A Systematic Review of Interventions Addressing Adherence to Anti-Diabetic Medications in Patients with Type 2 Diabetes—Components of Interventions

Background Poor adherence to anti-diabetic medications contributes to suboptimal glycaemic control in patients with type 2 diabetes (T2D). A range of interventions have been developed to promote anti-diabetic medication adherence. However, there has been very little focus on the characteristics of these interventions and how effectively they address factors that predict non-adherence. In this systematic review we assessed the characteristics of interventions that aimed to promote adherence to anti-diabetic medications. Method Using appropriate search terms in Medline, Embase, CINAHL, International Pharmaceutical Abstracts (IPA), PUBmed, and PsychINFO (years 2000–2013), we identified 52 studies which met the inclusion criteria. Results Forty-nine studies consisted of patient-level interventions, two provider-level interventions, and one consisted of both. Interventions were classified as educational (n = 7), behavioural (n = 3), affective, economic (n = 3) or multifaceted (a combination of the above; n = 40). One study consisted of two interventions. The review found that multifaceted interventions, addressing several non-adherence factors, were comparatively more effective in improving medication adherence and glycaemic target in patients with T2D than single strategies. However, interventions with similar components and those addressing similar non-adherence factors demonstrated mixed results, making it difficult to conclude on effective intervention strategies to promote adherence. Educational strategies have remained the most popular intervention strategy, followed by behavioural, with affective components becoming more common in recent years. Most of the interventions addressed patient-related (n = 35), condition-related (n = 31), and therapy-related (n = 20) factors as defined by the World Health Organization, while fewer addressed health care system (n = 5) and socio-economic-related factors (n = 13). Conclusion There is a noticeable shift in the literature from using single to multifaceted intervention strategies addressing a range of factors impacting adherence to medications. However, research limitations, such as limited use of standardized methods and tools to measure adherence, lack of individually tailored adherence promoting strategies and variability in the interventions developed, reduce the ability to generalize the findings of the studies reviewed. Furthermore, this review highlights the need to develop multifaceted interventions which can be tailored to the individual patient’s needs over the duration of their diabetes management.


Introduction
Diabetes, characterized by hyperglycaemia, is one of the major chronic conditions that impacts a significant proportion of the population worldwide [1]. Diabetes is projected to be the 7th leading cause of death in 2030 [2]; with type 2 diabetes (T2D) accounting for most of the cases [2].
Management of T2D involves weight loss, via reductions in caloric intake and increases in physical activity, oral anti-diabetic agents and/or insulin [3,4]. Non-pharmacological measures remain one of the key management recommendations, and are usually the starting point for patients diagnosed with T2D [3,4]. Different classes of anti-diabetic agents are available for treatment, and dual, followed by combination therapy is recommended if mono-therapy is insufficient. Patients may ultimately require insulin replacement therapy to adequately manage their diabetes [4].
While adequate management of diabetes is extremely important for the prevention or delay of complications arising from poorly controlled blood sugar levels, the 'chronic' nature of the disease, lifelong requirement for medications, requirement for changes in lifestyle, the need to cope with social, cultural and psychological distress that may occur with the disease, and the clinical manifestations and associated complications, make the management of diabetes very complex. There is also a need to individually tailor the therapy for diabetic patients depending upon their age, demographic variability, co-morbidities and risk of developing complications [3].
Amidst this complexity, remaining adherent to treatment recommendations, such as home glucose monitoring, adjustment of food intake, administration of medication(s), regular physical exercise, foot care and regular medical visits [5] may be a challenge [6]. Adherence to medications has been recognized as key for optimally controlled diabetes [7,8] in T2D patients.
Many factors affect adherence to treatment in diabetes such as disease and treatment characteristics and complexity, age, gender, self-esteem, stress, depression, quality of the relationship between patients and health care providers, social support, and patients' ability to remain adherent amidst changing circumstances in their daily life [5]. The World Health Organization (WHO) has recognized five factors which influence adherence to all medications. They are health system related factors, socio-economic factors, condition related factors, therapy related factors and patient related factors [5]. Adherence is therefore a multifactorial behaviour and solutions require acknowledgment of the range of factors.
