Background: Intellectual disability (ID) is described as a neurodevelopmental disorder which occurs during the developmental period and impacts intellectual as well as adaptive functioning across social, cognitive and practical domains. Approximately 3 percent of the South African population has mild to severe forms of ID. It is well described that caregivers of children with ID and other developmental disorders have higher levels of stress related to caring for their child than parents with typically developing children. However, little research has been reported in the South African context. Objectives: 1. To determine the nature and extent of self-reported symptoms of stress in caregivers of children with intellectual disabilities. 2. To establish which demographic variables and child factors are associated with carer stress. 3. To compare the nature and extent of self-reported stress and demographic and child variables associated with stress in carers of children with intellectual disability with the same measures reported by carers whose children are developing typically. Methods: This was a purposive, descriptive analytical study. Participants were 59 caregivers whose children attended Red Cross War Memorial Children’s Hospital outpatient clinics. The ID group comprised 35 caregivers of children with ID. The control Non-ID group had 24 caregivers of typically developing children. The children’s ages ranged from 2 to 10 years. Caregivers were administered a set of four questionnaires: a demographic questionnaire; the Parent Stress Index; the Hospital Anxiety and Depression Scale and the Aberrant Behaviour Checklist that rated the children’s behaviour. The data were analysed using IBM SPSS. Results: Both the ID and Non-ID groups showed elevated scores for self-reported anxiety. The ID group of caregivers reported significantly more problem behaviours in their children than the caregivers in the Non-ID group. Hyperactive behaviour predicted for depression in the caregiver ID group which also reported higher levels of stress and significantly higher levels of depression than the Non-ID group. 49 percent of the ID group caregivers reported levels of depression in the clinical range. Further, the ID group reported significantly more financial difficulties and also more appointments for their child at health facilities than the control group. Conclusion: Carers in this study setting experienced similar stresses and described similar child behavioural problems as those reported internationally. A significant finding was the high rate of caregivers of children with ID whose depression scores were in the clinical range. Child behaviour was a significant factor associated with caregiver stress and anxiety in both study groups and with depression in caregivers of children with ID. The study findings have implications for the mental health and behavioural support needs of both caregivers of typically developing children and caregivers of children with ID and for a range of services that provide this support.

Background: Research predominantly from high income countries suggests that depression and anxiety are highly prevalent among prison inmates. With limited available research from low and middle income countries, this study aimed to estimate the prevalence of possible anxiety and depression among Malawian inmates and identify factors associated with these conditions. Methods: This cross-sectional study was conducted at Thyolo prison situated in the southernmost region in Malawi, from February to March 2018. A total of 378 male prison inmates were interviewed face-to-face using a general questionnaire examining sociodemographic characteristics, prisoners’ previous and current involvement with the prison services, and self-reported mental health problems. Validated screening instruments were used, including the Center for Epidemiologic Studies Depression scale (CES-D) to measure depression; while the Beck Anxiety Inventory (BAI) was used to measure symptoms of anxiety. Multiple logistic regression analyses were conducted to determine the association between anxiety, depression and related variables. Results: The prevalence of possible anxiety disorder and depressive disorder was 62.7% and 72.5% respectively. After adjusting for the effects of the other variables in the model, a multiple logistic regression found that anxiety was significantly associated with no support (AOR 1.65 95% CI: 1.02-2.66) and a previous mental disorder diagnosis prior to imprisonment (AOR 2.07 95% CI: 1.05-4.07); while depression was significantly associated with no support (AOR 5.19 95% CI: 2.66-10.14), and having two or more traumatic events (AOR4.30 95% CI 1.65-11.23). However, being divorced (AOR 0.12, 95% CI: 0.04-0.35) was found to be a protective factor for depression. In addition to this, no support (AOR 2.24 95% CI: 1.36-3.69) and a previous mental disorder diagnosis prior to imprisonment (AOR 2.10 95% CI: 1.05-4.22) were significantly associated with comorbid depression and anxiety. Conclusion: This study provided valuable insights into the prevalence and risk factors associated with possible depression and anxiety among prison inmates in Malawi. Therefore, regular screening of prison inmates for early detection and treatment of mental disorders would be beneficial to improve the quality of life of prison inmates.

