Article Text

Original research
Evaluating the implementation of group empowerment and training (GREAT) for diabetes in South Africa: convergent mixed methods
  1. Robert James Mash,
  2. Darcelle Schouw
  1. Division of Family Medicine and Primary Care, Stellenbosch University, Cape Town, South Africa
  1. Correspondence to Professor Robert James Mash; rm{at}sun.ac.za

Abstract

Objective Type 2 diabetes is a leading contributor to the burden of disease in South Africa. Primary care is struggling to support self-management and lifestyle change. Group empowerment and training (GREAT) for diabetes is a feasible and cost-effective intervention in our setting. This study aimed to evaluate the implementation of GREAT for diabetes.

Design A convergent mixed-methods study evaluated a range of implementation outcomes: acceptability, appropriateness, adoption, feasibility, fidelity, reach and cost.

Setting Ten primary care facilities from a district in all nine provinces of South Africa.

Participants Descriptive exploratory individual semistructured interviews were conducted with 34 key stakeholders from national policy-makers to primary care providers. Three focus group interviews were held with 35 patients.

Results The National Department of Health saw GREAT as an acceptable and appropriate intervention, but only five of the nine provinces adopted GREAT. District-level and facility-level managers also saw GREAT as an acceptable and appropriate intervention. Factors related to feasibility included physical space, sufficient staff numbers, availability of resource materials, the health information system, adaptation to the model of care (selection of facilities, patients, adjustment of patient flow and appointment systems, leadership from local managers and the whole clinical team) and inclusion in systems for quality improvement. No major changes were made to the design of GREAT and fidelity to the session content ranged from 66% to 94%. Incremental costs were US$494 per facility. Due to disruption from the COVID-19 pandemic, only 14 facilities implemented and reached 588 patients at the time of evaluation.

Conclusion Key lessons were learnt on how to implement GREAT for diabetes in a middle-income country setting. The findings informed the design of a programme theory using a health system framework. The programme theory will guide further scale-up in each province and scale-out to provinces that have not yet implemented.

Question This study focused on evaluating how to implement GREAT for type 2 diabetes in primary care and to take it to scale in South Africa.

Finding The findings led to a programme theory on how to successfully implement GREAT for diabetes in the South African context.

Meaning The study demonstrates relevant contextual factors that need to be considered in the implementation of group diabetes education programme in a middle-income country.

  • primary care
  • general diabetes
  • health education
  • health services

Data availability statement

Data are available on reasonable request. Data can be made available on reasonable request to the authors and subject to the conditions of informed consent.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • The implementation of group empowerment and training (GREAT) and its evaluation were significantly disrupted by the COVID-19 pandemic for a period of 2 years.

  • After the pandemic, evaluation was conducted 3 months after reimplementation of GREAT instead of the intended 6 months.

  • It was difficult, therefore, to observe as many sessions as originally intended and quantitative data were limited in scale.

  • It was possible to interview stakeholders from national to local level and to reach saturation of data as well as interview stakeholders from provinces that did not implement.

  • It was not possible to evaluate sustainability, and this will need a further study.

Introduction

Diabetes is a major contributor to the burden of disease in South Africa and the leading cause of mortality among women.1 The new National Strategic Plan for non-communicable diseases (NCDs) recognises the importance of diabetes and sets new goals in the form of a cascade: 90% of adults will know if they have raised glucose, 60% of those with raised glucose will receive intervention and 50% of those receiving an intervention will achieve control of their diabetes.2 In the Western Cape, 75% of patients do not reach the glycaemic target (glycosylated haemoglobin (HbA1c)>7%).3

Achieving the glycaemic target in people with type 2 diabetes requires attention to both treatment and self-management.4 Treatment is specified by the essential drug list5 and the provision of medication is relatively good in South African primary care. Patient education and counselling, however, are limited by the high workload, lack of expertise in lifestyle modification6 and relative lack of resources for NCDs, when compared with other diseases, such as HIV and tuberculosis (TB).7 The National Strategic Plan emphasises the need to educate and empower people, and support health literacy and behavioural interventions.2

