Implementing Evidence-Based Design to Improve Adherence in Self-Administered Treatment Technology

Introduction

The healthcare system in the UK is facing increasing challenges, especially for chronic diseases (Wilson et al., 2005). These challenges include but are not limited to insufficient funding, lack of resources (Appleby et al., 2014), lack of staff (Buchan et al., 2017) and increasing demand for hospital admissions (Smith et al., 2014). These challenges became apparent during the COVID-19 outbreak that hit the world, including the UK, at the beginning of 2020 (Willan et al., 2020). The outbreak has shown the significant pressure on the healthcare systems around the world, which may suggest that a new look of the system and providing a new system paradigm is needed. This paradigm may involve a wider utilisation of medical technology and the move to self-administrated and patient-empowered interventions where both patients and clinicians move from one-way mentorship relationship to a concordance two-way communication relationship.

Several studies have shown the benefits of adopting medical technology systems, especially for treating chronic diseases (Cutler, 2007; García-Lizana & Sarría-Santamera, 2007; Mirza et al., 2008; Ganasegeran et al., 2017; Wickramasinghe et al., 2011). The self-administered treatment technology can be as effective as clinically administered treatments (Scogin et al., 1990). Additionally, medical technology can reduce medication costs (Hoppe et al., 2000; Scogin et al., 2003), hospital and General practitioners (GPs) admissions, the psychological impact of the disease, instances of surgical intervention and the side effects of using pharmaceutical treatment.

Poor adherence is a challenging problem for healthcare systems (van Dulmen et al., 2007). An estimated 20% to 30% of patients do not take their medications (Viswanathan et al., 2012). The level of non-adherence increases in chronic diseases to between 30% and 50% (Barnett, 2014; Bourbeau & Bartlett, 2008: Nunes et al., 2009; Sabaté, 2003). van Dulmen et al. (2007) reviewed 38 systematic reviews of the effectiveness of adherence interventions and defined three practices:

This paper introduces the development of a design-driven Adherence Framework and Adherence Canvas for self-administered treatment technology. The literature studies were overviewed the understand the factors that affect patient adherence. Practical adherence factors were identified through interviews with five SMEs designing and developing self-administered treatment technology. The data concluded from the literature and case studies were triangulated into the eDelphi study to build and assess an Adherence Framework that can be implemented as an evidence-based design tool for companies to guide, measure and test the consideration of adherence in self-administered treatment technology.

Impact of poor adherence

The study explored the related literature from two main aspects: 1) the challenging problem of adherence and adherence frameworks, and 2) the role of design in behaviour changes and the evidence-based design models and their implementation in healthcare.

The failure to achieve a significant level of adherence can have severe consequences for patient health, including relapsing, morbidity, health status, hospitalisation and mortality (Aldeer et al., 2018; Baum et al., 2012; Viswanathan et al., 2012). Mennini et al. (2015) investigated the cost of poor adherence to antihypertensive therapy in five European countries (Italy, Germany, Spain, France and England) over ten years. The study explored cardiovascular complications associated with hypertension (i.e. stroke, heart attack and heart failure). Improving patient adherence to antihypertensive therapy to 70% could reduce the cases of cardiovascular complications by 82,235 in the five European countries combined, with 6,553 fewer cases in England alone.

Poor adherence also results in financial and workforce burdens, such as increasing healthcare costs and negatively impacting healthcare workforce productivity (Bosworth et al., 2011; Conn et al., 2016). For chronic diseases, the financial cost to the EU healthcare systems was estimated to be £77 billion in 2006, which represents around 10% of their total healthcare expenditure. In England, the drug cost is around £140–210 million. This cost includes the burden of non-adherence to the medical intervention. Improving adherence can reduce costs by an estimated £25.3 million in England (Mennini et al., 2015).

Theoretical frameworks of adherence

Several theories were presented to understand patient adherence. These theories could be categorised into factor-based, psychology-based, and stage-based theories. As the aim of this literature is to identify the factors that influence adherence, only the first two categories were discussed in this study:

Factor-Based Theories.

Factor-based theories explain adherence, focusing directly on the factors that drive the change regardless of the underlying psychological frameworks. The first factor-based model is the WHO adherence factors, defining five dimensions affecting patient adherence. (Aldeer et al., 2018, Sabaté, 2003):

Theoretical Adherence Frameworks.

