Background: Community health worker (CHW) programs have been positioned as a way to meet the needs of those who experience marginalization and inequitable access to health care, and current global health narratives also emphasize their adaptable nature to meet growing health burdens in low-income settings. However, as CHW programs adopt more technical roles, the value of CHWs in building relationships with clients tends to be overlooked. More importantly, these programs are often reframed and redeployed without attending to the interests and needs of program clients themselves. We set out to gather perspectives of program and CHW engagement from clients of a maternal and child health program in rural South Africa. Methods: We conducted 26 interviews with pregnant or recently-delivered clients of the Enable Mentor Mother program between February–March 2018. After obtaining informed consent, a trained research assistant conducted all interviews in the clients’ home language, isiXhosa. Interviews, translated and transcribed into English, were organized and coded using ATLAS.ti software and thematically analyzed. Results: We found that clients’ home-based interactions with Mentor Mothers were generally positive, and that these engagements were characterized by two core themes, instructive roles and supportive relationships. Instructive roles facilitated the transfer of knowledge and uptake of new information for behavior change. Relationships were developed within the home visit setting, but also extended beyond routine visits, especially when clients required further instrumental support. Clients further discussed a sense of agency gained through these interactions, even in cases where they chose not to, or were unable to, heed their Mentor Mother’s advice. Conclusions: These findings highlight the important roles that CHWs can assume in providing both instructive and supportive care to clients; as deepening relationships may be key for encouraging behavior change, these findings pinpoint the need to bolster training and support for CHWs in similar programs. They also emphasize the importance of integrating more channels for client feedback into existing programs, to ensure that clients’ voices are heard and accounted for in shaping ongoing engagement within the communities in which these programs operate.
Ethical approval for this study was granted by Stellenbosch University’s Health Research Ethics Committee (N16/05/062). An additional file includes further ethical considerations and reflections from the authors. This study was descriptive in nature and utilized semi-structured qualitative interviews with pregnant women and recently delivered mothers who were clients of the Enable Mentor Mother program, further referred to as Enable. This study was linked to a larger study of the implementation of Enable during its first three years, part of ongoing efforts to strengthen the quality of care delivered to mothers and infants in the region where Enable operates [25, 26]. The O.R. Tambo District, where Enable is based, is among the poorest districts in South Africa. In addition to high unmet need for health services, many of Enable’s clients live in remote rural areas, where accessing health facilities can be challenging, and resources unevenly distributed [27]. Enable is the only program of its kind to operate in this specific area. With regards to maternal and child health, the O.R. Tambo District has significantly higher rates of maternal and neonatal mortality than the national average [27]. A large number of mothers and caregivers are able to access government child support grants, which many of them rely on for basic child care needs given high rates of unemployment and rural-urban migration [28]. Enable is a home visiting intervention focused on maternal and child health and nutrition; the program has been operating in the O.R. Tambo District since 2016. The program uses the Philani Mentor Mother model, which was originally designed and implemented by Philani Nutrition Project in peri-urban Cape Town [29]. In this model, Mentor Mothers (MMs) are identified by established community leaders as mothers who have managed to raise healthy children despite significant adversities. Recruited mothers undergo a six-week training, followed by an evaluation, before a subset are selected to work in their own communities. MMs recruit and follow-up on pregnant women, as well as specifically targeting families with underweight children, to provide supportive, preventive care in the home for up to five years. While Enable adheres to Philani’s model of intervention training and content, the program was adapted to be a “social franchise” of the original model, in which a model is applied in a new setting, supervised and implemented by a different organization [25]. While a MM may have up to 50 or 60 maternal clients, each client is enrolled and visited by only one MM, although these visits occasionally involve supervisors or program coordinators. MMs are distributed across a wide geographical area, each covering their own set of clients within a given village. A list of all clients in the Enable program was obtained from program staff. In order to ensure a diverse sample of interviewees, two clients per Mentor Mother were included as a target sample. The first author (CL) purposively sampled clients from each MM’s caseload, while ensuring variation in characteristics by age and number of children. Alternative clients were contacted in cases where a client was uninterested or unavailable. Interviews were conducted between February and March 2018, and were arranged and completed by an isiXhosa-speaking researcher with extensive qualitative interview experience (VN). CL devised a draft interview schedule and met with VN to discuss each question and refine or clarify phrasing where needed. The semi-structured interview schedule was then finalized and used to guide interviews. Topics covered included: client experiences with their own individual Mentor Mother; knowledge gained since enrolling in the program; possible avenues for improving the program; and experiences with the larger health system. Informed, written consent was obtained before interviews began, and all interviews were conducted in participants’ homes and audio-recorded with consent. Both informed consent and the interview itself were conducted in participants’ first language, isiXhosa. Interviews averaged one hour duration each. Rigorous quality control measures were taken to ensure consistency and alignment with core questions. In debriefings following the interviews, CL and VN reviewed interviews, and were satisfied that the target number of participants allowed for saturation. Saturation was assessed by the presence of rich transcripts, with coherent information across interviews, and the understanding that additional information gathering would likely lead to redundancy in the data [30]. Between April and October 2018, an experienced isiXhosa-speaking team reviewed audio recordings, and simultaneously translated and transcribed interview content into English. Specific words or short phrases were left in isiXhosa on a case-by-case basis to preserve meaning, and alternative definitions or explanations were added where appropriate. CL met with the transcription team regularly to discuss progress. A senior member of the transcription team checked 50% of transcripts over the course of the transcription period for quality and accuracy. Thematic analysis was employed, using the methods described by Braun and Clarke [31, 32]. Data were organized and coded using ATLAS.ti qualitative software. All transcripts were read closely before coding for familiarization, and were later coded inductively [33]. The topics covered, while aligned with questions in the interview schedule, varied substantially among interviewees. A total of 98 codes were initially developed to summarize content across the transcripts. Following several rounds of collation, review and refinement, 51 codes formed the final code list. Initial themes were generated as described by Braun and Clarke to examine viable candidate themes, and these themes were reviewed against the dataset to ensure alignment, before final themes were defined, named, and narratively written up [32]. A second independent coder (SG) read three interviews (11.6% of interviews) and discrepancies were resolved through discussion between SG and CL.