Type 2 diabetes patients’ and providers’ differing perspectives on medication nonadherence: a qualitative meta-synthesis

Background Poor adherence to medication regimens increases adverse outcomes for patients with Type 2 diabetes. Improving medication adherence is a growing priority for clinicians and health care systems. We examine the differences between patient and provider understandings of barriers to medication adherence for Type 2 diabetes patients. Methods We searched systematically for empirical qualitative studies on the topic of barriers to medication adherence among Type 2 diabetes patients published between 2002–2013; 86 empirical qualitative studies qualified for inclusion. Following qualitative meta-synthesis methods, we coded and analyzed thematically the findings from studies, integrating and comparing findings across studies to yield a synthetic interpretation and new insights from this body of research. Results We identify 7 categories of barriers: (1) emotional experiences as positive and negative motivators to adherence, (2) intentional non-compliance, (3) patient-provider relationship and communication, (4) information and knowledge, (5) medication administration, (6) social and cultural beliefs, and (7) financial issues. Patients and providers express different understandings of what patients require to improve adherence. Health beliefs, life context and lay understandings all inform patients’ accounts. They describe barriers in terms of difficulties adapting medication regimens to their lifestyles and daily routines. In contrast, providers' understandings of patients poor medication adherence behaviors focus on patients’ presumed needs for more information about the physiological and biomedical aspect of diabetes. Conclusions This study highlights key discrepancies between patients’ and providers’ understandings of barriers to medication adherence. These misunderstandings span the many cultural and care contexts represented by 86 qualitative studies. Counseling and interventions aimed at improving medication adherence among Type 2 diabetes might become more effective through better integration of the patient’s perspective and values concerning adherence difficulties and solutions.


Background
Medication adherence plays an important role in the clinical care of Type 2 diabetes because it directly contributes to the effectiveness of patients' treatment and wellbeing [1,2]. Diabetes affects a growing number of patients, and represents one of the primary causes of death among adult individuals [3,4]. Diabetes affects about 382 million people worldwide, of which 85 % to 95 % accountable to Type 2 diabetes in high-income countries, as well as in low-and-middle income countries [4]. The prevalence of Type 2 diabetes grows steadily, due to environmental and behavioural factors such as economic growth, urbanization, ageing populations, poor dietary habits, and decreased physical activity [4,5].
Diabetes is a disease with no specific cure and a demanding self-management regimen [4,5]. It is a progressive condition that requires continuous management as well as patient and provider collaboration in order to avoid both short-term and long-term life-threatening complications [4,5]. Diabetes management targets optimal blood glucose levels, thereby preventing the onset and progression of diabetes-related complications including cardiovascular complications, nerve damage, kidney failure, eye disease, and diabetic foot, all factors that can eventually lead to death [3][4][5]. Effective Type 2 diabetes management can include adherence to medication regimens (hypoglycaemic oral tablets and/or insulin injections), as well as adjustment of specific life-style behaviours, such as increased physical activity, adherence to specific dietary regimens, smoking cessation, and strict monitoring of blood glucose levels [1,5].
Although good glycemic control can help to prevent such complications, diabetes treatment regimens can be complex. Patients often do not adhere to medication regimens [1,2,[6][7][8][9][10][11][12]. Non-adherence represents burdens both for patients and for healthcare systems by increasing morbidity and mortality, reducing quality of life, and raising healthcare costs [1,2,6,[9][10][11]. Traditionally, non-adherence behaviours stem from a patient's failure or refusal to comply with the prescribed medication instructions due to a lack of knowledge or lack of motivation [7,[9][10][11]13]. In this tradition, researchers investigate why patients failed to comply with providers' recommendations [7,[13][14][15]. However, new perspectives on this topic acknowledge the beneficial effects on treatment outcomes of a more collaborative relationship between patient and provider that focuses on concordance rather than adherence or compliance with medication regimens. This perspective recognizes adherence as resulting from a broad set of factors, and linked to more than just knowledge and motivation [7,10,13,16]. The shift towards a more patientcentered model of care recognizes the "empowered autonomy" of patients as equal and active partners in care, contributing experiential knowledge to the decisionmaking process of care [7,10,13,16]. A patient-centered approach, then, encourages the use of a negotiated model of care to foster concordant treatment behaviours [7, 9-11, 13, 16].
Acknowledging patients' voices in the treatment decision-making process requires deeper understanding of patients' views of medications, and how these might differ from the assumptions or values of healthcare providers. This manuscript synthesizes numerous qualitative studies to distil broadly relevant and applicable insights into better medication adherence. We focus on patient and provider perceptions of patients' barriers to medication adherence, amongst community-dwelling adults with Type 2 diabetes. In particular, our research question asks: what barriers to medication adherence Type 2 diabetes patients and their providers identify? This synthesis includes 73 studies which include patient perspectives, 9 studies which include provider perspectives and 4 studies which include both patient and provider perspectives. Findings reveal the full spectrum of barriers and facilitators patients face in using diabetes medications as directed. The four existing studies comparing both patient and providers perspectives highlight some key incongruencies in attitudes and perceptions towards medication adherence barriers [17][18][19][20]. Research findings reveal discrepancies between providers' conceptualization of quality of health as opposed to the patient's idea of overall well-being, as well as different attitudes to the risk of medication adverse effects [17][18][19][20]. However, most of these studies address particular ethnic populations, or patient populations with specific comorbid conditions, or specific healthcare professional services, without providing an overall picture of the differences between patients and providers. This study adds to the under-researched literature on the differing perspectives on medication adherence between patients and providers. Further, analysis of the differences between patient and provider perspectives highlights areas for developing more patient-centered practices to improve medication adherence.
The topic of this study was informed by the Ontario Health Technology Advisory Committee's Expert Advisory Panel on Community Care for Type 2 Diabetes project on the improvement of access to, and quality of, diabetes services and care to enhance prevention and improving diabetes management. This agency commissioned a report on patient perspectives on barriers and facilitators to medication adherence. During our analysis of this data, we noted the discrepancies between patient and provider perspectives and so returned to our data to perform a secondary analysis on the current topic.

