Developing a Suite of Resources to Improve Patient Adherence to Compression Stockings: Application of Behavior Change Theory

Plain Language Summary

The best way to treat and prevent venous leg ulcers is by wearing compression stockings every day. However, some people do not wear their stockings as much as they should, demonstrating “low adherence” to compression stockings. Using behavior change theory, three new resources have been developed to attempt to improve patient adherence. These include a new questionnaire to determine individual patient barriers, a decision tree to guide clinicians in addressing the barriers, and answer sheets listing recommendations to address each barrier.

Background

Clinically prescribed compression stockings are considered best practice, for chronic venous insufficiency (CVI), to heal and prevent venous leg ulcers (VLU).^1–3 VLUs are wounds in the lower legs, occurring due to higher pressure within veins, often related to higher body mass index and/or impaired mobility.³ They are a significant problem for those over the age of 65, with international prevalence studies reporting rates between 1.5 and 3.0 per 1000 people.⁴ With demographics trending towards sedentary lifestyles and an ageing population, VLUs will likely remain a significant burden on society.⁵ VLUs have negative impacts at the individual level, contributing to pain, reduced quality of life and work productivity,⁶ while also burdening health-care systems due to wound management costs and hospitalization.⁷

Compression stockings must be worn consistently,³ since low adherence contributes to delayed healing⁸ and recurrence of VLU.^9,10 Unfortunately, worldwide adherence rates are low, reported to be between 12% and 52%.¹¹ Adherence to wearing compression stockings is a complex multi-faceted phenomenon, with over 200 barriers identified in the literature.^10,12 Many of these barriers relate to patients’ lack of understanding of why stockings are required, difficulty with getting them on and off, and discomfort whilst wearing them.^13,14

To determine each patient’s specific barriers to adherence, a suitable questionnaire is required. This is supported by Michie,¹⁵ a behavior theorist, who recommends that the reasons contributing to a “problem behavior”, such as poor adherence to medical advice, should be identified, prior to implementing any intervention. Several purpose-made questionnaires^16–20 designed to determine patient-reasons for discontinuing compression therapy have been developed, however these do not adequately address the multi-dimensional aspects of adherence. Further, they lack sufficiently detailed questions for meaningful and personalized intervention planning. For example, Allaert’s 2019 study¹⁶ evaluated the psychometric properties of a self-questionnaire on adherence, Benigni¹⁷ investigated difficulties with donning stockings through a survey, and Shannon²⁰ assessed adherence to wider clinical recommendations for preventing VLUs, but none of these included prior negative experiences,²¹ low mood²² or low self-belief^23,24 as potential barriers. In relation to the comfort of a stocking, Allaert’s¹⁶ only question was “I find it uncomfortable (heat, cold, pressure)” and Clarke-Moloney¹⁸ simply asked “How do you find the feel/sensation’ of the stocking on your leg?”

Unidimensional interventions that aim to improve behaviors associated with adherence to compression stockings have often been unsuccessful and none have demonstrated consistent superiority over others.²⁵ In contrast, research into improving adherence in other chronic health conditions including diabetes mellitus²⁶ and chronic kidney disease,²⁷ demonstrates more positive impacts from personalized and multidimensional approaches. A recent scoping review²⁵ of adherence to wearing compression stockings found that, while personalized and multi-dimensional interventions show promise, further high-quality research is required. It recommended that the design of a multidimensional behavior change intervention to improve adherence with compression stockings, must accommodate the multitude of adherence barriers and be sufficiently flexible to deliver tailored personal interventions.

Many authors^15,28,29 have employed theory in the development of other clinical interventions (prototypes) for behavior change, to promote clarity around planning, design, implementation, and evaluation. Theory can provide structure to complex knowledge, allowing for generalization across populations and settings.^15,30,31 This is particularly relevant in the context of the adherence phenomenon, where patients often present with several barriers at the same time, requiring a complex intervention.^28,32

In consideration of an overarching framework to guide resource development, a design thinking approach³³ was given due consideration, since it aligns with the intent to implement a collaborative and human-centered approach to a real-world problem. However, since this framework encompasses prototype testing, it was considered beyond the scope of this paper. Behavior change theory allows for a more concentrated focus on development of a behavior change intervention.

Behavior change interventions can be defined as “coordinated sets of activities designed to change specified behavior patterns aiming to promote uptake of clinical services to support healthy lifestyles”.³⁴ While there are several broad theoretical approaches available for developing interventions generally in health-care settings,^15,35–37 more specific theories have also been used in the context of promoting adherence to medical recommendations.^30,38 This section describes justification for the selection of a specific behavior change theoretical model.