Group empowerment and training (GREAT) for diabetes was developed as an intervention in the Western Cape.8 Local primary care providers recommended a group approach to deal with the large numbers of patients in the public sector.9 They suggested four sessions of group empowerment that would help people with poorly controlled type 2 diabetes and those newly diagnosed. The intervention was developed as suggested, and an initial evaluation found it to be cost-effective10–12 and feasible to implement.13 14 There is also good global evidence for the effectiveness of group education for diabetes.15 The World Diabetes Foundation funded the scale-up of the intervention to other provinces in South Africa with support from the National Department of Health (NDOH).

Most studies on diabetic group education focus on the effectiveness of the intervention and very few studies directly evaluate how to implement and take such interventions to scale, particularly in low-income and middle-income contexts.16–19 The Diabetes Education for Self-Management of Ongoing and New Diagnosed programme is a major group education programme in the UK, but the evidence library focuses on effectiveness and not how to implement.16 Evaluation of the implementation of the Diabetes Education and Prevention with a Lifestyle Intervention programme in the USA only considers the opportunity costs of the facilitators and patient satisfaction.17 Evaluation of implementation of a coordinated care model in patient-centred medical homes in the USA focused on measuring the reach of the different activities and which elements were sustained after the grant period.19 Only one study was found that adopted a systematic implementation science approach for a community-based group exercise and education programme for people with type 2 diabetes in New Zealand.20 No such studies were located within low-income and middle-income countries.

The main aim was to evaluate implementation of the GREAT for diabetes programme in primary care facilities in South African provinces and to identify key contextual factors that influenced implementation. The evaluation was intended to help refine the programme theory and further scale-up within these provinces.

Methods

Study design

Implementation science is ‘centrally concerned with getting evidence into practice’.21 Methodology, therefore, focuses either on developing process models that guide the process of translating research into practice or on explaining what influences implementation outcomes or evaluates actual implementation.21 This was a convergent mixed-methods study that evaluated a menu of implementation outcomes derived from the work of Peter and Proctor22 23: acceptability, appropriateness, adoption, feasibility, fidelity, reach and costs. The operationalisation of these outcomes in this study was defined as follows:

  • Acceptability: Do stakeholders perceive that it is worth doing? What are the factors for and against this? How does it align with emerging policy on chronic care?

  • Adoption: Do stakeholders decide to collaborate and adopt the intervention? What are the key factors they considered in making this decision?

  • Appropriateness: Do stakeholders perceive that the GREAT for diabetes programme is fit for purpose and relevant to the different settings? What aspects are a better fit or a worse fit?

  • Feasibility: When healthcare workers are trained to implement GREAT in different settings what happens? How feasible is it to implement successfully? What are the factors that enable and hinder implementation in different settings?

  • Fidelity: How do healthcare workers modify or customise the intervention to their local settings? Why was this necessary? What were the implications of this for the delivery of content, use of resources and communication style/skills?

  • Reach: How was the GREAT programme embedded in the organisation of care? How often is the programme run? Who is targeted? How many patients are engaged?

  • Costs: What were the incremental costs (costs that were in addition to the normal budget)?

Quantitative data were used to evaluate reach, costs and some aspects of feasibility and fidelity. Qualitative data explored acceptability, appropriateness, adoption, feasibility and fidelity.

Study setting

South Africa has 9 provinces and 10 departments of health (1 national and 9 provincial). Policy is set at a national level, but budget allocation and programmatic priorities are decided on at a provincial level. Implementation of policy and provincial plans is the responsibility of district management teams. There are 52 districts across the country.

The public sector primary care services are nurse driven with support from doctors and cater for the needs of approximately 80% of the population who do not have insurance.24 Medication is defined by an essential drug list and clinical management by the standard treatment guidelines and adult primary care guidelines.5 25

The intervention

The ‘Living GREAT with diabetes’ programme consisted of four sessions, led by a facilitator over 1–2 hours, with 10–15 patients with type 2 diabetes, in a venue that allowed interaction. The four sessions focused on (1) what is diabetes, (2) lifestyle change, (3) understanding medication and (4) avoiding complications. The programme expected facilitators to be health promotion officers or nurses.

