Mental health treatment outcome expectancies in Burundi

Abstract

Best practices in global mental health stress the importance of understanding local values and beliefs. Research demonstrates that expectancies about the effectiveness of a given treatment significantly predicts outcome, beyond the treatment effect itself. To help inform the development of mental health interventions in Burundi, we studied expectancies about the effectiveness of four treatments: spiritual healing, traditional healing, medication, and selected evidence-based psychosocial treatments widely used in the US. Treatment expectancies were assessed for each of three key syndromes identified by previous research: akabonge (a set of depression-like symptoms), guhahamuka (a set of trauma-related symptoms), and ibisigo (a set of psychosis-like symptoms). In individual interviews or written surveys in French or Kirundi with patients (N = 198) awaiting treatment at the clinic, we described each disorder and the treatments in everyday language, asking standard efficacy expectations questions about each (“Would it work?” “Why or why not?”). Findings indicated uniformly high expectancies about the efficacy of spiritual treatment, relatively high expectancies for western evidence-based treatments (especially cognitive behavior therapy [CBT] for depression-like symptoms), lower expectancies for medicine, and especially low expectancies for traditional healing (except for traditional healing for psychosis-like symptoms). There were significant effects of gender but not of education level. Qualitative analyses of explanations provide insight into the basis of people’s beliefs, their explanations about why a given treatment would or would not work varied by type of disorder, and reflected beliefs about underlying causes. Implications for program development and future research are discussed.

Keywords

Burundi mental health spiritual treatment traditional healing treatment expectancies

Burundi is a small, central African nation bordering Tanzania, Republic of Congo, and Rwanda. There is a recent history of genocidal conflict and violence, and Burundi’s economy and development has been very slow to recover in every sector. Healthcare, always scarce in Burundi, became even more limited with the outbreak of war. While new clinics and hospitals are being built and staffed, these offer only primary health services and mental healthcare is still largely neglected despite the widespread need for it, especially postwar. There are very limited human resources with almost no ongoing training or retraining (World Health Organization, 2008). There has been very little mental health research (Sharan et al., 2009), although a regional conference of health ministers noted the importance of moving the development of psychosocial interventions onto “the central stage of evidence based science” (Baingana & Ventevogel, 2008, p. 172).

In this context, research is needed to guide the development of mental health programs. Best practices in global mental health stress the importance of understanding local terms, beliefs, and values in developing mental health services (Bass, Bolton, & Murray, 2007; Belkin et al., 2011). In two previous extensive studies (Irankunda, Heatherington, & Fitts, in press), we documented the local vocabularies used by Burundians to describe the most common mental health problems, and their understandings about their causes and typical local treatments. That work identified and cross-validated three key syndromes: akabonge (a set of depression-like symptoms), guhahamuka (a set of trauma-related symptoms), and ibisigo (a set of psychosis-like symptoms).

Working from that foundation, the current study examines another kind of belief: expectancies about whether particular treatments are likely to work. Positive expectancies have been shown to be strongly associated with good outcomes in individual and group psychotherapy, above and beyond the treatment itself, and across various types of therapies and disorders (Constantino et al., 2007; Safren, Heimberg, & Juster, 1997; Schulte, 2008). Demonstrated mechanisms of action through which positive expectancies enhance outcome include increasing patients’ motivation to engage in the treatment and enhancing the development of a strong working relationship (alliance) with the therapist (Constantino, Glass, Arnkoff, Ametrano, & Smith, 2011; Dew & Bickman, 2005; Meyer et al., 2002). These findings derive largely from research in the US, Europe, Israel, and Australia, yet there is reason to believe that in Burundi and other developing African countries, expectancies are likewise important. It behooves program developers to understand the credibility and acceptability of current and proposed interventions in the local context.

Thus the current study examined the expectancies of Burundians about the likely outcomes of four different interventions for the three different types of mental health problems previously identified. The interventions included both local and western approaches: traditional healing, spiritual intervention, medication, and a western empirically supported treatment (EST)¹ for each disorder.

Methods

Setting and participants

The study was conducted at the Village Health Works clinic (VHW; www.villagehealthworks.org), a nonprofit primary health clinic in the rural village of Kigutu. It has a very good reputation for health care; thus, patients travel locally as well as from across the country to the clinic and are a diverse group. Participants were drawn from patients awaiting primary health care services (e.g., for malaria, TB, and chronic health conditions such as diabetes, heart problems, etc.). While previous findings revealed that people who come to the clinic are familiar with mental health problems (Irankunda et al., in press), there are no official mental health services offered at this clinic (or in most places in Burundi), just one BA-level counselor, hired to work with HIV/AIDS patients. We did not ask about the mental health status of the respondents or their own relatives as these questions are considered intrusive and embarrassing in the culture. Also, we did not want to imply that we were diagnosing people or offering mental health services. There were 198 participants (68% women; 32% men), Mean age = 30.48, SD = 12.19; 8% reported no schooling, 13% some elementary school, 39% some high school, and 40% some college; 96% were Christian, 1% Muslim, and 3% reported “no religion.”

