INTRODUCTION

Long-term medication adherence is a well-recognized requirement for positive health outcomes for many chronic diseases.1 While medication side effects, toxicities, and burdens of treatment for chronic diseases have improved dramatically in recent decades, patients often struggle to achieve adequate medication adherence rates. In the US alone, costs related to poor medication adherence are in the billions of dollars each year.2 , 3 Not surprisingly, efforts to better understand the determinants and facilitators of, and challenges to, adherence have proliferated in medicine and social science, contributing to a robust body of literature that can seem overwhelming.4

In response to the ongoing need to optimize work with patient medication adherence in a manner that leverages social-behavioral science, the University of Alabama at Birmingham (UAB) hosted a conference, “Understanding and Improving Treatment Adherence: An Interdisciplinary Approach,” in February 2016. Our review and discussion of social-behavioral models of medication adherence provides a rationale for their wide-scale use, a brief sampling of HIV medication adherence models that can be generalized to other conditions and regimens, a synthesis of salient factors identified across multiple models, and practical recommendations for using evidence-based and evidence-informed adherence models to guide practice and research. Here we summarize this portion of the working group meeting to bridge the gap between social-behavioral science and the diverse disciplines engaged in providing high-quality patient care. The objective of this review is to build confidence in the utility of fully leveraging social-behavioral theories, models, and frameworks in both clinical work and research.

Adherence to Medication Regimens Is Often Far from “Simple”

To date, a number of “reasons” for non-adherence have been well-documented, ranging from “just forgetting” to complex social negotiations of avoiding dosing to avoid disease-related stigma.5 , 6 From a social-science perspective, any reason has underlying causal mechanisms (e.g., forgetting resulting from lacking a skill set to dose in a timely manner) that relate to factors important in most decision-making and behavior change efforts (e.g., attitudes and beliefs—the lack of skills may be driven by beliefs that daily dosing is not really needed). Rather than assuming that adherence results from being advised to adhere by a medical provider, or knowing that non-adherence will be unhealthy in the short or long term, a social-behavioral perceptive identifies multiple underlying causes of medication adherence (and suboptimal adherence). Theoretical principles that come from diverse areas of study, including the fields of cognition, behavior change, motivation, coping, affect regulation, and decision making, provide a strong basis for the assertion that self-administration of medication regimens is a complex phenomenon. Like most human behavior, dosing behavior happens in a larger context of individual beliefs, social experiences, anticipated outcomes, and myriad “ways of being” that include culture perspectives and worldviews. While the act of self-administration of a medication may appear discrete and “simple,” the models of adherence we consider below suggest that this observable behavior has dynamic, but identifiable, sets of precursors.

Adherence from a Social-Behavioral Perspective?

One benefit of applying social-behavioral models is that such models offer a comprehensive mosaic of factors related to optimal and suboptimal adherence based on evidence accumulated through research on “real-world” experiences with medication adherence, each factor offering a potential path for intervention. Many providers have experienced the ineffectiveness of focusing on a single factor influencing adherence (e.g., information only—simply telling patients to take their medications as prescribed, and that not doing so will lead to complications). Appreciating other factors found to influence adherence, directly or indirectly, provides a greater opportunity to help patients optimize their adherence. The space between having medications in hand and benefiting from them is intersected by multiple social-behavioral facilitating and challenging pathways. Given the growing call for patient-centered and differentiated care, adopting and sharing complex perspectives of patient adherence that help to gain an optimal understanding of patient challenges and experiences, and that avoid centering on patient performance (e.g., good patients adhere and bad patients do not) can be facilitated through adoption of comprehensive conceptualizations of adherence outcomes.

HIV provides an example of the benefit of considering a comprehensive view of factors related to optimal and suboptimal adherence. In the case of HIV treatment, achieving and sustaining suppression of the virus (i.e., a desired clinical outcome) requires adequate adherence to antiretroviral therapy (ART). Shortly after the wide-scale introduction of ART in the US in 1996, structural and medication-related factors were identified as important barriers; however, many other important factors emerged over time, including acceptance of the disease, negotiation of stigma, perception of the regimen as needed/helpful, and positive attitudes towards the regimen and regimen effects. Social-behavioral models detailing underlying drivers of HIV medication adherence provided a framework from which practitioners and researchers could identify and work with the complexities of ART adherence, and support patients in achieving positive clinical outcomes.

