Barriers and facilitators to evidence based care of type 2 diabetes patients: experiences of general practitioners participating to a quality improvement program

Landmark studies have demonstrated that intensive management of hyperglycemia, hyperlipidemia, and hypertension significantly reduces morbidity and mortality in patients with type 2 diabetes mellitus (T2DM) [1‐9]. T2DM is a 'silent disease' until irreversible microvascular (e.g., nephropathy, retinopathy, diabetic foot) and/or macrovascular (e.g., myocardial infarction, stroke) complications become apparent. Prevention of these complications rests on timely institution of drug therapy by the prescribing physician, usually a general practitioner (GP), and the patient's compliance with the treatment regimen and willingness to make lifestyle changes. A proactive follow-up of diabetic patients is essential and should include foot examinations, blood and urine tests, and eye examination [10]. In addition, patients should be counseled about the dangers of diabetes and the importance of a healthy lifestyle, and impressed with the need for compliance with doctor's orders.

Unfortunately, many patients do not receive such level of care despite the availability of internationally-accepted treatment guidelines describing optimal management of patients with diabetes [11]. Optimal use of guidelines in general practice demands specific implementation strategies aiming at the reduction of barriers to high-quality care [12]. However, a clear understanding on how to overcome these barriers seems to be lacking [13‐15], despite previous studies which outlined the obstacles that prevent GPs from following the guidelines [16‐24]. Our study reports on 20 GPs who participated in an 18-month quality improvement program (QIP). The aim of this program was to improve diabetes-related patient outcomes through the implementation of evidence-based guideline recommendations. The different interventions of this QIP are described in the Appendix. The program resulted in significant improvements over time of HbA1c (-0.4%, CI 95% (-4;-3)), systolic blood pressure (-3 mmHg, CI 95% (-4;-1)) and LDL-C (-13 mg/dl, CI 95% (-15;-11)). However, results widely varied between participating GPs. Accordingly, we conducted a complementary, qualitative study (January to April 2008) nested in the controlled trial, to gain better insight into what changes the GPs had actually experienced. To fully understand these changes, we relied on an 'implementation model' based on the one described by Grol et al., 2004 [25‐27].

We conducted this qualitative research to acquire a better understanding of the barriers to high-quality diabetes care and into the mechanisms of change that eventually were induced by the QIP according to the experience of participating GPs. We opted for 'one-on-one' interviews in order to investigate the perceptions of the GPs about the QIP that essentially targeted the individual GP. We opted for semi-structured interviews in order to let the interviewees talk freely, as well as to deepen the interviewees' personal feelings about both the experienced barriers to high-quality care and facilitators of change.

To gain maximum information, the interviewees were randomly chosen from a stratified sample of participants according to clinical performance scores before and after the intervention. The clinical practices were divided in four strata relying on baseline performance (stronger versus weaker) and on the degree of improvement during the project (modest versus substantial). A researcher not involved in the interviews randomly chose five GPs within each stratum. If a selected GP refused to participate, the next GP on the list in that stratum was invited.

Subsequent discussions delved deeper into these topics by using an adaptation of 'reflective listening', a counseling technique that elicits a thorough disclosure of the interviewees thoughts and feelings [28]. It involves reflecting back to the interviewee what the interviewer believes was said in order to verify or clarify the interviewee's statements, and encourages interviewees to continue elaborating their views. In our interviews, not only were the assertions reflected back, the interviewees were also actively confronted with eventual inconsistencies in their answers. Throughout, the interviewers provided reassurance by intonation and body language in order to disclose the very personal feelings and experiences of the interviewees.

The interviews took 30 to 45 minutes and were conducted individually by two experienced researchers (GG and LBO), one a practicing GP and the other a community nurse specializing in health care consultancy. All interviews were taped and transcribed.

Two GPs refused to participate in the interview and were replaced by the GP next in line. In a post-hoc analysis, we found that few new themes were emerging after about 17 interviews, making it unnecessary to continue the interviewing after the 20 initially planned interviews. Table 1 shows the main characteristics of the interviewees that were felt to be typical of all 120 participants in the QIP. Table 2 shows the results of itemization that was obtained in commons consensus by the two researchers.

All but four of the GPs confirmed the importance of improved adherence to the evidence-based guidelines. The four GPs who did not experience improved adherence belonged to a stratum with a stronger baseline performance, and three of them also belonged to the stratum with weaker improvement during the project. Three of them revealed that they had previously followed an intensive course on diabetes management. The fourth GP is still collaborating with the medical faculty of the university. Most interviewees also reported improvements in follow-up procedures, evidence-based drug prescription practices, and referral rates. The more frequent follow-up visits included regular blood monitoring and general screening for complications. Several GPs mentioned better recordkeeping.