Ethical considerations

The study was approved by the Williams College IRB and reviewed and approved by the VHW Director of Clinical Programs and the VHW Medical Director. Written consent was obtained from the literate participants (literacy was reliably established by self-report and confirmed by coherent answers on the survey) and oral consent from the nonliterate by the first author. For all participants, the study goals, anonymity, voluntary participation, and independence of participation from their treatment at the clinic were explained. Participants were informed that the questions were unrelated to their own diagnosis or treatment, and that the questions were the same for everyone. Of those approached for consent, all agreed.

Procedure

The surveys were conducted individually by the first author, a Burundian, at a table to the side of the clinic waiting area, in Kirundi (the native language) or French. After obtaining informed consent, participants were given a choice of completing a written form of the survey (copies in Kirundi and French were available) or completing it as a structured interview in their preferred language. Subsequent analyses of the data collected via survey (n = 140) compared to interview (n = 58) showed no effect of the method of administration. Interviews took approximately 30 minutes each. It is highly unlikely that the participants were previously familiar with the cognitive behavior therapy (CBT), psychoeducation, and exposure treatments because of the state of development of mental health treatment in Burundi, as described earlier. Nonetheless, participants found the task easily understandable and readily shared their opinions in ways that made it clear that they understood both the treatment descriptions and the instructions.

Measure

The measure consisted of three parts repeated once for each of the three syndromes. The first part described the syndrome in commonsense language, using the terms for the syndromes (akabonge, guhahamuka, and ibisigo) and the symptoms that were documented in the earlier study (see the Appendix for symptom descriptions).² The second part described four different treatments for that syndrome: spiritual treatment (priest, pastor, church); treatment by a traditional healer; medication; and a well-validated empirically supported treatment (EST) for the symptoms that is widely used in the US and other western countries. For akabonge, this was cognitive behavior therapy (CBT); for guhahamuka, it was exposure therapy; and for ibisigo, it was combined family pychoeducation and medication.³ Key websites (e.g., Association for Cognitive Behavior Therapy [http://www.abct.org/Home/], the U.S. Dept of Veterans’ Affairs VA National Center for PTSD [http://www.ptsd.va.gov/], and the National Alliance for the Mentally Ill [http://www.nami.org/]) were consulted in writing the EST descriptions. The descriptions of medicine, spiritual healing, and traditional healing were drawn from the first author’s knowledge and from visits to key informants in Burundi (e.g., traditional healing centers, priests, the mental hospital in the capital) during prior research. None of the descriptions referred to the evidence-base of any of the treatments (see Appendix A for treatment descriptions). In the third part, following each treatment description, participants were asked if they thought the treatment would work for that disorder (response categories: yes, no, I don’t know) and to explain why or why not. Finally, they were asked demographic questions about age, gender, education, and religion. Copies of the measure are available in Kirundi, French, and English from the authors.

Qualitative coding of explanations about treatment efficacy

The explanations of why treatments were believed to work or not work were transcribed in entirety (most people provided short phrases) and translated to English by the first author. The corpus of explanations was then studied using a constant comparison method for examining the recurring themes and generating categories for content coding (Strauss & Corbin, 1990). Each of the two authors and a research assistant independently studied the first 25 transcripts and created a preliminary list of categories for the “yes” and “no” explanations for each treatment. These lists were compared and reconciled across the three coders, followed by more coding, elaboration, and refinement of the list of categories, until no new categories emerged. In this manner, the final set of categories captured all codable responses in the dataset. (Uncodable responses were those that simply reasserted that the treatment would or would not work, or that offered conditions, e.g., “It would work if the person was cooperative.”) Following a training period of 6 hours, reliability among the coders to a percent agreement criterion of 88% was reached. The entire body of data was then coded (the first 25 transcripts were mixed back in and recoded) for these themes, with coders overlapping on 15% of the transcripts; reliability during the training was sustained during the actual coding, with an average percent agreement of 94%.

Table 1.

Explanations of why treatments would work.

	EST			Medicine			Spiritual treatment			Traditional healing
	Aka	Guha	Ibi	Aka	Guha	Ibi	Aka	Guha	Ibi	Aka	Guha	Ibi
The treatment provides human contact, support, comfort	77	31	19
It normalizes the problem/person	17	25	8
It provide desensitization/helpful exposure	6	36	1
It provides understanding, insight, explanation, or expert advice	58	13	22
It provides instruction/teaching about how to behave, conduct oneself	8	5	14
It helps family members reconcile, connect, support the person	7	4	18
It manages the symptoms (sleeplessness, agitation, etc.)				16	24	17
It treats the root causes (bacteria, neurons, etc.)				7	7	1
It is done by those who are smart, trustworthy				9	7	2
Because the patient believes it will work (essentially, a placebo explanation)				5	10	3
Divine intervention, including intercessory prayer							60	62	54
Prayer itself helps provide comfort, hopes, distracts, etc.							51	56	60
Social support from prayer & church							15	11	10
Personal empirical evidence “because I’ve seen it work”							7	5	2
God /prayer/priest chases away ibisigo/abaganza							1	3	47
They know spirits/can harness them for good										0	0	42

Note. Aka = akabonge; Guha = guhahamuka; Ibi = ibisigo.