Each session was structured and described in detail in the training manual. It required the facilitator to adopt a guiding style derived from motivational interviewing. This guiding style emphasised collaboration and interaction, evoking ideas and solutions from the group rather than telling people what to do, exchanging information rather than just giving information and respecting people’s viewpoint and choices.

A number of resource materials supported implementation. The facilitator received a training manual and two other resource materials to support their knowledge of diabetes.26 27 The facilitator was trained to use a variety of pictures (a flipchart), card games and tools as part of the group sessions. Facilitators were provided with a limited supply of patient handouts from the ichange4health programme to support behaviour change and reinforce learning in the group.28 These were not considered essential to the delivery of the sessions.

Implementation strategies

The implementation strategies are described in online supplemental table 1. The table uses Powell’s typology of strategies to classify and describe the intended strategies.29 30 Powell identified 68 discrete strategies that are organised into 6 implementation processes: planning, educating, financing, restructuring, managing quality and attending to policy context. Our eight strategies described in online supplemental table 1 were from the planning and educating processes: build buy-in, obtain formal commitment, obtain a mandate for change, conduct educational meetings, train the trainer, identify and prepare champions, provide ongoing consultation and conduct ongoing training.

Facilitators were very positive about the training workshops and commented on a paradigm shift in terms of their approach to behaviour change and communication skills. They also improved their knowledge of diabetes. Online supplemental table 2 summarises their feedback. The educational approach was effective by including theory, modelling and simulated practice with feedback. Trainees were also able to discuss the challenges they experienced with behaviour change counselling.

The intended ongoing training did not happen because the COVID-19 pandemic led to a lockdown just days after the initial training workshops. The budget was then used to revisit each province at the end of 2021 and beginning of 2022 to retrain and relaunch GREAT. Data on implementation were then collected 12 weeks later.

Data collection

The research coordinator (DS) made an initial visit to each province to assess the extent of implementation and interview key stakeholders at district and provincial levels. A research assistant was then employed to collect data on reach, observe actual sessions, and interview staff and patients at the facility level in each province.

Descriptive exploratory qualitative interviews

People’s experience of the GREAT intervention was explored by means of either focus groups (face to face for patients) or individual semistructured interviews (for health services staff, virtual or face to face) using interview guides. The interview guide for health service staff included open questions to explore acceptability, adoption, appropriateness, feasibility, fidelity, reach and to identify key costs. The interview guide for patients included open questions to explore acceptability and appropriateness of the GREAT sessions, any issues that impacted its feasibility from their perspective and feedback on how the sessions had changed their self-management and behaviour.

We planned 5 focus group interviews with patients and 26 key informant individual interviews: 1 national policy-maker, 9 provincial managers and then 1 district manager, facility manager and facilitator in each of the 5 provinces that implemented (15 in total). In addition, we interviewed the lead trainer and coordinator of implementation for the GREAT project.

Course evaluation

Facilitators completed an end of course questionnaire on their experience of the training (particularly what they had learnt and what could be done better).

Observation of GREAT

Whenever possible the research assistants observed sessions as they were visiting the districts. They used a structured tool to observe fidelity to the content and structure of the sessions as well as facilitation and communication skills. The observer rated the delivery of content in each session according to the manual on a scale from not done (1), partly done, (2) or fully done (3). The observer rated the facilitation skills on a Likert scale from poor (1) to excellent (5) for ‘deals effectively with loud versus quiet people’, ‘minimises negative interactions and conflict’ and ‘enhances motivation to change’. The guiding style (evocation, empathy, choice and control, collaboration) and communication skills (use of questions, reflective listening and information exchange) were observed and evaluated on a scale from 1 to 4 with clearly defined anchors for each point on the scale. The observation tool is included in online supplemental material.

Evaluation of reach

Research assistants visited each facility to collect data on reach. They used a structured questionnaire to collect data on how GREAT was implemented and how many groups/patients had been reached. Where possible they collected the registers of GREAT sessions from facilities.

