By 2017, an estimated 37 million people were living with HIV and AIDS (PLWHA) globally and as of June 2017, 20.9 million PLWHA were accessing antiretroviral therapy (ART) [1]. Nevertheless, more than one third of patients are not adhering to treatment - an estimated 62% of PLWHA are adhering to their ART (taking ≥90% of ART) worldwide [2]. It has been reported that adherence to ART is better in sub-Saharan Africa compared to North America (77% versus 55%) [3].
In 2016, it was estimated that one in every seven (14%) people in South Africa were living with HIV/AIDS [4]. With an estimated 7 million PLWHA, South Africa has the highest number of PLWHA in the world [5]. South Africa, in 2011, had a 75% increase in access to ART, becoming the largest ART programme in the world with an estimated 3.3 million PLWHA currently initiated on ART [6]. Managing a large number of patients within a large ART programme poses various challenges. Prominent among these challenges are the problems of sub-optimal retention in ART care (high levels of lost-to-follow-up), poor adherence to medication and overcrowded health care facilities [7].
The impact of the growing numbers of patients in care was demonstrated by Médecins Sans Frontières (MSF) at the Ubuntu Clinic, a public health clinic in a densely populated, low-income residential area in Cape Town. This site was the largest ART clinic in the Cape Town Metro District, situated in an area with an extremely high HIV prevalence [8]. From 2006, the capacity of the facility to enrol new patients on ART showed a decline. As the clinic became saturated, the loss-to-follow-up rates of patients 12 months after enrolment increased with each successive annual cohort initiated on ART from 2005 to 2008. The decline was largely attributed to the facility-based and staff intensive model of care that was used for the management of PLWHA on ART [9]. The context of scarcity of nurses and doctors, accentuates the need for a more efficient model for managing large cohorts of patients on ART and specifically with the effective use of community-based strategies.
In the search for effective long-term retention models, the adherence club intervention, a differentiated care model [10] – consisting of streamlined HIV treatment and care adapted to the needs of a targeted group – was developed and piloted at Ubuntu Clinic. The adherence club is comprised of a group of patients whose appointments have been harmonised. Patients attend sessions that are modular and that can theoretically be placed outside of the clinic to reduce further congestion.
While the original conceptualisation of the adherence club intervention was facility-based (conducted within the premises of the facility) [11], community-based [12] and home-based [13] (out-of-clinic adherence club models) have also been piloted and implemented to further decentralise ART services. The goal of the out-of-clinic versions of the adherence clubs is to remove important health system barriers to retention in care and medication adherence such as long distances to clinic. Our study focused on the facility-based adherence club intervention.
The adherence club model demonstrated promising outcomes in terms of improved patient flow, an increase in the monthly enrolment of patients on ART and decreased loss-to-follow-up while increasing the overall number of patients in care [8]. Two years after the first enrolment of patients in the adherence clubs, only 2.4% of club patients had a negative outcome – 0.7% had died and 1.7% were lost to follow-up [11]. Altogether, 97.6% of patients were still in care: 89.5% remained in the club system, 4.8% had returned to mainstream care at Ubuntu clinic, and 3.3% had been transferred to other clinics [11].
Based on these results, the adherence club model was selected as a potential intervention to address the challenges of poor patient retention in care, suboptimal adherence to ART and health care facility congestion [14]. In 2011, the model was rolled out as a system improvement intervention aiming at streamlining the treatment and care of ‘stable’ patients in the Western Cape Province of South Africa. The rollout and implementation of the adherence club model was conceived and executed through the collaboration between the Western Cape Provincial Department of Health (WC DoH), the non-governmental organisation Treatment Action Campaign (TAC), the Cape Town Municipality City Health department (CoCT DoH), MSF, and the Institute for Healthcare Improvement (IHI).
During the first phase of the rollout, from January 2011 to March 2015, 77% of ART sites in the Cape Metro area of the Western Cape Province implemented the adherence club intervention [14]. The graph (Fig. 1) below shows the progressive coverage of patients on ART in the Western Cape Province by the adherence club care model from December 2010 to June 2016.
This article reports on the results of the first phase of a study called “A realist evaluation of the antiretroviral treatment adherence club programme in selected primary health care facilities in the metropolitan area of Western Cape Province, South Africa” [15]. While other papers describe how the adherence club is organised and executed [12, 13, 16,17,18], we provide, in this paper, a comprehensive description of the programme based on a document review and participant observations of adherence club sessions in six primary health care facilities. Furthermore, we present the initial programme theory of the adherence club intervention based on the realist logic – i.e. exploring how, why, for whom and in what circumstances the adherence club intervention is expected to work.