Table 2.

Explanations of why treatments would not work.

	EST			Medicine			Spiritual treatment			Traditional healing
	Aka	Guha	Ibi	Aka	Guha	Ibi	Aka	Guha	Ibi	Aka	Guha	Ibi
It is not helpful or it is even harmful 1to talk about problems or to remember (“gusimbura”)	1	15	1
Cites an inconsistent alternative belief about what works^a	3	5	17	47	31	17	9	9	1	27	22	20
Patient can’t or won’t even listen or attend to the treatment	3	1	16
Lack of personal empirical evidence (e.g., no one with ibisigo goes to hospital and returns cured)				4	2	8
Because this is not a clinical disease or physical illness (e.g., it’s a disease of the heart but not of the body, it’s a possession not a medical illness)				58	44	48
Because it is a disease that wounded the person’s heart (biology implied, doesn’t help you understand the disease)							4	5	1
Because the patient is incapable of participating in or cooperating with the treatment (e.g., won’t listen, won’t sit still, doesn’t believe in prayer)							7	6	5
Sorcerers are incapable or incompetent, have no knowledge of or skills for these diseases										50	53	20
Sorcerers are malevolent or evil (e.g., they lie, turn people against each other, only want to make money)										79	61	31
Sorceres are in league with Satan, or use the spirits for the worse or are sick themselves with ibisigo										5	1	18
Sorcers use a lot of noise, frighten people, retraumatize										1	8	1

Note. ^aExamples: “In prayer there is no conversation and you need conversation to find solutions to depression”; “Only God not a sorcerer can help”; “A patient with trauma cannot be treated with medicine, but needs a conversation with a clinician so he can get advice about how to cope.”

Results

Outcome expectancies

The percentage of people who answered “I don’t know” to the efficacy questions was low; for akabonge, it was < 6% for each treatment; for guhahamuka, it was 5% for spiritual treatment, 7% for traditional healing, and 9% for each of the medication and exposure therapy treatments. The treatments for ibisigo, especially medication and the EST (medication and family psychoeducation combined) yielded more “I don’t know” responses, 15% and 13% respectively, and a wider range across treatment types; that is, just 3% of people indicated uncertainty about the efficacy of spiritual treatment, compared to 11% for the traditional healer.

Figure 1 displays the positive outcome expectancies (percentages of people who said “yes” it would work) to allow a snapshot of each of the four treatments by syndrome. These data are presented descriptively; chi-square analyses are not appropriate as the data are not independent across treatments, given the repeated measures design. The data show some clear patterns. For akabonge, the majority of respondents expected cognitive behavior therapy (89%) and spiritual treatment (73%) to work, with medicine (30%) and traditional healing (12%) trailing far behind. For guhahamuka, a majority of people (77%) thought the spiritual treatment and exposure therapy (70%) would work, with medicine (43%) and especially traditional healing (10%) again garnering lower expectancies. For ibisigo, spiritual treatment was expected to be efficacious by a large majority of participants (88%), while just over half (57%) expected the combined medication and family psychoeducation treatment to work. In the case of ibisigo, however, traditional healing was expected to work by relatively more people (39%) than for the other syndromes, with medication expected to work by only about a third of respondents (30%).

Figure 1.

Expectancies about the efficacy of each treatment, by disorder.

Gender and education effects

Chi-square tests of the proportion of “yes,” “no,” and “I don’t know” answers by gender, for each of the four different types of treatments were conducted for each syndrome to examine gender effects on treatment expectancies. To minimize experiment-wise error rates, p values of less than .01 were considered significant and those less than .05 as tending toward significance. For akabonge, men were more likely than women to say that medication, χ² (2, N = 198) = 8.31, p = .01, and spiritual treatment, χ² (2, N = 198) = 8.35, p = . 004, would work. For treatments for guhahamuka, men expressed higher expectancies than women that medication, χ² (2, N = 198) = 9.52, p = .009, and spiritual treatment, χ² (2 N = 198) = 13.47, p = . 001, would work. There were no significant effects of gender on expectancies about the efficacy of the EST or traditional healing treatment for either syndrome, and no gender effects in expectancies for any of the treatments for ibisigo.

Since gender and education are highly confounded in Burundi (men are more likely to be educated), another set of chi-square analyses assessed the effects of education, splitting the sample for this purpose between those with no or primary education and those with high school or college education. The results showed only a few weak associations of education level and expectancies. There were statistical tendencies for more educated people to have higher expectancies for traditional healing, χ² (2 N = 198) = 6.39, p = .04, and spiritual treatment, χ² (2 N = 198) = 7.28, p = .02, to be effective for akabonge. However since the overall number of people who thought the traditional healing would be effective for depression was low, these results should be interpreted with caution. All other tests yielded nonsignificant differences.

