Summary of main findings
The practices exposed to our team-based CQI intervention within the context of the present study initiated and completed significantly more quality improvement projects than the practices in the control group. The projects ran in a more structured way and in significantly more projects the self-set goals were met. As practices were free to select their own topics for improvement and set their own objectives, the fact that the intervention group met a significantly greater number of self-defined improvement objectives than the control group is an important finding.
In terms of the VIP dimensions the intervention practices were also more successful, but the differences between intervention and control practices were small and non-significant.
Strengths and limitations of the study
The results of this carefully randomised study show a modest but undeniable contribution of the team-based CQI intervention to a culture of improvement. We performed the first randomised controlled trial on continuous quality improvement in general practice. The VIP instrument, which was used to provide the practices with an extensive overview of their weak and strong practice management aspects as well as an outcome measurement, has shown to be able to discriminate between practices22 and to show trends in time.23–25
Some possible limitations on the present study are as follows. To start with, both the intervention and control groups in our study consisted of practices which volunteered to participate and therefore practices that were probably very motivated to change. A motivated practice appears to be a prerequisite for successful improvement and the effective implementation of improvement efforts. Use of the team-based CQI intervention examined here with other, less motivated, practices may therefore require more extensive support and perhaps financial or other incentives.
Second, although we performed a careful randomisation of the practices that were interested with the help of a computer program and an independent researcher, the control practices were doing better regarding CQI at the beginning of the study than the intervention practices. We do not know whether a part of the improvement in the intervention practices is due to ‘regression to the mean’.
Third, as a consequence of the fact that practices could freely select the topics for organisational change, the number of practices for a particular topic was in many cases quite small (that is, the number of practices for a particular change topic was only a fraction of the total number of practices) and thus too small to produce statistically significant change at the level of the group. Besides, although the practices made use of the VIP feedback and indicators to select topics for improvement, the improvement objectives formulated by the practice teams did not have to match the VIP indicators, which may have obscured many intervention effects. As the practices were free to define their own improvement objectives, we could not know — a priori — just which aspects of practice management might improve. As a result of this situation, different aspects of the intervention practices may have improved than measured by the VIP and a significant effect in terms of the VIP therefore not detected. For example, the VIP does not measure dimensions of practice management related to staff workload, and reducing staff workload was often the main objective when practices selected practice accessibility and availability as starting point and improvement objective.26
In fact, access by phone was in some practices limited as a consequence of more structured working day for the practice nurses. As a consequence of this free topic choice and therefore a multifactorial design, a power analysis was complicated as it was very difficult to predict which topics would be chosen. Despite this uncertainty and the lack of figures from similar studies, we did perform a power calculation before we began our study. To find a difference of 25% in outcome between intervention (75%) and control group (50%), with a 5% two-sided significance level and a power of 80%, we needed 55 practices per group. Although 25% of the practices that were invited to take part agreed, which is much more than in similar studies,16,17 we were not able to include more than 49 practices in total. The most important barrier for practices to take part in our study was the randomisation: very many practices were enthusiastic about the intervention, but they did not accept the chance to be allocated to the control group and therefore decided not to take part in the study.
Finally, the present study was undertaken in a period when practice nurses were first being introduced into Dutch primary care. All of the practices in our study were preparing for the introduction of such a new professional or had recently appointed a practice nurse. The introduction of such a new function certainly cost both the intervention and control practices considerable time and energy. Alternatively, the introduction of such a new function may have motivated both the intervention and the control practices to examine their management and adjust this as needed. Relative to the control practices, thus, the intervention practices may not have changed enough and thus as a result of our intervention, to produce statistically significant differences in the management of primary care services. Yet, general practice is always in a state of change so any intervention has to show its worth in that context.
Implications for clinical practice and future research
The team-based CQI model will be offered to practices in the Netherlands as one of the support tools in the accreditation procedure. During the first accreditation assessment, practices need to define improvement goals, which will be evaluated during a reassessment after 3 years. As this provides a new incentive to improve, the effect of this should be evaluated. Besides, as many practices in our project chose improvement topics related to staff workload, we recommend that in future evaluations workload of staff and the team climate inventory (TCI) will be included.27