Evaluation of costs

Data were collected on setup and incremental operational costs from the project’s financial records. District and facility managers were asked about incremental costs. Costs were estimated per facility.

Data analysis

Qualitative data analysis

Recorded interviews were transcribed verbatim and checked for accuracy. Transcripts were analysed by RJM and DS with the help of Atlas-ti software (V.8), using the five steps of the framework method31:

  1. Familiarisation: The researchers read the transcripts, listened to the tapes and identified key issues and ideas.

  2. Coding index: The researchers each identified potential codes from the issues and ideas identified in step 1. They integrated and defined the codes and organised them into categories.

  3. Coding: The researchers coded all the transcripts using the coding index and if necessary added additional codes.

  4. Charting: Codes were grouped together in categories and reports derived from Atlas-ti that collated all the data together.

  5. Interpretation: The researchers interpreted the reports for key themes that related to the implementation outcomes.

Quantitative data analysis

Data were collected on paper and then captured by DS on Excel sheets. Data were checked for accuracy and, if necessary, verified with the research assistants. Data were then descriptively analysed in Excel. Data from the training workshop questionnaires were categorical and analysed as frequencies and percentages. Sessions were observed and the subcomponents of each session scored on a scale of not done (score 0), partly done (score 1) and fully done (score 2). An overall fidelity percentage score was calculated for each session based on the maximum possible score if all components were always fully done (100%). A mean score for each component of each session was also calculated and interpreted as poor (score<0.5), moderate (score 0.6–1.4) and good (score>1.5). Categorical data from the facilities on implementation are reported as frequencies (denominator is <20) and numerical data as medians and ranges.

Patient and public involvement

Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research.

Results

Overall, we interviewed 34 key informants from the health services, which included the lead implementer for the GREAT project, a national policy-maker, 16 facilitators, 7 facility managers, 6 district managers and 3 provincial-level managers (online supplemental table 3). In addition, we conducted 5 focus group interviews with a total of 35 patients (online supplemental table 3). In the last few interviews with key informants, we did not identify any new themes in the data and concluded that data saturation was achieved. The four provincial managers in the provinces that did not implement declined to be interviewed but were asked via email to explain why they did not adopt GREAT and these emails were included in the qualitative data.

Appropriateness, acceptability and adoption

The NDOH saw the initiative as appropriate and acceptable, as it would empower people and improve quality of life. Five of the provinces agreed to adopt the initiative because it enabled person-centredness, social support, patient education, behaviour change and prevented complications. It was seen as evidence based and well aligned with national policy and primary care guidelines. They commented that the correct protocol was followed by coming to them via the NDOH:

So you came at a time when we were saying it’s possible for NCDs to do the same, use the same strategy that I used by HIV/AIDS. And GREAT was the best in terms of how you can assist our healthcare workers to be able to feel at ease and also be approachable from the patient perspective. (PDOH KZN)

The four provinces that did not adopt the programme, agreed that it was worthwhile implementing, but articulated a number of reasons for not participating: they did not receive the communications from NDOH and the GREAT team, there was no NCD coordinator, internal politics thwarted the decision-making process and not having enough funds:

The reason why the Department did not participate on the program is that the person that coordinated the program went on retirement and there was not coherent coordination of the program. (Director Health Support, Free State)

At the district level, managers were aware that the number of patients with diabetes was increasing and that more needed to be done to empower people. They anticipated that the initiative would also improve adherence to medication and even reduce the amount of medication prescribed.

Meetings with the facility managers ensured buy-in and addressed organisational issues. Health professionals perceived the group approach as more efficient and less repetitive than trying to counsel people individually. Clinicians supported the approach, particularly family physicians, who were responsible for clinical governance. The guiding style of communication was seen as more appropriate than the directive approaches currently used.