Participants’ reasoning about treatment efficacy: Qualitative analyses

Table 1 displays the categories of explanations that participants provided about why a given treatment would work and the numbers of respondents whose answers included each type of explanation; Table 2 displays this information for respondents’ explanations of why a given treatment would not work. As the tables show, in general, explanations of both kinds were unique to the type of treatment, suggesting that participants understood and were responding thoughtfully to the questions.

Some clear themes emerged from these data. Many people made references to the mechanisms of social support, comfort, and normalization from the therapeutic relationship itself as being operative in CBT and exposure therapy, for example: “because when you get someone to comfort you, it makes you feel better; a patient can feel better when someone explains that what happened can happen to anyone else” (M, 26, HC).⁴ On the flip side, there were parallel references to why medication would not be effective, for example, “because depression is a pain from the heart … cannot be treated by pills or injections” (M, 25, HC). Indeed, there were strong themes in the qualitative data about the pain of social isolation that attends mental problems and the therapeutic effect of social support (including physical presence) by helpers.

Some respondents who thought the CBT would work also focused on the mechanism of cognitive understanding or changed ways of thinking, which is in fact a core mechanism in CBT theory, for example: “because it helps to get an explanation of why she or he is experiencing this” (M, 26, HC); “because this treatment emphasizes explaining, understanding and advising” (M, 17, HC). Similarly, some people who thought the exposure therapy would work also made references to the mechanisms of exposure and normalization that are a core part of the theory of that treatment, for example:

The present is shaped by the past, so you better know well your past and be able to see it as normal. When things happen to you and you start trying to forget and not want to confront it, another time comes when you have to remember what happened; when you talk about the things that happened to you, it helps the patient not be afraid of the things that happened. (M, 45, HC)

Interestingly, this mechanism (exposure) was also cited by other participants who thought that it would not work, with frequent references among the group that doubted its efficacy (which was nonetheless a minority) to gusimbura, the Burundian tradition of not talking about or mentioning the name of the dead. One person noted: “because the more you remember, the more it hurts, you would not get peace by talking about it” (F, 18, NE). Another noted that it would be like piercing a scar.

Explanations for why spiritual treatment would work are clear in the table and referred strongly to both the power of divine intervention and the act of prayer itself, for example:

Prayer can work well because it in fact acts directly where the problem is—in the heart. The clients regain peace and joy in the heart and also hope. Prayer soothes and helps in everything, the client can be able to laugh again, and have trust in people starting with God. (F, 23, HC)

In the case of ibisigo, the dominant belief linking it with the supernatural (ibisigo is the local name for delusions, hallucinations, uncontrolled behavior, and the spirits which are thought to cause them) was evident. But again, these references sometimes gave rise to competing expectancies regarding traditional healing: a sizable minority reported that the healer could harness and use that power (e.g., “a sorcerer can treat ibisigo because it is said that s/he gets powers from ibisigo, so s/he can know your secrets and know how to treat them” [M, 25, HC]), while a larger number of people reported that the healer was co-opted by it and in league himself with the bad spirits (e.g., “how can you be attacked by demons and then go to a sorcerer who possesses demons. A sorcerer works with Satan” [M, 27, HC]). Some responded that traditional healing was “sorcery” and antithetical to Christian beliefs and healing practices, while others thought that traditional healing was very relevant for this set of symptoms but not others. In any case, very few respondents indicated that ibisigo was irrelevant to the treatment of psychosis. Respondents distinguished between ibisigo as a “traditional” disease requiring “traditional” treatments and akabonge and guhahamuka as conditions caused by difficult life experiences or biological disorder, and thus “clinical” diseases. The qualitative findings also shed light on the relatively low expectancies about the efficacy of medicine revealed in the quantitative data. People mentioned not knowing if there were any medicines for this condition, or being doubtful that if there were medicines they would work, as they had never seen them work for their friends or family. Others who believed that medicine would not work cited the belief that these were not medical conditions and thus medicine would not treat the underlying cause.

Discussion

The results yielded some unexpected findings. Interestingly, the data revealed that most participants thought that the empirically supported treatments would work, especially CBT for akabonge/depression and exposure-based therapy for guhahamuka/PTSD, even though it is likely that they had no prior knowledge of them. This general result was surprising, given that the empirically supported treatments (ESTs) as described all involve, to varying extents, an interpersonal relationship in which the client talks about his/her troubles and shares inner feelings or thoughts and that Burundian cultural practices discourage sharing inner troubles and talking about past negative events like death and conflict. In particular, given the cultural notion of gusimbura, we had anticipated that most people would not endorse exposure therapy. It is possible that because of the war and attendant suffering, people’s awareness of psychological suffering and need for help may be eroding whatever cultural barriers to “talk” therapy may exist. An account from psychologist Ervin Staub on his arrival in postgenocide Rwanda for psychological interventions notes that:

In a country where people were known for not showing their feelings and for talking about personal matters only to close relatives, many Tutsis we met immediately began to tell us [their stories] … whether taxi drivers, the staff of NGOs or government ministers. (Staub, 2011, p. 302)

In addition, the “common factors” (support, hope, normalization, and in particular the counseling relationship itself) found to be operative in psychotherapy outcomes in the West (Norcross & Wampold, 2011), also figured prominently in people’s thinking about the psychosocial treatments described in this study.