Different theories have been proposed to understand patient adherence to treatment or behaviour change (Rosenstock et al., 1988). Examples of the adherence factors frameworks are briefly highlighted to build an understanding of the influence of the literature informing the adherence factors in the study:

Information, Motivation, Strategy

In this model presented by Martin and DeMattio (2013), the adherence drivers are classified into three main factors: information, motivation and strategy (I-M-S). The model addresses three main aims:

Ensure the patient has the right information related to how to adhere to the treatment.

Help the patient to believe in the treatment and commit to it.

Help the patient to overcome the barriers to treatment adherence and develop strategies for chronic disease management (DiMatteo et al. 2012).

The I-M-S model didn't focus on medical technology, especially for new companies working on delivering medical technologies. This model becomes challenging to adopt when manufacturing treatment devices and measuring adherence factors' efficiency. The second observation is that the model did not consider the other factors affecting self-administered treatments adherence.

Figure 1 presents a visualisation of examples of 1) the theoretical frameworks of adherence, 2) the factors in each theory, 3) communication-related factors, and 4) design-related factors. Furthermore, the figure links and re-orders the factors into categories as described below:

The adherence theoretical frameworks (blue) are presented on the right side. Some theories are linked to each other (grey lines). In some instances, the Social Learning theory overlaps with other theories, such as the Health Belief Model.

In linking different factors and their relationship to each other, three main categories can be identified: Motivation, Knowledge and Communications. The communication between patient and clinician can lead to a higher level of adherence as highlighted in the I-M-S model and the case studies interviews highlighted later in this paper. Factors such as the environmental cue (Social Learning) and subjective norms (Reasoned Action/Planned Behaviour) have communication aspects as they involve social interaction.

In medical technologies, the patient experience extends beyond behaviour change factors to involve design aspects (Fogg, 2009). The lack of linkage (grey lines) between the Experience category and the adherence theories confirms the design aspects' lack of consideration in physical and digital treatments.

Investigating adherence-related theories and mapping the adherence factors to the four categories highlighted in Figure 1 contributes to building the Adherence Framework that is assessed in round one of the eDelphi study highlighted in 6.0 Building the Adherence framework: Delphi Study. The framework is used in the Adherence Canvas used to evaluate companies' adherence consideration as will be discussed in 7.0 Design-Focused Adherence Canvas.

Evidence-based design

The evidence-based design practice has been recently established in the last 25 years, driven by the need for an evidence-based approach to drive the decision-making process (Ulrich et al., 2011). Sackett et al. (1996) defined evidence-based design as ‘the conscientious, explicit, and judicious use of current best evidence in making decisions. Evidence-based design is a growing trend in the healthcare system, especially in space design and hospitals design (McCullough, 2010). As this trend is relatively new, it has recently evolved to consider design elements such as environment, visual design, safety and sustainability.

Several models have been presented to categorise the different types of evidence and measure their rigorousness. Examples of these models are the Oxford Evidence-Based Medicine Levels method and the American Institute of Architects Guidelines for Healthcare Design method. While the former method is concerned with medical interventions, the latter focuses on architecture and space design. Figure 2 shows the Health Environments Research & Design Journal (HERD) model, which consists of six levels that classify the research's rigorousness to dictate the design process (Stichler, 2010). These models guided the development of an evidence-based design tool to help companies measure adherence consideration highlighted in 7.0 Design-Focused Adherence Canvas which align with.

Case studies interviews

The second part of the research, interviews with case studies SMEs, contributed to understanding how companies consider adherence while developing self-administered treatment technology regimens. It also contributed to understanding how adherence is considered during the design process. To this goal, purposive and snowball non-probability sampling was used to select and recruit the companies. A total of 353 companies were overviewed and filtered based on the scope of the research to five SMEs. The data collected from the case studies were anonymised and replaced with randomly generated names (YMX, ESA, LW7, 3AB, Z5W and DE7).

Companies have a wide range of experience and the consideration of adherences. Companies with more extended experience in medical technology development (YMX) showed a better consideration for adherence than the other companies. Table 1 below lists the adherence factors and the companies that considered them during the treatment device design. While all the interviewees acknowledged the important role of adherence, the table below shows that most companies considered only a few factors.

Analysing the data from the five companies showed that while self-administered treatment technology can provide solutions to many of the healthcare system's challenges, there is a gap between the theoretical knowledge, industry application, and evaluation (clinical trials). The analysed data from the case studies is triangulated with the literature adherence theories on the first round of the eDelphi study where the panellists' consensus on the Adherence Framework factors.

Building the Adherence framework: Delphi study

The literature review and case studies interviews aimed to build a clear understanding of the problem of adherence and the factors that affect patient adherence to self-administered treatment technology. Data observed from different perspectives was triangulated and fed into a Delphi method to identify and assess the adherence factors.

