Exploring ‘generative mechanisms’ of the antiretroviral adherence club intervention using the realist approach: a scoping review of research-based antiretroviral treatment adherence theories

Realist evaluation is a member of the family of theory-based evaluation approaches [36]. It operationalises the causal mechanisms that are likely to operate for a programme to work and the contexts in which these mechanisms are triggered to lead to the outcomes. Realist evaluation begins by eliciting an initial programme theory, which is subsequently tested, often by employing qualitative and quantitative research methods. The initial programme theory is revised based on the empirical findings and can lead to a middle range theory that explains in which contexts the intervention works by triggering specific mechanisms. Realist studies do not only answer the question ‘what works?’ but ‘how or why does this work, for whom, in which circumstances?’ [36].

While conducting a realist evaluation, identifying the ‘generative mechanism’ – the underlying social drivers of behaviour – is key, as it is one of the central elements of its explanation strategy. Pawson and Tilley explain that when programmes are implemented, they provide a resource, an opportunity or a constraint of some kind that influences the target person’s decision-making [36]. Therefore, a generative mechanism is “the process of how subjects interpret and act upon the intervention (or components of the intervention)” [37]. Lacouture and colleagues offered another description of ‘mechanism’ from a public health perspective: “a mechanism is hidden but real. It is an element of reasoning and reaction of agents in regard to the resources available in a given context to bring about changes [outcome] through the implementation of an intervention and evolves in an open space, time and open system or relationships” [38]. Based on these definitions, three main characteristics of mechanisms can be identified: (1) they are usually invisible; (2) sensitive to variations in context, and (3) generate outcomes [39].

The other two components required to complete theory building in realist evaluation are context (of action) and outcome [40]. According to generative causality in realist evaluation, a generative mechanism can only be triggered in specific context conditions to cause the expected outcome. This causal relationship is represented thus: an outcome (O) is generated by a mechanism (M) being triggered in a specific context (C). In realist evaluation, this relationship is explored using the context-mechanism-outcome (CMO) configuration heuristic [36, 41]. To facilitate understanding, the intervention (I) and the actors involved in the intervention (A) are also being represented, leading to the Intervention-Context-Actor-Mechanism-Outcome (ICAMO) configuration [42].

From our previous work on framing the concepts of ART adherence and retention in care using realist logic, we developed the framework presented in Fig. 1. This framework illustrates that the resources (information, skills, material resources, support) provided by the adherence club intervention, works through mechanism(s) acting through the actors (patients, lay counsellors and clinicians), which [the mechanisms] are sensitive to various contexts (distal and proximal), to cause the intended outcomes (adherence to medication and retention in care). Our conceptualisation considered both psychological and relational (social) mechanisms. Examples of psychological mechanisms include motivation, self-efficacy and empowerment as these emerge from an individual’s cognition. The relational mechanisms were considered with regard to the relationship that the patient shares with the care providers (e.g. trusting relationship and social support) and fellow group members (group dynamics and culture that exist within each group). These are in essence relational mechanisms that causally affect people by virtue of the individual’s position relative to his/her social relationships. Therefore, with regard to group-based adherence models, mechanisms can play out at the level of individuals, dyadic relations (between provider and patient, for instance) and/or the group.

Various reviews have been conducted to identify and investigate the roles of health behaviour change theories and health psychology theories in developing strategies to improve HIV/AIDS medication adherence [46]. With this review, we aimed at searching for research-based (peer-reviewed) application of behaviour change theories to provide explanations of how adherence-enhancing interventions work, with the overall goal of developing an initial programme theory of a group-based adherence intervention – the adherence club intervention.

The first author and a research assistant searched for articles for possible inclusion in the review. We applied two methods to search for papers: electronic database searching (PubMed, EBSCOhost, CINAHL, PsycARTICLES and Google Scholar) and manual reference list search. We searched each database using the following Medical Subject Headings (MeSH) search phrases “Adherence [OR] compliance to antiretroviral therapy [AND] theories [OR] models.” We also searched the reference list of the papers identified through the database search. We defined the inclusion criteria for the review using the PICOT mnemonics for reviews.

We excluded non-English papers. We assessed the search hits and selected only the articles with relevant titles. We stopped searching for articles when we reached saturation, i.e. when no new papers were identified. Table 3 below illustrates the data search process and the final number of studies that were retained for review.

The screening of the 84 articles was conducted by the first author and the research assistant and proceeded in three stages: (1) screening by title, (2) screening by abstract, and (3) reading the full article. Twenty-six (26) articles were obtained for the inclusion in the review process. Figure 2 outlines the data screening and selection process following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram protocol.