Competing alternative explanations for the relatively positive expectancies about these psychotherapies must be considered; specifically, it is possible that the novelty of the EST treatments or their slightly longer descriptions made them attractive relative to the other treatments. However, the fact that respondents’ expectancies about the ESTs differed across disorders counters these alternate explanations. Further, the qualitative data provide compelling evidence of substantive and nuanced responses to each of the ESTs described, as well as clues as to what it is about the treatments that participants responded to favorably. Overall, the qualitative data provide evidence of the internal validity of the quantitative results, that is, evidence that people understood the task, responded thoughtfully to the questions, and discriminated between the nature of the three syndromes in answering the questions.

In general, expectancies about medication were surprisingly low; the qualitative data shed light on the reasons for this finding. Participants associated medication with physical illness, not with problems whose causes were attributed to suffering or spirit possession, echoing the findings of Muga and Jenkins (2008) in Kenya. This finding illustrates a general theme: participants’ beliefs about the effectiveness of the treatments were closely linked to their perceptions of the causes of mental health problems. Further, since psychoactive medicines are largely unavailable in Burundi, people had not heard of others taking them and benefitting from them. This observation is interesting given recent critiques of Western influences and the pharmaceutical industry for “medicalizing” mental distress (Watters, 2010)—a trend that has obviously not reached rural Burundi.

An even more striking pattern was the low expectations for the efficacy of traditional healing. With the exception of ibisigo, there was very little faith in traditional healing. Indeed, there was considerable suspicion and negativity, with traditional healers being described as charlatans and/or sorcerers: “a sorcerer only lies and does nothing but takes money from the patient and puts people in conflict” (M, 45, HC). The strength and specificity of these opinions, in conjunction with the quantitative data and the lack of a strong inverse relationship with education, shows that the results cannot be explained simply by people’s unwillingness to admit to a belief in traditional healing (which is considered, especially by more educated people, as “backward”). Rather, it seems related to a combination of personal (negative) experiences with traditional healers and teachings of Christianity, which eschews these kinds of supernatural beliefs. The latter is well articulated by the response, “Prayer is the most important remedy for a person with ibisigo. Ibisigo are Satan’s spirits and that’s why a prayer works” (M, 25, HC). This resonates with observations by Stroeken (2012) in Tanzania that, especially in urban areas, “church sermons about the magic foolishly embraced by so-called ‘backward’ villagers reinforce the idea of traditional medicine belonging to an illicit realm” (p. 120). Further, as discussed by Irankunda et al. (in press), there is a complex history in Burundi in which a historical set of traditional healing beliefs and practices (albeit different from the current traditional healing practices) were erased by missionaries in the early mid-1900s. In any case, the fact that a substantial minority of people did expect traditional healing to work for psychotic-type conditions (but only those) was further evidence that participants discriminated between different disorders in their considerations of treatment efficacy, and that their expectations were closely related to their causal attributions about the problems.

In this sample, there was a consistent pattern for men to report higher expectancies than women with regard to medication and spiritual treatment for PTSD. It is not clear why. Perhaps men have a more optimistic outlook in general. These explanations as well as certain methodological factors (the interviewer was male) need to be evaluated in further research. However, these differences cannot be attributed to different levels of education, as education was not significantly associated with expectations. The lack of effect of education is important to note because better educated people were overrepresented, relative to the general population, in the sample.

Limitations and suggestions for future research

In research such as this, there is always the chance of response bias, such that respondents tell interviewers what they think the interviewers want to hear. Sabuni (2007), in a study in the Democratic Republic of Congo about perceived causes of physical illness, notes that their respondents shared one cause of illness when talking with a religious leader or health professional, another with friends, and another (often involving sorcery) with family and clan. He cautions that the efforts of health educators and treatment providers who are not cognizant and respectful of local beliefs can be rejected passively as well as actively. In our study, this concern was mitigated somewhat by having a Burundian researcher who was born and raised in the area. It is possible, however, that his being a well-educated person might have biased the results in some way.

The descriptions of the empirically supported treatments were by necessity slightly longer (only limited information was needed to describe treatments provided by priests, traditional healers, and using medication). This may have made the descriptions of ESTs relatively more persuasive. This concern is partially mitigated by the fact that respondents discriminated between different disorders when evaluating the usefulness of these treatments and showed no consistent bias for the longer descriptions.

Another limitation is that it was possible to know what particular elements of the treatments were most relevant to participants’ views of them. As with treatment research in the US and elsewhere, effective (or perceived effective) mechanisms of change might be specific to the particular treatment (i.e., the intercessory prayer itself, the desensitization to traumatic memories as a function of repeated exposure) or nonspecific (the power of belief, the effects of an empathic helping relationship). The body of psychotherapy research in the US suggests that while both are important, nonspecific factors are particularly powerful (Wampold, 2013). Also, there are many different types of traditional healing rituals (cf. Janzen, 1978; see also Devisch, 1993), and our description of the traditional healing modality was, by necessity, general.