Studies have shown that medication adherence rates among patients with T2D is not optimal and is comparable to those with other chronic diseases [9]. Adherence to oral diabetes medications in T2D patients is 36 to 93% and to insulin is around 60% [10]. Treatment nonadherence in diabetes is well recognised, and interventions to promote adherence, improve glycaemic control, self-care behaviours and other key outcomes have been designed and implemented in T2D patients. Several reviews and meta-analyses published over the last decade have addressed these interventions in T2D patients [11][12][13][14][15][16][17][18][19][20][21][22][23].
Some reviews have focused on the wider aspect of treatment adherence in diabetes, incorporating not just adherence to anti-diabetic medications but treatment adherence as a whole [14,15,20]. Where adherence to medications are specifically discussed [12,17,22,23], the reviews have included studies that analysed adherence to a range of medications taken by T2D patients, rather than being specific to anti-diabetic medications [12]. Only a limited number of reviews have specifically evaluated interventions assessing adherence to anti-diabetic medications [17,22,23]. However, there has been very little focus on the characteristics of these interventions and how well they address factors that impact non-adherence.
Coincidentally, a review [24] examining interventions to improve medication adherence in patients with T2D was published at the same time as the current review was being finalized. However, the review has broad inclusion criteria and does not focus specifically on adherence to anti-diabetic medications. Furthermore, there was no in-depth categorization of the intervention components or an exploration of non-adherence factors addressed by the interventions. These are key gaps identified in the existing reviews of the literature. Therefore, the current review aims to assess the characteristics of the interventions, focusing on their components and evaluating how they address the factors that contribute to non-adherence to anti-diabetic medications. The specific objectives are to: • identify and categorise the intervention components which have led to improved medication adherence to anti-diabetic medications; • determine the WHO non-adherence factors that the interventions have addressed; and • evaluate changes in intervention designs and implementation techniques over time.

Literature search
A review of the literature was conducted to identify research articles that have evaluated the impact of interventions on adherence to anti-diabetic medications in T2D patients. Studies were searched in the following databases: Medline, Embase, CINAHL, International Pharmaceutical Abstracts (IPA), PUBmed, and PsychINFO. Each database was searched using the appropriate terms for medication adherence (concept 1), type 2 diabetes/ anti-diabetic medications (concept 2) and intervention studies (concept 3). Key words/ terms to denote these concepts were used and then combined using 'and' operator (concept 1 and concept 2 and concept 3) (S1 Fig). The search strategy was limited to articles published from January 2000 to April 2013 (inclusive) and published in English. The references of relevant publications (all studies included in this review and relevant systematic reviews [12,14,15,17,20,21,23] were hand searched to find additional studies that met the inclusion criteria (S1 Table). The search strategy as well as the study inclusion and exclusion criteria have also been illustrated previously [25].

Data extraction and analysis
For each study, the "intervention" and its characteristics were evaluated in-depth. The intervention components were summarised and the impact of each intervention was recorded. The interventionist(s), provision of training to the interventionist(s) before delivering the intervention, and the method of evaluating the delivery of the intervention were assessed. The interventions were categorised as recommended by Roter et al. [26]. Further information about the study characteristics can be found elsewhere [25].

Operational definitions
For the purposes of this review, the following operational definitions were used: • The term 'medication adherence/ adherence' has been used throughout the review to indicate the extent to which individuals take their medication, despite the alternative terms used in the studies included in the review.

• Types of Interventions
• Interventions have been classified into Educational, Behavioural and Affective, where appropriate, following the classification used by Roter et al [26]. Two additional categories were also used: • Educational: "pedagogic interventions, verbal or written, with a knowledge based emphasis designed to convey information. Specific strategies included one-to-one and group teaching, the use of written and audio visual materials, mailed materials, and telephone instructions" [26].
• Behavioural: "interventions that were designed to change compliance by targeting, shaping, or reinforcing specific behavioural patterns. This included strategies such as skill building and practice activities, behavioural modelling and contracting, packaging and dosage modifications or tailoring, rewards, and both mail and telephone reminders" [26].
• Affective: "strategies that attempted to influence adherence through appeals to feelings and emotions or social relationships and social supports. Included were family support, counselling, and supportive home visits" [26].
• Economic: interventions that dealt with economic or cost related issues pertaining to medications.
• Multifaceted: interventions that had components that could be categorized into more than one of the above categories.

