Determinants of mental illness and care seeking behaviours in Northern Haiti: an assessment of demographic, social, and religio-cultural factors among patients at the first mental health clinic in the region

Abstract

Populations in countries such as Haiti demonstrate a high level of need for mental health care despite a lack of services and trained professionals. In addition to the dearth of biomedical services, local belief systems and explanatory models contribute to a majority of the population relying on traditional medicine as their first option for care. Using a mixed-methods approach, we aim to characterise mental illness at the first mental health clinic in the region – Sant Sante Mantal Mòn Pele (SSMMP) – by interviewing 96 patients with a demographic questionnaire as well as Anxiety, Depression, and Functionality Scales. Multivariate logistic and linear regression models were conducted examining the impact of demographic variables on whether patients believed their illness was caused by sent spirits or previously visited a Vodou priest for treatment, as well as Depression, Anxiety, and Functionality Scale scores. Factors associated with mental illness in this sample included sex, number of traumatic events, physical health status, and number of sessions attended at SSMMP. Factors which impacted traditional beliefs or practices related to mental illness included sex, age, and income.

Introduction

Background and significance

According to recent research, the global disease burden of mental illness has been significantly underestimated until recently. Current figures suggest that 13% of disability adjusted life years (DALYs) worldwide are caused by mental health issues while up to 85% of individuals with serious psychiatric conditions living in low- and middle-income countries (LMICs) do not receive the treatment they need (Vigo et al., 2016). This is largely due to the fact that many low-income countries have less than one psychiatrist per million inhabitants and few trained psychologists or other biomedical mental health practitioners (Ventevogel, 2016). Research in the field of global mental health is currently focused on how to address this high burden of mental illness while improving access to high-quality, sustainable, and cost-effective services in LMICs (Patel et al., 2018).

Haiti is a country that exemplifies these gaps with high levels of mental health disease burden and low levels of treatment or access to biomedical care (Galvin & Michel, 2020; IESM-OMS, 2011; Smith-Fawzi et al., 2012; Wagenaar et al., 2012). Like many populations residing in LMICs, Haitians face harsh social vulnerabilities that increase risk for poor mental health, such as high levels of poverty, violence, and social stressors, poor physical health and nutrition, and low levels of education (Kieling et al., 2011; Lund et al., 2012; Smith-Fawzi et al., 2012)

Today, mental health services in Haiti remain largely limited to the 60-bed Mars & Kline Centre and the 120-bed Beudet Hospital, both located in the capital Port-au-Prince. However, even services at these institutions have been deemed inadequate and of poor quality (Raviola et al., 2020). Additionally, high levels of violence, gang activity, and unrest in the capital prevent many from being able to access services there. Besides Mars & Kline and Beudet, non-governmental organisations (NGOs) such as Zanmi Lasante provide significant mental health services in the Central Plateau region to the northeast of Port-au-Prince, yet few services exist outside of these areas, leaving the majority of Haitians with few options for mental health care (Nicolas et al., 2012). More high quality mental health treatment services are urgently needed to fill these gaps in services, particularly in regions where little to no biomedical treatment options are available.

Sant Sante Mantal Mòn Pele (SSMMP) – or Mental Health Center at Morne Pelé in English – was created in 2016 just outside of the city of Cap-Haïtien with the goal of bringing evidence-based, culturally-appropriate, and community-centered outpatient mental health treatment to Haitians living in the north of the country for the first time (Galvin & Michel, 2020). At the time of this study, clinical staff included a physician, three psychologists, a nurse, and a lab technician all specialised in mental health issues. Before the arrival of SSMMP, Haitians with chronic mental illness such as clinical depression, anxiety, bipolar disorder, schizophrenia, addiction, and epilepsy had few options outside of consulting traditional healers such as Vodou priests (ougan), or travelling to the south of the country for care. Yet, with a dearth of current research about mental health services in Haiti – and in the north of the country in particular – there is a significant knowledge gap regarding the basic concepts of treatment in the Haitian social and cultural context and how these services can best meet local needs (Michel, 2019).

Religious beliefs and explanatory models of illness in Haiti

Culture and systems of belief are essential factors to consider when trying to understand health behaviours in a given context (Kleinman, 1978). With regards to Haiti, it is important to note that many Haitians maintain a “cosmocentric” perspective of health in which people see themselves as nested within and impacted by a vast world of spirits (Sterlin, 2006). This is particularly the case due to the influence of Vodou, a mix of West African animism and French Catholicism that resulted in a unique syncretic religion specific to Haiti (Fils-Aimé, 2016; Métraux, 1958; Pierre et al., 2010). For this reason, many studies consider participants who report their religion as Catholic to also be followers of Vodou; however, the vast majority of Haitians continue to maintain belief systems strongly influenced by Vodou (Meudec, 2007). Therefore, Haitians’ unique religious and spiritual beliefs may be a mediating driver of mental health outcomes and of particular importance in designing and improving mental health services.

Religious belief is an important component in the functioning of daily life in Haiti (Cadichon, 2019; Jean-Charles, 2017). One researcher described Haiti as a “labyrinthian socio-religious space” that encompasses Protestantism, Catholicism, and Vodou (Tremblay, 1995). In the last century however, Catholicism has lost significant influence to Protestantism which has gained many converts in Haiti (Barthélemy, 1991). While many Haitians – particularly “born again” Protestants – often claim to despise Vodou and refuse any compromise with proponents of these practices, the vast majority of Haitians have a relationship to Vodou beliefs (Meudec, 2007; Métraux, 1958; Tiberi, 2016). It is important to highlight that Haitians have maintained these beliefs despite a long history of attacks against and stigmatisation of Vodou (Métraux, 1958; Pierre et al., 2010). As described in one of the most prominent historical works on the subject of Vodou, Le Vaudou Haïtien, author Alfred Métraux writes,

All Haitians, whatever their social status, have trembled in their youth at stories of zombi and werewolves and learnt to dread the power of sorcerers and evil spirits. Most of them, under the influence of school or family, react against such fancies but some give in to them and consult a Vodou priest in secret.