A second important contribution of social-behavioral models of medication adherence centers on providing a clear architecture for intervention implementation and effect monitoring. In an environment of limited resources for practice and research efforts, having a clear, evidence-informed model for why an intervention program or strategy would be expected to have a positive effect is essential. Social-behavioral models identify the causal factors that must change for an intervention to be effective. Change in theory-identified factors (e.g., self-efficacy, skills in managing side effects) as a result of a new service or intervention provides explanations for why an effect is achieved. Alternatively, failure to see changes in factors underlying medication adherence may explain where an intervention or service failed. Simply stated, focusing on just the primary endpoint (i.e., adherence) without unpacking the drivers of intervention effects fails to fully explain supportive or null findings.

Social-Behavioral Models of Medication Adherence (Examples from HIV Medication Adherence)

Specific to HIV medication adherence, there are a number of social-behavioral models that have been evaluated, used for intervention design, and adapted over time. From these models, core “drivers” of medication adherence can be identified, and likely generalized to medication adherence for other conditions.

Munro and colleagues provided a comprehensive overview of models in the evidence base through 2006.7 These include established health behavior models, such as the Health Belief Model8 , 9 and Information, Motivation, Behavioral Skills (IMB) models,10,11, – 12 as well as Social Cognitive Theory (SCT),13 , 14 the Theory of Reasoned Action15 , 16 and Planned Behavior,17 , 18 Protective Motivation Theory,19 and the Transtheoretical Model (TTM).20 , 21 While all approaches offer a benefit to understanding ART adherence, no single model has emerged as superior.

Across these models, there are several common components. Core constructs shared across explanatory adherence models include a level of accurate information about or knowledge of the disease; the role of treatment and the specifics about following the treatment recommended; awareness or some sense of need for the treatment; motivation to adhere (such as attitudes and beliefs about adherence and non-adherence, with different models detailing various core drivers of motivation); and a skill set needed to execute adherence behavior across diverse situations and settings. Models differ in how community- and structural-level factors are incorporated. Many models include such factors as relevant to the extent that they are incorporated into individual beliefs and experiences, while others characterize systematic levels of influence as contributing independently to adherence, as the environment in which care is negotiated sets the context for any individual-level experiences with care.

Table 1 presents factors that are considered critical across several models, as well as the constellation of factors needed to establish and sustain optimal levels of medication adherence according to each theory. Readers are encouraged to use the references and resources provided to further explore the details and literature pertaining to specific models of interest. Most of the models and frameworks in Table 1 are explanatory or predictive in nature, focusing on causes of medication adherence, but typically do not specify how to change adherence behavior (except the IMB model,23 which does specify a process for intervention development). Thus these models articulate the factors that should be optimized or reduced, while the specifics on how to do that requires additional steps.

Table 1 Extrapolated Core Constructs and Predictions of Leading Models of Adherence
Full size table

Practical Guidance for Working with a Social-Behavioral Model of Adherence

Optimizing research and practice through the use of a social-behavioral model to ground interventions involves (1) selecting a model, (2) tailoring the model, (3) operationalizing factors in the model, and (4) actually putting the model and model-based interventions or programs into action by using it. Each step is summarized below in terms of considerations and action items. We also provide an example of each step in Table 2, for the development, implementation, and evaluation of the iENGAGE intervention targeting viral suppression among people living with HIV newly entering clinical care, based on a situated IMB model. Note that there are a number of strategies to move from model to intervention programs, and the steps provided here are meant to highlight the general process of using social-behavioral models.

Table 2 Application of Recommended Practical Guidance for Working with a Social-Behavioral Model of Adherence: the Situated IMB Theory-Based Integrating ENGagement and Adherence Goals Upon Entry (iENGAGE) Behavioral Intervention Consisting of Four In-Person Counseling Sessions Over 12 Months Among Persons Newly Initiating Outpatient HIV Medical Care at Four HIV Clinics in the United States (5R01AI103661–05)
Full size table

SELECTING A MODEL

From the models of HIV medication adherence, one or a combination of models often resonates with a service provider’s or researcher’s experiences in working with ART or other medication adherence. Reviews of models have been conducted in an attempt to establish empirical superiority of one model over others. A recent meta-analysis34 reviewing 124 theory-based adherence interventions across diseases found that motivational interviewing35 (more a communication approach that also identifies sources for behavior change36 than a theory or model per se) was the most commonly used, followed by SCT, and then the TTM, self-regulation, cognitive theory, IMB, and self-management. Common use, however, does not speak to efficacy of a given model in relation to others or in relation to no model at all.

