Discussion
Diabetes is a public health priority requiring adequate management and monitoring at the primary care level to reduce prevalence and prevent complications. Mechanisms to assess progress need, therefore, to be robust and effectively monitored. In April 2017 participation in the NDA became a contractual obligation, and STPs and clinical commissioning groups (CCGs plan to use the NDA outcomes to help monitor progress locally.
The study aimed to explore GP-related factors which may influence NDA participation and the impact that participation in the NDA may have on diabetes management and patient care. The results indicate that smaller practices from certain areas of the country and those that have a higher proportion of elderly patients, greater diabetes prevalence, and practices with patients from more deprived areas were less likely to participate in the NDA. Evidence from the NDA suggests that a lack of understanding of the benefit of participating, a onerous data submission process, timing of the data submission window, competing priorities and uncertainty around consent, and submitting patient identifiable data may explain why practices with large diabetes and elderly case loads, smaller practices and those from certain areas of the country or from deprived neighborhoods may be less likely to participate in the NDA.23 It is important that CCGs and STPs are aware of these inequalities when offering support and/or resources (eg, local incentives or the appointment of a local clinical champion) in the future to local practices for diabetes management and monitoring.
It is possible also that additional factors may account for local participation rates. Despite accounting for, and adjusting for, several demographic variables, it was not possible to account for other factors that are not routinely collected. It is possible that these factors could influence NDA participation and diabetes management, which could in turn introduce bias and confound the results. Evidence suggests that other practice-related factors, including practices with a dedicated diabetes recall system, mini-clinic or specialist staff (eg, nurses, dietitians or chiropodists), or those that have fewer patients attending hospital clinics, may have a positive impact on diabetes care and outcome.12 13 16 Although these data were not available within the current data set, these factors should be explored locally in the future.
The results also suggest that there are statistical differences in diabetes management outcomes between practices that participated in the NDA and those that did not. HbA1c, but not blood pressure and cholesterol, was independently related to NDA participation when accounting for practice characteristics. This suggests that participation in the NDA may be related to better diabetes outcomes and patient care. However, these results need to be interpreted with caution as differences are small or marginal and the clinical significance is difficult to determine. It is possible therefore that certain biases may exist which could confound the results. One possible explanation for these results may be the perverse incentive created by the QOF points achievement thresholds. In the clinical areas of the QOF, practices are rewarded with points if their achievement is over a certain threshold. Thresholds for diabetes indicators are in the range of 38%–78% (DM003), 40%–75% (DM004), and 35%–75% (DM007). Practices will generally receive maximum points for an upper threshold and proportionately between those limits.19 As a consequence, a floor and ceiling effect may occur, because once practices achieve the lower limit of the threshold there may be a lack of further attainment incentive beyond the upper payment thresholds. This can result in little variation in achievement rates across practices and as such may lead to small differences being observed.24 The small, although significant, association between NDA participation and diabetes management may also be explained by the large number of cases in the regression analysis which may have resulted in the model being overpowered. The impact of type 1 versus type 2 diabetes on HbA1c results must also be considered as it is possible that patients with type 1 diabetes have more variation in HbA1c; however, these data were not available for the current study. In addition, only 1 year of data was used in the study. The 2014/2015 data set was selected as it was the first year that a voluntary opt-in participation process was used. This meant that the process was more similar to the contractual obligations now adopted and ensured equal grouping and few missing data which may otherwise have biased the analysis. Further investigation should replicate the study with data from the first year of compulsory participation once this becomes available. Lastly, caution must also be used when interpreting these findings, particularly when using the NDA to help monitor local progress and commissioning as there are subtle differences in NDA and QOF metadata. For example, there are differences in HbA1c parameters used, and differences in the timing and inclusion of patients in the NDA and QOF data sets; NDA covers all patients with diabetes over a 15-month period, while QOF covers patients aged 17 years and older over a 12-month period. While overall there is a significant relationship between the NDA and QOF outcome data at the practice level, these subtle differences should be considered when interpreting the results of these analyses.
Despite the cautious interpretation of the relationship between NDA participation and diabetes outcome using HbA1c as a marker, it is important to consider the generalizability of the study’s findings to participation in other mandatory and non-mandatory activities/clinical audits and the wider impact on patient outcome. It may also be important to consider whether patients should be made aware of, or encouraged to inquire, about practice participation in the audit to increase practice accountability given the impact that participation may have on diabetes care offered.
In conclusion, the results indicate that variation in NDA participation exists. Despite the fact that submission to the audit is now a contractual requirement, it may be possible that practices with certain identifiable characteristics may struggle to collect and submit sufficient data of an adequate quality to the audit. It is suggested therefore that some GPs may need additional support and resources to address these inequalities and to aid data collection and submission. In addition, although the markers of diabetes management used in the NDA are well established, their ability to discriminate between areas at a population level, in order for STPs and other government organizations to measure progress with care, is less certain. Given the changes to contractual obligations and the importance of effective diabetes management to population health, further evaluation is required.