Feasibility and fidelity

Running the group sessions

Table 1 shows the fidelity of facilitators to the content of the group sessions. Fidelity ranged from 66.0% of the expected content for session 3 (on medication) to 93.8% for session 4 (on avoiding complications). Most of the core content was delivered with good fidelity (mean score≥1.5). The card game on myths and beliefs, closing the sessions, and discussing alcohol, tobacco use and stress, only scored moderate fidelity. A language barrier was only reported once in the Western Cape, as the facilitators did not speak Xhosa. The sessions in KwaZulu-Natal scored the lowest as the facilitators did not follow the session structure and it appeared that untrained facilitators were also used.

Table 1

Fidelity to the GREAT sessions

Facilitation skills were rated as good (N=10). Facilitators dealt effectively with loud versus quiet people (mean score 3.7), minimised negative interactions and conflict (mean score 3.7) and enhanced motivation to change (mean score 4.2). Key communication skills were observed most of the time with evocation scoring the highest (mean score 3.5), then respecting choice and control (mean score 3.4), exchanging information (mean score 3.2), empathy (mean score 3.1), collaboration (mean score 3.1), a balance of open as well as closed question (mean score 3.0) and use of reflective listening (mean score 2.9).

It was easy for facilitators to present the materials in the local language. Facilitators had to use simple language and avoid jargon, but overall they reported that people could easily relate to and understand the content. This was enabled by the commitment to a more collaborative style of communication. Sessions were more fun, relaxed and conversational than the usual talks. Facilitators also mentioned that participants learnt from the interaction in the group, which was often very appropriate to the culture and context:

The content it’s very appropriate because we are able to reach out to different age groups, we are able to reach out to different ethnic groups as well that attend to the facility and in a language they can understand. (Facilitator GP)

No major changes were made to the content. Some respondents reported that certain elements were emphasised more than others in the local context. Some facilitators also incorporated their own materials. The elicit-provide-elicit format was seen as a cornerstone of the approach, even if facilitators became more flexible in navigating the content. Using visual materials, demonstrations and group activities all helped to engage learners.

Using the resource materials

The flipchart pictures and card games worked well in explaining diabetes and healthy eating. Visual materials were particularly useful and some suggested creating video materials:

If we’re showing them the medications we have examples of all of them. I don’t think they really see the relevance of that. But I think with the flipcharts and stuff, that is quite helpful coz its more visual and then they can understand. (Facilitator WC)

In each district, there were complete sets of materials for 10 primary care facilities. However, some respondents reported that the materials were not available. Sometimes managers attended the training and took the materials away. In one district, it appeared that another diabetes project retained the materials. In KwaZulu-Natal they wanted to implement immediately in more facilities, but this resulted in materials being shared, lost or not returned:

So the intention was to only have one District. Then they wanted to have two Districts. So we had more participants [at the training] than we actually bargained for which led to that uncomfortable conversation of who’s getting the resources because we never agreed that two Districts must come. So people were unhappy. (GREAT coordinator)

The materials given to the facilitators to ensure they had sufficient knowledge and expertise were seen as very useful. The optional patient handouts from ichange4health had a mixed reception, with some issues related to literacy and contextualisation.

Space, time and staff limitations

Observers and interviewees all agreed that space was a problem. Many facilities used unsuitable spaces, for example, outside areas that were weather dependent. In some cases, people organised venues in the community:

Because we don’t have a space where we can you know ensure that they are not exposed to the harshness of the weather. So that’s the threat. Maybe they may end up not coming because they know that they will be having a meeting outside the facility where it’s not warm. (Facilitator GP)

Some smaller, often rural facilities, had very few staff and it was difficult for them to implement GREAT alongside their other duties. Some of these small rural clinics also had very small numbers of people with diabetes:

Some of the facilities they are like one-man professional nurse then he has to do all the other issues and then you find out that they actually see it as they don’t have time actually. (District manager NC)

It was also important to have at least two or more facilitators trained per facility to share the load, cover for leave and staff turnover. Higher level professionals, such as doctors or dieticians, struggled to allocate time in their schedules.

One manager felt that staff might have negative attitudes towards patient education and counselling while another respondent felt that health promotion was not monitored, which reduced its perceived importance.