Study selection
The literature search identified 6,662 citations. A total of 230 articles appeared to meet the review inclusion criteria and were retrieved in full text. However, only 49 studies actually met the study criteria and were selected for review. Three more studies were identified from hand searching. Thus, a total of 52 studies  were included [25] (S2 Fig). [ 28, 29, 31, 34, 35, 37, 39, 43, 47-50, 56, 59, 60, 62-64, 66, 67, 69, 72, 75-77]. Participant inclusion criteria varied across the studies in terms of age, HbA1c value, duration since diagnosis, sample size and duration of study. Most of the interventions were carried in community settings (67.3%), and most were conducted for a maximum of 1 year, with the impact of a majority of the interventions being assessed for a duration of 6 months. In addition to assessing anti-diabetic medication adherence as a primary (73.1%) [27-33, 35-38, 40, 42-50, 53-57, 59, 61, 64, 68-72, 74-77] or a secondary (26.9%) outcome, the studies assessed a range of other clinical or patient specific outcomes. The study characteristics have been discussed in detail previously [24].
The most widely used method for measuring anti-diabetic medication adherence was selfreport, and the commonest tool reported was the Summary of Diabetes Self-care Activities (SDSCA) questionnaire. Overall, the methods for measuring anti-diabetic medication adherence varied across the studies, and the implications have been discussed elsewhere [25].
The interventionists involved in delivering these interventions varied. Pharmacists and nurses were involved in six [51,61,63,70,75,76] and three [36,53,72] interventions, respectively. A nurse was also involved in another, along with a dietician and another professional [54]. Student level researchers were involved in three studies [31,50,69] and in two, they were referred to as 'integrated care managers' [50,69]. Two studies mentioned respectively 'health care professionals' [44] and health care providers [74] were involved in leading the intervention sessions, without specifying who they were. "Coaches" [56] and Community Health Workers (CHWs) [52] were interventionists in other studies. Three studies did not specify who the interventionist was [49,57,77], and in two an interventionist was not required as these studies either evaluated the effect of implementing the Medicare part D program on anti-diabetic medication adherence [57], or the patients were delivered their OHAs in Real Time Medication Monitoring (RTMM) medication dispensers, which helped in registering their medication real time [77].
The interventionists were trained for the delivery of the study intervention in eight studies [50,52,61,67,69,70,72,75]. Two studies [50,72] assessed the delivery of the intervention process, while two others mentioned providing supervision of [67], and consultation with [75] the interventionists during the intervention delivery.
3.7. Characteristics of interventions reporting improvement in both medication adherence and HbA1c. Out of the 34 studies that assessed the impact of interventions on both medication adherence and HbA1c, nine [36, 50-54, 56, 67, 69] reported a positive impact on both medication adherence and HbA1c. However, in one study, the improvements in HbA1c were seen in patients with HbA1c>7%, rather than all participating patients [56]. The interventions were all multifaceted. In one study, the educational component was the primary component; however, it appeared that there was also a behavioural component, though it cannot be stated with certainty without obtaining more detailed information about the program [51].
Three studies [50,53,69] addressed condition-related, patient-related, therapy-related and socio-economic factors. Two others [56,67] targeted the former three factors. One [54] was focused only on patient-related factors. Two addressed condition and therapy-related factors [36,51], and one [52] addressed four: patient, condition, therapy and socio-economic-related factors. None addressed the health care system related factors.
Two studies [50,69] included patients who had diabetes and depression and involved integrated care management for diabetes and depression. Both of these studies, published by the same authors, employed the same intervention strategy. The intervention was multifaceted, delivered by trained 'integrated care managers' and addressed patients' individual, social and cultural needs, and was designed to improve medication adherence, glycaemic control and depression outcomes.
Two studies consisted of telephone-based interventions to improve diabetes control [53,67]. One involved telephone calls conducted by nurses to deliver education about of diet, exercise and medication adherence [53]. In the other, participants received telephone calls from health educators primarily about diabetes medication adherence [67].