Methods
We provided the methods for the search in detail in a technical report written with the same data on medication adherence among Type 2 diabetes patients for the Ontario Health Technology Advisory Committee's Expert Advisory Panel [21]. The report focused only on patients' perspectives to barriers and facilitators of medication adherence. We summarize those methods here. Figure 1 summarizes the systematic bibliographic search process. We developed a search filter that combined existing published qualitative filters [22][23][24] with a diabetes-topic-specific filter. Because qualitative methodology filters have poor specificity, we used exclusionary terms to improve the precision of the filter (we describe details elsewhere, available as a CHEPA Working Paper online at http://www.chepa.org/research-papers/working- Fig. 1 Flow diagram selection process papers or from corresponding author upon request) [21]. We searched OVID MEDLINE, EBSCO Cumulative Index to Nursing, Allied Health Literature (CINAHL), and ISI Web of Science Social Sciences Citation Index (SSCI), for studies published from January 1, 2002 to August 10, 2013August 10, . 2002 was chosen to produce a manageable number of results, and to reflect that the knowledge before this time was well summarized in the WHO's 2003 report on medication adherence [1]. We included papers in English, available online through McMaster University's library system, reporting primary qualitative empirical research, involved or addressed adults with Type 2 diabetes mellitus (including papers with both Type 1 and Type 2 diabetes), and conducted in Canada, the USA, Europe, Australia, or New Zealand. These countries were chosen because they have similar levels of resource availability (e.g. diabetes health care, medications) to Canada. When papers were not available through the library system of our large, research-intensive university we made attempts to contact the authors to request a copy of the paper through information available in the abstract/citation or a Google search. Only 1 paper was unavailable after these attempts (as shown in Fig. 1). We excluded papers that were unpublished (e.g., reports, theses), not in English, reported secondary or non-empirical studies, used non-qualitative methods, or were off-topic (that is, not addressing the topic of medication adherence). Our search terms were designed to find qualitative studies about diabetes; further refinements of the search (e.g. topic of medication adherence, like health care context) were performed manually. Examples of exclusionary terms include "coefficient" and "p value". At least two reviewers independently reviewed titles, abstracts, and later full papers to determine eligibility. We reviewed titles and abstracts to identify findings related to medication adherence, medication and selfmanagement. We then reviewed the full text of the papers before inclusion to identify any findings related to medication adherence. Data extraction was performed by two authors; all authors participated in analysis. Discrepancies were resolved through conversation between the two authors with a third author participating when an additional perspective was needed. Studies that included either Type 2 diabetes patients OR both Type 1 and Type 2 diabetes patients were included. When analyzing studies that included participants with both types of diabetes, we considered the data related to Type 2 diabetes patients when the authors provided this separately. When no distinction was made between the data from Type 1 and 2 participants, we included all data. We included a total of 86 papers in this synthesis, summarized in Table 1. For a detailed list and description of the main focus of each study see Table 2.