There is support for adopting a model to guide intervention development versus not using a model. For example, a review of 85 internet-based health behavior interventions engaging over 43,000 participants suggested that better intervention effects were achieved through reliance on a health behavior model to guide intervention components than with no model.37 Meta-analyses of interventions are somewhat limited in their ability to sort out effects of model-based versus atheoretical adherence intervention approaches because of heterogeneity. This heterogeneity may underlie failures to identify a significant advantage of interventions with an articulated theoretical basis as reported in some reviews.38 , 39 Empirical support for each model included in Table 1 is readily available in the literature. Once a model is selected, additional considerations are needed before the model can offer full utility in a given application; these consideration include tailoring the model to a specific context, operationalizing core factors, and using the model.

TAILORING THE MODEL

Models identify the core factors and pathways that facilitate or deter the adoption of the behavior needed to achieve the health outcome, but they often lack specific details about what the factor really looks like for a specific group or population. For example, the IMB model specifies that medication adherence-related information is a critical component in medication adherence. However, additional elicitation work within a given cohort or community is needed to identify the specific kinds of information that promote adherence and the kinds of misinformation that impede medication adherence. Similarly, in several models, negative adherence-related attitudes are identified as problematic, as they create motivation not to adhere. To intervene, one must first identify the specific attitudes or beliefs that are posing the greatest problems in a given sample, cohort, or patient. Models identify highly generalizable constructs, such as awareness, motivation, or self-regard, and propose the structural relationships between these factors (what leads to what, what mediates which relationships); intervention strategies try to optimize adherence by impacting these broad factors in positive ways. Once you have your model of which factors matter the most at a general level, you next need to get very specific. We recommend using models to focus efforts to get specific (see below) and evaluate, propose, or offer tailored and targeted, theory-informed intervention strategies.

OPERATIONALIZING FACTORS

Whereas a model or theory identifies areas that are important in medication adherence, how each factor operates, exists, or is operationalized in a given cohort can be articulated a number of ways, including basic elicitation work and the use of process planning models. Elicitation work can be thought of as a qualitative and quantitative evaluation of the theory-related needs of a given patient, patient group, or larger group. The IMB model of adherence,23 for example, suggests that effective interventions require focus groups or interviews to identify which kinds of information, motivation, and skills are most pertinent to the target population or person. An application of the IMB model to initiation and retention in care for chronic diseases further emphasizes the critical role of this process, noting that all core constructs in the model must be situated to a specific group of patients or community for the model and model-based interventions to have utility (a situated IMB model [sIMB]).24 While constructs in social-behavioral models help to narrow the scope of all possible facilitators and barriers to those that are identified by the model as most critical, it is possible to use current literature, patient and stakeholder groups, surveys, and interviews with current patients to articulate the specific needs in a given community in each area.

USING IT

Once a model is selected and the model-based factors operationalized in reference to a specific disease, regimen, or population, and the core general constructs have been operationalized, the model should be evaluated for utility by using it in practice and research. Several process approaches can inform the use of a model in quality improvement efforts. The plan, do, check, act (PDCA),40 the PRECEDE-PROCEDE model,41 the six Steps for Quality Intervention Development (6SQuID),42 and Intervention Mapping43 all use models in their approaches to intervention development, implementation, and evaluation. Readers are encouraged to consult these strategic approaches in the development of intervention strategies. However, even in the absence of these evaluation methods, any clinic or program seeking to enhance client-centered adherence care can use an adherence model to identify potential model-based gaps, challenges, and resources of relevance in patient populations.