(1958, p. 87)

This continues to remain largely true today, with high levels of belief in magic, sorcery, and Vodou ritual practices (Cénat et al., 2021; Michel, 2019).

A 2012 estimate found that roughly 70% of Haitians rely on traditional medicine as their first option for care (OPS/OMS, 2012). However, this same report described how biomedical care can be difficult to access for many Haitians, which partly explains the reliance on traditional medicine. Traditional medicine however is primarily used by Haitians due to widely accepted explanatory models of illness that emphasise sent spirits (Auguste & Rasmussen, 2019; Jean-Jacques, 2019; Jean-Jacques, 2019; Meudec, 2007; Méance, 2014). A sent spirit is viewed as a spirit intentionally sent by someone supernaturally, used to explain an experience of misfortune such as an accident, illness, or death (Kaiser & Fils Aimé, 2019). As in other societies influenced by traditional African beliefs, many Haitians believe humans do not simply die without a reason; rather, if someone dies someone must have killed them, usually through act of sorcery such as a sent spirit (Graeber, 2011; Meudec, 2007). Vodou priests are often sought to treat psychological distress by mediating relationships between the living, the dead, and ancestral spirits (Wagenaar et al., 2013). The authors of this study have also conducted in-depth research examining how Vodou priests diagnose and treat sent spirits, including detailed case studies of these afflictions (Galvin et al., 2022). While there is debate about the impact of these beliefs on mental health stigma and outcomes, few other studies to date have examined the extent to which Haitians with mental disorders believe their illness is caused by sent spirits and seek treatment with traditional healers.

Goal and objectives

The overall goal of this study is to examine the experiences of patients with mental illness at the first mental health clinic in northern Haiti, as well as the relationships between demographic, social, religious and economic factors and mental health outcomes. More specifically, this study sought to (1) assess the demographic factors that impact traditional beliefs and practices with regards to mental illness etiology and treatment. Secondly, we (2) measure traumatic events experienced by patients and the relationship to mental health among this population. Lastly, we (3) study the factors associated with three mental health outcomes: depression, anxiety, and functional ability. This study was conducted with a view towards gaining deeper understand of patient populations at SSMMP with the ultimate goal of improving services as well as increasing access to high quality mental health services throughout the region.

Methods

Study sample and design

As few have studied the lives of individuals living with mental illness in northern Haiti, this purposive cross-sectional study surveyed patients about their experience with mental illness at SSMMP between 31st August 2020 and 15th February 2021. We determined that a sample of roughly 100 participants would be sufficient and in line with similar mental health pilot studies (Eustache et al., 2017; Kaiser et al., 2013; Legha et al., 2020). SSMMP clinic data between 2018 and 2019 data indicates 61% of patients were female and 39% were male (Michel, 2019). We, therefore, expected a similar breakdown in our study; however, due to difficulties recruiting men, the final month of the study was used to interview solely male patients. In addition, as male patients present with more than double the rates of psychosis as female patients at SSMMP – and often arrive for care in chains or other physical restraints put on by fearful family members – many male patients had to be excluded from participation due to inability to respond to the study questionnaire. Psychosis is determined by staff psychologists and encompasses five primary elements: confused thinking, delusions, hallucinations, changed emotions and disturbed behaviour, including violence (Ventevogel, 2016). Other studies globally have confirmed that males often represent the most serious cases – including florid psychosis – as they are the most likely to delay treatment (Bijl et al., 2003).

The main inclusion criteria for participating in this study comprised: (1) being a person of Haitian nationality of at least 18 years of age, (2) having a history of mental illness, (3) being physically and mentally capable of answering questions in the survey, and (4) having given informed consent. Patients presenting with epilepsy but no symptoms besides seizures were excluded. All patients who presented for treatment at SSMMP during the study period who met these criteria were offered the opportunity to take part. As all participants were recruited from SSMMP, researchers were also given access to the patient’s medical chart at the clinic, if agreed to by the patient. This provided additional information on diagnostic history as well as qualitative information shared during sessions with the psychologist, allowing researchers to better understand the patient’s history and background related to mental illness and treatment.

Measuring mental illness

The methods presented here are similar to other studies conducted on mental health in Haiti as well as in other LMICs (Ventevogel, 2016; Wagenaar et al., 2012). This means using a combination of existing culturally validated scales and demographic variables to assess prevalence and severity of depression, anxiety, and functional ability. In addition, other elements were included, such a measuring the number and type of traumatic events experienced. Many studies have demonstrated the association between experiencing traumatic events and mental illness (Belik et al., 2007; Martsolf, 2004; O’Brien, 1998). In the context of high levels of extreme violence as well as severe political and social unrest in Haiti over the last few years, there is a high rate of trauma-related disorders in Haiti (Auguste & Rasmussen, 2019). In particular, one study found that nearly 40% of Haitian youth have experienced traumatic events such as kidnapping, gang violence, or physical/sexual assault (Jaimes et al., 2008).