Implementation of evidence-based treatment was evident in more timely adjustments in therapy if target criteria fell short, and in greater attention to cardiovascular risk factors, above and beyond conventional glycemic control. Finally, more patients were treated with insulin.

Some interviewees reorganized their practices to better comply with the guidelines. Others instituted regularly scheduled office visits, and some split the visits into two parts: one part dedicated to routine follow-up and the other to discussions of treatment and lifestyle. The interviewees noted better medication compliance and improved adherence to follow-up schedules by the patients.

Previous studies have disclosed a significant gap between the quality of diabetes care commonly encountered and recommended evidence-based guidelines [14]. To date, most research on barriers to and facilitators of high-quality care has been done before the start of improvement programs. Our study was based on interviews with GPs who actually participated in a project aimed at optimizing diabetes care. This approach, combined with the 'reflective listening' technique, elicited disclosure of very personal feelings and experiences related to changes in performance. As such, qualitative research nested in an experimental trial may clarify the improvements that a QIP brings about in a general practice.

The primary finding was that the project accomplished more than merely improving the quality of care. It also impacted the emotional and motivational status of the GPs. Previous focus group-based research had revealed that GPs working in the 'usual' setting in our country felt frustrated, partly because they felt inferior to specialists [29]. We showed that role-redesign and delineation of responsibilities vis-à-vis the specialists enhanced a GP's self-esteem and sense of responsibility. All interviewees were unanimous that this project was very beneficial because it added value to their jobs, even though some were concerned that QIPs could have manipulative ends or lead to sanctions.

Second, most of the GPs reported a major improvement in their diabetes care. According to the theory of planned behavior, decisions are made according to personal models and beliefs about the changes about to be made, and the perceived benefits and risks associated with them [30]. Several GPs indicated that the changes resulted from a conscious decision based on interconnected key elements during the quality improvement process. Reported key elements were the need to keep up with knowledge, the increased awareness that their practice needs improvement, and that their attitude needs adjustment. The GPs also observed attitudinal changes in their patients, e.g., better adherence to drug regimens and follow-up visits.

Finally, the QIP also altered interpersonal relationships. Most GPs confirmed that the QIP strengthened their relationships with their patients and improved communications with specialists and other healthcare providers. They also perceived a change in attitude on the part of the endocrinologists toward them, which markedly enhanced the GPs' motivation and sense of responsibility. These findings substantiated various theories and research findings that a positive relationship among healthcare providers is an important component of high-quality patient care [31, 32].

Nevertheless, limitations of the QIP were also described. First, according to the interviewees, a significant minority of patients remained refractory to change, with many refusing to see a nurse educator. Most patients found it difficult to change their lifestyle, and even in the case of motivated individuals the changes were often minimal and temporary. These findings confirm previous findings that sustainable lifestyle changes are hard to implement in clinician-centered models of patient education [18, 33‐35]. Moreover, these models are labour- and resource-intensive [36] and traditionally put the emphasis on imparting knowledge [37]. Yet, in even the most successful trials of face-to-face education, many participants are not willing or able to attend the sessions [38, 39]. Therefore, ongoing research evaluates the effect of new models that are based on peer support. These models put the emphasis on coping with illness, rather than managing it [40]. Peer support seeks to build on the strengths, knowledge and experience that peers can offer. Greenhalgh et al. has tested the effect of a narrative method (a person telling a story) versus conventional nurse-led education in a minority ethnic group of people with diabetes [40]. The results show that unstructured storytelling is associated with improvement of patients' enablement and comparable changes in biomedical markers. Other self-management programs evaluate the effect of other peer support interventions, like telephone counseling or web-based peer support. Future QIPs may incorporate peer support interventions replacing or complementing the traditional clinician-centered patient education interventions.

At GP-level, four interviewees affirmed not having experienced a major impact of the QIP on their quality of care. In fact, they experienced the QIP somehow as superfluous because they already paid special attention to evidence-based diabetes care before the start of the project. The study also revealed that some GPs were reluctant on to reorganize their practices to comply with the project's requirements, or even to find the time for efficient patient follow-up. Accordingly, future QIPs should specifically address such issues. Moreover, while the project was indeed able to induce a change in attitude with regard to medical diabetes treatment, some other deeply rooted attitudes were more difficult to change. For example, several GPs asserted that nurse educators and other personnel in the so-called 'soft sector' are of little value in good diabetes care. Collaborative shared care with specialists also remains a concern, despite the improvement that was observed during the project. One GP reported persistent problems with one local endocrinologist who was blamed for his disdainful attitude to general practice. Other GPs described minor remaining difficulties with endocrinologists despite overall satisfaction with the arrangements. These findings complement previously reported difficulties in collaborative shared care. One of the major reported issues about shared care is the problem of suboptimal communication between the involved providers [41]. This problem is associated with discontinuity in care and lower quality of care [42]. Other problems are related to lack of clear division of tasks and responsibilities between the involved providers, eventually leading to overlap and competing interests [29, 43]. Despite these problems, we think that shared care is necessary to guarantee high-quality diabetes care because the management of this disease is too complex and too broad to have it provided by one person. However, the aforementioned problems are a real point of concern. Moreover, as our research shows, providers are not always willing to collaborate. Thus QIPs should pay special attention to eventual relational problems, to communication issues and to the distribution of rights, responsibilities and tasks between patients, GPs, nurse educators and specialists.