Certain characteristics of the sample potentially limit the generalizability of the results. The respondents were not mental health clients and did not necessarily have personal or family experience with the syndromes. Future research should assess outcome expectancies and/or preferred treatments in a sample of mental health patients or patients’ family members, and at different points in the course of treatment. This can be difficult, however, in a setting in which few treatment choices are actually available. Also, as noted earlier the sample was not representative of the education level in the country as a whole, although it was representative of the patients who find their way to the VHW clinic. Finally, the interrater agreement index may be an overestimation of interrater reliability; unfortunately, the data necessary for a calculation of Cohen’s kappa (a preferable index) were not preserved.

Implications for the development of mental health services

This study is the first of its kind to our knowledge, and begins to lay some groundwork for much needed mental health services in Burundi by informing mental health workers and program developers about the expectations that patients bring to treatment and revealing the reasons why people may be open to or wary of various treatment options. Although in this study it was not possible to correlate expectancies with actual outcomes (given the lack of mental health treatment), the results highlight the importance of evaluating, even if informally, the credibility of treatments being planned and developed, as well as those that are being offered. With regard to current practice, the findings may aid clinicians in deciding how to adapt or even talk with patients about the rationale for various treatments, including medication, to ensure mutual collaboration. Constantino (2012) and others have found that pretreatment expectancy enhancement strategies have been effective, thus adaptations of these might prove useful if new treatments are to be introduced. Further, although there were clear trends in expectancies, there was considerable variability within participants’ expectancies, that is, there was not always strong consensus on treatment efficacy, even within this culture. This suggests that clinicians should attempt to get a sense of individual patients’ particular understandings about the nature of the disorder and about their treatment expectancies.

The findings also suggest that, at least for depression and PTSD, treatments that have been empirically supported in the US should not be ruled out as being culturally unacceptable, and may in fact have some potential benefits. To be clear, we are not suggesting that CBT and exposure-based therapies be imported whole-scale to Burundi, rather that they be more closely considered and, if they are to be used, adapted as needed to fit the culture. Although some adaptations of counseling interventions in the Burundian context have proven to be ineffective or even detrimental (e.g., Tol, Komproe, et al., 2014), there is empirical evidence that this kind of adaptation is possible and effective if done thoughtfully. For example, in a randomized clinical trial, interpersonal therapy adapted for use in Uganda was shown to be successful (Verdeli et al., 2003). There are also case reports of successful adaptations of cognitive-based psychoeducation for ihahamuka (the noun form of guhahamuka) with Rwandan clients (Hagengimana & Hinton, 2009).

In such adaptations, it is especially important to take the local service delivery context into account. In Burundi, this includes a dearth of trained mental health professionals, inaccessibility of health clinics especially in rural areas, and low literacy rates. These are not, however, insurmountable challenges. Best practices in both community psychology and global mental health stress the importance of local collaborations and incorporation of existing community resources in the development of mental health services. Our findings suggest that local clergy are a particularly important resource, as alternate treatment providers, gatekeepers, referral sources, and/or stakeholders in mental health service planning. Further, programs that harness the strengths of the community are also recommended. The strong themes in the qualitative data about the pain of social isolation that attends mental illness and the therapeutic effect of social support (including physical presence) by helpers echoes the Burundian tradition of “sitting with” a person in mourning; people have an intuitive sense of the importance of human contact as a “common factor” in successful treatment. It also helps explain the strong, sometimes angry responses about some traditional healers who “turn people against each other.” And, it resonates with calls of government ministers and mental health NGO leaders to help restore the valuable social networks and community structures that have been destroyed by the recent conflicts and ongoing poverty, by building healthy individuals (Baingana & Ventevogel, 2008). Loss of social support is a risk factor for PTSD in civilians exposed to violence (cf. Hall, Murray, Galea, Canetti, & Hobfoll, 2014), and just as importantly, the presence of social support resources as well as cognitive social capital (the perception of trust and reciprocity within one’s community) are protective factors in postconflict settings, for both adults and children, including in Burundi (Hall, Tol, Jordans, Bass, & de Jong, 2014). A recent analysis of 160 reports of psychosocial support and mental health interventions in diverse disaster settings found that facilitation and promotion of various community support efforts for vulnerable people was near the top of the list (along with individual counseling) of what is actually being done in these settings, although both are seriously underrepresented in research (relative to studies of specialized treatments, e.g., for PTSD; Tol, Barbui, et al., 2014).

Further, another existing resource is the network of community health workers who are currently engaged in primary health care and who are knowledgeable about and respected in the local villages they serve as “outreach workers” from clinics, including VHW. Hanlon et al. (2014), working in five diverse countries (including Uganda, South Africa, and Ethiopia) present a strong case for the potential application of existing models of primary health care (with their established mechanisms of engaging and keeping people in treatment, outreach, and adherence support) to mental health care. Further, group-based treatments, especially those focused on psychoeducation and mutual support, may be a viable, more cost-effective mode of treatment, and consistent with cultural values of social support (de Jong, Scholte, Koeter, & Hart, 2000).