A 2011 systematic analysis³⁴ of 19 frameworks of behavior change interventions concluded that none covered the full range of intervention functions required for effective behavior change. Subsequently, Michie developed the Behavior Change Wheel (BCW).³⁴ This is an iteration of the popular Theoretical Domains Framework (TDF).³⁹ The BCW claims to be more flexible than the TDF alone, however the TDF is often used alongside it.⁴⁰

The BCW has at its core, a “behavior system” involving three essential conditions: capability (C), opportunity (O) and motivation (M) to influence behavior (B). These form the basis of Michie’s COM-B model³⁴ of behavior change. It proposes that at any given moment, a particular behavior will occur only when the person has the capability and opportunity to engage in a certain behavior, and when they are more motivated to perform that behavior than any other.^34,41

The BCW on its own has demonstrated success and broad applicability in behavior change programs. A 2015 scoping³⁰ review of adherence promotion theories in pelvic floor muscle training critiqued 12 theoretical models/theories and identified Michie’s BCW as helpful in informing future research into adherence interventions. BCW has also been used successfully to improve adherence to medical advice in diabetes management^42,43 and heart failure.⁴⁴ Furthermore, Gould et al⁴⁵ employed the BCW to systematically inform an intervention for health provider care of Australian indigenous pregnant women regarding smoking cessation. Similarly, a theory-based 2018 study⁴⁶ used the BCW to embed particular clinical behaviors into adult hearing aid fitting consultations, while Sinnott et al⁴⁷ described the use of the BCW to develop an intervention for use by general practitioners to improve medication management in multimorbidity.

Use of the COM-B model together with the 14 domains of the TDF facilitates a more detailed transition from a behavioral analysis of the problem to the selection of the most contextually and effective intervention functions, including modes of delivery.^41,45,46 The successful application of BCW and the TDF to other complex chronic conditions requiring multifactorial solutions argues well for its extension to the development of programs to improve adherence to compression stockings for the management of venous leg ulcers. Therefore, this paper describes the application of behavior change theory, specifically the COM-B model described by Michie in the BCW, supplemented by the additional details embedded in the TDF.

Methods

Michie’s BCW was used to guide selection of intervention functions to address barriers to target behaviors, identified using the COM-B model and TDF domains; a target behavior being one that has been chosen, or “targeted” for change.⁴⁸

Initially, a local metropolitan hospital expert clinical panel was established by the lead investigator. Individuals were approached for inclusion based on their clinical relatedness to the topic and availability for subsequent collaborative phases of brainstorming, reframing, and refinement of ideas to inform the selection of appropriate resources for development. The final panel, connecting through email and personal interviews, consisted of occupational therapists, vascular nurse practitioners, vascular surgeons, and university academic staff.

The steps involved in resource development reflect similar BCW methodology described by earlier authors.^46,47,49 Figure 1 illustrates the steps undertaken by the panel through the stages of defining the problem through to identification of appropriate behavior change techniques and their mode of delivery.

Figure 1 Overview of steps used in resource development using the BCW approach.

Firstly, the problem behaviors were defined, informing selection of target behaviors according to COM-B defined terms. The source (clinician or patient) of the desired behavior was then described alongside its linking TDF domain.^39,50 “Clinicians” comprises occupational therapists and occupational therapy assistants, and “patients” are adults who are referred to outpatient clinics for compression stockings, with a history of venous leg ulcers and reduced adherence to stockings. To avoid repetition, affirming language was used. For example, “the patient can’t put stockings on and off by himself” became, “the patient can put stockings on and off by himself.” The dominant barriers to the target behavior were identified and linked to potential enablers within the context of occupational therapy (OT) outpatient clinics. Potential recipients of a new intervention were considered, and included hospital administrators, vascular surgeons, vascular nurse practitioners, general practitioners, patients, and clinicians.

The second step was to select, develop, and tailor the intervention functions to achieve the target behavior. The COM-B model was applied to the target behaviors and barriers. Potential strategies were linked to the desirable patient behavior. COM-B intervention functions with potential to elicit behavior change^51,52 were considered by the panel in relation to each barrier, considering contextual relevance, and anticipated target acceptability. Kneebone’s Impact-Likelihood Matrix for behavior prioritization⁵³ (Figure 2) prompted analysis of their likely impact and ease of implementation and was adapted with permission. Interventional strategies most likely to have a greater impact on stocking adherence were identified as those requiring the least effort and having the highest likelihood of adoption by clinicians.

