Developing a task-sharing psychological intervention to treat mild to moderate symptoms of perinatal depression and anxiety in South Africa: a mixed-method formative study

Background: Symptoms of depression and anxiety are highly prevalent amongst perinatal women in low-resource settings of South Africa, but there is no access to standardised counselling support for these conditions in public health facilities. The aim of this study is to develop a task-sharing psychological counselling intervention for routine treatment of mild to moderate symptoms of perinatal depression and anxiety in South Africa, as part of the Health Systems Strengthening in sub-Saharan Africa (ASSET) study. Methods: We conducted a review of manuals from seven counselling interventions for depression and anxiety in low- and middle-income countries and two local health system training programmes to gather information on delivery format and common counselling components used across task-sharing interventions. Semi-structured interviews were conducted with 20 health workers and 37 pregnant women from four Midwife Obstetric Units in Cape Town to explore perceptions and needs relating to mental health. Stakeholder engagements further informed the intervention design and appropriate service provider. A four-day pilot training with community-based health workers refined the counselling content and training material. Results: The manual review identified problem-solving, psychoeducation, basic counselling skills and behavioural activation as common counselling components across interventions using a variety of delivery formats. The interviews found that participants mostly identified symptoms of depression and anxiety in behavioural terms, and lay health workers and pregnant women demonstrated their understanding through a range of local idioms. Perceived causes of symptoms related to interpersonal conflict and challenging social circumstances. Stakeholder engagements identified a three-session counselling model as most feasible for delivery as part of existing health care practices and community health workers in ward-based outreach teams as the best placed delivery agents. Pilot training of a three-session intervention with community-based health workers resulted in minor adaptations of the counselling assessment method. Conclusion: Input from health workers and pregnant women is a critical component of adapting existing maternal mental health protocols to the context of routine care in South Africa, providing valuable data to align therapeutic content with contextual needs. Multisector stakeholder engagements is vital to align the intervention design to health system requirements and guidelines.