With regard to ibisigo, the findings also suggest that clinicians might need to accompany counseling with traditional healing practices, or at least an acknowledgement and acceptance of them as an adjunct, especially for individuals with strong beliefs about supernatural causes. Stroeken (2012) and others note that the traditional healer is often the “first port of call” (p. 120) for Africans suffering from mental illness. Given current and previous findings on causal beliefs and epistemologies this would seem particularly true for psychotic-like conditions, and Stroeken suggests that this may in fact be based in experience, that is, encounters with doctors and medicine having proved particularly ineffectual for these problems.

Finally, other researchers (e.g., Janzen, 1978; Muga & Jenkins, 2008; Ngoma, Prince, & Mann, 2003) note that in the developing world, pluralistic beliefs about the power of traditional healing, western medicine, and Christian prayer coexist in the culture, and can be successfully mixed in treatment packages (Hagengimana & Hinton, 2009; Ngoma et al., 2003), just as biomedical and alternative medicine practices are being jointly used in the US by some practitioners and clients.

This study revealed both great need and great interest in mental health services, even in a culture in which it is atypical to talk about one’s problems and especially about traumatic past events. The emerging field of implementation science (National Institutes of Health, 2014) encourages global health workers to use the results of empirical research to inform and evaluate workable, on-the-ground, evidence-based programs and services. In that regard, we hope that these findings will prove useful both in informing the development of mental health services and in stimulating much needed further research.

Footnotes

Acknowledgements

We gratefully acknowledge the support of Williams College Department of Psychology and Molly Magavern/Mellon Mays Program, Village Health Works Cofounder and Director of Clinical Programs Dr. Dziwe Ntaba for welcoming us to collect the data at Village Health Works, Drs. Laurie Pearlman and Marlene Sandstrom for research consultation, Jessica Fitts for research and editorial assistance throughout, and Yedidya Erque for research and editorial assistance on the revision of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Pacifique Irankunda, BA (Psychology & Political Science), is a Burundian writer and research associate at Williams College, Williamstown, Massachusetts. He was a finalist for the Elie Wiesel Ethics Essay Contest, and won a Pushcart Prize in 2014 for his essay Playing at Violence, which was published by American Scholar. He is a contributing editor at Pushcart Press. His first forthcoming book, The Morals of the Civilized, is to be published by Random House.

Laurie Heatherington, PhD, is Edward Dorr Griffin Professor of Psychology at Williams College, Williamstown, Massachusetts. She researches processes of change in family therapy, the therapeutic alliance, and treatment outcomes in community mental health and residential treatment, with recent work in global mental health. She is former President of the North American Society for Psychotherapy Research; with collaborator Myrna Friedlander, she received the 2010 Distinguished Contribution to Family Systems Research Award by the American Family Therapy Academy.

Appendix

References

Baingana

Ventevogel

(2008) Mental health and psychosocial interventions and their role in poverty alleviation. In: Proceedings of a conference. Intervention: International Journal of Mental Health, Psychosocial Work & Counselling in Areas of Armed Conflict 6(2): 167–173.

Bass

J. K.

Bolton

P. A.

Murray

L. K.

(2007) Do not forget culture when studying mental health. The Lancet 370(9591): 918–919.

Belkin

G. S.

Unützer

Kessler

R. C.

Verdeli

Raviola

G. J.

Sachs

Eustache

(2011) Scaling up from the “bottom billion”: “5x5” implementation of community mental health care in low-income regions. Psychiatric Services 62(12): 1494–1502.

Constantino

M. J.

(2012) Believing is seeing: An evolving research program on patients’ psychotherapy expectations. Psychotherapy Research 22(2): 127–138.

Constantino

M. J.

Glass

C. R.

Arnkoff

D. B.

Ametrano

R. M.

Smith

J. Z.

(2011) Expectations. In: Norcross

J. C.

(ed.) Psychotherapy relationships that work: Evidence-based responsiveness, 2nd ed. New York, NY: Oxford University Press, pp. 354–376.

Constantino

M. J.

Manber

Ong

Kuo

T. F.

Huang

J. S.

Arnow

B. A.

(2007) Patient expectations and therapeutic alliance as predictors of outcome in group cognitive-behavioral therapy for insomnia. Behavioral Sleep Medicine 5(3): 210–228.

De Jong

J. P.

Scholte

W. F.

Koeter

M. W. J.

Hart

A. A. M.

(2000) The prevalence of mental health problems in Rwandan and Burundese refugee camps. Acta Psychiatrica Scandinavica 102(3): 171–177.

Devisch

(1993) Weaving the threads of life, Chicago, IL: University of Chicago Press.

Dew

S. E.

Bickman

(2005) Client expectancies about therapy. Mental Health Services Research 7(1): 21–33.

10.

Hagengimana

Hinton

D. E.

(2009) “Ihahamuka,” a Rwandan syndrome of response to the genocide. In: Hinton

D. E.

Good

B. J.

(eds) Culture and panic disorder, Stanford, CA: Stanford University Press, pp. 205–229.