Other interventions consisted of SMS delivered by nurses to educate the patients and aid in reinforcement of diet, exercise and medication adjustment [36]; a diabetes education program delivered by community health care workers [52]; pharmaceutical care program [51]; integrated health coaching [56] and a cognitive behavioural therapy for adherence and depression [54]. Educating patients, individually or in groups, on various behavioural and medicine related issues has remained one of the most widely used strategies to promote adherence over the past 13 years. Similarly, using telephones has remained a popular strategy for delivering interventions and conducting follow-ups. However, more recently, mobile phones have been used to deliver educational messages as well as send text reminders as behavioural prompts (eg in collecting refills and taking medications) [36,64,78].
Interventions published, particularly after 2005, were more intensive and complex. The interventions not only addressed the patients' knowledge gap and problems related to self-management practices, but also incorporated a range of other pertinent issues, for example 'cognitive improvement' [35], 'cultural issues' [37,40], 'inclusion of support person' [37], 'stress and stress management' [37] and 'self-efficacy' [37,47]. Concepts such as 'integrated health coaching' [56], 'cognitive behavioural therapy' [54,62] were also implemented and evaluated. Goal setting as an intervention strategy was seen to emerge in the studies published after 2007 [43,48,56,67]. Thus, over time the interventions have become more patient-centred attempting to address specific patient needs with focus on adding more affective approaches.
Use of technology for designing interventions was reported more in the studies published after 2005. Such, technology driven interventions, varied from simple strategies, for eg. using a cell phone or an internet to input blood glucose data and receive feedback [36], to the use of complex computer programs, such as NICHE technology [38] and Well Doc's Proprietory Management Software System [39] for diabetes management.
Three studies, published since 2009 [46,57,68] have addressed the economic component, which had not been addressed previously.
Although interventions have become more complex in recent years, more 'straightforward' strategies such as delivering education materials, providing routine education sessions, and displaying multimedia programs in waiting room settings, are still used in some interventions. Moreover, the complex interventions have involved the addition of more patient oriented components to these 'simple' strategies, and these have proven slightly more successful than the 'simpler' ones.

Discussion
There has been a significant increase in the number of studies implementing and evaluating interventions aimed at promoting adherence to anti-diabetic medications, with the majority being conducted in the USA. Overall, the systematic review identified a range of intervention techniques which have been employed in an effort to improve self-care behaviours including adherence to anti-diabetic medications, in patients with T2D. The interventions were primarily directed at patients and ranged from simple educational interventions, including supply of educational materials, to complex interventions which took into consideration patients' attitudes, practices and preferences in addition to educating and aiding them to adhere to therapy. In fact, a majority of the interventions were multifaceted, consisting of a combination of more than one educational, behavioural or affective strategy. The interventions were delivered mainly by nurses and pharmacists; only a few studies reported having trained the interventionist on delivering the intervention, and even fewer reported assessing the delivery of the intervention by the interventionists.
Only a few of the interventions reviewed reported a significantly positive impact on adherence to anti-diabetic medication, and even fewer had a positive impact on both medication adherence and HbA1c levels, a parameter assessed in most studies as the measure of glycaemic control. Most of these successful interventions were multifaceted, employing a combination of strategies. However, interventions with comparable approaches showed varying results, and made it very difficult to deduce the effective intervention components. It was outside the scope of this review to investigate the components and delivery of the interventions in-depth, to determine whether the interventions were developed appropriately and/or adequately delivered by the most suitable healthcare professional. This level of investigation requires access to process implementation data which may not have been collected by the original researchers.
Combination strategies have been recognized as more likely to enhance outcomes and reduce costs for patients with T2D [79]. Whereas this aspect is undoubtedly supported by the findings in the review, a few interventions which consisted of an educational, behavioural or economic component only, have also had an impact on anti-diabetic medication adherence. Overall, educational content was the most common component of the interventions. This indicates the importance of educating patients about their condition and its therapy as a first step in increasing awareness about the importance of adherence to treatment and motivating patients to take their medications to ensure improved disease management and better health outcomes. Moreover, educating patients through the provision of written materials, electronic media, and verbally, could perhaps be regarded as one of the simplest and most efficient strategies to implement when the goal is to promote patients' adherence to their therapy. However, effective education and determining that patients have understood the information and that the information has not only increased knowledge but resulted in a behavioural change, presents a greater challenge to researchers, clinicians and healthcare providers involved in delivering interventions to patients with T2D.
