical variables, and a chisquared test for trend was used to compare ordered categorical variables. All significance tests were two-tailed. We sought to identify independent predictors of use of one service or the other using logistic regression with the type of service as the dependent variable. In a further logistic regression model, with common mental disorder as the outcome, we then adjusted the association between attendance at a traditional healer centre and common mental disorder for any independent predictors of traditional healer attendance. Finally, we compared participants in the two settings for attributions regarding the presenting problem, for its chronicity, for the extent of previous consultations with biomedical providers for the same problem, and for ICD–10 diagnoses and symptom profiles.
Altogether 182 attenders were approached in each setting. Of these, 178 at the primary health clinics and 176 at the traditional healer centres agreed to participate. The response rate was therefore over 96% in both settings.
Prevalence of mental disorder and patient characteristics
Forty-two of 178 primary health clinic patients (24%) and 85 of 176 traditional healer patients (48%) scored 12 or more on the CIS–R, and were thus classified as having a common mental disorder. The 95% confidence intervals around the observed difference in prevalence of 25% were 15% to 34% (χ2=22.408, d.f.=1, P<0.001). Women outnumbered men in both settings ( Table 1). Attenders at the traditional healer centres were older and better educated than the primary health clinic patients. They were more likely to be employed, to be saving money and also to be in debt. They were more likely to be Christian, to be divorced or separated and to have children. In a logistic regression analysis, being older, Christian, better educated and divorced, separated or widowed were all independently associated with traditional healer centre attendance. After adjusting for these patient characteristics the association between attendance at a traditional healer centre and common mental disorder (crude odds ratio 3.03, 95% CI 1.92–4.76) was little changed (adjusted odds ratio 2.99, 95% CI 1.77–5.05).
Further exploratory analysis revealed large differences in the pathways to care and attributions at presentation among patients in the two settings ( Table 2). Those attending traditional healers were twice as likely to be ‘unable to say’ what was wrong with them, and proportionately less likely to present with a distinct physical illness or specific physical symptoms. In around 10% of those attending traditional healers there was a supernatural attribution, but no attenders at primary care made such an attribution. For attenders of traditional healers their main complaint was more likely to have been chronic, and they were more likely to have previously consulted several biomedical care providers for the same complaint. These last two effects were more prominent among those without common mental disorders compared with those with such disorders; the interaction term was not, however, statistically significant in either instance.
Associations with common mental disorder
Neither gender, age, education nor any of the indicators of socio-economic status was substantially or statistically significantly associated with common mental disorder in either setting. An exploratory multivariable analysis also did not identify any independent associations with potential risk factors. Those identified as cases by the CIS–R assessment were more likely to have had presenting complaints of more than 1 year’s duration (90.0% v. 59.6% for non-cases, χ2=11.4, P=0.001) and to have consulted previously with four or more practitioners for the same complaint (35.0% v. 13.2% for non-cases, χ2=17.3, P<0.001).
Phenomenology and diagnoses
Table 3 shows the proportion of patients in each setting reporting clinically significant symptoms (two or more symptoms) in each of the 14 domains of the CIS–R. Symptom profiles were quite similar in the two groups. Overall, the most common symptoms were fatigue, obsessions and depression among primary health clinic attenders, and obsessions, worries regarding physical health and depression among traditonal healer centre attenders. A similar pattern was seen among cases. Panic and compulsions were rarely reported in either setting. Forty-seven participants (26%) in the primary health clinic group and 87 participants (49%) in the traditional healer group were allocated an ICD–10 diagnosis by the PROQSY computer program. In both settings mixed anxiety–depressive disorder accounted for slightly over half of all ICD–10 cases. Depression and obsessive–compulsive disorder were relatively more common in the traditional healer group, and phobias were more common in the primary health clinic group.
Our main hypothesis was confirmed, as the prevalence of common mental disorders recorded among those attending a traditional healer centre in Dar-es-Salaam (48%) was twice that recorded in those attending a primary health clinic (24%). There is no published comparable study from Tanzania; however, the primary health clinic prevalence is consistent with the average prevalence of 21.8% in 14 studies from sub-Saharan African countries reviewed by Stock ( 1995) and the prevalence of 27% reported from Harare ( Patel et al, 1995). The figure is also comparable with the World Health Organization multinational study of common mental disorders in general medical settings, in which the mean prevalence recognised by the Composite International Diagnostic Interview was 24%, ranging from 7.3% to 52% ( Goldberg & Lecrubier, 1995). There are few studies of the prevalence of common mental disorder among those attending traditional healer centres; Patel et al ( 1995) reported a prevalence of 40% in Harare, a relatively high figure, although not as high as that seen in our study.
