CHAT SA: Modification of a Public Engagement Tool for Priority Setting for a South African Rural Context

Background: Globally, as countries move towards universal health coverage (UHC), public participation in decision-making is particularly valuable to inform difficult decisions about priority setting and resource allocation. In South Africa (SA), which is moving towards UHC, public participation in decision-making is entrenched in policy documents yet practical applications are lacking. Engagement methods that are deliberative could be useful in ensuring the public participates in the priority setting process that is evidence-based, ethical, legitimate, sustainable and inclusive. Methods modified for the country context may be more relevant and effective. To prepare for such a deliberative process in SA, we aimed to modify a specific deliberative engagement tool – the CHAT (Choosing All Together) tool for use in a rural setting. Methods: Desktop review of published literature and policy documents, as well as 3 focus groups and modified Delphi method were conducted to identify health topics/issues and related interventions appropriate for a rural setting in SA. Our approach involved a high degree of community and policy-maker/expert participation. Qualitative data were analysed thematically. Cost information was drawn from various national sources and an existing actuarial model used in previous CHAT exercises was employed to create the board. Results: Based on the outcomes, 7 health topics/issues and related interventions specific for a rural context were identified and costed for inclusion. These include maternal, new-born and reproductive health; child health; woman and child abuse; HIV/AIDS and tuberculosis (TB); lifestyle diseases; access; and malaria. There were variations in priorities between the 3 stakeholder groups, with community-based groups emphasizing issues of access. Violence against women and children and malaria were considered important in the rural context. Conclusion: The CHAT SA board reflects health topics/issues specific for a rural setting in SA and demonstrates some of the context-specific coverage decisions that will need to be made. Methodologies that include participatory principles are useful for the modification of engagement tools like CHAT and can be applied in different country contexts in order to ensure these tools are relevant and acceptable. This could in turn impact the success of the implementation, ultimately ensuring more effective priority setting approaches. The CHAT tool was modified for use in the Agincourt Health and Socio-Demographic Surveillance System (HDSS) study area (https://www.agincourt.co.za/), located in Bushbuckridge municipality in Mpumalanga province. The site, typical of rural areas in SA, is comprised of 31 villages, 20 000 households and a population of approximately 111 500. 35,36 There are 2 health centers and 6 satellite clinics in the site and 3 district hospitals within 20-60 km. Pipe-borne water is not available to most households and sanitation systems are poor. Electricity is available in all villages, but is unaffordable for most and few tarred roads exist. Every village has at least one primary school and most have a high school but the quality of education is poor 37 and unemployment rates are high with labour-related out-migration commonly occurring. Life expectancy at birth is 61 for males and 70 for females, 35 with significant socioeconomic disparities across different indicators. 38 In order to modify CHAT for use in Agincourt HDSS study area we followed a 5-step approach: First a rapid review 39 of national health policy documents was conducted to identify national health topics/issues and related interventions that were a priority for the country. The starting points were the most recent national health policy documents in 2017: the transcript of the 2017 SA treasury budget speech where health spending was mentioned, and the NHI White Paper of 2015. 40 We then selected other national health documents that included any of the health topics/issues identified in the budget speech or the NHI White paper. Eleven documents were included. Finally, we included provincial and district level policy to identify any health topics/issues and related interventions specific to the Bushbuckridge rural context which had not been identified from the national documents. Three additional documents were included. The documents that were included in the review are shown in Supplementary file 1. The documents were scanned in their entirety and any specific interventions related to the health topics/issues were identified and captured in a Microsoft Excel sheet. Next, we conducted a focus group discussion (FGD) with each of the following groups: home-based care (HBC) service providers in the Agincourt HDSS study area, provincial and district level experts, and national policy-makers, in order to identify which health topics/issues and related interventions (solutions) each group thought were important. The health topics/issues related to categories of disease (eg, HIV/AID and tuberculosis [TB]) or focus area ( eg, mental health) or broader issues like “access,” while the interventions (solutions) related to specific activities and services that would address the health topics/issues, for example, the provision of contraceptives at schools. The first FGD (HBC FGD) comprised HBCs from Bushbuckridge who were selected from 3 home-based carer organizations. We selected home-based carers as representatives of the community because in the South African context the carers are members of the communities who either volunteer or are paid a small stipend to perform basic care and support services within the home environment. The 3 specific organizations were convenience samples based on proximity to facilitate transport to a central location. Only HBCs who could speak English were selected but most HBCs who work in Bushbuckridge do speak some English. Seventeen people were invited and the final group comprised 13 participants including 11 females and 2 males. For the second FGD (Prov/Distr FGD) 12 participants were originally invited and final group comprised of 7 (6 females and 1 male), 3 were provincial-level policy-makers, 3 district-level decision-makers and one public health specialist located in Bushbuckridge. Participants were selected using purposive sampling to ensure diversity across directorates within the Department of Health (DOH). For the third FGD (National FGD) 11 people were invited and the final group comprised 8 senior national-level policy-makers selected using purposive sampling to ensure a broad representation from different directorates within the National DOH. There were 2 females and 6 males. We provided a board with a blank wheel that comprised coloured slices, and participants were given sticky notes to write down 2 major health topics/issues. Each participant had an opportunity to present their topics/issues and add them to the board. Health topics/issues that overlapped were grouped together and given an overarching title (eg, Access) by agreement amongst the participants. Additional topics/issues identified