A model-based, well-articulated understanding of relevant and modifiable constructs in the pathway of factors influencing medication adherence can guide evaluation of programs and the construction of new ones. At the most basic level, supports for adherence and related tools or strategies should address one or more of the model-based factors. The need to address model-based factors is of particular importance in the case of interventional research, where model-based interventions can be evaluated for effects on adherence and the model-based factors that lead to adherence (e.g., does participation in an intervention increase motivation, shift attitudes, or enhance skills?). The use of models in clinical care is also useful.

As an example of the utility of using models in clinical care, clinics may provide patients with pillboxes to facilitate adherence. From an IMB model perspective, providing pillboxes to patients would be expected to assist patients who have deficits in organizational and memory skills for dosing. The model suggests that by reducing the difficulty of dosing, one would need fewer behavioral skills to be able to dose successfully. The easier it is to dose, the lower the demands on motivation will be. The more complex a regimen, the more well-informed and highly motivated one needs to be to bring multiple skills to bear for successful adherence. Whether a pillbox would be expected to help depends on where the deficits causing non-adherence are most acute. If pillboxes are provided without attention to other critical determinants of adherence, their use may have a minimal or even negative impact on adherence.

The IMB model identifies motivation as a critical factor in adherence. In the case of patients struggling to develop positive beliefs about the effects of a regimen, reducing the difficulty of dosing through organizational tools is not the issue. In the case of patients who have a system of dosing in place that works well for them (e.g., has a bottle that is kept in view as a cue) but have deficits in other areas, introducing a pillbox misses the mark, and may actually further erode adherence by perturbing a dosing strategy that would work fine if other drivers of adherence were in place (e.g., high levels of knowledge and positive motivation).

The point of the prior discussion is that client-centered care relies on identification of specific deficits in factors that drive adherence. Identifying which relevant gaps are causing problems is facilitated by exploration of model-based factors (in this case, the kinds of information, attitudes, and beliefs about the personal and social consequences of adherence and non-adherence, and pertinent skills needed to execute dosing in commonly occurring situations). Rather than eliciting every possible reason for a missed dose, model-based discourse focuses on relative strengths and challenges in the social, contextual, and personal factors identified as requisite for dosing. Asking about model-based factors can make discourse more efficient and targeted.

Using a model to guide research and practice offers opportunities to better appreciate the complexities of adherence, engage in exploration of gaps and resources in a more efficient manner, and identify adherence promotion strategies that bridge specific gaps in model-based factors, and ultimately can be evaluated formally and informally to determine whether the model-based factor and adherence outcomes change as expected. Moreover, patient-centered care may be positively impacted by adopting an approach to patient discourse that highlights the importance of the myriad factors that contribute to optimal, long-term patient adherence.

CONCLUSIONS

Social-behavioral models can assist research and practice by consolidating the multitude of “reasons” for non-adherence into superordinate factors that drive adherence. These factors can then be operationalized and serve as the focus of intervention efforts and patient–provider discourse. We have provided a sampling of adherence models for readers to consider, as well as a series of steps that can be used in “actioning” a model, from a description of adherence in general to a powerful tool in understanding, intervening in, and monitoring adherence for specific conditions and cohorts. There is no consensus on a single “most effective” model, although many models share common core factors (e.g., motivation and skills).

Although models have high generalizability in their constructs, the steps we provide here for using a model highlight the need to bring broad constructs (e.g., motivation) down to the level of intervention target (e.g., the kinds of beliefs most impactful to a given cohort). Tailoring models may result in limiting their generalizability across cohorts. However, with client-centered interventions, one would expect to have highly generalizable constructs (e.g., regimen knowledge), with highly specific tailoring used to identify deficits and promote optimization (e.g., knowledge of dosing requirements for insulin-dependent diabetes). Further, if research were to develop higher standards for reporting on the factors that an intervention targeted to improve adherence and whether improvements were achieved on those factors, some strategies developed for one condition, regimen, or population could be considered and evaluated in other programs, using similar models but with different conditions, regimens, or populations.

There is growing consensus on the value of using well-articulated models to conduct and evaluate research—as the measurement of changes in factors targeted by an intervention can helped to unpack intervention effects. The use of an adherence model to provide clear, evidence-informed formulation of how the core factors of the model operate within specific communities can facilitate a systematic evolution of intervention programs that effectively support medication adherence and other health behaviors.