Other demographic factors that can have strong links to mental health outcomes on a population level include sex, income, education level, location of residence, marital status, number of children, and religious affiliation (Ventevogel, 2016; Wagenaar et al., 2012). As significant associations exist between physical and mental health, it is also essential to assess any ailments patients may be experiencing outside of mental health concerns, as well as a general assessment of overall physical health (Ohrnberger et al., 2017). These factors impact individuals experiencing mental illness on multiple levels (Figure 1).

Figure 1.

Factors associated with depression, anxiety, and functional ability. Text in bold refers to factors that are examined for the purposes of this study.

Data collection and measures

All instruments in this study were designed as self-report questionnaires. Due to the high illiteracy rate in Haiti – particularly in the north of the country – researchers used the questionnaires as a structured interview starting with informed consent, and then reading each survey question aloud to the participant. The average time for this interview was 45 min to one hour. To determine if patients suffer from depression, researchers used the Zanmi Lasante Depression Symptom Inventory (or ZLDSI) – a locally developed scale that borrows from several existing depression screeners plus idioms of distress and is clinically validated in Haiti (Figure A1) (Legha et al., 2020; Rasmussen et al., 2015). To determine if patients suffer from anxiety, researchers used a version of the Beck Anxiety Inventory that has been translated into Haitian Creole (Krèyol) and culturally validated (Figure A2) (Kaiser et al., 2015). Lastly, researchers used the Krèyol Functional Assessment (KFA) that was developed and culturally validated in Haiti using Bolton and Tang’s (2002) approach (Figure A3) (Kaiser et al., 2013). These scales were selected as they are among the only mental health assessment scales which are not simply translated but also culturally validated for use in the Haitian cultural context, incorporating local understandings of mental illness and idioms of distress.

In addition to these three validated scales, significant demographic data related to mental health outcomes were collected related to gender, age, location of residence, mental health diagnosis, marital status, physical health, number of children, monthly income, years of education, religious beliefs, health history, spiritual beliefs related to diagnosis, and previous visits to traditional healers for treatment (Das et al., 2007; Wagenaar et al., 2012). Lastly, traumatic events were measured using an adapted list from similar studies (Ventevogel, 2016). Location of residence was captured as the name of the town or nearest town in where the individual lived, as well as whether it is an urban or rural setting. Marital status was denoted as either single, married, or in plaçage. Haitian sociologist Laënnec Hurbon describes plaçage as “a compromise between African polygamy and Western forms of marriage” (Hurbon, 1987, p. 83). Others describe it as a form of common law marriage, however, one in which women are at a relative disadvantage in terms of regulating terms of the relationship and inheritance (Meudec, 2007; Régulus, 2012). Physical health was captured as being either “good” or “bad”. Number of children, monthly income, and years of education were noted in numerical terms.

Few other mental health studies in Haiti have collected data regarding the relationship between patient spiritual beliefs and illness etiology (Wagenaar et al., 2012). Following formative research conducted in 2019, researchers determined that it was important to capture these data among patients at SSMMP. For this reason, patients were asked if they believe their mental illness was caused by a sent spirit. In Krèyol, the phrase “Do you believe this was something someone did to you?” (Eske ou kwè se fè y’ap fè’w?) was determined to be the best formulation. Immediately after, patients were asked if they had visited a Vodou priest for treatment (Eske ou te wè ougan?) so as to make clear the visit was related to this illness.

Patients were told that they could share any additional information throughout the interview. Furthermore, at the end of the survey, patients were asked again if they had anything else they would like to share. This qualitative data, combined with health history information in the patient medical chart, allowed for supplementary data to inform and interpret quantitative analyses. Finally, participant observation at SSMMP was an important element of this study as by observing the day to day work, interacting with practitioners and clients, and observing community interactions/attitudes towards the clinic, researchers were able to gain insight into overall clinic operations.

Training and cultural validation

During two weeks in August 2020 prior to starting research, a training took place for the four researchers on the study team – one American licensed clinical social worker (LCSW) and the three Haitian psychologists at SSMMP who were going to conduct interviews with participants. This training ensured mutual understandings of study protocols and intent, as well as to confirm the final questionnaire was comprehensible for patients, culturally valid, and relevant to the study’s overall goals. The American researcher on the team is a practicing mental health provider with seven years of practice experience, and fluent in Krèyol with extensive previous research experience in Haiti, including prior experience volunteering at SSMMP.

During the first week, all researchers were provided a copy of the questionnaire and requested to provide feedback. Through these validation exercises, several necessary adaptations to the questionnaire were identified. For example, some language in the Depression and Anxiety Scales was altered so as to ensure full comprehension by patients in northern dialects of Krèyol. Additionally, other discrepancies came to light among psychologists’ understandings of “urban” and “rural”. To ensure consistency in data collection, only residents of the large cities of Cap-Haïtien, Port-au-Prince, and Gonaïves were recorded as “urban”, with residents of smaller towns recorded as “rural”. Other issues treated during these exercises related to how best to calculate of “years of education”, with psychologists having differing interpretations of how to best to quantify this based on the Haitian education system. Ensuring common sets of understandings, comprehensibility, and cultural relevance of each item adjusted for potential threats to validity in the process of developing and adapting this questionnaire. Following the training, surveys were pilot tested with patients. This process ensured that local perceptions and understandings guided instrument development.