Literature search
We used the integrative technique of qualitative metasynthesis to analyze our data [25][26][27]. Qualitative meta-synthesis aims to both summarize a range of findings across studies while retaining the original meaning and to compare and contrast findings across studies to produce a new integrative interpretation [28]. Analytical integrative meta-synthesis combines and synthesizes findings in new interpretative ways, while preserving the differences and complexities of the topic under study. Congruent with this meta-synthesis methodology, we started with a pre-defined topic and research question, which guided data collection, extraction of findings, and analysis. We retrieved all qualitative research relevant to this research question. Critical appraisal remains controversial for qualitative research methodology, in part because there is a lack of consensus in the field about what constitutes high quality research [29]. Procedural detail is typically under-reported, but even when reported and achieved, methodological procedures do not always guarantee to useful results [27,29]. Accordingly, we followed current conventions in qualitative meta-synthesis and neither appraised nor excluded papers on the basis of any          superficial indicators of quality other than excluding papers which did not provide evidence to support their stated findings [25,26,28,[30][31][32][33][34]. The data extraction phase involved identifying findings relevant to the topic, focusing on the authors' secondary interpretationsi.e., the authors' "data-driven and integrated discoveries, judgments, and/or pronouncements researchers offer about the phenomena, events, or cases under investigation" [26]. Primary data makes ad hoc appearances in qualitative reports; while we did not focus our analysis on these excerpts per se, we did extract participant quotes when useful for illustrative purposes.
We analyzed our data using a staged coding process similar to grounded theory, [35,36] breaking findings into their component concepts and then grouping and re-grouping those findings across studies according to inductively identified themes. First, FM, MV, and DH coded the same sources of data (the 86 articles) and identified the preliminary categories. Categories were formed based on both prevalence of information across a large number of studies and usefulness or importance of information in a smaller number of studies. These categories provided the foundation for our interpretive insights of medication adherence across the body of research. We used a constant comparative and iterative approach, in which we compared preliminary categories with the research findings, raw data excerpts, and coinvestigators' interpretations of the studies. FM, MV, and DH met regularly to discuss the analytical findings and the next analytical steps. Finally, all the authors jointly negotiated the final emerging analytical themes.
All authors participated in the overall analytical process, meeting regularly to discuss the iterative process of analysis, compare findings and interpretations, and decide how to move forward.

Results
The 86 included studies involved 2797 individuals with Type 2 diabetes, 40 caregivers, and 356 clinicians. The integrative analysis of these studies provides rich findings concerning how patients and providers perceive barriers to medication adherence. We organize these findings into 7 categories of barriers and facilitators: (1) emotional experiences as positive and negative motivators to adherence, (2) intentional non-compliance, (3) patient-provider relationship and communication, (4) information and knowledge, (5) medication administration, (6) social and cultural beliefs, and (7) financial issues. For each, we describe how patients and providers understand the barriers, and highlight key areas of congruent vs. divergent understandings.
Healthcare professionals peripherally address the theme of emotions in conversations of motivation, explicitly attributing poor adherence to patients' lack of motivation, even when providers do not explicitly discuss the impact of emotions on motivation [75,[82][83][84][85].
While patients' motives are deep rooted and difficult to modify [85], providers echo patients' perspectives on the influence of symptoms in adherence behaviours [78,82]. Asymptomatic patients adhere less consistently to medication [58,83,86]. In contrast, patients who feel unwell, thus frightened, convert to insulin therapy more willingly [86]. Some providers use insulin as a threat to motivate their patients to improve adherence [41,58]. Providers, as patients, also recognize that symptom improvement motivates patients [82], acknowledging the motivating effect of positive emotions related to empowerment and success [78].