The main objective of the study was to determine the prevalence of common mental disorders among users of traditional healers and primary health clinics. Traditional healers are frequently ignored in studies of help-seeking and care provision although they cover the health needs of a substantial proportion of the African population ( Gessler et al, 1995). The selection strategy for the traditional healers was critical to the avoidance of bias. Purposive sampling would have led us to well-known and more expensive traditional healers, who were also likely to be familiar with biomedicine. Wealthy, better-educated clients might have been overrepresented with respect to the clientele of the less prominent healers providing the bulk of the service in this sector. An attempt was made to reduce this bias by asking the district chairman of CHAWATIATA to provide a complete list of healers with affordable (middle-range) consultation fees and at least three patients attending per day. The healer sample was then selected at random from this list.
The CIS–R was used for the first time in Tanzania in this study. However, as noted earlier, this instrument has proved to be a feasible and valid assessment of common mental disorders in a variety of other developing country settings, including Zimbabwe. Efforts were made to reduce ascertainment bias arising from cultural and clinical invalidity through scrupulous translation and back-translation procedures. The advice of an independent local psychiatrist at the final stage of translation should have increased the local validity of the Swahili version. Previous research has shown that if careful attention is given to translation and conceptual validity, etic instruments can be used with reasonable confidence across cultures. ⇓
The study was conducted in the capital city, Dar-es-Salaam, and its findings may not be generalisable to the mainly rural and agrarian population of Tanzania. Moreover, only two primary health care centres and eight traditional healers (two from each of the four main categories of healer) were included. Despite the random selection we cannot claim that this is a truly representative sample of all healers in Dar-es-Salaam, nor by extension that the participants necessarily represent all those who use such services. Biases might have resulted that could explain the differences in prevalence of common mental disorders between the two settings. Nevertheless, we believe that the study provides reasonably valid and reliable data on the extent and distribution of common mental disorders among patients in these settings, which may be used as a starting point for guiding the development of mental health care and for further research on common mental disorders.
Patterns of use of traditional and biomedical services
In the developed world, common mental disorders are costly and disabling disorders, which present often in primary care but are rarely recognised or treated ( Ormel et al, 1994; World Health Organization, 1995). In Africa, traditional healers are generally more accessible than biomedical practitioners; in rural Tanzania the ratio of doctors to population is 1:20 000 whereas that of the traditional healers is 1:25 ( Swantz, 1990). The quality of biomedical health services in Africa (including Tanzania) has been criticised ( Van der Geest et al, 1990; Gilson et al, 1994). In contrast to traditional healers, staff in primary care clinics are often found to show little concern or respect for their patients. Formal community mental health services are yet to be realised in Tanzania and there are only ten psychiatrists in the whole country, four of whom work in the main hospital in the capital city.
Traditional medicine and biomedicine differ in their concept of the nature and causes of illness. In Africa, mental disorders are often perceived as a source of misfortune; ancestors and witches are believed to have a crucial role in bringing them about. Such disorders may be viewed in terms of magical, social, physical and religious causes, but rarely as diseases within the Western biomedical paradigm ( Ndetei & Muhangi, 1979). Traditional healers’ rituals are linked to the maintenance or restoration of well-being in the whole community. Treatment and prevention focus upon the quality of human relationships and social interaction; healers provide their patients with moral and social guidelines to prevent them from catching the same illness again.
For the Tanzanian centres included in this study, the traditional healer and primary health clinic patient populations were strikingly different. This was particularly true as regards their pathways to care and their presenting complaint. The traditional healers’ patients had long-standing complaints, and were multiple and frequent consulters prior to the index consultation. The primary care patients had more acute complaints, and for two-thirds (compared with only 7% for the patients of traditional healers) the index consultation was their first for this complaint ( Table 2). Although caution is needed when making dynamic inferences from cross-sectional data, traditional healers would seem often to be a last resort for patients with persistent problems who were presumably dissatisfied with the outcome of previous consultations with biomedical providers. It would be tempting, although incorrect, to seek to explain the excess of common mental disorder among the patients of traditional healers on this basis. Chronicity and multiple help-seeking are as evident in traditional healer centres for those without as for those with common mental disorder. Rather, it seems that intractable conditions in general may cluster in these centres, among them a substantial proportion of persistent common mental disorders. Other observed differences in patient characteristics also did not explain the high rates of common mental disorders in those attending traditional healer centres. A multivariate analysis indicated that being older, better educated, widowed or separated and avowing a Christian faith were each independently associated with attending a traditional healer. In Zimbabwe, a contradictory pattern was observed, in that clients of traditional healers were more likely to be female and also to be unemployed, with less education ( Patel et al, 1997). In Tanzania, people who are educated, older and in employment are probably better placed to afford to pay for consultations and treatment at traditional healer centres. Primary health clinic consultations are free, but consultation time is short (approximately 3 min), with little opportunity to discuss symptoms or receive explanations about health problems. The association between Christian faith and attending traditional healers indicates that traditional explanatory models continue to exert a strong influence upon help-seeking behaviour, notwithstanding the efforts of missionaries to devalue traditional medicine. The Church in Tanzania officially rejects traditional medicine, which has its roots in animistic religion. However, in the face of hardship many believe that ancestral spirits or evil spirits have been used by jealous persons to inflict illness, misfortune or to generate conflict. The solution is thought to be to consult traditional healers. In Tanzania, modernisation, formal education and economic development seem to have had little impact upon these traditional ways, even among those who are most exposed to modern influences.