Extraction of data from the identified papers was done under the following topics: (1) Study citation and setting, (2) Study purpose and theory of interest, (3) Study design and methods, (4) Study limitations, (5) Conclusions/Recommendations. Refer to Additional file 1 for the data extraction process.

We examined each theory to identify context, mechanism and outcome components as defined in realist terms (See Additional file 2 - ‘Data code manual). We then assessed to what extent the theories could inform a generative causality using the intervention (I) leading to outcome (O) by triggering a mechanism (M) in context (C) for specific actors (A) conceptualisation. Following this assessment, we identified the possible mechanisms – the underlying social drivers of behaviour – from each of the theories that showed potential to explain ART adherence in the realist sense.

We summarised the characteristics of the studies under the following topics: Evidence type, research approach, study design, objective of theory inquiry, and study setting (continent). Table 4 below shows the general characteristics of the 26 studies included in the review. It should be noted that 17 out of the 22 primary research papers used a quantitative approach and only five research papers used other approaches.

The graph below (Fig. 3) represents the various theories that are discussed, tested or developed in the identified studies.

According to our judgement, three of the 11 theories applied to explain adherence to ART identified in the review featured conceptualisations that are consistent with the realist logic of causality: Information-Motivation-Behaviour [47], Ewart’s Social Action Theory [48] and the Health Behavioural Model [49]. In the section that follows, we look into the framing of these theories.

The Information-Motivation-Behaviour (IMB) model was discussed in 11 of the 26 articles that were included in the review. Some of the articles applied the IMB model to explain adherence to ART, or to conceptualise the factors affecting patients’ adherence to ART while others used the IMB model to develop an explanation on how ART adherence occurs [50‐60]. Based on the IMB model, “well-informed, well-motivated patients who possess adequate skills for enacting complex patterns of adherence-related behaviour will adhere to their ART regimen optimally over time” [57]. However, what is missing is how Intervention, Context and Mechanism interact to cause that Outcome. Figure 4 indicates how the IMB model is used to explain adherence to ART.

In Fig. 4, adherence information is (part of) the intervention that the patient receives. This information relates to the regimen, correct use of ART and the importance of adequate adherence. Additional information offered to the patients relates the side effects of the medication and possible drug interactions. According to the IMB model, this information can be a source of motivation for the patient (motivation in the psychological sense). Social support (also labelled social motivation by IMB authors) is based on how the patient perceives the support of his or her significant others. The patient’s motivation (psychological or social) and self-efficacy are moderated by actor-related elements such as the psychological health of the patient (e.g. depression), unstable living situation, poor access to medical care services and substance use. The patient’s motivation can be influenced by these context conditions.

According to the IMB model, adherence behavioural skills (including objective ability and perceived efficacy for performing critical skills such as acquiring self-administering ART) play a vital role in determining whether a patient adheres to their medication or not. This notion was applicable when ART medication regimens were very complex to self-administer and required some skills. Today, antiretroviral medication is less complicated to self-administer, as it is in most regimes, a single tablet taken once a day. The IMB model also specifies that adherence information and adherence motivation may be directly related to ART adherence in cases where medication-taking behaviours are not complex or demanding [52]. However, the IMB model is not clear about what connects each of the boxes, for instance why is it that a person with adherence behavioural skills changes his/her behaviour and becomes adherent? Neither does it present a configurational approach that builds an explanation of the outcome by linking all of the precedents in a coherent causal configuration. Therefore, the IMB model somehow remains a descriptive model of precursors and not a causal model.

Ewart’s Social Action Theory (SAT) was explored in two studies [61, 62]. The theory focuses on behaviour change and factors such as social context and support that can assist to foster and maintain that change. According to SAT, health behaviours result from an interplay of three domains: (1) context, in the sense of the social-environmental and the specific personal attributes of the individual, (2) processes of self-change that create new or modified action scripts and (3) self-regulation as an action state (in this case adherence to medication) [62].

SAT holds that health behaviours are a result of self-change or self-regulatory processes by which an individual makes the transition from old actions to adopting new behaviours [62]. Specific mechanisms of behaviour change that are identified in SAT include (1) problem-solving, (2) motivation, (3) generative capabilities (empowerment) and (4) social interaction processes. Remien et al., [62] suggested that the ability to problem-solve is related to social and emotional adjustments. According to SAT, patients can become motivated to adhere to their treatment based on their outcome expectancy – an evaluation of their personal capacities to carry out the intended behaviour – as well as when they generate goals that could stimulate the expected behaviours. Empowerment as an essential generative capability refers to a sense of personal control, mastery, and power to effect change. Social interaction processes between the patient and health care providers and with other significant others are important elements in the self-change processes.