We conducted the study in four Midwife Obstetric Units (MOUs) situated in Community Health Centres (CHCs) in Cape Town. These facilities serve four low socio-economic communities situated in urban and peri-urban areas in the Cape Town Metropolitan health district, located between 20 and 40 kms from the city centre, with a population size ranging between approximately 33,200 to 44,400 per facility catchment area [32, 33]. The majority of the individuals in these communities describe themselves as ‘coloured’. In South Africa the term ‘coloured’ is commonly used to refer to individuals who are of mixed-race ancestry [34]. While the predominant language is Afrikaans, isiXhosa- and English-speaking community members from neighbouring areas also attend these facilities. Antenatal care is predominantly provided at MOUs by nursing staff and non-specialist health workers. The Department of Health WBOTs, managed by professional nurses, receive referrals from designated CHCs and provide community-based integrated health care services at household level. We used a qualitative study design with triangulation of data from multiple sources to inform the development of a task-sharing counselling intervention (see Fig. 1). Overview of the four-phase study process We used a four-phased process. The first phase involved a review of task-sharing counselling intervention and health system training manuals. For the second phase we conducted qualitative interviews with health workers and pregnant women; and the third phase entailed engagements with the Western Cape Department of Health (WCDoH) managers from each of the four health sub-districts, facility-based management teams and NPO co-ordinators. Findings from phase one to three informed the draft version of the counselling intervention and training manual. Phase four then consisted of a four-day classroom-style pilot training with CHWs and supervisors from one NPO, followed by a feedback session. Learnings from the pilot study informed the final version of the intervention design and training manual. For phase two, we recruited pregnant women and health workers through purposive sampling at four MOUs. At each facility, pregnant women were screened by trained fieldworkers. Those women who screened positive for symptoms of depression and/or anxiety (using the Edinburgh Postnatal Depression Scale) [35] or experiences of violence (using a bespoke questionnaire), during the recruitment period, were invited to participate in a 30-min audio-recorded semi-structured interview. The mean age of the pregnant women was 29 years. Half of them were either employed or self-employed while the other half were unemployed and dependent on family to meet their basic needs. All except one woman had experienced a previous pregnancy. Professional and lay facility-based health workers who provided care to pregnant women, were invited to participate in individual 30 to 60-min audio recorded interviews. Twenty health workers [including operational managers, antenatal care nurses, lay counsellors (consisting of HIV and breastfeeding counsellors), health promotion officers and mental health nurses] participated in the study. In phase four we selected CHWs and supervisors from one NPO (providing community-based support to the pilot study site) to take part in training for the intervention protocol. The inclusion criteria for CHWs were: minimum of three years’ experience, completion of the WBOT in-service training programme and interest in mental health counselling. Five CHWs with two of their supervisors (professional nurses) met the inclusion criteria and we invited them to a 3-day class-room style training. The supervisors were then invited to participate in an additional day of supervision training. In phase one we reviewed two systematic reviews of task-sharing psychological treatments for low- and middle-income countries. In addition, we reviewed and reported on the content of seven task-sharing counselling manuals that met our inclusion criteria. The inclusion criteria for the manual review were: individual perinatal psychological interventions for the treatment of depression and anxiety, designed for delivery by non-specialist health workers, in LMICs with proven effectiveness. First, we inspected studies included in the Clarke et al. systematic review of perinatal psychosocial task-sharing interventions for CMDs [15]. Second, we inspected studies in the Singla et al. systematic review of task-sharing psychological interventions for LMICs [15, 36]. In addition, we supplemented our findings from these reviews with maternal mental health intervention manuals developed for the South African context, obtained from experts in the field such as the ASSET principal investigator and co-investigators of the African Focus on Intervention Research for Mental Health (AFFIRM) and Perinatal Mental Health Project (PMHP)). Further to expert recommendations, we sought to broaden the scope of the manual review to include broader task-sharing depression counselling treatments and interventions that integrated transdiagnostic evidence-based task-sharing approaches adapted to local cultures and contexts in LMICs. District and facility-based health system stakeholders shared in-service training programme guides, that allowed us to explore standard training methodologies and guidelines. The final manual review included six manualised evidence-based task-sharing counselling interventions for LMICs (AFFIRM, Thinking Health Programme, Friendship Bench, Healthy Activity Programme, Problem Management Plus, Trauma-Focused Cognitive Behavioural Therapy) [37–42] and one basic perinatal counselling skills guide for health workers in South Africa (PMHP basic counselling guide for health workers) [19]. In addition, two health systems in-service training guides for HIV and community-based counselling were included in the review. The conceptual model for understanding the cultural and contextual determinants and manifestations of depression and anxiety was built on two areas of previous work: (1) on Kleinman’s theory of explanatory models that suggests that social and cultural contexts have a strong influence on the explanatory models that individuals or groups use to make sense of their illness experience [43, 44] and (2) on an understanding of the critical role of the social determinants of mental health in shaping the mental health of populations [45]. Therefore, it is essential to consider context-relevant social determinants and the mental health views of health service providers and users when planning a mental health counselling intervention. In phase two, we conducted semi-structured interviews with 37 pregnant women and 20 health workers at the four MOU sites from September 2018 to February 2019. The interviews were conducted in a private room in the participants’ preferred language (English, Afrikaans or isiXhosa). Interview questions for pregnant women included: the participant’s understanding and experiences of depression and anxiety, perceived causes of distress, experiences of violence or abuse, perceived support and personal coping methods. Interview questions for health workers included: the participant’s work experience and mental health and counselling training, understanding and description of depression and anxiety, perceived causes of these symptoms and common forms of violence as observed in their work with women during and after pregnancy (perinatal period). In phase three we engaged various stakeholders through presentations, workshops and meetings. The stakeholders included: (1) a DoH working group that consisted of managers of each of the four health districts in the city of Cape Town, (2) the facility- and community-based managers, and (3) managers and staff from the four MOUs. Feedback from these interactions were recorded in meeting minutes and notes. In phase four, we conducted a four-day pilot training that consisted of 3 days counselling training with CHWs and one-day training with supervisors. The training programme included activities such as scenario exercises, trainer demonstrations, participant role-plays and group discussions. Through these activities, the trainer was able to observe how well the participants engaged with the manual content, the terminology used and the level of difficulty of counselling concepts. Two additional post-training feedback meetings with supervisors informed further content adjustments to the intervention protocol. For each of the manuals identified in the manual review, we extracted data using the following preselected categories: therapeutic goals, counselling techniques, number of sessions, frequency of delivery (monthly or weekly), delivery method (group or individual) and additional in-session techniques. For phase two, we used a thematic framework analysis approach to analyse the interview transcripts [46]. The semi-structured interviews with health workers and pregnant women were transcribed verbatim by experienced bilingual transcribers. Interviews conducted in Afrikaans and isiXhosa were first translated into English before analysis began. The thematic framework was guided by the interview topic questions and further emergent themes were captured through in-depth reading of the transcripts. Transcripts and data were managed using NVivo 12 Pro qualitative data analysis software (QSR International Pty Ltd). Three researchers were involved in the coding process of the health worker interviews and each analysed a third of the transcripts. Two researchers were involved in the coding process of pregnant women interviews and each analysed half of the total number of transcripts. In both coding processes, 10% of all transcripts were randomly selected for double-coding and analysed to establish inter-coder reliability. In phase three, we incorporated feedback from managers and healthcare workers into the draft version of the counselling intervention. This version was then used during the pilot training. In phase four, we documented the pilot training findings in the form of training observations by the trainer during counselling role-plays, verbal participant feedback during group discussions, a debriefing session with supervisors after training and personal training notes. Minor manual content adjustments were made throughout the training process and tested during training. Learnings from the pilot study informed the development of the final version of the counselling intervention.

N/A

• Dima Health