The role of EBM in daily practice remains a point of controversy. While many GPs accepted the existing guidelines, some did not. Some GPs fundamentally disagreed with EBM. Others accepted EBM as background support, but were afraid that EBM would be used to impose coercive instructions for daily practice. Several GPs questioned the feasibility and desirability of the American Diabetes Association guideline-based recommendations in the elderly or immobile people. Indeed, elderly patients are particularly sensitive to the adverse effects of drugs and polypharmacy, putting constraints on the classic diabetes treatment. In particular, hypoglycemia is an important topic in the diabetes treatment of elderly people. Recent studies [44, 45] clearly indicate that hypoglycemia may be a contributing factor to morbidity and mortality in older patients. As such, strict adherence to guidelines for younger patients could be deleterious for the frail elderly [46]. Geriatric guidelines on the management of type 2 diabetes accentuate that treatment should be holistic, targeting all important aspects of the geriatric patients with priorities in the treatment scheme. Diabetes-related targets should be individually adapted to the frail patients with special attention to avoidance of side effects [47‐49].

This qualitative research presents some limitations. A first possible bias concerns the researchers who conducted the interviews. They were previously involved in the QIP, and thus they are known by the interviewees as promoters of this program. As a consequence, GPs in disaccord with some issues of the QIP-process may have been discouraged to mention them. The GP cohort selected for the study represented an additional limitation. The participants were part of a larger sample of volunteer GPs who were particularly interested in the project. This selection bias may well be reflected in their answers. In order to generate a broad spectrum of answers regarding barriers to change, we employed a targeted sampling procedure that took into account the performance of the GP's practice. Only their subjective feelings and views are covered here, although a more balanced picture would have emerged if a joint patient-provider perspective had been offered. It remains for future research to include interviews with patients and, perhaps, employ mixed focus groups, and audio- or video-record observations of the clinician-patient encounters. However, despite the possible bias, we feel this qualitative study has provided a very balanced overview of the QIP's strengths and weaknesses, and validated the quantitative findings that had been obtained.

Previous research revealed numerous barriers to high-quality diabetes care at the level of provider, patient, and healthcare organization. However, most of this research was done outside the context of quality improvement. Our research reveals the viewpoints of physicians who experienced a quality improvement process and it allows for evaluating the complex interactions between barriers and facilitators during this process. It has become obvious that implementation of a QIP encounters an array of cognitive, motivational, and relational barriers that are embedded in a patient-healthcare provider relationship. As their success may depend on overcoming key barriers, QIPs should incorporate mechanisms to actively detect and overcome these barriers or to cope with them. Moreover, several barriers appear to be interdependent, developing several 'chains of barriers'. This phenomenon may be a reason why multifaceted QIPs acting on different barriers in a chain are likely to be more effective than single interventions.

Our research particularly revealed the GPs feelings on collaborative shared care. While some of them disagree on the added value of diabetes educators, many GPs feel some uneasiness regarding the competition with specialist care. These feelings may be reinforced by the typical Belgian healthcare setting, but we believe that they are the expression of a very human nature and thus not unique to the Belgian situation. Literature on this issue, however, is very scarce. Our research also showed that these negative assumptions and feelings can be overcome by paying attention to them and by enhancing the personal contact and communication between the people involved.

The interviews also revealed the limits of a clinician-centered model of patient education and self-management, and confirmed the quantitative results of the study on this issue. Future QIPs could incorporate and test innovative patient-centered methods, like different models on peer support for patients.

Finally, several interviewees reported real concerns on the applicability of the 'traditional' diabetes guidelines in a subset of the patient population, namely the elderly. These concerns have been joined by specific geriatric guidelines. These findings show that quality improvement is not a unidirectional process from guideline to practice. Often, several practitioners express the same difficulties with implementing a guideline. In that case, it might actually reveal a flaw in that guideline rather than a barrier related to the practitioners. And thus QIPs should also be used as instruments to test the feasibility of guidelines as well as to highlight any flaws.