Background The assessment of a patient's individual needs offers many benefits and it is essential for planning and implementing services and interventions. Need is a subjective concept and may be defined from several perspectives. Patient, carer and staff interests may differ, influencing their perspectives in defining needs. Traditionally, the staff perspective on needs has taken priority but the steady growth of the 'user movement' and 'recovery philosophy' has led to this being challenged. This study aimed to establish patient, carer and staff perceptions of patient need, the extent to which these perceptions were homo- or heterogeneous, and what factors were associated with local perceived needs. Methods The study was informed by a systematic review of the literature focused on the individual needs of people with severe mental illness assessed from multiple perspectives. Patients, carers and staff on the Valkenberg Hospital assertive community treatment (ACT) service were assessed using the Camberwell Assessment of Need Short Appraisal Schedule to evaluate needs. Patient global functioning and current levels of psychopathological symptoms were assessed using the Global Assessment of Functioning scale and the Positive and Negative Syndrome Scale respectively. Kappa statistics were computed to assess agreement in the participants' perspectives.

Background: Intimate partner violence (IPV) during pregnancy is a common phenomenon across the world. The present study sought to determine the prevalence and predictors of intimate partner violence among pregnant women attending a midwife and obstetrics unit in the Western Cape. Methods: A convenience sample of a hundred and fifty pregnant women (n= 150) attending antenatal appointments at the Mitchell's Plain Midwife and Obstetrics Unit (MOU) were asked to participate in this study. Consenting women participated in an interview where they were asked questions concerning interpersonal violence and other psychosocial constructs, such as history of childhood trauma, exposure to community violence, depressive symptoms and alcohol use. Frequency distributions and descriptive statistics were calculated for categorical and continuous variables. Multivariable logistic models were developed to control for socio-demographics and psychosocial constructs. The first model was based on report of any form of IPV over the previous 12 months, while the remaining three models investigated the disaggregated forms of IPV: physical abuse, sexual abuse and emotional abuse. Results: Overall, the lifetime and 12-month prevalence rate for any IPV was 45% and 32%, respectively. For 12-month IPV, 32% reported general abuse, 29% physical and 20% reported being sexually abused. After adjusting for the effects of the other variables in the model, depressive symptoms, and reporting that this pregnancy was unplanned were significantly associated with the reporting of any IPV in the past 12 months. Looking specifically at 12 months general IPV, women who had depressive symptoms were more likely to experience some form of general IPV (OR= 6.42, CI 2.51-16.41) than women not at risk. Also, women of 'coloured' race were more likely to experience general IPV than Black African respondents (OR= 1.46, 95% CI 0.64-3.34). The model exploring associations for 12-month physical IPV found women who were at risk for depression were more likely to experience physical IPV (OR= 4.42, CI 1.88-10.41) than women not at risk, while the model exploring associations for 12-month sexual IPV found that women who reported experiencing community violence were more likely to report sexual IPV than women who reported no exposure to community violence (OR= 3.85, CI 1.14-13.08). Conclusion: This is the first study, which illustrates high prevalence rates of IPV among pregnant woman at Mitchells Plain MOU. A significant association was found between 12-month IPV and unintended pregnancy. Also, woman who are at risk for depression were found to have an increased chance of experiencing both general and physical IPV. Sexual IPV was associated with high levels of exposure to community violence. Further prospective studies in different centres are needed to address generalisability and the effect of IPV on maternal and child outcomes. Greater recognition of IPV in pregnancy could contribute to improved antenatal care, as well as enhanced policy development for appropriate intervention strategies. Key Words: Intimate partner violence; Interpersonal violence; Domestic violence; Abuse; Pregnancy; Antenatal; Postnatal depression and Community violence