Effective teamwork

It was important that information about the initiative was shared with all staff involved as implementation required a team approach. For example, the clinician needed to identify patients, the clerk to predraw their folders, the pharmacist to fast-track their prescriptions, the facilitators to run the sessions and the facility manager to coordinate and plan. Facilitators could involve other members of the team in specific sessions. For example, the pharmacist to help with medication or the dietician to help with healthy eating. Facilities also needed a clear sense of priority and accountability to the district and subdistrict management:

The team had to be multidisciplinary. There had to be a MO, I think, a clinical nurse practitioner, all those clinicians who actually work in the club setup with the diabetics, and there had to be a health promotor, and there also had to be in the team, somebody, like the pharmacist, the physio, and all of those people who are the specialist in the modules, they had to be aware of it. (NCD coordinator WC)

Patient-related factors

Patients were very positive about the group sessions and appreciated the guiding style. How to eat healthily on a limited budget and with limited access to a variety of foods was a key issue:

Everything we did was better, there is no other thing I can say maybe we want it to happen on top. Everything was better, we were treated very well. We were helped, we arrived here my child saying we were dying, but after we entered this session we heard that there is no death if you take your treatment accordingly, you eat your food accordingly, and you won’t die. (Patient WC)

Patients were reluctant to attend if this increased the length of their visit or perceived that it might do so. One or two facilities reported that patients were reluctant to engage in group activities or did not seem motivated to understand their diabetes.

Patients did not always keep appointment dates and attendance at subsequent sessions might fall away. Staff thought the sessions might have less impact if the sessions were too far apart. There was a sense that attendance might be better in those who committed through a shared-decision making process vs being instructed to attend:

I think it’s more based on their own motivation. I think a lot of patients are forced to come to these kinds of things when it’s not actually what they want to do. And I think those are the ones who don’t come back. (Facilitator WC)

Facility organisation and management

The GREAT coordinator commented that implementation needed to be driven and supported by policy-makers and decision-makers from all levels. A lack of support could reduce motivation and ability to implement. Some districts had a subdistrict layer of management, which helped with implementation. At the facility level, there needed to be a champion to drive implementation. It was difficult for less influential staff members, such as health promotion officers, to do this. Most success was found when implementation was driven by a health professional, such as a dietician, who had more agency and was also motivated to improve care for diabetes:

Our biggest benefit was having the buy in from our manager so we could start immediately. We had obviously all our resources from the training. We had an existing [register] full of patients so we could easily recruit patients. And then like I said, myself and the occupational therapist, we kind of have been doing groups together. (Facilitator WC)

Developing an implementation plan was part of the training and a template was provided to help people think through the organisational issues. Facility managers were invited to this session but did not always attend. The capability and commitment of the facility and subdistrict management to implement GREAT was a key factor.

Results on the organisation of GREAT from 16 facilities that did implement in early 2022 are shown in table 2. Most facilities (10/16) agreed that their facility manager was supportive of implementation and just over half of the facilities had a suitable space (9/16). The facilities were mostly small with no permanent doctor. People with uncontrolled diabetes were mostly targeted with a median of 10 people per group. All the sessions were offered, but often only one session type had been delivered in the previous 3 months. Facilities mostly sent nurse practitioners for training as facilitators, followed by health promotion officers and allied health professionals. No doctors or community health workers facilitated sessions.

Table 2

Feedback on implementation and organisation of GREAT (N=16)

Facilities mostly organised sessions to correspond with the next appointment, which could be one to 6 months later. Facilities that required additional visits thought that this gave them extra time to run the groups and increased their value to patients, however, patients might be disadvantaged by having to make additional visits:

I realized that in our clinic, it won’t be possible for us to facilitate the group session on the day of their appointments. So what we did, we took their information normally using the referral form and then we opted to calling the patients and give them a date in which they are going to come for the session. (Facility manager GP)

Most facilities ensured patient flow by having their files available and fast-tracking them for their medication afterwards. In some facilities, patients were disadvantaged by queuing for longer at the pharmacy after their sessions. Sessions could go up to 2 hours depending on the size of the group and level of engagement. Some facilities combined sessions 1–2 and 3–4 so that they only needed two longer sessions with patients. Most facilities targeted patients with poorly controlled diabetes or newly diagnosed. One or two appeared to just select people conveniently, others focused on those that lived close by or had been living with diabetes the longest.

