Discussion

This study explored whether offering protocolised primary diabetes care in a care group is related to improvement of people with type 2 diabetes being monitored as recommended. We found that after 1 year of collectively organised and facilitated primary diabetes care, monitoring of people in line with GP recommendations increased substantially. In addition, we found in experienced practices, participating at least 3 years in the care group, no significant differences in recommended monitoring as compared with new practices, participating for 1 year. These findings indicate that participating in a care group has a rapid and enduring effect on the quality of monitoring of people with type 2 diabetes.

To our knowledge, this is the first study in Europe to explore the relationship between care group participation and registration concerning monitoring of essential biomedical and lifestyle diabetes indicators. As demonstrated by previous work,12 appropriate registration of diabetes monitoring is associated with significantly better HbA1c levels. Similarly, a meta-analysis established that appropriate self-monitoring of blood glucose was associated with better HbA1c levels.20 Thus, in our view, adequate monitoring is clinically relevant. Our findings underpin the outcomes of a longitudinal evaluation regarding the first Dutch initiative on collectively supported implementation and delivery of structured primary diabetes care. This study revealed a trend reflecting improved measure of indicators such as systolic blood pressure and LDL.21 In addition, our results support the conclusions of previous annual national benchmarks which were based on aggregated data of care groups between 2011 and 201322 and which suggested that monitoring of people in line with professional GP guidelines has improved. Furthermore, our findings are confirmed by a British evaluation of GP support by diabetologists and nurse specialist concerning diabetes care, which showed that the number of appropriate referrals to secondary care increased significantly.23 In the USA, the CPC initiative has key characteristics in common with the Dutch care group approach. Our findings show a greater increase in monitoring than found in the evaluation of the first year CPC24–26 which detected only small improvements in monitoring. This difference might be explained by the recent introduction of the CPC programme, since an in-depth evaluation of US practices participating in the CPC programme revealed that practice staff appreciated advice adjusted to their job roles and practice organisation, and the electronic health record system and other digital systems used in their practice27—indicating that a quality transition had been initiated. In addition, an evaluation of the first year of the Dutch care group approach reported much room for improvement of individual monitoring, hardly any significant improvement of diabetes-related health outcomes and missing data due to registration problems28; also, in an evaluation of the second and third years, only modest improvements in monitoring were found.29 In other words, the better outcomes of our study might be explained by a broader experience with the care group approach.

In our view, one important strength of this study is the design. In general, a randomised clinical trial (RCT) might be useful to eliminate bias. However, in RCTs achieving adequate powering is a common problem. In contrast, observational studies generally allow inclusion of large-scale study populations. To illustrate, in the case of our study, meeting the powered study population within an RCT design would have been severely hindered by logistical barriers. That is, finding sufficient practices that were willing to be assigned to a randomisation procedure concerning care group participation or a control condition would virtually have been impossible. This problem can be avoided with an observational design. Thus, when using an observational design in this field, barriers with regard to the external generalisability of the findings might be alleviated.30 In addition, since our design typically does not interfere with the daily organisation of GP practices, adequate reliability of our findings can be assumed. Moreover, in our study, the observational real-life setting reflects the reality of diabetes monitoring in this specific study population. The design we used is in line with other studies that also used a pragmatic design to conduct diabetes-related studies in primary care.31–35

Nevertheless, some limitations warrant discussion. First of all, our findings are only generalisable to people younger than 80 years participating minimally 1 year in the care protocol. Second, the number of new practices was relatively low, which might have influenced our findings on the effect of care group participation. For example, two new practices lacked baseline data, indicating weak registration of diabetes monitoring, and were thus excluded for our first research analysis; in addition, in the new practices, a considerable number of people were excluded because of missing information on essential personal data (age, gender and diabetes duration). Missing data are a common challenge when using routine registry data.36 This implies that our results on the effect of care group participation are primarily applicable to people with registration of elementary diabetes-related information. Second, since no control group could be included, we cannot proof a causal relation between the observed increase in the monitoring of people and participation in a care group. In addition, it should be noted that given the observational design, our findings might be affected by residual confounding. Third, concerning the second analysis, different groups that varied in size were compared. Therefore, our findings might have been influenced by other factors (eg, size and organisation of the GP practice, or characteristics of the practice population) even though we did correct our analyses for the level of GP practice and additionally for age, duration of diabetes and gender of the individuals.