In the antiretroviral era children perinatally infected with HIV (PHIV+) are surviving into adulthood and are at risk for emotional and behavioural problems. Few studies of these problems have been conducted in low and middle income countries (LMIC) and even fewer in sub - Saharan Africa where the burden of the HIV epidemic remains heaviest. The aims of this study were to provide a quantitative description of emotional and behavioural problems in a group of children and adolescents with perinatally acquired HIV in South Africa compared to a group of well - matched HIV - negative controls and to identify demographic, biological, cognitive and contextual correlates of emotional and behavioural problems. A cross-sectional descriptive, analytical study was conducted. Participants were recruited from community and hospital based clinics. Wi thin the HIV - infected group, children were further divided into three subgroups: PHIV+ children who had never been on antiretroviral therapy (ART naïve), PHIV+ children on ART and PHIV+ children on ART with a confirmed diagnosis of HIV - related encephalopathy (HIVE) . Emotional and behavioural problems were assessed using the Child Behaviour Checklist (CBCL). Several measures were used to assess demographic, biological, cognitive and contextual correlates of problem behaviours: socio - demographic questionnaire, clinic records, neuropsychological test battery, Family Resource Scale, Family Support Scale and Center for Epidemiologic Studies - Depression Scale. Children were compared by HIV status on demographic, cognitive and contextual variables as well as the total and subscale scores of the CBCL. Multivariate comparisons of the influence of contextual and cognitive variables on CBCL total problems was performed using a hierarchical step - wise linear regression analytic procedure. The final sample (N=108) for data analysis included 78 PHIV+ children and 30 HIV - negative children. Groups were comparable with respect to demographic and contextual variables. Cognitive performance scores were significantly lower in the PHIV+ children when compared to HIV - negative controls (p<0.001). Rates of caregiver depression were higher in both groups than the lifetime prevalence rates reported in the South African population.

Background Mental and substance use disorders cause significant disability worldwide. In spite of the availability of evidence-based treatment, non-adherence rates remain high in people with severe mental illness. Mental health services are however under-resourced, especially in low- and middle-income countries. Interventions that employ task shifting, the delegation of health care delivery tasks to less specialized health workers, have the potential to address this resource shortage. Community health workers, while an established and important delivery agent for task shifting in many forms of chronic illness, including mental illness, have lacked access to standardized structured training in mental health. Together with novel approaches such as mobile health, task-shifting interventions have the potential to improve adherence and clinical outcomes for MHSU, thus reducing the burden on stretched mental health resources. While the evidence for the effectiveness of task shifting interventions is growing, it is unclear whether the combination of a task shifting intervention with mobile health would be acceptable and feasible in low resource settings. It is also unclear to what extent a structured mental health training programme would result in improved knowledge, confidence and attitudes amongst community health workers. Methods First, I conducted an appraisal of current evidence for interventions delivered by non-specialist workers for mental illness in Sub-Saharan Africa. The aim was to characterize the types of such interventions that have been carried out in Sub-Saharan Africa, to ascertain extent of use of non-specialist workers; the outcomes explored; any acceptability and feasibility findings; as well as any efficacy outcomes. Second, I developed and piloted two task shifting interventions geared at improving care for severe mental illness in Cape Town, and evaluated their acceptability, feasibility and preliminary effectiveness. Systematic review: For the systematic review, eligible studies published prior to 21 June 2017 were identified by searching the Cochrane library, PsychInfo, and Medline databases; as well as the World Health Organization International Clinical Trials and Pan African Clinical Trials Registries. The bibliographies of study reports for all eligible trials were scanned for additional studies. Included trials were those of interventions a) delivered by non-specialist health workers for b) adult populations (18-65 years) with c) psychiatric disorders diagnosed in line with ICD or DSM classification systems in d) Sub-Saharan Africa. No restriction was placed on the nature of the psychiatric disorder. Pilot randomized controlled trial: A pilot randomized controlled trial was conducted, in which 77 participants with severe mental illness were recruited from Valkenberg psychiatric hospital in Cape Town, with 42 randomized to receive the intervention and 37 to receive treatment as usual. In the intervention arm, a treatment-partner selected by the participating MHSU underwent a psychoeducation and treatment-partner contracting session. The intervention pair then received two text message reminders of clinic visit appointments monthly. The primary