Costs

Table 3 presents the setup and incremental operational costs. Setup costs are for one workshop with 20 facilitators and 2 trainers. It is assumed that this would be for 10 PHC facilities in a district or subdistrict. Costs include the possibility that a venue might need to be hired, although most venues were free. Costs also include the possibility that some participants in rural areas might need accommodation if the distances are too far to return home. The setup costs also include the costs of bringing two trainers from another province. It is assumed, however, that in the future provinces will have their own in-house trainers. Nevertheless, if these costs are included the total setup costs were ZAR9315 per facility (US$494; at ZAR18.8–1US$).

Table 3

Set up and incremental operational costs

There were no incremental operational costs for facilities. There might be some costs incurred if venues needed to be hired or there was a graduation event. The opportunity costs were very difficult to determine but could be important in terms of the time taken by facilitators and other staff. The incremental costs do not include the cost of printing and providing patient handouts on diabetes as these were not essential.

Reach

Table 4 shows the number of facilities that had been implemented at the time of evaluation and the early reach of the initiative. Overall, 54 facilities were trained, 16 had implemented, 34 were planning to implement, 2 decided not to implement and 588 people with diabetes were reached.

Table 4

Reach of GREAT

Discussion

The key findings are summarised as a programme theory in figure 1 using a health system lens derived from the WHO’s framework for primary healthcare.32 The key factors to be considered in implementation are summarised under health system structures and inputs as well as service delivery activities. Outputs, outcomes and impact were not measured, but the theory summarises what is expected from the findings within such a logic model.

Figure 1

Summary of key findings and programme theory. GREAT, group empowerment and training; NCD, non-communicable disease; NDOH, National Department of Health; No, number; PDOH, Provincial Department of Health.

The latest Medical Research Council (MRC) framework for developing and evaluating complex interventions emphasises the importance of developing such a programme theory.33 This evaluation has enabled the construction of a programme theory based on the key findings that can now guide further implementation while simultaneously being further tested and refined. The MRC framework lists several core elements to consider with regard to implementation: context, stakeholders, key uncertainties, refined intervention and economic considerations.33 Each of these will be discussed in relation to the findings.

In terms of the context, implementation was clearly disrupted by the COVID-19 pandemic. Not only did the pandemic make group interventions for people with diabetes dangerous, but it distracted healthcare workers from quality improvement in NCDs.34 In addition, healthcare workers were traumatised and exhausted by responding to the pandemic and were not able to implement new initiatives.35 Another key issue in the context was the inequity in resources and capacity for NCDs relative to communicable diseases such as HIV and TB.7 For example, at national and provincial levels, there were many people dedicated to services for HIV and TB while those responsible for NCDs, often had other duties and lacked capacity to respond.

In terms of stakeholders, national policy-makers were supportive and easy to engage with. Provincial policy-makers were difficult to engage with in four provinces and this appeared related to issues with accountability, communication and decision-making, rather than a rejection of the intervention itself.

The implementation strategy needed to allocate more time to engaging district and facility-level stakeholders so that it was clearly on the agenda of local managers; there was a clear plan around the required inputs and model of care; and open communication to create buy-in from the whole clinical team. Gaps in the capability of middle-level managers have been identified as a key issue in the successful reform of the South African health system.36 Implementation also needs to be monitored through the health information system and supported by individual performance management.

Healthcare workers recognised the need for more effective patient empowerment and appreciated the guiding style. However, there is also a degree of clinical inertia37 and reluctance to innovate, especially if this goes against the prevailing culture or threatens to interrupt patient flow.38 39 Greater engagement with more senior clinicians may be needed to ensure the support of the whole team.