Delphi Round Two

The questionnaire (four questions) was based on the Likert Scale question type to afford the chance to measure the importance of every subcategory collected from round one. Each category identified in round one required one matching Likert scale question. Each question presented a category from round one's content analysis for adherence factors. The rating scale included five levels, in ordinal scale, for the importance of the factors. The levels were: Very High (5), High (4), Moderate (3), Low (2), and Very Low (1).

The N/A option was available for panellists who thought that the factor was not applicable.

A comments section let panellists provide further thoughts about the rating system. While these comments may have contributed to guiding the third round, their main purpose was to get the opportunity to collect further opinions from the panellists which could be used to build the final Adherence Canvas.

Results overview and analysis

The data analysed has showed a rough agreement on the level of importance between the panellists. The figures (3, 4, 5 and 6) provide a visual overview of the answers and how the majority of the answers fell on the right side of the scale (High and Very High).

In Table 4, the SD values reveals there is high agreement on some of the factors such as: understanding of patient behaviour, benefits motivators, patient-centred design and design research. For these values, the minimum level of importance is 2 and the maximum level is 5.

TABLE 4. The SD of the 20 factors in round two and round three.

	Round 2 SD	Round 3 SD
Patient-Clinician Relationship	0.74322	0.63994
Communication Effectiveness	0.72375	0.63994
Clinician-Side Implementation	0.86189	0.70373
Intervention Usage	0.78446	0.79881
Confidentiality and Privacy	1.01419	1.03280
Adequate Shared Information about Intervention	1.09978	0.74322
Shared Adherence Consequences	1.12546	0.73679
Understand Patient Behaviour	0.63994	0.63246
Patient-Centred Shared Knowledge	1.16292	0.81650
Patient Information	1.09945	0.96115
Goals Motivators	1.29099	0.74322
Risks Motivators	1.11270	1.12122
Benefits Motivators	0.65465	0.65465
Individual Motivators	1.08233	0.74322
Social Motivators	0.91548	0.79881
Product Design Characteristics	0.67612	0.70373
Patient-Centred Design	0.61721	0.63994
Design for Adherence	0.91026	0.91026
Design Research	0.61721	0.70373
Device Usage	0.82808	0.51640
Valid N (listwise)	15	15

TABLE 5. Wilcoxon signed-rank test on SD.

		N	Mean Ranking	Sum of Rankings
Round_3_SD - Round_2_SD	Negative Rankings	11a	8.73	96.00
	Positive Rankings	3b	3.00	9.00
	Ties	1c
	Total	15

• a Round_3_SD < Round_2_SD.
• b Round_3_SD > Round_2_SD.
• c Round_3_SD = Round_2_SD.

TABLE 6. The test statistics for the Wilcoxon ranked-signed test.

	Round_2_SD - Round_1_SD
Za	−2.731b
Asymp. Sig. (2-tailed)	0.006

• a Wilcoxon Signed Ranks Test sing the Z statistic.
• b Based on positive ranks.

TABLE 7. General adherence factors and their importance (median) based on the Delphi results.

Category	Sub-category	Level of importance
Collaboration and Partnership	Patient–clinician relationship (e.g. monitoring tech, consistency, follow-ups, feedback, confidentiality, adaptability)	5
	Effective communication (e.g. suitable language, style of communication)	5
	Clinician side implementation (e.g. clinical implementation, clinical trials feedback)	4
	Usage intervention (e.g. support, reduced communication)	4
	Confidentiality and privacy (e.g. reliability, trust)	5
Knowledge and Information	Adequate information shared about intervention (e.g. clear treatment information, simplicity of information)	5
	Shared adherence consequences (e.g. connecting personalised choices and consequences, measuring targets fpr the health outcome)	5
	Understand patient behaviour (e.g. identifying adherence instances, mismatch in perception of urgency, actual patient behaviour)	4
	Patient-centred shared knowledge (e.g. patient knowledge, tailored information, usage)	5
	Patient information (e.g. demographic information, technology literacy, treatment literacy)	3
Motivation	Goals (e.g. setting achievable goals)	4
	Risks (e.g. perceived severity (suffering–drawbacks)	4
	Benefits (e.g. perceived benefit (reinforcement–reward)	4
	Individual (e.g. self-efficacy, health condition)	4
	Social (e.g. social expectancy, environmental cues, how others perceive the patient)	3
Patient Experience	Product design characteristics (e.g. aesthetic, packaging and usage process)	4
	Patient-centred design (tailoring) (e.g. usability, patient characteristics, psychological factors)	5
	Design for adherence (e.g. compliance measurement, concordance, patient participation in treatment process, guidance on factors influencing adherence)	5
	Design research (e.g. field research and testing, prototyping iteration)	4
	Device usage (e.g. practicality, affordability, comfortability, self-administration of treatment, social support, facilitation)	5

The analysis overview has shown evidence of a level of agreement between the panellists; this observation is supported by the value of the standard deviation around the mean of each factor. Therefore, the third round of the Delphi process aimed to identify the consensus between the panellists on the results from round two.