outcomes were acceptability and feasibility of the intervention, measured through qualitative interview and process evaluation at 3 months post-discharge. Secondary outcomes for efficacy were 1) adherence to the first clinic visit; 2) any readmission in the 9 months following discharge; 3) quality of life; 4) symptomatic relief; and 5) medication adherence. These efficacy measures were conducted at baseline and again at 3-month study review. Between-group comparisons were done using an intention to-treat ANOVA analysis for efficacy outcomes. Community Health Worker Training Intervention: My second task shifting intervention was a quasi-experiment evaluating whether structured mental health training would improve the knowledge and skill of community health workers while improving their confidence and attitudes towards mental illness. A training programme was developed in partnership with the Western Cape Department of Health, and piloted with 58 community health workers who had not previously received mental health training. Mental health knowledge and skill were measured though the use of case vignettes and the Mental Health Knowledge Schedule (MAKS). Confidence was measured using the Mental Health Nursing Clinical Confidence Scale (MHNCCS), while attitudes were measured using the Community Attitudes towards the Mental Ill Scale (CAMI). Measures were conducted at baseline, at the end of the training, and again 3 months after the end of training for the knowledge and skill measures. Daily evaluation questionnaires were used to establish acceptability, and a training evaluation questionnaire was used to obtain further acceptability data, as well as to establish feasibility of the training intervention. T-tests and regression models were used to test changes in questionnaire scores before and after each intervention, adjusting for baseline scores. Quantitative data were entered and analysed using STATA 10.0 for the pilot randomized controlled trial and the R statistical programme for the CHW intervention, while qualitative data were managed and analysed using NVIVO 8, a qualitative analysis programme for all analyses, for which a grounded theory approach was used, followed by thematic analysis. Ethics and registration: Ethical approval was obtained from the University of Cape Town Human Research Ethics Committee, Faculty of Health Sciences for the treatment partner and mobile health intervention (HREC REF: 511/2011) and for the community health worker training intervention (HREC 913/2015). Both interventions were registered on the Pan African Clinical Trials Registry (PACTR201610001830190 and PACTR201610001834198 respectively). Finally, Health Impact Assessment Unit clearance was obtained from the Western Cape Department of Health for both trials (RP168/2011 and WC_2016RP59_635 respectively). The systematic review was registered on the International prospective register of systematic reviews (PROPSERO) (CRD42017065190)). Results Systematic Review: Due to heterogeneous methods and treatment outcomes, a meta-analysis was not possible. A narrative synthesis is thus presented. Fifteen trials of interventions delivered by non-specialist workers (5087 participants) were identified. In each of the trials, the intervention was acceptable and feasible, with preliminary efficacy findings favouring the interventions. Pilot randomized controlled trial: The treatment partner and text message intervention components were acceptable. While the treatment partner and psychoeducation components were feasible, the text message component was not, as a consequence of several socioeconomic and individual factors. While efficacy outcomes favoured the intervention, they did not reach statistical significance due to the small sample size. Community Health Worker Training Intervention: Mental health knowledge improved as demonstrated by improved diagnostic accuracy on case vignette response. Sixty-three percent of participants demonstrated improved accuracy in making a diagnosis, with a roughly two-fold increase in performance in these individuals. There was a significant increase in the average scores on the Mental HeAlth Knowledge Schedule pre- to post training (t = -4.523, df = 55, p < 0.001, N=56). This improvement was sustained at 3 months after the end of training assessment scores (t = -5.0, df = 53, p < 0.001, N = 54). There was a significant increase in the average Confidence scores pre-intervention (mean SD): 45.25 (9.97) to post-intervention 61.75 (7.42), t-test: t = -8.749, df = 54, p < 0.001, N=58). Attitude scores (n=45) indicated no change in authoritarian attitudes [mean (SD): Pre 27.87 (2.97); Post 26.38 (4.1), t = 2.720, p-value = 0.995], while benevolence [mean (SD): Pre 37.67 (4.46); Post 38.82 (3.79), t = -1.818, p-value = 0.038] and social restrictiveness [mean (SD): Pre 24.73 (4.28); Post 22.4 (5.3), t = -2.960, p-value = 0.002] attitudes showed improvement pre- and post-training, as did tolerance to rehabilitation of the mentally ill in the community (t = 2.176, p-value = 