Behavioural strategies, which are increasingly being used in adherence support interventions, range from adherence aids (eg dose administration aids) which address non-intentional adherence, to motivational interviewing and other more complex strategies, which aim to address several factors impacting patients' medication taking behaviour. Non-adherence may be due to a multitude of factors, and a combination of educational and behavioural strategies provide a greater likelihood of addressing the factors that impede adherence in a broader group of patients. Additionally, diabetes management involves management of lifestyle issues, and therefore, it becomes quite important for patients to adapt to the changing circumstances. Behavioural component(s) would aid the patient in the process of adaptation, and educational strategies will provide the information needed to facilitate the behavioural changes. Thus, it is not surprising that most of the interventions designed over the last few years have consisted of both educational and behavioural components.
Affective components were employed from the earlier years although affective components have become more common in the later years, and are present in most of the interventions. Newer approaches such as inclusion of 'peer supporters' as interventionists, addressing cultural issues and ethnic beliefs, adding novel techniques like integrated health coaching, and addressing cognitive issues in the more recent years, has added to the 'affective' component, while also making the newer interventions more patient-centred. While simple individual counselling or an education session could be regarded as being focussed on a single patient, the impact would however depend upon the extent to which an individual's psychosocial needs are taken into consideration. Psychosocial issues, such as attitudes about illness, affect, mood, diabetes related quality of life, resources, diabetes-related distress, and cognitive abilities have been recognised in recent years as important predictors of overall diabetes management [80]. Furthermore, from a global perspective, cultural issues and societal orientation have also been found to influence diabetes management [81,82]. Issues such as these have been recognised as important predictors of adherence and diabetes management and been included in interventions, thereby increasing the 'affective components' in recent years.
Interactive technology is increasingly being explored for health promotion [83] and was also employed in adherence promoting interventions identified in this review, in order to deliver educational, behavioural and affective components, either singly or in combination. In its simplest approach, interactive technology was used to educate, remind, or provide advice to patients. Other examples of interactive technology used were the Well-doc System, NICHE technology, tele-health devices, and RTMM technology. These technologies enabled the collection and transfer of patient specific data/ information across to different professionals/ personnel, who could then deliver the tailored feedback and reminders to the patients. The increasing advancement in technology and benefits that can be received particularly with regard to patients being in contact with their care providers from home is an appealing prospect. Furthermore, technology driven interventions could have a greater reach, better adoption and implementation, and increased sustainability; thus having a greater positive heath impact [84]. However, more research is needed to establish the sustained effects of such technologies and to evaluate how such technologies can really be useful in the short and long term in promoting adherence to medications.
Human behaviour in itself is a complex phenomenon. It is therefore more likely that any intervention designed to influence human behaviour, such as modifying medication adherence or other self-care behaviours in patients with T2D, would be more successful if multiple factors that intertwine to aid the change in complex human behaviour are addressed. Combined interventions comprise of different components, which may act both independently and inter-dependently [85], to address the change(s) desired, and may be more effective than using a single component in isolation. However, complexity involved in designing, implementation, evaluating and replicating 'combined intervention(s)', often complicates the practicalities and interventions involving a single component may be preferred as they are easier to design, implement and replicate, and often times are successful in influencing a behaviour change.
It is recommended that the diabetes services be led by the needs of patients [80]. Hence, it is logical that interventions designed for a behaviour change are guided by patients' need. Individualizing or tailoring strategies for behaviour change to the needs of the individual patients is more effective than a 'generic form' of any strategy; for example, tailoring health behaviour change messages have been found more effective than generic communications [86]. In tailoring an intervention, the strategies tend to be more patient centred. More than half of the interventions included in this review considered the individual patient's circumstance(s) while delivering the intervention; however, the extent of individualization varied. Each component (educational, behavioural, affective or economic component) needs to be individualised, and rendered more patient centred. Williams et al have also emphasised the need for 'tailored interventions' to promote medication adherence in patients with T2D [24].
The factors affecting adherence to anti-diabetic medications have been inconsistently reported [87]. However, WHO has identified five factors that interact to influence patients' adherence behaviour [5]. WHO patient-related, condition-related, and therapy-related factors contributing to non-adherence were addressed by most of the interventions identified in this review. The socio-economic factors were addressed by fewer interventions and hardly any covered the health-care system related factors. However, there was a growing trend in the later years in incorporating socio-economic factors into the intervention designs. The impact of socio-economic factors and health care system related factors on medication adherence has been established particularly in cases of chronic diseases [81,88], though their influences have been found to be 'inconsistent' [5]. As the evidence increases, it is likely that more interventions will include strategies to address socio-economic and health care system factors impacting adherence, as was seen in the more recent interventions identified in this review.