According to SAT, the intended health behaviours are shaped by broader social-environmental systems and intrapersonal factors of the individual, representing contextual influences that can either facilitate or hinder behaviour change. Physical settings and social systems both affect and interact with biological structures and processes within the individual to create intrapersonal contexts that influence goals and generative capabilities [48]. Ewart suggests that these larger social-environmental systems “contextually determine how personal change mechanisms operate” [48]. This is in accordance with the realist logic that suggests that the mechanisms can only be fired in the right contextual environment, which leads to the intended outcome. These contextual conditions according to SAT include patient’s demographics, living circumstances, social network, organisational systems and physical environmental factors [62].

Ewart [48] suggests that SAT could potentially be used to specify mediating mechanisms linking organisational structures to personal health. According to Ewart, behavioural interventions (such as the adherence club intervention) strengthen self-regulatory systems in individuals that foster capacity for self-protective action. Figure 5 represents the conceptualisation of SAT to explain ART adherence.

Notably, while SAT presents a precedent of behaviour in the form of processes of self-change and stipulates that context matters, it conflates environment with personal (psychological) factors. Secondly, it does not explain which of the sub-elements of environment and self-change are linked to which other. Therefore, although SAT presents three categories (context, mechanisms and outcomes) that can be useful to describe an adherence intervention, it falls short of explaining exactly why and how these factors combine to explain positive or negative outcomes.

The Health Belief Model (HBM) also offers a potential explanation for adherence to ART and was explored in one study [60]. The HBM aims at assessing health behaviour of individuals through the perceptions and attitudes a person may have towards disease and negative outcomes of certain actions. The HBM was conceptualised around the individual’s beliefs and attitudes captured in four constructs representing the perceived threat and net benefits [63]. These constructs are perceived susceptibility and perceived severity that make up perceived threat and perceived benefits and perceived barriers representing the net benefit [49]. In relation to ART adherence, perceived severity refers to an individual’s subjective assessment of the severity of the consequences of non-adherence to ART. Perceived susceptibility refers to the individual’s assessment of the personal risk of developing problems with regard to ART medication non-adherence. Perceived benefits, on the other hand, relate to an individual’s assessment of the value of adhering to ART. Finally, perceived barriers relate to an individual’s assessment of the obstacles to taking the medication. These constructs represent possible mechanisms that could be triggered to enforce adherence behaviour.

The 1950 model of HBM was consolidated when Becker et al. [64] published a paper that considered a range of alternative approaches to understanding the social and psychological determinants of health and illness behaviour [63]. The authors added the influence of the patient’s demographic variables and psychological characteristics on the identified mechanisms. These demographic variables and psychological characteristics of the individuals are important conditions that could trigger the mechanisms identified. Finally, to complement the HBM, Becker et al. [64] included the concept of ‘readiness to be concerned about health matters’ which represents an individual’s general health motivation. Figure 6 indicates the operationalisation of the identified concepts relevant to HBM.

When conceptualised in the context to chronic medication adherence such as ART, the HBM translates to the desire to achieve a better quality of life and the belief that adherence to ART will ameliorate the health of the patient and this would influence whether a patient adheres to their medication or not. According to Janz and Becker [49], if an individual has high perceived threat towards a disease or a health issue; low barriers to adopting healthy behaviours; and high perceived benefits to action that would help avoid the health issue; then there is an increased likelihood of the individual engaging in the recommended behaviour.

We would say that the HBM usefully describes demographic and psychological factors that influence health motivation, and more interestingly identifies the factors (candidate mechanisms) that perpetuate actual adherence behaviour. However, it also fails to operationalise a realist causal explanation. That is, it fails to identify what mechanisms or sets of mechanisms are triggered under what context of action to cause the expected outcome of interest.

The review was limited to an extent by the fact that most of the studies did not fully describe the theories that they applied, but mostly conceptualised these theories into variables to test for associations. In addition, most of the studies included in this review applied the cross-sectional study design that could not fully accommodate dynamic theoretical propositions that capture the notion of adherence to ART or make inferences on the causality of effect [46]. These limitations were minimised by returning to the original theories to understand their explanatory power in relation to adherence to ART.