0.018). Participants responded well to training, appraising it as acceptable and appropriate to their work. They expressed a need for a longer training programme with further training on substance use and geriatric disorders. Stakeholder participation was consistent and contributed to the feasibility of the intervention. Conclusions A review of task shifting interventions by non-specialist health workers indicates that these have yielded positive outcomes for mental health service users in published trials. Such interventions have the potential for reducing the mental health treatment gap in low and middle income countries in a cost-efficient way. Further work is however required to develop specific treatment approaches for particular disorders, and to assess the outcomes of such interventions, including cost-efficiency measures. The measures of outcome used in this field remains somewhat disparate; the development of a common research agenda may assist in developing and replicating further investigations and generalising findings. A treatment-partner intervention is acceptable and feasible in a low- and middle-income setting such as ours. Careful work is, however, needed to ensure that any additional components of such an intervention, such as mobile health, are tailored to the local context. Appropriately powered studies are needed to assess efficacy. Structured training in mental health is acceptable and feasible in our setting. The training intervention led to an improvement in knowledge and skill amongst community health workers while improving confidence and attitudes. Participation of policy stakeholders was key in ensuring the success of the intervention. There is a need for interventions evaluating the outcomes of community health worker training to provide more detailed descriptions of their training interventions. More focus must be placed on measuring service and end-user outcomes to improve the rigor and quality of such investigations, with well-powered randomized controlled trials being best placed to answer questions regarding efficacy and cost-effectiveness. In summary, my systematic review, and my pilot task-shifting interventions in the South African context indicate that task shifting interventions such as these are acceptable and feasible, offering a promising solution to addressing the under-resourcing of mental health care. However, interventions should ideally be tailored to the specific communities they target, taking into account specific individual, community, technological, and sociodemographic factors. Future training interventions should provide more detailed descriptions of programme components and focus on measuring patient outcomes, while all task shifting interventions may benefit from incorporating an evaluation of cost effectiveness. Task shifting presents a viable and accessible opportunity for creative innovation and as we work towards achieving mental health for all.

Background: Domestic homicide (killing of a person aged 16 or older by a family member or a current or former partner) accounts for 50% - 70% of homicides perpetrated by offenders with mental illness. Despite these statistics, surprisingly little is currently known about the characteristics of domestic homicides perpetrated by those with severe mental illness. To the best of our knowledge, domestic homicide in the context of severe mental illness has not been researched in South Africa. Objective: To investigate domestic homicides by offenders with severe mental illness referred to the Forensic Mental Health Service at Valkenberg Hospital for forensic psychiatric observation. Methods: A five-year retrospective folder review was conducted to obtain data on the characteristics of offenders and victims, as well as the circumstances surrounding the homicide. Results: The majority of the offenders in our sample were young (mean age of 31), single, unemployed males who were known to mental health care services. Substance use disorders and non-adherence to medication were common. Psychotic disorders were the most prevalent diagnoses. The majority of victims were male and a significant minority of the domestic homicides were parricides (28.6%). The incident took place at the victim's residence or the victim and perpetrator's shared residence in most cases. Stabbing was the most common method used. Almost half of the perpetrators were psychotic when the incident took place and 60% of these were first episode psychoses. In spite of the high prevalence of substance use disorders (66.7%), only 23.8% of the sample reported that they were intoxicated when they committed the offence. Conclusions: The majority of our sample was known to mental health care services. This implies that there were potential missed opportunities to prevent these lethal assaults. Our research identified treatment adherence, comorbid substance use disorders and aggressive treatment of first episode psychosis as a possible focus of future interventions in order to prevent domestic homicides due to mental illness.