Ethical considerations

The population of persons with lived experience of mental illness in Haiti represents a vulnerable group. The ethical issues posed by conducting research with this population was addressed at several levels in the study. First, the enrolment was conducted by the Haitian psychologist treating the patient at SSMMP. The psychologist asked patients about their willingness to be contacted by a researcher only after checking all the predetermined criteria, which includes the cognitive ability to understand the nature of the research and to consent. The psychologist also judged whether the study would potentially be harmful to the psychological well-being of the patient. Only after this preliminary screening were participants asked if they were willing to be contacted. If the answer was “yes”, they were included on a list for the researcher’s review.

Second, the consent form represented a dual consent process. The researcher explained to each person with lived experience of mental illness that to ensure that their rights would not be violated, they can designate another person who takes care of them to consent for their participation in the study. If requested, this same caregiver was also required to consent in a separate process. Third, during the interviews or observations, the researcher reminded the informants that they can choose not to answer any question, or to stop the session at any time, for any or no reason. In this sense, participation was completed voluntary, and several patients declined when offered the opportunity to participate. Patients were compensated with a 500 Haitian Gourdes ($7 US) reduction in the cost of their visit or medications for participation. This amount was determined in collaboration with Haitian psychologists at SSMMP. Institutional Review Board (IRB) approval was obtained from the Haiti National IRB – or Comité National de Bioéthique (IRB #1920-51). All research was conducted with respect to local COVID-19 guidelines.

Analysis procedures

Data analysis was performed using STATA software (version 15.0; StataCorp, College Station, TX). Univariate and bivariate analyses were used to examine the relationships between social, demographic, economic, religious and cultural factors, and Depression, Anxiety, and Functional Assessment Scales (StataCorp, 2017). Kernel density estimation plots were used to assess the normality of distributions on outcome variables. Student’s t-tests and chi-square tests were used to assess relationships that included categorical variables, and Pearson correlations were conducted to examine relationships between continuous variables. Logistic regressions were performed to measure the impact of demographic characteristics on the belief that the patient’s illness was caused by a sent spirit, and if the patient had previously visited a Vodou priest for treatment of this illness. The Hosmer-Lemeshow Goodness of Fit Test was performed to ensure a good fit for the model. In addition, model fit checks were performed to detect outliers, and the variation inflation factor (VIF) test was used to test for multicollinearity among independent variables in the model.

Ordinary Least Squares (OLS) regression models were used to assess the effects of demographic variables and exposure variables on the continuous variables: total Anxiety, Depression, and Functional Assessment Scale scores. To check for regression assumptions we ran several tests including the Breusch–Pagan/Cook-Weinberg test for heteroskedasticity, the VIF test for multicollinearity among independent variables, as well as P–P plots and Q-plots to check for normality of the error terms. With regard to identifying influential data, we performed the Cook’s distance (or Cook’s d) test to examine for influential data which indicates outliers that may have an impact on results. Using Cook’s d, four participants had extreme values that seemed to influence regression coefficients. While average number of sessions per patient was only two, these four patients had attended between 14 and 17 sessions. No other patients had attended more than nine. Thus, to examine the influence of these values, two separate models were run – with and without the four values. The results indicated significant differences between the two models and as such the final model did not retain these four outliers.

Results

Sample characteristics

This study interviewed 96 patients, including 64 women and 32 men (Table 1). Patients ranged between 19 and 79 years of age with an average age of 43 years old. Most participants in this sample live in rural areas, with a minority from urban areas, primarily the city of Cap-Haïtien. Six patients came from other parts of Haiti however, with one from Gonaïves, one from Ouanaminthe, one from St. Marc, one from St. Raphael, and two from Hinche. Patient medical charts listed a majority of patients with a diagnosis of depression or anxiety – usually a combination of the two – while 19% listed a diagnosis of bipolar disorder. Six patients also had co-morbid epilepsy. Overall education levels were low, with the average patient having only an 8th grade education. Eleven patients had no formal education, and roughly one-third of patients could not sign their name.

Table 1.

Sample characteristics (n = 96).

	N (%)	Mean (SD)
INDIVIDUAL CHARACTERISTICS
Age (min/max: 19-79)	–	43 (13.4)
Sex
Male	32 (33%)	–
Female	64 (67%)	–
Years of Education (min/max: 0–18)	–	8 (5.2)
Physical Health
Good	41 (43%)	–
Bad	55 (57%)	–
Marital Status
Single	41 (43%)	–
Plaçagea	17 (18%)	–
Married	38 (39%)	–
HOUSEHOLD CHARACTERISTICS
Location
Urban	39 (41%)	–
Rural	57 (59%)	–
Average Monthly Income in USDb (min/max: 0–322)	–	50 (77)
Number of Children (min/max: 0–8)	–	2 (1.9)
Religion
Catholic/No Religion	20 (21%)	–
Protestant	76 (79%)	–
MENTAL HEALTH CHARACTERISTICS
Diagnosis on Medical Chart
Depression/Anxiety	78 (81%)	–
Bipolar	18 (19%)	–
Number of Sessions (min/max: 0–17)	–	2 (2.4)
Number of Traumatic Events (min/max: 0–8)	–	3 (1.7)
Believe Illness Caused by Sent Spirit
Yes	72 (75%)	–
No	24 (25%)	–
Visited Vodou Priest for Treatment
Yes	40 (42%)	–
No	56 (58%)	–
Depression Scale Score (min/max: 0–42)c	–	21 (11.8)
Patients with Depression	73 (76%)	–
Patients without Depression	23 (24%)	–
Anxiety Scale Score (min/max: 0–60)d	–	25 (15.7)
Patients with Anxiety	43 (45%)	–
Patients without Anxiety	53 (55%)	–
Functionality Scale Score (min/max: 0–36)	–	15 (9.3)

^aPlaçage is a more informal, open form of marriage common in Haiti.