Intentional non-adherence
Some patients intentionally and purposefully do not follow their medication regimens. We conceptualize intentional non-adherence as the patients' refusal to adhere to a specific medication regimen. Patients' beliefs and attitudes toward the health care system can promote informed and not informed intentional non-compliant behaviours [17,43,50,78,[87][88][89].
Intentional non-adherence sometimes results from denial about the seriousness of diabetes [47,57,65,68,77,90]. Denial of the severity of diabetes may relate to the belief that "everybody's got it" [90], or to the underlying scepticism and lack of trust about the effectiveness of the treatment coupled with the fear that the prescribed medication is unnecessary, unhealthy, or dangerous [44,50,80,89,91]. Most commonly patients decide not to adhere to medication regimens as an effort to avoid side effects [50,67,68,78,79,89,[92][93][94]. This type of intentional non-compliance often takes a trial and error approach, with the patient self-adjusting medication (i.e. doses and timings) [17].
Providers in several studies describe a scenario where patients agree to take the medication, but then do not follow through for unclear reasons despite provider's "best detective work" [75,95]. Providers ascribe different motivations to this behaviour, including cultural motives (e.g. preference for traditional medication), financial constraints, depression, and poor cognitive ability [17,85]. However, we found no evidence that providers recognize that intentional non-adherence may result from a patient's attempt to mitigate unpleasant medication side effects.
Providers, while acknowledging the contributions of a collaborative model of care, address systemic, structural, cultural, and linguistic barriers to patient-provider relationships that impact medication adherence [17, 66, 75, 82-85, 95, 98, 110-112]. In particular, providers recognize different ways in which they may affect patient adherence, including poor "detective work" when devising treatment regimens, poor negotiation abilities, delay in starting insulin therapy, cultural insensitivity, incorrect a priori assumptions about patient knowledge and understanding of the treatment, as well as feelings of powerlessness and frustration which affects the healthcare professionals' ability to provide adequate recommendations [17,75,84,85,112].
Sometimes providers address the negative impact of structural and language barriers to patient-provider communicationwhich in turn hinders medication adherence [17,20,85,95,110]. Systemic problems include long waiting lists, busy schedules, and practice organization barriers, which limit physicians' available time to communicate with patients [20,95,110]. In particular, time constraints and systemic barriers delay their decision to start treatment therapies (e.g. insulin), which need several clinical encounters to adequately instruct patients [84,95,110].

Information & knowledge
Patient accounts of how they negotiate their medication regimens offer explanations for why they choose to manage their condition in a way that suits their personal circumstances and understanding of their health, body, and diabetes [60,68,73,79,81,102,108]. Overall, studies present contradictory findings about the relationship between understanding and adherence [38,44,46,57,63,65,66,70,81,94,103,107,113]. For some patients, a lack of understanding and inadequate knowledge about the medication [47] and prevention of a complication [50,52,59] results in poorly controlled blood glucose levels and poor adherence [43,50,52,56,69,71,82,91,101]. In other instances, patients report that they understood medications' importance, but not how the medications work [17,40,44,53,59,78,85,91,97,103]. Thus, patients often report abstaining from medication when asymptomatic, or they consciously decide to take medication according to how they feel [17,62,69,76]. Other studies describe patients as knowledgeable, but unable to translate this knowledge into appropriate action (e.g. "what to do when things go wrong") [40,91,103,108,114]. This kind of understanding may improve with experience [44,51,94,97], necessitating a set of problem solving strategies [51,102,115], including creative solutions to diabetes self-management [68,94,97,100,102]. In light of these contradictory findings we may conclude that the role of information and understanding varies in importance depending on individual circumstances.
Providers describe patients' lack of sufficient knowledge about the disease as one of the primary reasons underlying poor medication adherence [17,84,110]. As this provider reports, 'Oh, I probably said that it [the cholesterol] was alright and then she thought is was alright to stop, something like that, that's possible? That happens: they think everything is in order again' [95].
Providers also report a strong empathy for patients around ideas such as the complexity and impact of diabetes as an unpredictable, frustrating, and longterm disease, identifying the importance of involving and integrating all aspects of the patients' life [75,83]. Providers identify the following information priorities for patients: integrated knowledge acquisition about the nature of the disease, medications used and how they work, lifestyle factors (diet, nutrition, exercise), self-care, monitoring procedures, underlying processes of diabetes, and the relationship between diabetes symptoms, medication, and long term consequences [19,20,75,82]. Providers consider discussions about medication-related issues, and improvement of patients' medication knowledge, important to promoting medication adherence [66,75,82,84,85,95]. Providers also recognize language barriers, which limit a more in-depth conversation about a patient's circumstances and health beliefs [20,110].