Includes abstract. / Includes bibliographical references. / Electroconvulsive therapy (ECT) involves the administration of an electrical current to the brain in order to produce a tonic-clonic seizure which is deemed therapeutic. It is an effective and safe procedure for the treatment of severe mental illnesses such as major depression, mania and schizophrenia. Currently little is known about the characteristics of ECT practice in South Africa. This study aims to determine current electroconvulsive therapy (ECT) practice and to compare it with reported ECT practice internationally. This is a retrospective, descriptive study, to determine the characteristics of ECT practice in South Africa; data was collected using a self-report questionnaire. The study population consisted of doctors and nurses who practiced ECT in any 12 month period between 2011 and 2012. Both private and state facilities were included in the study. Initially contact was made with hospital mental health facilities to ascertain whether an ECT machine was present on site. Once formal approval was obtained from the appropriate designated bodies, questionnaires were sent to clinical staff involved in ECT at active sites. The 36-item questionnaire covered relevant questions on: utilization rates, equipment, staffing, practice and monitoring parameters, and indications for use. Forty two institutions had an ECT machine on site, of which thirteen institutions reported non-use. Questionnaires were sent to the 29 active ECT sites. Facilities responding to the questionnaire amounted to 83% (n=24), but of these, 21 units responded to the ECT utilization questions. ECT is performed as a modified procedure in six provinces by psychiatrists, registrars, medical officers and general practitioners. In-and outpatient ECT is offered in 79% (n=19) of hospitals. The number of persons treated with ECT/10 000 population per year (ppy) is 0.22 while the number of ECT procedures/10 000 ppy is 1.19. More patients in the private sector receive ECT as a treatment modality than in the public sector (U = 22, p = 0.045). ECT is performed in a minor theatre/operating room in 79% of units, while the rest is performed in a treatment room. All but one unit had a separate recovery room. Informed consent or assent was used in all institutions. Pre-ECT work-up most commonly involved a physical examination (95.5%, n = 21) and basic blood work investigations (87%, n=20). Bilateral, unilateral and bifrontal electrode placements are used, while various dosage- determination and monitoring methods are employed. The vast majority of patients (89.22%, n=869) receiving ECT are between the ages of 18 and 59. The most common indication for ECT is depression (84.77%, n=796). The utilization rate in South Africa is similar to that of countries like Bulgaria, Poland and India, but less than that of some high-income countries. Even though ECT practices in South Africa generally follow international guidelines, standardisation of practice is still recommended.

