Implementing quality indicators for diabetes and hypertension in family medicine in Slovenia

Out of 584 family medicine practices that were included in this project at the end of 2015, 568 (97.3%) had complete data and could be included in this analysis (Table 2). The mean duration of participation in the project was 28.7±18.7 months, with a median of 33.0 months. The mean age of the family physicians was 51.4±9.0 years, median 53.0.

Table 2

Characteristics of family medicine practices included in the project.

The mean value of this quality indicator was 84.2±53.6, with a median of 84.0 (Table 1).

The results of a multivariate analysis showed that the variables ‘IT provider No. 1 and 4’ and ‘Duration of participation in the project’ were significantly correlated with a higher value of this quality indicator, and ‘Working in the Maribor region’ was significantly correlated with a lower value of this quality indicator (Table 3).

Table 3

Characteristics of family medicine practices included in the project.

For continuous variables regression coefficients indicate the change of the observed variable with each increasing unit of explanatory variable.

For categorical variables regression coefficients indicate the changes of the observed variable in comparison to a reference category.

The mean value of this quality indicator was 25.6±22.3, with a median of 21.2.

The results of a multivariate analysis showed that the variable ‘Duration of participation in the project’ was significantly correlated with a higher value of this quality indicator, and ‘IT provider No. 3 and 4’ were significantly correlated with a lower value of this quality indicator (Table 4).

Table 4

Multivariate analysis for higher value of the quality indicator ‘Measured HbA1C in patients with diabetes at least once a year’.

For continuous variables regression coefficients indicate the change of the observed variable with each increasing unit of explanatory variable.

For categorical variables regression coefficients indicate the changes of the observed variable in comparison to a reference category.

The mean value of this quality indicator was 18.2±14.3, with a median of 16.3 (Table 1).

The results of a multivariate analysis showed that the variables ‘Duration of participation in the project,’ ‘Female gender,’ ‘Self-contractor,’ ‘Working in the Maribor region’ and ‘Having IT No. 2’ were significantly correlated with a higher value of this quality indicator. Variables ‘IT provider No. 3’ and ‘IT provider No. 4’ were significantly correlated with lower values of this quality indicator (Table 5).

Table 5

Multivariate analysis for higher value of the quality indicator ‘Percentage of patients with hypertension with a systolic blood pressure 140/90 mmHg or lower.’

For continuous variables regression coefficients indicate the change of the observed variable with each increasing unit of explanatory variable.

For categorical variables regression coefficients indicate the changes of the observed variable in comparison to a reference category.

High values of structure quality indicators are an important sign that the conditions for measuring quality have been established. On the other hand, lower levels of process quality indicators were found and could be attributed to several reasons. The lack of continuous quality control with feedback could be one of them; the project continued without continuous data analysis, feedback to practices on the quality of their work, or benchmarking (22), and no staff were employed to carry out these tasks.

It is also possible that adherence to the guidelines was low. A recent study from Slovenia showed that the introduction of this different model of chronic patient management in family medicine improved the process of quality of care, but the desired level of quality has not yet been achieved (23), which probably points to low adherence to the guidelines. Other studies have also shown that primary care physicians’ adherence to hypertension and diabetes guidelines is low (24-27). For example, adherence to hypertension guidelines was found to be between 10 and 50% (24, 25). On the other hand, another study showed that HbA1C was measured in almost all the patients (26).

It is also possible that the selection and development of the quality indicators themselves was not optimal. The development of quality indicators must be based on a systematic evidence-based approach, and expert consensus and guidelines should be considered. The indicators should be acceptable, feasible, reliable, sensitive to change, and valid (16). The quality indicators in our project have not yet been evaluated according to these features, and therefore it could be possible that the low levels of quality indicators are associated with their suboptimal nature.

The most important variable that was shown to be associated with higher levels of quality indicators was the duration of participation in the project. This indicates that the practical introduction of quality indicators in the everyday work of practices might be associated with some problems that could influence the quality. These problems are yet to be recognised. The region of practices was also recognised as important. Some previous data from Slovenia indicate that there are differences between regions in terms of quality (10). Which factors contribute to that is unknown and a subject for further studies. In addition, IT providers proved important in our research. Ensuring quality of data during electronic data capture is always a problem, and several ways of reducing errors must be applied (28, 29). There are no reports about the quality of data gathering in this project, and it therefore seems that there are some problems which still need to be recognised and addressed.

The quality indicator ‘List of patients with diabetes’ indicates the prevalence of diabetes in registered patients of family medicine practices in Slovenia. It should be mentioned here that its value could also be affected by the actual prevalence of diabetes in the region of Slovenia (30), and does not only depend on the quality of work.

Other factors can also contribute to the quality management of chronic patients and were not included in this study. These might include systemic factors, reimbursement, service organisation and capacity, cultural factors, disease epidemiology, practice systems in terms of incentives, practice information capacity, access, the use of teams in quality policy, detection of quality and safety problems, staff and patient safety, inclusion of patients’ perspectives, and the length of consultations (31-34). These factors should be considered when studying this topic further.

We chose only a few quality indicators because we focused on the management of chronic diseases and not on their prevention. This was done to ensure the clarity and focus of the study. It could, however, be possible that by excluding some quality indicators we overlooked some other differences.

The other problem is the quality of the data, as the electronic database was only established in 2015; our analysis was carried out at the end of 2015, and we can, therefore, anticipate that some technical problems had not yet been resolved. Therefore, the reliability of the data might not be as good as if we had done the research later.

Our study did not include other factors that could influence the quality of management of chronic diseases, especially the characteristics of other team members, particularly nurse practitioners. Since nurse practitioners are very involved in the management of chronic patients, the inclusion of their characteristics could have helped us to build a more comprehensive view of this matter, and produced more reliable statistical models. Other limitations are the exclusion of patient data, and the cross-sectional nature of this study, which prevents us from detecting causality. This would further increase the comprehensibility of the results.

In our study, we did not analyse the characteristics of those family medicine practices that did not participate in the project. Therefore, we do not know if they differ significantly from those included in the study.

The study showed that the quality assessment was a challenge, because the values of the quality indicators were set empirically while preparing the new model of family medicine practices in Slovenia. The quality indicators should be re-evaluated and changed if necessary according to the established methodology (16). It was also recognised that expert supervision at the location is necessary to assess the process of work of the family medicine team at different levels (e.g. following protocols, regularly measuring the parameters of chronic diseases, recording them in the electronic database, and their reporting and analysis), and to discuss the obstacles directly with care providers. Continuous quality control and benchmarking should be established in order to improve the quality of chronic patient management in Slovenian family medicine practices. The study also indicated the need for collaboration between different professionals (e. g. IT specialists and health care providers) to adopt the IT system to fully support patient management. The quality indicators should be a basis for financing the practices according to quality standards. It may also be important to inform each team about their quality results every month, and ask for their feedback.

The regional differences in quality which emerged from our analysis should be further explored. Other possible factors that could contribute to quality should be studied. The quality indicators should be reviewed each year, which would allow a comparison of indicators between two points in time in the same practice, or between practices using the new approach and those using the classical approach, and to observe the trend of change according to the existing circumstances.