Medication administration
For many patients, the administration of medication poses a significant barrier to adherence. Patients describe fear as a common barrier to insulin administration, in particular fear of needles, fear of consequences of administering insulin incorrectly, and the pain of injection or blood testing [37,38,41,44,46,49,50,58,69,71,73,83,88,113,117]. Other patients specifically mention they were not afraid of needles and did not find insulin injection painful [58]. Patients relate insulin administration to other psychological barriers, such as a feeling of stigma around the possession and use of an injectable medication because of the possibility of being mistaken for an illicit drug [37,38,41,50,58,68,73,88,100,102].
In general, providers do not recognize the administration of medication as a potential barrier to adherence, except in the case of patients with physical or cognitive impairments [75,82,84], co-morbid conditions [75,83,84], or related to treatments, for example fear of needles upon initiation of insulin treatment [20,82,84,86,110]. Healthcare providers perceive family support as crucial for patients with poor cognitive and physical resources, [82,110] for reinforcing providers' medication instructions, and for holding the patient accountable for his/her selfmanagement [17,82,84,110].

Social and cultural health beliefs
Health beliefs about medication and diabetes are often linked to social or cultural understandings about the body, diabetes and medication, which in turn can affect medication adherence in many different ways. Multiple factors shape these health beliefs, such as the information sources used by the patient, past experiences, attitudes of others, faith and religious beliefs, education, and cultural community [38,41,44,48,71]. A patient's health belief system may affect the way he or she decides to approach medication adherence, and how to integrate (or not) the requirements of the medication regimen into everyday life [44,76].
A patient's health beliefs and cultural background will also affect the relationship s/he desires with the prescribing physician [44,48,50,59,60,62,69,71,72,74,89,91,117]. For instance, patients who are members of historically oppressed communities by the dominant culture can be suspicious of medical advice [69,71,74,89,117]. Patients from cultures that perceive physicians as high status individuals with significant authority may feel uncomfortable asking questions [20,93,98]. Several papers recommend including the patient in a culturally appropriate way as an active partner of care to improve medication adherence [39,41,57,58,70,93]. However, providers should adapt to the patients' beliefs and preferences, as some patients may refuse to work with clinicians in this way.
Provider perspectives rarely address the issue of traditional medication alongside or instead of allopathic medication [95]. Providers are more likely to mention challenges linked with the patient's cultural background and beliefs, such as aversion to insulin, fatalistic attitudes, the perception that fat is healthier or a desire to please the physician [20,83,84,110,112].

Financial issues
Patients widely mention the cost of medication as a barrier to medication adherence [17-19, 37, 42, 44, 46, 50, 64, 68, 70, 75, 76, 81, 89, 91-93, 97, 119], although studies involving participants with access to public health insurance less likely to mention cost as a barrier [47,73]. Financial barriers can extend beyond the cost of medication and physician services. Even patients with health insurance can struggle to afford testing supplies, syringes, and non-physician supportive care [17,19,37,70,75,76,81,91,92,97]. Patients living in poverty also face other structural and material constraints such as low health literacy, poor quality housing, shift work, stress, inability to access healthy food etc., that affect their ability to adhere to medication regimens [37,68,70,89,119]. When faced with financial constraints, patients may use tactics including: taking medication less often than recommended, choosing the most "important" medication to pay for, sharing pills with other people, drawing on personal capabilities and social networks, and asking their doctor for help [18,50,68,89,92].
Providers interviewed in some projects understand the financial barriers that patients may face, but commonly do not identify this issue as a barrier to medication adherence [17,18,75,83,112]. In some cases, while providers may recognize that some patients struggle with the cost of medication, 'these were clearly secondary concerns from the clinicians' perspective' [18]. In several instances, clinicians acknowledge the cost of medication as a contextual barrier along with other struggles related to low socio-economic status [83,112]. While some clinicians perceive that these struggles are outside of their realm of influence [75], others offer creative strategies for alleviating financial burden such as prescribing generics, giving samples, changing the regimen to accommodate constraints or helping patients participate in patient assistance programs [18,75].