in step 1 (desktop review) that were not mentioned by the group, were described by the facilitator and the group decided whether they wanted them included. Following this, another round was conducted where participants wrote solutions (interventions) to address the topics/issues using sticky notes in a similar manner (Supplementary file 2). The approach fostered strong engagement and allowed the research team to refine the health topics/issues and related interventions identified in step 1. The FGDs were recorded, transcribed and analyzed qualitatively to identify topics/issues and interventions (solutions). During the FGDs participants were asked to vote on each health topic/issue. Topics/issues that received the highest number of votes were included in a follow up ranking process using email to each individual participant from HBC FGD and Prov/Distr FGD in order to reconcile the differences between the FGDs and to refine the list of topics/issues for the Bushbuckridge context. Three out of 6 participants from the Pov/Distr FGD completed the follow up ranking while 6 out of 13 participants responded from the HBC FGD. We used the Borda count method to determine the overall ranking score. 41 The Borda count method is considered a plausible approach in aggregating individual ranked preferences. The ranking was 1-13; we counted how many times each topic/issue was ranked 1-13 by all the participants. We multiplied this number by the ranking number (1-13) and then added this up to determine the total Borda count. The total Borda counts closest to zero were the ones that were ranked the highest. We initially selected the top 10 health topics/issues but some were combined and/or dropped based on being too broad or already featuring within other identified topics/issues. The top 7 health topics/issues were selected for the final CHAT board. The specific interventions under each of the topic/issues were refined and finalized by referring back to the qualitative data from the FGD and the desktop review. A total number of 70 interventions across the 7 health topics/issues were costed. Costing included both program and patient-related costs. Actuarial costs were estimated by factoring in likely utilization rates. For population estimates and epidemiological parameters we searched the literature from the Agincourt Research Unit. Ad hoc searches were also undertaken in PubMed, Embase, and Science direct for relevant literature on population health in Bushbuckridge. The literature were collated and assessed for relevant information that could be used to populate our costing template. Prices/unit costs of the components were collected from a variety of sources. Where cost information for the public sector was not available, we relied on the latest legal tariff document for the public sector. Public sector costs were assumed to be 70% of private sector costs. Medication costs were extracted from the South African National Health Laboratory Service Report 2018. For Bushbuckridge-specific parameters we relied on district level policy documents. Cost components and unit costs were developed based on researching existing programs that offered similar interventions. For information on costs of education and information provision we relied on expert consultations with stakeholders at the National Department of Health and with non-governmental organizations who have extensive experience in public awareness campaigns. An Excel sheet was developed in Microsoft Excel to aggregate cost components. Interventions were expressed as a percentage of total costs and were converted to a relative number of stickers. Our starting point was 0.5% = 1 sticker using an existing actuarial model developed by Milliman (https://us.milliman.com/en), an international actuarial company that has experience in the adaptation of CHAT, but some allocations were revised based on the judgement of the authors. This was reasonable because the costing exercise already relied on some assumptions due to lack of data and it was important that the final sticker value represented the relative costs of interventions as accurately as possible. Where expert opinion was needed in order to verify the relative costs we reached out to individuals who were familiar with these costs. The stickers represented the monetary resources that would be required when specific interventions were selected. An overview of key data categories and sources are shown in Supplementary file 3. Interventions were grouped together and categorized using the common classification of level of care for health interventions used in SA: Health promotion, prevention, diagnosis (screening), treatment, rehabilitation, and palliative care. 30,40 Where categories overlapped they were merged (eg, prevention and screening). The final CHAT SA board was developed based on the results from the steps above and was translated into the most widely spoken vernacular language in the study area (Shangaan). The icons and design elements were developed in conjunction with a medical artist at the National Institute of Health in the United States (Figure 1). Final CHAT SA board. Abbreviations: CHAT SA, Choosing All Together South Africa; TB, tuberculosis. A user manual was also developed which explained each category of intervention in detail by listing the specific interventions as well as context specific scenario cards to demonstrate the consequences of choices made during the CHAT exercise. These cards were developed drawing on the qualitative data that emerged from the FGD with the home-based carers to ensure the scenarios were appropriate for and relevant to the context. The materials were translated into the local language of Shangan and were checked by 2 individuals familiar with this language. CHAT SA was tested with a group of 11 community members (9 females and 2 males) from Bushbuckridge. This test phase was carried out in order to observe how participants interacted with the tool and the supporting materials and to ensure the tool allowed for meaningful rationing considerations. Materials were adjusted slightly and finalized post the test. Data were analyzed using thematic content analysis for the desktop review and qualitative analysis for the FGDs. Initial themes were developed by identifying the topics/issues that emerged as priorities from the Budget Speech and NHI White Paper. Following this, the policy documents were reviewed to identify themes (topics/issues) that related to those from Budget Speech or NHI White Paper as well as any that were specific to Bushbuckridge municipality or Mpumalanga province. Codes were reviewed by all authors. The FGDs were recorded, translated and transcribed in English. Data were analyzed qualitatively using thematic coding and codes were reviewed by all authors. For the FGDs we first identified themes that corresponded with those from the desktop review then identified any new topics/issues that emerged as separate themes. Sub-themes were developed to identify specific solutions/interventions under each theme and were classified according to promotion (education); prevention; diagnosis (screening); treatment; rehabilitation; palliative care.