Figure 2 The impact-likelihood matrix for behavior prioritization.53 Notes: Reprinted from Kneebone SL, Fielding K. The impact-likelihood matrix: a policy tool for behaviour prioritisation. Environ Sci Technol. 2017;70:9–20, Copyright (2017), with permission from Elsevier.53

This process led to the third step, where the specific details about the resources to be developed were determined using a narrative and problem-solving process.

In development of the first resource (a questionnaire), barriers to stocking adherence were grouped thematically, and a list of possible questions considered capable of revealing these barriers was assembled by the lead investigator. These questions were then refined through consultation with the local expert clinical panel to provide a level of satisfactory detail. Recognizing, for example that there are multiple reasons for lack of comfort in stockings, this new questionnaire comprises 11 questions that relate to this issue alone.

In designing the second resource (a decision tree), the COM-B component and TDF domains were considered in relation to the source of the enabling behavior, linking each barrier to multiple levels of existing local resources, including videos and handouts.

Potential clinical recommendations regarding each question (barrier) on the questionnaire were explored in development of the third resource (clinical answer sheets) through a review of the literature, supplemented by several phases of feedback from the expert clinical panel. These were revised until a consensus was reached regarding the final version of each corresponding answer sheet.

Results

The following section presents the results from the application of behavior change theory as described above. Initially, the expert clinical panel agreed that only patients and clinicians were appropriate recipients of the intervention. The step of defining the problem and specifying the target behaviors found that, of 14 potential TDF domains, the target patient- and clinician-behaviors and barriers were associated with the following eight domains: Knowledge, skill, environmental context and resources, social/professional role, memory/attention/decision process, beliefs about consequences, beliefs about capabilities, and social influences. Table 1 presents this information.

Table 1 Mapping of Barriers and Enablers to the Wearing of Compression Stockings, Linked to Michie’s COM-B Functions, Grouped by TDF Domains, and Rated on Their Justification for Inclusion

The second step resulted in the selection, development, and tailoring of the intervention, while also identifying intervention functions to achieve the target behavior. Application of the COM-B model to the target behaviors and barriers revealed recommendations in six sub-divisions isolating potential facilitators across multiple domains. The intervention functions were identified and prioritized and are also included in Table 1. Nine possible COM-B intervention functions were considered by the panel for contextual relevance and acceptability. Five intervention functions were excluded (persuasion, incentivization, coercion, restriction, and modelling) considered difficult to implement due to practical limitations of prebooked outpatient appointments which are inflexible (30- or 60-minute appointments) and patients are routinely only reviewed every six months. Those considered feasible were education, training, environmental restructuring, and enablement. Of 16 potential contextually relevant patient- or clinician-actioned enablers mapped onto a prioritization tool, four were excluded. Justification for omission is described in Table 1.

This process identified the need to develop a suite of resources to comprehensively address the variability and combined complexity of the intervention strategies and enablers identified. These resources were modelled around published evidence, strengthened by the expert panel’s shared knowledge and experience.^54,55 A new patient questionnaire, sufficiently specific to determine individual patient barriers to adherence was created. Secondly, a decision tree was designed to assist clinician’s efforts to systematically address each patient barrier to the target behavior. Finally, recommendations to address each barrier were agreed upon, and assembled conveniently as “clinical answer sheets”. This suite of resources is shown in Figure 3.

Figure 3 The suite of resources.

The “Barriers to Compression Questionnaire” comprises 24 questions, thematically grouped into five categories. It prompts the treating clinician to make clinical observations and ask probing questions to identify participant-specific barriers to adherence. Each question requires a “yes” or “no” response. If a “yes” response is recorded, the clinician or participant rates the significance of this score on a Likert scale of 1 (mild) to 5 (severe). Table 2 demonstrates the link between each barrier, the question relating to the barrier, the relevant COM-B component and the TDF domain and source of enabling behavior.

Table 2 Barrier Identification Expanded for Detail, Linked with Its COM-B34 Component, TDF39 Domain and Source of the Enabling Behavior

The clinical decision tree links each barrier to multiple levels of resources. For example, if a patient’s primary barrier is related to lack of skills in applying compression stockings, the decision tree will link to the relevant clinical answer sheet, recommend targeted skills-based training and/or enablement via assistive devices, and suggest pertinent patient handouts and videos. If the barrier relates to discomfort whilst wearing stockings, then the answer sheet will suggest a modified stocking type/sizing based on the reason for the discomfort and suggest further resources if appropriate.

A separate clinical answer sheet corresponds to each of the 24 questions. Clinicians are instructed to identify their patients’ top three barriers before designing and implementing a treatment plan, as per usual clinical practice.⁵⁶

To illustrate the process from identification of a barrier through to selection of a behavior change intervention, using appropriate intervention functions, three examples are shown in Figure 4.