^bConverted from Haitian Gourdes (*.013 on February 25, 2021)

^cPatients scoring 12 and above are considered depressed.

^dPatients scoring 26 and above are considered to have moderate to severe anxiety.

Participants were also asked questions about their household, however as the concept of household is not readily understood in Haitian culture, interviewers informed participants that the household is composed of people who share living space, finances and food, and can include a boyfriend or girlfriend or a relative’s child that lives with you and that you care for (Kolbe & Hutson, 2006). Participants reported a monthly household income ranging from 0 to 240,000 Haitian Gourdes (or $0 to $322 US) with an average of 3,827 Haitian Gourdes ($50 US). Yet, over one-third of all participants reported having no income at all.

In terms of religion, the vast majority of participants reported being Protestant, including a significant minority among more radical “born again” sects such as Pentecostal, Jehovah’s Witness, Adventist and Apostolic Protestant. The remaining participants reported being Catholic or having no religion, except for one participant who identified Vodou as his religion. Three-quarters of participants said they believed their mental illness was caused by a sent spirit. In addition, 42% of patients in the sample had previously visited a Vodou priest for treatment of their mental illness. Among the patients who believed their illness was caused by a sent spirit, over half had previously visited a Vodou priest for treatment, almost always multiple times.

The mean Depression Scale score among this sample was 21. Patients scoring 12 and above are considered depressed (Rasmussen et al., 2015). Over three-quarters of participants scored 12 or above. This sample’s mean Anxiety Scale score was 25. Patients scoring 26 and above are considered to have moderate to severe anxiety, however, cut-off scores for the Beck Anxiety Inventory have yet to be fully validated for use in Haiti (Beck et al., 1996). Just under half of participants scored 26 or above. The mean score on the functional assessment was 14.

Number of traumatic events

The number of traumatic events that impacted participants’ mental health highlight significant levels of trauma among this sample (Table 2). The most common traumatic events experienced were related to deaths of loved ones with two-thirds reporting the sudden death of a loved one as a trauma contributing to poor mental health. The vast majority (87%) of participants reporting a sudden death of a loved one also reported a loved one being murdered. Additionally, one-fifth of participants reported being injured in a fight. Over one-quarter of the sample experienced a natural disaster with many having lived through the Port-au-Prince earthquake of 2010, and others reporting hurricanes or other weather-related disasters such as floods. Lack of food or water, or lack of medical care were also traumas reported by roughly a quarter of the sample. Approximately one-sixth of the sample reported lack of shelter. Over one in five women in the sample reported trauma from a difficult pregnancy and almost one in ten women reported having been raped.

Table 2.

Traumatic events experienced by participants that impacted mental health (n = 96).

Traumatic event	Male (%)	Female (%)	Total (%)
Sudden Death of Family or Close Friend	19 (59%)	44 (69%)	63 (66%)
Murder of Family or Close Friend	16 (50%)	39 (61%)	55 (57%)
Separation of Family	7 (22%)	24 (38%)	31 (32%)
Natural Disaster	12 (38%)	14 (22%)	26 (27%)
Lack of Medical Care	11 (34%)	14 (22%)	25 (26%)
Lack of Food/Water	7 (22%)	16 (25%)	23 (24%)
Injury Due to Fight	10 (31%)	8 (13%)	18 (20%)
Difficult Pregnancy	1 (3%)	14 (22%)	15 (16%)
Lack of Shelter	4 (13%)	10 (16%)	14 (15%)
Victim of Gang Violence	5 (16%)	7 (11%)	12 (13%)
Loss of Property	1 (3%)	4 (6%)	5 (5%)
Rape	1 (3%)	5 (8%)	6 (6%)
Imprisonment	1 (3%)	1 (2%)	2 (2%)

Illness etiology and traditional healing

Logistic regressions were conducted regarding the belief that a participant’s illness was due to a sent spirit and whether a patient had previously visited a Vodou priest for treatment. Results show that age was a significant predictor of both (Table 3). Specifically, younger participants were more likely to report believing their illness was due to a sent spirit and were more likely to visit a Vodou priest for treatment. With regards to believing illness was caused by a sent spirit, participants with less monthly income were more likely to believe. Lastly, men were more likely to believe their illness was caused by a sent spirit than women, and married individuals were less likely to believe compared those who are single. Hosmer-Lemeshow Goodness of Fit Test was performed on the belief illness caused by a sent spirit and whether the patient visited a Vodou priest, indicating no significant lack of fit.

Table 3.

Factors associated with believing illness is caused by Sent Spirit and visited Vodou priest for treatment (n = 96).^a

Variable	Believe illness is caused by Sent Spirit			Visited Vodou priest for treatment
	OR	SE	pb	OR	SE	p
Sex (ref: female)	8.48	7.59	.02	1.18	.61	.75
Age	.91	.03	.004	.93	.02	.01
Location (ref: rural)	2.24	1.52	.24	.74	.38	.55
Number of Children	1.33	.28	.18	1.29	.23	.16
Years of Education	.95	.07	.51	.96	.06	.53
Monthly Income	1.00	.00	.01	1.00	.00	.051
Marital Status (ref: single)	–	–	–	–	–	–
Plaçage	.39	.41	.38	2.94	2.42	.19
Married	.22	.17	.04	.82	.50	.75
Religion (ref: Catholic/No Religion)	–	–	–	–	–	–
Protestant	4.06	3.28	.08	1.02	.64	.98
Constant	116.53	219.85	.01	17.92	25.63	.04
Pseudo R2	.30	–	–	.17	–	–
X2	31.86	–	<.001	22.21	–	.01

^aLogistic Regression Model.