Our study shows that providing protocolised primary diabetes care in a care group context is associated with a rapid increase in monitoring of individuals with type 2 diabetes. This might be explained by the three cornerstones of the care group support. First, in the context of a high workload and competing priorities in daily GP practice,3 the support provided to GPs and nurse practitioners with regard to implementation and delivery of a diabetes care protocol might encourage essential organisational changes in individual practices. This is supported by a Canadian study showing that in the view of GPs, supporting access of GPs to other health professionals in primary care such as nurse practitioners facilitates interprofessional collaboration and improves diabetes care.37 To illustrate, although the collaboration process between GPs and nurse practitioners in daily practice is sometimes perceived as challenging,29 within care groups, different stakeholder groups report clarity about one another’s expertise, roles and tasks.38 Accordingly, process coaching by an experienced staff nurse might ameliorate the functioning of the GP team and subsequently care delivery. More effective functioning of the GP team and improved care delivery might result in development of a team-based approach to realise timely invitation of people for diabetes consultations at ward or a team-based approach to reduce no shows.

Second, effective use of a CCDSS enables systematic and appropriate monitoring of diabetes-related health outcomes. Because the accessibility of IT systems is known to be a barrier in primary diabetes care,3 39 appropriate coaching concerning the use of these systems is required.40 Care group-related support with regard to the use of a CCDSS stimulates up-to-date oversight of individual monitoring, thus contributing to a higher number of people being monitored as recommended. Third, the mandatory educational diabetes courses enable GPs and nurse practitioners to keep their knowledge and skills up to date. As a result, optimal benefits from the collective approach might be derived.

In other words, the care group approach tackles several internationally reported barriers on the delivery of diabetes care and thus contributed to improvement of care quality. Therefore, the benefits of collectively organised logistic and quality support might also be relevant for other protocolised diabetes care settings, such as the CPC+ programme in the USA.

From the perspective of individuals with type 2 diabetes, quarterly consultation in a care group setting, which is characterised by systematic and ongoing attention for diabetes-related self-management and lifestyle support, is associated with an increase in being monitored as recommended, although for certain subgroups of people, a more flexible ‘care protocol’ might be sufficient.41

For future research, further examination of factors that might affect relations between care group participation and outcomes within participating practices—such as local geographical and socioeconomic characteristics or practice organisation—is needed to gain a better understanding of the association between care group participation and monitoring of people. To add, previous studies have shown that structured primary diabetes care and structured monitoring of diabetes target indicators are associated with improved diabetes-related health outcomes, including HbA1c,12 13 which in turn affects the risk of fatal and non-fatal myocardial infarctions.14 42 However, more detailed exploration of the relationship between monitoring of individual diabetes indicators in line with professional recommendations, diabetes-related changes in treatment and health outcomes (eg, meeting treatment targets, cardiovascular complications, hospital admissions) might enhance our understanding of adequate, collectively supported primary diabetes care. Next, evaluating the financial costs and benefits of this diabetes care approach might be interesting for policymakers. Finally, although we found that protocolised primary diabetes care with collective support is associated with better monitoring, little is known about the personal perspective of the individuals themselves with regard to participation in a structured care protocol.

To summarise, in practices that started with protocolised primary diabetes care within a care group setting, the monitoring of people as recommended increased considerably after 1 year. In experienced practices, the odds of being monitored in line with professional guidelines did not significantly differ from new practices participating 1 year in the care group. Thus, collectively organised logistic and quality support of GP practices is associated with improvement of primary diabetes care monitoring. The association between care group participation and diabetes health outcomes needs further research. More insights into the personal perspective of the stakeholders (GPs, nurse practitioners and individuals with diabetes) are recommended.