Financial or economic factors could become a major issue for patients with chronic diseases. Management of long term conditions such as diabetes and the consequences of poorly controlled diabetes could have an astounding economic impact on the patient and on each country globally [89,90]. Although the magnitude of the impact is more likely to depend on the economic status of the patients, reduced out of pocket expenses have been identified as a factor that could improve medication adherence [91]. Only a few studies have looked into this aspect, with three identified as part of this review. One evaluated the impact of US 'Medicare Part D' program on adherence, another assessed the impact of generic substitution in reducing costs, and the third focused on the impact of a value based insurance program. Whereas the former was successful in improving medication adherence to anti-diabetic medications, the latter two were not. Nothing conclusive can be drawn about how cost impacts adherence from these studies. However, expenses in terms of medication taking, diabetes management, and the consequences of not adhering to medications could have a staggering economic impact [90]. More studies are therefore needed to explore the impact of a range of economic interventions. As mentioned, health-care system related factors, including healthcare professionals, were the least addressed WHO adherence factors in the interventions included in this review. Although health care providers like nurses, pharmacists and sometimes clinicians were involved in delivering the intervention, whether they had sufficient training in the matters relating to patient adherence behaviour was unknown. Their involvement could largely be observed as a supplier of patient-related, therapy-related and condition-related information and have been classified accordingly. Health care systems globally are intricate; effectively addressing health care system during trials could pose practical difficulties, which could be a reason for the small number of interventions addressing health care system related factors. However, as they cover quiet a significant expanse in patients' overall health care behaviour, it is necessary that these factors be explored and addressed effectively when designing future interventions for patients with T2D.
In summary, while most of the interventions addressed multiple factors, none were able to cover all five WHO factors. An ideal intervention would be one that explores and addresses all interacting factors through a tailored strategy to bring about a change in patient medication adherence [87]. However, nothing conclusive could be drawn from the studies included in this review, as to the successful intervention, as the majority were found to address only one or two of these interacting factors. The factors and the extent of their influence could vary between people, and is likely to be addressed by tailoring of the intervention. Further research also needs to focus on the 'extent' of the influence of the different factors. Ample consideration is needed in terms of how such interacting factors could be effectively addressed and incorporated to improve medication adherence.

Conclusion
Multifaceted interventions appear to be a more effective approach in improving medication adherence and glycaemic target in patients with T2D than individual interventions, as they provide a range of strategies to address a number of factors that may impact an individuals' adherence to their medications. Educational component formed the most widely used component in the interventions, followed by behavioural and affective components. While educational components have remained popular throughout, affective components have become more common in the later years. Nonetheless, it is extremely difficult to conclusively state which strategies are more effective, because interventions with similar components have proved to be successful in some studies and unsuccessful in others. Additionally the variation observed in research designs, and methods of outcome assessments, together with differences in patient characteristics prevented a detailed comparison of findings or a meta-analysis.
Patient-related, therapy-related and condition-related factors were addressed by most of the interventions. Health care system related factors were least addressed. While 'social' factors are increasingly being addressed, the focus on the 'economic' aspect is still lagging behind. In addition to effectively addressing all these factors, future research needs to focus on exploring the extent to which such factors influence an individuals' medication taking behaviour throughout the medication taking journey.
Overall, future research investigating development and evaluation of effective and sustainable adherence promoting interventions needs to include standardized methods and tools, and prioritize the needs and therefore tailor interventions to the needs of individuals.

Limitations of the Review
The diverse nature of the studies included in the review presents an important limitation preventing a detailed comparison of the interventions (and a meta-analysis). Moreover, studies included were not selected based on their quality. The review was conducted with a broad approach, and intended to include all studies addressing anti-diabetic medication adherence in patients with T2D irrespective of their quality. The components of the interventions and their categorization into the WHO factors for each study have been based only on the information provided in the manuscripts. This could have had an influence on those classifications where the interventions were not explained in sufficient detail. However, the authors have done their best to ensure proper identification of intervention components using the information provided.