Key uncertainties that remain include the ability of primary care facilities to continue implementing after this initial evaluation. As the health system ‘bounces-back’ from the COVID-19 pandemic, we need to investigate whether the facilities that promised to implement, actually did. Further uncertainties include the ongoing provision of the resource materials, and our current strategy is to allow provinces to print these under contract. Maintaining momentum for scale-up is also uncertain as the initial grants come to an end and provincial budgets are under increasing economic constraints. Although the South African sugar-tax was meant to support health promotion and patient education,40 the NDOH lack capacity to manage the funds. A further uncertainty is the development of more trainers of facilitators in each province. Ideally, these should be people who have experience of facilitating GREAT, but on the other hand, need to be embedded in official provincial training centres.

The intervention itself was well received and did not appear to need major revision. As facilities are refurbished or newly built, attention should be given to space for working with groups. The training workshops may benefit from a longer duration to enable acquisition of the guiding style and further simulated practice.

During the COVID-19 pandemic, the content of GREAT was also embedded in a WhatsApp Chatbot and if sustained this could be a useful adjunct to the group sessions.41 There was also experimentation with virtual groups, but in the South African context, this may be very challenging in poorer communities.42 There may be a need to develop further content, for example, to assist people with initiating insulin.43

Although costs were not prohibitive, provinces would need to budget for provision of the resources, training and in some cases space. Provincial budgets are under constraint and NCDs are not prioritised. The rising burden of disease from NCDs and the new National Strategic Plan may start to change this perspective. Opportunity costs were important, and managers must be willing to prioritise time from existing staff for the delivery of the intervention. This raises a particular conundrum in that this may require spending more time now on patient empowerment to improve control and reduce the workload in the longer term, versus managing the high workload in the short term by neglecting patient empowerment as it takes extra time.

The evaluation of implementation was disrupted by the COVID-19 pandemic and conducted only 12 weeks after retraining the facilitators when many facilities had not yet implemented. This was necessary to meet the deadlines set by funders. It was difficult, therefore, to observe as many sessions as originally intended and quantitative data were limited in scale. On the other hand, it was possible to interview stakeholders from the national to local level and to reach saturation of data as well as obtain feedback from stakeholders in provinces that did not implement. We did not interview patients in every province and might have elicited different experiences elsewhere. It was not possible to evaluate sustainability, and this will need a further study.

Future research should test and refine the programme theory (figure 1) but may also want to explore more aspects of individual change listed in the theoretical domains framework (eg, professional role and identity, beliefs about capabilities, motivation and goals).44 These more intrinsic psychological issues in the healthcare workers’ and managers’ motivation and belief systems were underexplored and yet could be key to understanding why certain facilities implemented and others did not.

Conclusions

GREAT for diabetes was implemented in five provinces of South Africa and the evaluation of implementation informed the design of a programme theory. The programme theory highlights the key factors needed for implementation in terms of health system structures and inputs, and service delivery activities as well as the expected outputs, outcomes and impact. A number of specific recommendations will be incorporated into the further scale-up of implementation in all provinces of South Africa. Further evaluation of scale-up will be needed to continue to refine the programme theory.

Data availability statement

Data are available on reasonable request. Data can be made available on reasonable request to the authors and subject to the conditions of informed consent.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and ethics approval was obtained from the Health Research Ethics Committee of Stellenbosch University (N19/09/127), which is an accredited national ethics committee in South Africa. Ethical approval requires a commitment to the Declaration of Helsinki. All ethical guidelines and regulations were followed, and all human participants provided written informed consent. Participants gave informed consent to participate in the study before taking part.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors RJM conceptualised and led the implementation and research teams and is the guarantor for the study. The research was coordinated by DS who led the research assistants and collected the data. Both RJM and DS analysed and interpreted the data. RJM drafted the final manuscript which was revised by DS. All authors approved the final version.

  • Funding Research reported in this publication was supported by the South African Medical Research Council via its Biomedical Research and Innovation Platform with funds received from the South African National Department of Health and the Department of Science and Innovation. Funding was provided through the Global Alliance on Chronic Diseases.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.