Design-Focused Adherence Canvas

The Design-Focused Adherence Canvas is a one-page Likert Scale Radial Graph.

that allows companies to evaluate how adherence is considered in the treatment device during and after the design process. It evaluates the consideration of the general factors which affect adherence. There are two versions of the Adherence Canvas:

Design-Focused Adherence Canvas

The generic model is based on the consensus of Delphi process (Figure 7) on an Adherence Framework for treatment technology including the factors affecting patient adherence and the importance of each factor. It can be used in to address major diseases, evaluate new products and by companies who do not know the exact adherence factors affecting their treatment intervention are.

Advanced Design-Focused Adherence Canvas:

This model does not include any factors nor any levels of importance. It can be used by experienced companies who have a clear idea of the factors which affect their patients' adherence (Figure 8).

How the Adherence Can it be Used?

The below example provides a visual guide to the usage of the Design-Focused Adherence Canvas. This example will use placebo (dummy data) answers in the Adherence Canvas to visualise how it works in a real organisational setup.

The voting process

The company needs to determine the voting method to use. Usually, there are two methods: 1) Voting using dots and counting the dots, or 2) If anonymity is required, using a simplified version of the eDelphi method consisting of two rounds.

Once the above practice is defined, the team may move forward as below (the steps below use a two-round Delphi method):

1. The team facilitator codes the factor by colouring it based on its consideration. Irrelevant factors are coded with the colour black (Figure 9).
2. The team votes on how adherence factors are considered in the design idea. In Figure 7, the votes are marked with an X.
3. After finishing the first round, the team votes again in a second round, this time with the ability to see each-others' votes. If more than one consideration level shares the same importance, the voters should repeat the voting by considering that each level should have a unique level of importance (Figure 10).

Post the voting process

The results of the first round of voting are used to iterate the product to improve adherence consideration. The process is repeated in different stages to improve the product, then the team can visualise the different votes to overview how the adherence factors were considered in each prototype (Figure 11 and Figure 12).

The Adherence Canvas help and documentation are available for companies through a dedicated website: https://www.adherencecanvas.com. The Adherence Canvas can be downloaded and used under the creative common: Attribution-ShareAlike 4.0 International.

Conclusion

Poor adherence presents one of the challenges facing the healthcare system in the UK and worldwide. This poor adherence level increases in chronic diseases and self-administered treatments. While there is a gap in considering factors that affect self-administered treatment technology, this paper introduces an Adherence Framework and Adherence Canvas to evaluate, measure and improve the consideration of adherence in self-administered treatment technology. The study presents an Adherence Framework and the Adherence Canvas to 1) guide companies to consider adherence factors, especially new companies or companies developing new technology and 2) help companies evaluate the consideration of adherence factors during the design and development of the treatment technology intervention. The improvement of patient adherence can have significant benefits for the healthcare system, including improving patient health, increasing the intervention's effectiveness, and reducing the cost and the stress on the healthcare system.

Theoretical frameworks of adherence and interviews with five SMEs case studies were investigated to explore the factor-based and psychological-based theories, and design factors that affect behaviour change were investigated. Five SMEs were interviewed to understand the consideration of adherence during the treatment technology design and development. The factors were mapped into motivation, knowledge, communication, and experience categories.

The data was triangulated and assessed using the eDelphi method by panellists from the healthcare system and health technology designers. The eDelphi method involved three rounds. The process aimed to explore the panel's consensus on the Adherence Framework. The first round was analysed using content analysis and validated using inter-rater reliability. The agreement between raters was moderate based on Cohen's Kappa's Landis and Kock Agreement measure showing the possibility of 50% agreement based on chance. The final questionnaire was evaluated using the Wilcoxon test resulting in the assumption of a significance of 0.006, which means that the ratio of chance agreement is lower than the significance level standard (ϴ = 0.05).

Further research will test the Adherence Framework and Adherence Canvas in self-administered treatment technology for chronic diseases. This research's contribution to knowledge includes identifying the gap between knowledge and practice related to adherence to self-administered treatment technology, which opens the door for further research in this domain.