Risk factors for common mental disorder
In Harare, female gender, unemployment and poverty were each associated with common mental disorder ( Patel et al, 1997). Strikingly, none of these factors was associated with common mental disorder in our study. In Tanzanian society, women have a status that is perhaps unusual in sub-Saharan Africa. The Government, through the Ministry of Women and Children, together with non-governmental organisations, provides women with microcredit by means of grants and interest-free loans, education on running small businesses, and free legal advice. Most women live in large extended families in which child care and household tasks are shared. Any suggestion that these factors might explain the lack of a female excess of common mental disorders is speculative, and the findings themselves need to be treated with caution. In particular, the potential biasing effect of help-seeking needs to be considered; factors such as gender or poverty may influence help-seeking differently in those with and those without common mental disorder. Associations present in the general population may not be evident among those accessing care, and vice versa. In short, studies of this kind are not suitable for the identification of risk factors for common mental disorders; only population-based epidemiological studies provide information about aetiology.
Our research suggests that patients with common mental disorder constitute a large part of the workload of both primary health care clinics and traditional healers in Dar-es-Salaam. We cannot infer from this that most people with common mental disorder necessarily present to primary health care clinics and traditional healers. The first filter between onset of common mental disorder in the community and effective care is presentation to a primary care provider. Potential barriers are failure to seek help and inaccessibility of services. In the developed world this first filter is relatively permeable, but we cannot assume that the same will hold true for other cultures and health care systems. Population-based epidemiological studies are needed to improve our understanding of help-seeking behaviour and barriers to care at this level. We also provide indirect evidence that people seen by traditional healers may have particularly chronic conditions following the failure of primary health care, after multiple contacts, to resolve their problems. Problems may arise at the levels of both recognition and management. As in previous studies, most patients presented with somatic symptoms and a somatic attribution. There is little time, particularly in primary health clinics, to explore the possible psychological basis of these complaints, or to investigate the wider family and social context of the disorder. Traditional healers may be better placed in this respect; it would certainly be interesting to know more of their approach to these cases in terms of their formulation, their management and the treated outcome. Naturalistic prospective studies are indicated. These may suggest opportunities for productive alliances between Western and traditional medicine ( Green, 1988). With the support of the formal health system, indigenous practitioners might become important agents in organising efforts to improve the mental health of the community. Better understanding of the prevailing indigenous models and idioms of expression for common mental disorder should aid diagnosis and treatment. Traditional healers may have much to offer, and could usefully participate in joint training programmes in medical schools and with primary health care workers.
In Tanzania the Colonial Witchcraft Ordinance 1928, which was used to suppress the practices of traditional healers, is still in force. The Ministry of Health plans to establish a Traditional Health Care Practitioners Act with the objectives of registering, controlling and regulating the use of traditional healers, their practice and the medicine they utilise. In the meantime, traditional healers have taken the initiative. The work of registering healers has begun in coordination with CHAWATIATA. A main objective, stated in their constitution, is to encourage collaboration between Western-trained doctors and traditional healers, and to increase communication between these two sectors through meetings and seminars. Tanzanian biomedical practitioners tend to view this communication as a one-way street, aiming only to train poorly educated traditional healers and recruit their support in the pursuit of public health goals. True collaboration will occur when each is ready and willing to learn from the best of the others’ practices.
Clinical Implications and Limitations
Common mental disorders affect many of those attending primary health care and traditional healer centres in urban Dar-es-Salaam,Tanzania.
The overwhelmingly somatic presentations may have militated against recognition by clinicians, particularly given the short short consultation time in primary care.
The prevalence of common mental disorder was particularly high in patients of traditional healers; these patients usually presented with chronic symptoms and had previously consulted a number of biomedical care providers. This is likely to represent a failure of primary care clinics to detect and treat these disorders.
The results may not be generalisable to rural areas.
We cannot infer from the high prevalence of common mental disorder in primary care that most people with common mental disorder present in these settings. Population-based studies are needed to elucidate patterns of help-seeking and barriers to care.
Given the clinical settings for the study, help-seeking might have biased estimation of associations between potential risk factors and common mental disorder. Only population-based studies can provide information about aetiology.
We thank the Commonwealth Universities for their support for this project and their assistance and sponsorship of M.C.N. for his study at the Institute of Psychiatry, King’s College London. We also thank P. A. Rwegaikamu for her work on this study, the Tanzania Traditional Health Practitioners Association for allowing us access to traditional healers, the Ministry of Health and the Dar-es-Salaam Health Department for allowing us access to their clinics. Special thanks are due to the patients, nurses, medical assistants and traditional healers for their enthusiastic participation in this study.
- Received January 27, 2003.
- Revision received June 9, 2003.
- Accepted June 19, 2003.
- © 2003 Royal College of Psychiatrists