^bSignificant p-values are represented in bold at the p < .05 level.

Depression, Anxiety, and Functionality Scales

Linear regressions were performed for predictors of Depression, Anxiety, and Functional Assessment Scales (Table 4). Participants with poor physical health had higher levels of depression, anxiety, and lower functional ability. An increase in the number of traumatic events experienced by a patient was also associated with increased depression and anxiety. Female patients also had increased depression scores. A decrease in monthly income was also associated with a decrease in functional ability. However, as the number of sessions patients attended for treatment increased, there was a decrease in both depression and anxiety. Harmful levels of multicollinearity or heteroskedasticity were not detected in the models, revealing no significant lack of fit. R-squared values were .46, .34, and .35.

Table 4.

Factors Associated with Depression, Anxiety, and Functionality (n = 92).^a

Variable	Depression			Anxiety			Functionality
	b	SE	pb	b	SE	p	b	SE	p
Sex (ref: female)	−4.46	2.07	.03	−5.89	3.16	.07	.71	1.82	.70
Age	−.01	.09	.94	−.17	.14	.24	−.07	.08	.41
Location (ref: rural)	1.17	2.11	.58	1.64	3.21	.61	−1.48	1.85	.43
Number of Sessions	−1.69	.50	.001	−1.94	.76	.01	−.15	.44	.73
Number of Children	.03	.68	.97	.53	1.03	.61	.95	.59	.11
Years of Education	.26	.22	.23	.86	.33	.78	−.22	.19	.25
Monthly Income	−.00	.00	.05	−.00	.00	.96	−.00	.00	.001
Number of Traumatic Events	1.17	.57	.04	1.79	.87	.04	.73	.50	.15
Physical Health	−9.61	2.30	<.001	−10.61	3.50	.003	−7.48	2.02	<.001
Marital Status (ref: single)	-	-	-	-	-	-	-	-	-
Plaçage	3.28	3.02	.28	4.21	4.60	.36	.62	2.65	.82
Married	.87	2.34	.71	−.85	3.57	.81	1.83	2.06	.38
Constant	25.37	5.02	<.001	35.17	7.65	<.001	20.17	4.41	<.001
Adjusted R2	.46	-	-	.32	-	-	.35	-	-
F-value	8.07	-	<.001	4.87	-	<.001	5.46	-	<.001

^aOrdinary Least Squares (OLS) Regression

^bSignificant p-values are represented in bold at the p < .05 level

Discussion

To our knowledge, this is the first clinic-based study to examine patients living with mental illness in northern Haiti. The results highlight key findings with regards to the relationship between beliefs and explanatory models of mental illness, with large numbers of patients attributing their illness to sent spirits and visiting Vodou priests for treatment. In addition, key factors were identified with these variables in relation to age, sex, and income that impacted forms of belief and care seeking behaviours. Next, this study also exhibited important results depicting high levels of mental illness and impairment, with particularly elevated depression scores among participants. While this is to be expected in a clinical sample, the large number of severe traumatic events reported by patients in addition to physical ailments and poverty appear to play a particularly significant role in the mental distress of these patients.

Other recent studies have examined mental illness in the south of the country, particularly following the 2010 earthquake (Blanc et al., 2016; Cadichon et al., 2017; Hagaman et al., 2013). Our findings depict a population that is mostly rural, very poor, with little formal education, and living primarily in the area surrounding the city of Cap-Haïtien. The vast majority of these patients have a diagnosis of either depression, anxiety, or a combination of the two, with severe symptomology. Most of the patients in this sample also reported waiting months or years prior to seeking proper treatment, which partly explains the gravity of illness manifestations seen at SSMMP. For example, it was not uncommon for patients in this sample to report being sick for up to ten years before seeking care at SSMMP. As biomedical care has only recently become more available in the region, patients went to traditional healers as one of the only options for treatment. While many patients previously sought care from traditional healers for years at a time, some of these patients reported experiencing abuse and saw symptoms worsen rather than improve. Common forms of abuse experienced by patients at SSMMP include deep cuts, burns, or inflammation on the face and body from ritual “healings” performed by ougan to chase away harmful spirits, and were observed in several patients over the course of this study (Michel, 2019). several of these patients who experienced abuse had developed psychotic symptoms as well, several of whom were excluded from participating in this study due to an inability to respond to study questions.

Recent scholarship in the field of global mental health has highlighted the importance of differentiating clearly between disorder and distress in mental healthcare (Ventevogel, 2016). For example, many clinicians in the United States tend to see patients in distress due to social suffering related to problems in their personal or professional lives (Yapko, 1998). However, at SSMMP few patients seek care when problems are at this low level. Rather, patients who come to SSMMP virtually all suffer from a disorder – or psychiatric condition – meaning the level of care needed is much higher. Services such as those provided by SSMMP are important for meeting the needs of these patients, who are often suffering from severe mental illness.

Traditional beliefs and healing

Other studies have explored the systems of belief surrounding mental illness in the Vodou cosmology as well as cures that are performed by local traditional healers in the treatment of mental illness (Galvin et al., 2022; Meudec, 2007). In particular, those findings show how conceptions of mental illness and explanatory models in Vodou maintain perceptions and understandings of illness and disorder, as well as treatment modalities, that differ starkly from current biomedical models. For this reason, it was determined to be essential to capture mental illness etiologies and care seeking behaviours in relation to traditional healing that are common among this patient population, as belief in Vodou explanatory models remains widespread (Jean-Jacques, 2019).