Methodological approach
Realist evaluation
Realist evaluation seeks to understand how and why, for whom, and under what circumstances a programme works (or not) [19, 20]. The philosophical basis of realist evaluation is scientific realism [19]. The realist evaluator aims to identify the context-mechanism-outcome (CMO) causal relationship in order to explain “how, why, for whom and under what conditions a programme works” [21]. Following the understanding that people are not passive recipients of innovations [22] and that programmes can only work when the relevant actors adopt either all or parts of the intervention modalities, Van Belle [23] and Mukumbang et al. [24] elaborated on the CMO configurational logic proposed by Pawson and Tilley [19] to include components of the ‘Intervention’ and the ‘Actors’. These authors suggested that an intervention-context-actor-mechanism-outcome (ICAMO) configuration would provide a better analytical tool because aspects of the intervention, which provide the mechanisms and the actors through whom the intervention works are accounted for. Thus, in this article, we adopted the ICAMO heuristic tool.
Because realist evaluation is focused on providing explanatory models, it has the potential to open the “black box” of programmes by making explicit the generative mechanisms to explain how the programme modalities lead to the intended outcome(s) [25]. For this reason, realist evaluation is recommended as an appropriate approach for capturing the complexities of health care interventions during their evaluations [26, 27].
The realist methodology is a ‘theory driven, interpretative approach to uncovering underlying middle-range theories (or logics) driving interventions and their multiple components, as well as illuminating the contextual factors that influence mechanisms of change to produce outcomes’ [28]. The interpretive approach is driven by various forms of reasoning; deductive methods (based on testing specific hypothesis) and inductive reasoning (formulating general inferences), but central to the realist method of inquiry is abductive or retroductive reasoning [29]. Retroduction is a mode of inference in which events are explained by postulating (and identifying) mechanisms which are capable of producing outcomes [30]. According to Wynn and Williams [31], retroduction is characterised by the use of causal mechanisms as the basis for explanation, the possibility for multiple potential explanations, and the understanding that these causal mechanisms may or may not be observable empirically.
Typically, realist evaluations start with an initial programme theory (hypothesis) and end with a more refined theory. Therefore, the evaluator hypothesises in advance the intervention (I) (or its components), the relevant actors (A), mechanisms that are likely to operate (M), the contexts in which they might operate (C) and the outcomes that will be observed if they operate as expected (O). This hypothesis is formulated by conceptualising the components of a programme implementation process (programme modalities, context, actors, mechanisms and outcomes) to form theories about the underlying causes to arrive at explanations of what we observe. Based on this, realist evaluators seek to understand how and why programmes work by formulating programme theories.
According to Westhorp [32], “Realist evaluation is most appropriate for evaluating new initiatives or programmes that seem to work but where ‘how and for whom’ is not yet understood; programmes that have previously demonstrated mixed patterns of outcomes; and those that will be scaled up, to understand how to adapt the intervention to new contexts.” Since the adherence club intervention fulfils all the above conditions, we adopted the realist evaluation approach for the evaluation of the adherence club intervention.
The Programme theory
One of the central elements of realist evaluation is the programme theory. Realist evaluation starts and ends with a theory or with theories. Thus, eliciting an initial programme theory is a “pre-requisite of realist evaluation” methodology [32]. Developing a programme theory follows from the notion that programmes are theory-incarnate [19]. A programme theory is described as “a set of explicit or implicit assumptions of how the programme should be organised and why the programme is expected to work” [33].
Programme theories link activities and outcomes to explain how and why the desired change is expected to take place and represent how “the mechanisms introduced by the programme into pre-existing contexts can generate outcomes” [34]. This process is guided by a ‘generative’ model of causality, in which causal links are demonstrated through a fine-grained explanation of what happens between cause and effect.
Developing programme theories serves two main purposes: as a planning tool and/or as an evaluation tool. If used for planning, Sharpe [35] suggests that it is beneficial to develop a programme theory prior to the start of the programme. In this instance, the programme theory indicates how different elements of the programme are intended to work together and to identify the intermediate outcomes of a programme or an intervention [36]. This gives a clear indication of the goals and objectives of the programme and of the pathways through which they could be attained.
Programme theories are also used to guide monitoring and evaluation [36]. They are especially important for the evaluation of complicated and complex aspects of programmes [37]. To this end, the goal of the evaluator(s) is to understand not only patterns related to the outcome of the intervention but also to reveal how and why the intervention attains the outcome of interest [38]. It is noteworthy that the evaluation of the programme’s theory is an evaluation of the programme rather than the evaluation of the programme theory [35]. Whatever their use, programme theories should be concrete enough to be tested and refined through empirical research, and abstract enough to generalise from the case-specific theories [39].
What differentiates the programme theory in realist evaluation from programme theories as conceived in other theory-driven approaches, such as Theory of Change [40] or Theory-driven evaluation [41, 42], is that realist evaluation specifies what mechanisms will generate the outcomes in what context. Thus, a realist programme theory provides a conceptual framework for putting the underlying CMO components centre stage [43].