11.

Hall

B. J.

Murray

S. M.

Galea

Canetti

Hobfoll

S. E.

(2014) Loss of social resources predicts incident posttraumatic stress disorder during ongoing political violence within the Palestinian Authority. Social Psychiatry and Psychiatric Epidemiology 49: 1–8.

12.

Hall

B. J.

Tol

W. A.

Jordans

M. J.

Bass

de Jong

J. T.

(2014) Understanding resilience in armed conflict: Social resources and mental health of children in Burundi. Social Science & Medicine 114: 121–128.

13.

Hanlon

Luitel

N. P.

Kathree

Murhar

Shrivasta

Medhin

Prince

(2014) Challenges and opportunities for implementing integrated mental health care: A district level situation analysis from five low- and middle-income countries. PloS ONE 9(2): e88437.

14.

Irankunda

Heatherington

Fitts

(2016) Local terms and understandings of mental health problems in Burundi. Transcultural Psychiatry. 54(1), 66–85.

15.

Janzen

J. M.

(1978) The quest for therapy in Lower Zaire, Berkeley: University of California Press.

16.

Meyer

Pilkonis

P. A.

Krupnick

J. L.

Egan

M. K.

Simmens

S. J.

Sotsky

S. M.

(2002) Treatment expectancies, patient alliance and outcome: Further analyses from the National Institute of Mental Health Treatment of Depression Collaborative Research Program. Journal of Consulting and Clinical Psychology 70(4): 1051–1055.

17.

Muga

F. A.

Jenkins

(2008) Public perceptions, explanatory models and service utilisation regarding mental illness and mental health care in Kenya. Social Psychiatry and Psychiatric Epidemiology 43(6): 469–476.

18.

National Institutes of Health (2014) Implementation science information and resources. Retrieved from http://www.fic.nih.gov/researchtopics/pages/implementationscience.aspx.

19.

Ngoma

M. C.

Prince

Mann

(2003) Common mental disorders among those attending primary health clinics and traditional healers in urban Tanzania. The British Journal of Psychiatry 183(4): 349–355.

20.

Norcross

J. C.

Wampold

B. E.

(2011) Evidence-based therapy relationships: Research conclusions and clinical practices. Psychotherapy 48(1): 98–102.

21.

Sabuni

L. P.

(2007) Dilemma with the local perception of causes of illnesses in central Africa: Muted concept but prevalent in everyday life. Qualitative Health Research 17(9): 1280–1291.

22.

Safren

S. A.

Heimberg

R. G.

Juster

H. R.

(1997) Clients’ expectancies and their relationship to pretreatment symptomatology and outcome of cognitive-behavioral group treatment for social phobia. Journal of Consulting and Clinical Psychology 65(4): 694–698.

23.

Schulte

(2008) Patients’ outcome expectancies and their impression of suitability as predictors of treatment outcome. Psychotherapy Research 18(4): 481–494.

24.

Sharan

P. P.

Gallo

C. C.

Gureje

O. O.

Lamberte

E. E.

Mari

J. J.

Mazzotti

G. G.

Saxena

S. S.

(2009) Mental health research priorities in low- and middle-income countries of Africa, Asia, Latin America and the Caribbean. British Journal of Psychiatry 195: 354–363.

25.

Staub

(2011) Overcoming evil: Genocide, violent conflict, and terrorism, New York, NY: Oxford University Press.

26.

Strauss

Corbin

J. M.

(1990) Basics of qualitative research: Grounded theory procedures and techniques, Thousand Oaks, CA: SAGE.

27.

Stroeken

(2012) Health care decisions by Sukuma “peasant intellectuals”: A case of radical empiricism. Anthropology and Medicine 19: 119–128.

28.

Tol

W. A.

Barbui

Bisson

Cohen

Hijazi

Jones

van Ommeren

(2014) World Health Organization guidelines for management of acute stress, PTSD, and bereavement: Key challenges on the road ahead. PLoS Medicine 11(12): e1001769.

29.

Tol

W. A.

Komproe

I. H.

Jordans

M. J.

Ndayisaba

Ntamutumba

Sipsma

de Jong

J. T.

(2014) School-based mental health intervention for children in war-affected Burundi: A cluster randomized trial. BMC Medicine 12: 56. doi:10.1186/1741-7015-12-56.

30.

Verdeli

Clougherty

Bolton

Speelman

Lincoln

Bass

Weissman

M. M.

(2003) Adapting group interpersonal psychotherapy for a developing country: Experience in rural Uganda. World Psychiatry 2(2): 114–120.

31.

Wampold

B. E.

(2013) The great psychotherapy debate: Models, methods, and findings, New York, NY: Routledge.

32.

Watters

(2010) Crazy like us: The globalization of the American psyche, New York, NY: Free Press.

33.

World Health Organization (2008) Rapport WHO-AIMS sur le Système de Santé Mentale au Burundi, Bujumbura: Burundi: Author.