There were significant findings with regards to religion and explanatory models of illness and healing in this study, a subject which few others have examined in Haiti (Blanc et al., 2016). With three-quarters of patients reporting their illness being caused by sent spirits, this differs significantly from other mental health studies in rural Haiti which reported only 10% of respondents saying spirits could cause mental illness (Wagenaar et al., 2012). This is likely due to the fact that this study was conducted in the more isolated and rural north of the country where Vodou holds significantly more sway.Methodological differences could also partly explain the difference in findings compared to Wagenaar et al., who asked a community-based sample about specific, milder forms of distress, as opposed to a clinic-based population. Some studies also highlight how Protestants in Haiti are much less likely to believe in sent spirits compared with Catholics and other groups, due to negative feelings toward Vodou (Louis, 2011; Meudec, 2007). However, in this study over three-quarters of Protestants thought their illness was due to sent spirits compared with 70% of Catholics and other groups. Most interestingly, among more radical protestant sects – such as Pentecostal, Jehovah’s Witness, Adventist and Apostolic Protestant – who are deemed to be the most intolerant of Vodou, 96% (23 out of 24 participants) thought their mental illness was due to sent spirits. This highlights the way in which the vast majority of Haitians practice a syncretic mix of Christianity and Vodou, regardless of claimed religious affiliation (Galvin et al., 2022).

With regards to visiting a Vodou priest for treatment, 42% reported seeking treatment through an ougan prior to coming to SSMMP. This is a much higher rate compared with other studies in Haiti that found only 5.8% had sought care through traditional healers for mental health problems (Eustache et al., 2017). Catholics and other groups were more likely to seek treatment from ougan, with 55% visiting a Vodou priest for treatment compared with 38% of Protestants. Another study investigating this question similarly found that Catholics and other groups were more likely than Protestants to seek care from Vodou priests (Wagenaar et al., 2013). Only one-third of more radical Protestants visited Vodou priests for treatment indicating that while they almost always believed they were a victim of sent spirits, they often did not follow through with a visit to a Vodou priest for treatment. However, SSMMP psychologists argue that more radical “born again” Protestants may have felt uncomfortable revealing their visit when they had in fact previously gone.

In terms of other findings related to traditional beliefs and healing, younger people were significantly more likely to believe their illness was caused by sent spirits, with an average age of 41 among those who believe and 49 who don’t. Similarly, younger people were also more likely to have visited a Vodou priest for treatment, with an average age of 39 among those who visited one and 46 who had not. The literature on this subject often emphasises that less educated people are more likely to visit Vodou priests (Meudec, 2007). However, as young people in this sample were significantly more educated than older – and education level among groups who believe in sent spirits and visited Vodou priests were roughly the same – there seems to be more complexity with regards to the relationship between age, education, and traditional beliefs and healing than in much of the literature to date. Yet, one relationship that is confirmed by the literature is that poorer people are more likely believe in traditional healing methods (Clérismé et al., 2003). In this study, the average income of those who believed their illness was caused by sent spirits was less than half of those who did not. Additionally, men were more likely to believe their illness was caused by sent spirits than women, likely due to a proclivity to see their mental illness as caused by someone else via sorcery as opposed to occurring naturally – and thereby potentially bearing some responsibility themselves.

The results of this study exhibit high rates of belief in mental illness etiology related to sent spirits compared to other studies which examined this subject, as described above. While this may be due to the fact that northern Haiti is a place where traditional beliefs and Vodou remain strong, it is also important to note that – like many populations in a globalised world – Haitians can often hold multiple or hybrid explanatory models for any given set of symptoms (Pierre et al., 2010). In this sense, patients can often simultaneously hold heterogenous and dynamic understandings of their mental distress (Lichtenberg et al., 2021). Additionally, simply because patients believe their illness may be caused by sent spirits does not necessarily mean they believe a Vodou priest will be able to cure them. In particular, many patients who reported believing sent spirits were the cause of their illness said they sought relief through prayer at church, arguing that they believed God could cure them (se bondye ki konn geri). Once again, this reflects the syncretism that exists between Christianity and Vodou in Haiti.

Determinants of depression, anxiety, and functional ability

Over three-quarters of this clinical sample was clinically depressed and nearly half had moderate to severe anxiety. In addition, there are vulnerabilities at each level of the theoretical framework, as noted above in Figure 1, which function as determinants of mental illness among adults in northern Haiti.

First and foremost, contextual factors at the society level of the theoretical framework highlight important effects on mental well-being. In particular, rural Haiti suffers from extreme levels of poverty (Cénat et al., 2021; Singh & Barton-Dock, 2015; Sletten & Egset, 2004). Other studies have established strong associations between low socio-economic status and psychological distress (Das et al., 2007; Smith-Fawzi et al., 2012; Wagenaar et al., 2012). A majority of patients in this study told interviewers that they didn’t have any money (pa gen lajan), reporting that they had to borrow from friends or family to pay for treatment. Many also said they couldn’t work due to their illness, and were supported by friends or family members. Through participant observation, researchers also witnessed many patients who did not have enough money to pay for the entirety of their medications, leaving them to pick and choose which ones to purchase and which to forego. In addition, patients often did not return for their follow-up visits due to a reported inability to pay.While the clinic director often provided discounted treatment or medication when patients reported they were unable to pay, it was not possible to decrease fees for everyone in need.