Discussion
Extensive qualitative research exists on the topic of barriers to medication adherence amongst community-dwelling adults with Type 2 diabetes. Our synthesis of this research to date suggests that providers and patients share some common understandings of these barriers, as well as facilitators to overcome them. However, the qualitative research also identifies many points of misunderstanding, miscommunication, and missed opportunities for intervention. In general, providers tend to limit their focus to clinicallyoriented issues, while patients describe a much wider range of problems with medication adherence that arise from the personal, social, and practical challenges of living with diabetes. To the extent providers understand and address these wider concerns (possibly through a multidisciplinary professional, patient-centered approach to care), they will potentially improve both medication adherence and patient experiences. This reflects what the literature addresses as patient-models of care, where patients adhere to medications and not comply (emphasis added) overtaking past paternalistic models of care [6,16,120].
Our synthesis on medication adherence may contribute to define more clearly the key dimensions of the personcentered (PC) model of care and illustrate how this model may improve medication adherence among Type 2 diabetic patients. According to Bower and Mead [16], PC care is based on the following dimensions: inclusion of biopsychosocial factors, viewing the patient as a person, enhancing patient's empowerment and autonomy, involving the patient in the decision-making process through a two-way communication process and negotiation, encouraging a collaborative and mutual trusting relationship between patient and provider, and emphasizing the doctor as a person.
Our study shows that patients and providers often agree on the importance of medication adherence for symptom improvement, the benefit of a collaborative and responsive relationship between patient and provider, and effective communication of information. The integration of patient's perspectives in the clinical relationship, based on a mutual and trusting relationship, broadens the scope of the explanatory model of illness by addressing different 'dysfunctional' states and the possible interventions areas to improve adherence [7,10,16,[120][121][122]. As the PC model of care describes, these factors help patients integrate their medication regimen into their own system of health beliefs, the individual context of their everyday lives, and their changing circumstances [10,16,120,121,123,124].
Conversely, patients commonly cite providers' lack of collaboration, lack of interest in the patient's life and context, poor communication, or time constraints as interfering with their medication adherence. Providers on the other hand, as documented in recent studies [125][126][127], tend to address systemic barriers, such as limited time consultations and lack of inter-professional collaboration, as obstacles challenging the prioritization of patients' medical and psychosocial needs.
Providers and patients differ significantly on how best to influence patients' self-management practices. Providers tend to focus on patients' knowledge about the physiology of the disease and role of medical and lifestyle interventions: i.e., the nature of the problem, what needs to be done, and how. While some providers do recognize the importance of emotions and psychosocial factors, providers relate these more to the motivation, than to the capacity, to use medication properly. Recent studies corroborate these findings, indicating that providers consider motivation as crucial for patients' understanding of the illness and effective medical education [126,128].
Patients, however, describe diabetes and medication self-management as a multi-dimensional experience. Practical aspects include finances, daily routines, and the need for instrumental support. Psychosocial aspects include health beliefs, emotional impacts, social and cultural understandings of diabetes and medications. Selfmanagement models based on behaviour and integration theories reflect this perspective on medication adherence behaviours and self-care activities, acknowledging both the psychosocial and biomedical nature of the medication regimen [129,130] and highlighting that perceptions may differ between patients and providers on treatments goals and strategies [131]. Providers' focus on biomedical problem-solving can also leave patients feeling "reduced to their disease". This runs counter to the person-or patient-centered care approach that calls for treating people holistically, with their attention to their disease placed in context of attention to other factors in their lives [16,121,123]. All of these factors influence the way that patients interpret and apply medical advice.
This synthesis of the qualitative research on patient and provider perspectives underscores the recommended shift from the traditional medical view of medication "compliance" to the more patient-centered view of medication "concordance" with patients' many other needs, pressures and demands [7,10,13,16,121,123].