Over the past decade, calls to address the increasing burden of mental, neurological and substance-use (MNS) disorders and to include mental health care as an essential component of universal health coverage (UHC) have attracted mounting interest from governments. With the inclusion of mental health in the 2015 Sustainable Development Goals (SDGs) there is now a global policy commitment to invest in mental health as a health, humanitarian and development priority. Low and middle-income countries (LMICs) such as South Africa, contemplating mental health system scale-up embedded into wider SDG- and UHC-related health-sector transformations, must address a number of key mental health financing policy considerations for attaining population-based improvements in mental health. Despite ongoing transformations in the South African health sector, there has been an implicit neglect of the integration of mental health services into general health service development. This has been driven in part by a lack of locally-derived evidence in several areas, including: the economic basis for investing in mental health, the current resourcing of the mental health system, opportunities for improved efficiency and equity, and how reforms may be structured and paid for in light of the country's ongoing efforts to implement a National Health Insurance (NHI) scheme. This thesis therefore attempts to address these gaps and aims to generate new knowledge on the economic costs, impacts and financing strategies for mental health in South Africa. This aim is achieved by fulfilling the following research objectives: 1. To examine the impact of social, national and community-based health insurance on health care utilization for MNS disorders in low- and middle-income countries. 2. To examine the policy context, strategic needs, barriers and opportunities for sustainable financing for mental health in South Africa. 3. To quantify public health system expenditure on mental health services, by service level and province, and to document and evaluate the resources and constraints of the mental health system in South Africa. 4. To examine the household economic costs and levels of financial risk protection associated with depression symptoms in South Africa. In the first part, the systematic review reports on the impact of social, national and community based health insurance on health care utilization for MNS disorders in LMICs, published until October 2018. As a secondary goal, the systematic review identifies whether there are any specific lessons that can be learnt from existing approaches to integrate mental health care into financing reforms towards universal health coverage. In the second part, a qualitative examination of the policy context, strategic needs, barriers and opportunities for sustainable financing for mental health in South Africa was conducted through a situational analysis that was complimented with a synthesis of key stakeholder consultations. The findings provide recommendations for how scaled-up mental health services can best be paid for in a way that is feasible, fair and appropriate within the fiscal constraints and structures of the country. In the third part, the thesis then empirically quantified public health system expenditure on mental health services, by service-level and province for the 2016/17 financial year, and documented and evaluated the resources and constraints of existing mental health investments in South Africa through a national survey; achieving one of the highest sample sizes of any costing study conducted for mental health in LMICs. In the fourth and final part, a household survey study was conducted to determine the level of financial protection for persons living with depression symptoms in the Dr. Kenneth Kaunda health district of South Africa, which is serving as a pilot site for the NHI. The household economic factors associated with increased depression symptom severity on a continuum are reported; and demonstrate that financial risk protection efforts are needed across this continuum. The thesis concludes by synthesizing findings towards an improved understanding of the key lessons that can be learned from other LMICs toward sustainable financing for mental health; the economic burden of inadequate mental health care to households in South Africa; and the efficiency of existing mental health investments and inequities in resourcing and access. Through this lens, and borrowing from the experiences of other LMICs, recommendations for key priorities for health service and financing reforms towards the scaled-up delivery of mental health services in South Africa are generated. The thesis is presented as papers embedded in a narrative that includes an introduction and synthesis discussion. Four papers (3 published and 1 under review) form the basis of the results chapters.

Background: More than 13% of the global burden of disease is estimated to be due to neuropsychiatric disorders, with over 70% of this burden in low- and middle-income countries. Characterised by severe shortages of human and material resources, formal mental health services alone are inadequate to meet the burden of mental disorders in low- and middle-income countries. New community models and innovative ways of increasing community participation and systematic delegation of specific tasks to other community level professionals have been recommended. Available evidence documents historic clergy involvement in health and wellbeing issues, but they have rarely been viewed as a partner in community mental health care. Aim: This study examines the clergy's conception, recognition of and responses to people with mental illnesses. The purpose of the study is to inform the potential roles and contributions of the clergy to community mental health either as the only contact or as a step in to formal mental health care. Method: Twenty eight in-depth interviews were conducted with clergy from ten church denominations in Harare, Zimbabwe. A framework analysis approach was used for thematic analysis. Nvivo 10 qualitative data software was used to organise the data. Results: Mental illness was conceived as a multifactor phenomenon attributed to both natural (biological and psychosocial) and supernatural (malevolent and benevolent spiritual) causes. Spiritual factors were a dominant theme in both the clergy's views on the causes of, and in their management of mental illness. The clergy were regularly consulted on a variety of emotional and psychological problems. Assistance was readily provided for these problems by all denominations, despite professed capacity gaps in the recognition and management of mental illness, and lack of appropriate training in basic mental health issues. Basic mental health training was recommended by the clergy to enhance clergy capacity for mental health awareness raising, recognition of mental disorders, brief problem focused counseling, and for improving collaborative management for initial and continued informal and formal health care and support. Implications of clergy conceptions, current responses and the perceived role of the church in community mental health are discussed.