The types of traumatic events indicated also highlight important aspects of the larger societal context in which the patient lives. For example, two-thirds of patients reported experiencing the sudden death of a family member or close friend which negatively impacted their mental health. In addition, 87% of these sudden deaths – or 55 of 63 individuals – also reported the murder of a family member or close friend. This statistic may reflect high levels of extreme violence in Haitian society that have been growing in recent years (Brewis, Wutich, Galvin, & Lachaud, 2020), and has spiralled out of control since the death of President Jovenel Moïse in July 2021. On the other hand, as described previously, Haitians often believe that loved ones who die are killed by sent spirits and thus could be reporting loved ones as “murdered” when there was in fact no violent crime to be witnessed. However, other indicators such high levels of injury due to fights, gang violence, and several cases of rape highlight that there are in fact high levels of violence affecting mental health among patients at SSMMP. Extreme violence has been documented in other mental health-related studies in Haiti as well, and continues to severely impact the country today (Bolton et al., 2012; Martsolf, 2004; Pierre et al., 2010; Smith-Fawzi et al., 2012). Lastly, over one-quarter of patients reported being a victim of a natural disaster – usually the Port-au-Prince earthquake or a hurricane – exemplifying Haiti’s precarious ecological position in the Caribbean and the far reaching effects that natural disasters can have on mental health (Tiberi, 2016; Wagenaar et al., 2012).

Moving inward on the theoretical framework to the household and individual levels, sex was a determining factor in relation to depression, with women more likely to have higher rates of depression than men. This is a finding that has been confirmed repeatedly in Haiti and around the globe, and though there is still some debate as to exactly why (Patel et al., 2002; Rasmussen et al., 2015; Wagenaar et al., 2012; Yapko, 1998). In addition, there was also a strong relationship between lower income and decreased functional ability. Other studies using the KFA similarly found that lack of economic means is related to lower functionality (Kaiser et al., 2013). Weaker relationships were found between low functionality and number of children and years of education, however, these factors have been associated with increased poverty in Haiti (Sletten & Egset, 2004; Verner, 2008).

At the individual level, poor physical health was strongly associated with depression, anxiety, and poor functional ability, and over 60% of patients complained of a physical illness in addition to mental illness. Patients often complained of physical ailments due to incidents not always captured in traumatic events, such as head injuries from auto-mobile accidents, strokes, Parkinson’s, drug use, anemia, malaria, typhoid fever, diabetes, hypertension, acid reflux, ulcers, fever, or undiagnosed chronic pain or inflammation. Many of these patients reported believing their physical ailments were also caused by sent spirits – particularly when they were chronic – and had not sought appropriate medical care. Others complained that quality medical care was not accessible from where they lived. However, SSMMP has a primary care physician on staff and requires all patients to see the doctor at each visit, after the psychological assessment has been completed. This ensures that physical illnesses can be treated alongside mental illnesses, thereby decreasing the impact that physical disorders or pain have on mental distress. Importantly, this study also identified that the more patients return for follow-up care at SSMMP, the lower their depression and anxiety symptoms. While this study did not examine the effectiveness of mental health care treatment at SSMMP, these data are a first indication that treatment is likely having a positive impact overall.

Limitations

The findings of this study should be viewed in the context of several limitations. Firstly, there was a potential for selection bias in this study, as only patients who came to SSMMP during the study period were offered the opportunity to participate. On the one hand, since this population is overwhelmingly poor and rural with few resources to pay for mental health services, the sample may include more severe cases. However, on the other hand, as cases of psychosis and other more extreme illness were excluded from participating due to inability to respond to the questionnaire, this study perhaps captured the least severe cases seen at SSMMP. Additionally, because of the cross-sectional design of the study, a causal link between demographic factors and current mental health symptoms cannot be established. As indicated in the theoretical model, multiple factors affect depression, anxiety, and functional ability.

Secondly, with only 96 participants, the study’s sample size is relatively small. In addition, with half as many male patients as female patients, it was difficult to determine significant relationships in the data with regard to gender. However, despite difficulties recruiting male patients due to high rates of psychosis among men seeking care at SSMMP, the last month of the study was spent recruiting only men so as to increase this portion of the sample. Next, other studies using these instruments in Haiti have mentioned the ambiguity inherent in many terms used in Depression, Anxiety, and Functionality Scales as clinicians and researchers encounter different understandings among patients based on context (Kaiser et al., 2013). Additionally, as this study interviewed patients who often had severe mental illness, ensuring mutual understandings was at times difficult. Lastly, as previously described, social desirability bias may have influenced some participant responses, particularly with regards to traditional beliefs and experiences seeking treatment with Vodou priests.

Conclusion

This study uniquely examined the characteristics of patients at the first and only mental health clinic in the North of Haiti. We found evidence of strong associations between factors that influence depression, anxiety, and functional ability on multiple levels in this population. In particular, high levels of traumatic experiences or other adverse events influenced many of the patients’ illnesses. In addition, there were important phenomena related to traditional beliefs and healing that impacted views of illness etiology and care seeking behaviours. As many patients believed their illnesses – both mental and physical – were due to sent spirits, Vodou priests were often the first recourse when seeking treatment, with some patients delaying appropriate biomedical care for years. Overall, our study points to the importance of quality mental health treatment such as that offered by SSMMP in Northern Haiti for the first time. While further studies are needed to better elucidate factors impacting mental health in Haiti, current high levels of violence and unrest present significant obstacles to researchers thereby preventing additional investigation.

Appendix

Figure A1.

Zanmi Lasante Depression Symptom Inventory (ZLDSI).

Figure A2.

Beck Anxiety Inventory.

Figure A3.

Krèyol Functional Assessment (KFA).