Strengths and the limitations
A number of strengths and limitations of this study are worth noting. First, this study provides an updated and comprehensive qualitative systematic synthesis of Type 2 diabetes patients' and providers' different perspectives on medication adherence. Existing systematic reviews of qualitative research focus on diabetes management more broadly [122,[132][133][134], and quantitative reviews on medication adherence elicit different types of information. For example, quantitative studies concentrate on measuring medication adherence rates among Type 2 diabetes patients, or measuring medication adherence rates for specific drug therapies, nutrition regimens and educational interventions designed to improve adherence, rather than addressing the reasons and experiences of trying to adhere to medication regimes [12,[135][136][137][138][139]. By comparing patient and provider perspectives on this issue, we are able to provide an interpretive synthesis of the constellation of challenges that patients may face when prescribed a medication regime for their diabetes.
Another strength of this study is the large body of qualitative research available for synthesis on this topic: we were able to include 86 studies that together captured thousands of patients' experiences. The integrative meta-synthesis method also allowed us to distil robust thematic findings, each supported by a number of studies and therefore more transferable across settings. The rigor of the synthesis allows establishing generalizability and consistency of the findings among a large number of studies and across different countries, patient population groups, and patient demographic differences. A significant portion of the included articles (33 out of 86) focused on very specific populations of diabetic patients, such as indigenous groups, immigrants, and minorities. We used a constant comparative technique to examine medication adherence barriers across such groups in our analysis and found our main categories to be consistent across groups, with some variation in the sub-categories. For example, we noted the consistency of the theme of the importance of the patient-provider relationship across studies that included marginalized and nonmarginalized populations. Comparative analysis of this theme revealed that the patient-physician relationship was consistently mentioned as important by authors of both types of studies, although they tended to focus on different aspects of this relationship. For example, marginalized participants spoke about the difficulty in communicating with language and cultural barriers and the impediments that this provided to a strong, supportive relationship with their physician. Non-marginalized participants were more likely to focus on barriers such as wait times and short length of appointments. Both groups emphasized the importance of this relationship on their ability to self-manage their diabetes. Similarly, the results did not present consistent differences by age or gender across study populations, as the populations were overall consistent in terms of age and gender. We found that barriers to medication adherence were consistent across the different diabetic and demographic populations considered in the 86 studies. Therefore we are not presenting the results as stratified along these lines. Findings concerning culture in particular may be less transferable to jurisdictions and cultures not captured in this body of research. A limitation of this study is its focus on English-language research reports.
This study was limited also in other ways. First, this review includes only research conducted between 2002-2013. These dates reflect an attempt to include a manageable body of current literature. Given the depth and complexity of qualitative data, 86 studies provide a large body of data to describe and interpret. The WHO report on medication adherence published in 2003 [1] offers an authoritative summary of the state of knowledge before 2002; we have provided a review of this foundational literature in the introduction of the current manuscript.
Second, this meta-synthesis retrieved a great number of patient experiences reflecting a limitation of the underlying body of research: the relatively fewer qualitative research into providers' perspective on patient-barriers to medication adherence among Type 2 diabetes patients. Recent studies corroborate our results reinforcing the sense of saturation of our data [125][126][127][128]131], however, because studies on patient, not provider, perspectives continue to dominate the field, we highlight providers as an important population for future qualitative investigation and possibly multi-methodology research syntheses.

Conclusion
This study synthesized Type 2 diabetes patients' and providers' views about medication adherence, highlighting that these groups have different medication adherence priorities. While providers tend to focus on patients' motivations and medication administration practices, patient accounts emphasize their experience of diabetes as part of a holistic consideration of their whole lives. Providers shared clinically oriented perspectives, detailing rich and sophisticated conditional reasoning about their efforts to persuade patients to adhere to their medication regimen. Patients' accounts of medication adherence describe individual experiences of diabetes medication that are deeply embedded within the context of the individual's particular life circumstances, emphasizing that medication self-management practices are built upon more than just knowledge and motivation for change. The conceptual divide between patients and providers on the topic medication adherence enriches our understanding of why medication adherence may be experienced as an intractable issue by both patients and providers. The findings of this synthesis may assist providers in identifying potential factors that affect a particular patient's medication practices. Taking a patient-centered approach to medication self-management may encourage increased understanding the priorities and experiences patients, encouraging providers to identify the multiple underlying factors that promote or inhibit medication adherence in their patients creating the opportunity for patients to voice their questions or concerns about their medication regimens. Interventions that aim to improve medication adherence will benefit from considering the issue of adherence from a patient-centered model of care by tailoring the medication regimen to patients' life contexts, preferences and self-management practices.