Figure 4 Three examples demonstrating links between barriers, intervention function and intervention prompted by the clinical decision tree.

Discussion

This paper describes the application of an evidence-based, behavior theory-driven methodology for improving patient adherence to wearing compression stockings. It presents a suite of resources developed using Michie’s BCW and the TDF. A recent scoping review²⁵ found that multidimensional approaches showed greater impact on adherence to stockings but have not yet been developed for testing. These new resources were developed to meet this gap.

These resources have potential to demonstrate positive impacts on health economics and patient outcomes. It is predicted that the development of a program to improve stocking adherence may have greater impact on this population than any improvement in medical or wound treatments.⁵⁷ In a review of the health economics regarding compression stockings, a 2018 Australian study⁵⁸ highlighted the important contribution of compression therapy in the management of VLUs. This study proposed that the provision of compression stockings to affected individuals nationally would initially cost the health system an additional AUS$270 million over 5 years but ultimately save AUS$1.4 billion over the same period of time. Further supporting the allocation of government funding, Pacella⁵⁹ reported that when patients were reimbursed for any costs related to compression stockings, their health-related quality of life improved. Importantly however, this economic modelling assumes that patients are adhering to compression stocking regimes.³ Given the well documented low adherence to best-practice treatment with subsequent negative impacts on wound healing and recurrence,^13,14 improving compression stocking adherence may contribute to substantial cost savings and favorable patient outcomes^3,18 by allowing their full benefits to be realized.

It is also anticipated that these new resources will be transferable to other settings with a role in providing compression stockings for medical conditions where adherence is similarly low, such as lymphoedema, deep vein thrombosis, scarring and burns,⁶⁰ suggesting potential impact beyond venous leg ulcers.

The clinical information and guidance provided within these resources has the potential to influence clinician skill, confidence, and job satisfaction. Many may consider the specialized area of compression stockings outside their scope of practice⁵⁷ while clinical frustration related to the care of patients who are considered non-adherent to other forms of medical advice, has been linked to clinician stress, burnout, and general dissatisfaction with the health profession.⁶¹ Competence in influencing patient adherence may therefore impact clinician satisfaction in caring for this patient cohort with potential generation of interest in gaining further expertise in this clinical area.

Strengths and Limitations

The strength of this new suite of resources lies in the breath of analysis completed prior to their development and the multifactorial aspects considered. The underpinning theory-based approach is augmented by input from highly experienced local clinical experts. The developed resources meet the requirements of an innovative personalized and multidimensional approach to intervention, as recommended by earlier authors.¹⁴

A substantial evidence base supports the selection of behavior change theory to guide the development of a new intervention. Other models were considered, for example, a design thinking approach,^33,62 however this is a relatively new framework. The BCW and TDF have been employed extensively^42–44 to explore barriers and enablers to behavior change for other medical recommendations and they provide the scope to encompass the complexity surrounding compression stocking adherence.

Several limitations are acknowledged. Complex problems lend themselves to non-empirical approaches; however, empirical methods are required to draw conclusions and test their efficacy. Repeatability of research methodology is hindered by variability of site-specific factors including local clinical context, patient demographics and panel expertise. Further challenges exist because of the novelty of the resources and absence of similar comparative models. It is also acknowledged that the developed resources are required to be used together which, in consideration of their combined complexity, would require a behavior change for the clinicians using it. A paper-based delivery format could be considered cumbersome and, therefore, implementation of developed resources require attention regarding their delivery method.

Directions for Future Research

Given the known benefits of medical software applications, including increased access to point-of-care tools, better clinician communication, portability of information resources⁶³ and improved clinical decision-making,⁶⁴ the combined resources could be delivered on an App platform. A pilot study to investigate user acceptability, feasibility, and potential early proof of concept of these resources is recommended. This could then inform a larger randomized controlled trial with health-economics analysis, to test if these resources can improve stocking adherence and determine if there is a cost-benefit related to their use.

Conclusion

Behavior change theory was used to inform the design of a behavior change intervention to improve adherence to compression stockings. An analysis using Michie’s BCW³⁴ and the TDF³⁹ considered expert opinion and available literature, to determine barriers to wearing compression stockings, and identify potential intervention methods. These were rated on their ease of implementation and target acceptability within occupational therapy clinics, culminating in the development of a suite of resources including a questionnaire, a clinical decision tree, and clinical answer sheets for novice and experienced clinicians to improve patient stocking adherence.