Effectiveness of interactive dashboards to optimise prescribing in primary care: A systematic review

Abstract

Background Rising levels of both high-risk and low-value prescribing have the potential for adverse effects on patients, healthcare systems and society. It is thus necessary to develop effective and cost-effective interventions to support safe, effective and cost-effective prescribing. Advancements in technology, including machine learning coupled with the vast amounts of routine prescribing data available in primary care have supported the development of novel approaches to provide prescribers with ongoing and comparative prescribing data feedback. This systematic review aimed to explore the characteristics of interactive dashboard interventions in primary care that provide visual and longitudinal feedback on prescription data and to explore the effect of these interventions on prescribing-related outcome measures. Methods and Findings This systematic review was registered prospectively and reported in line with PRISMA guidelines. Multiple databases and grey literature were searched in November 2023 to identify interventional studies, including quasi-experimental designs that explored the effect of interactive dashboards on prescribing-related outcomes in primary care. Identified records were assessed for inclusion and data extraction and risk of bias assessment were completed by two independent researchers. Interventions characteristics and effects were described narratively. A meta‐analysis using a random‐effects model was performed where at least two studies were comparable in terms of participants, study design and outcomes. Twelve studies, reported across eleven different papers were included, eight randomised controlled trials, one controlled before and after study and three interrupted time series analyses. Nine papers were assessed to be of low risk of bias. Six studies reported a significant effect on prescribing-related outcomes, with an effect seen more often for studies focusing on potentially inappropriate or high-risk prescribing (four out of six studies). Two of the six studies that focused on antibiotic prescribing demonstrated a significant effect. A meta-analysis of three RCTs involving 406 general practices and 337,963 patients demonstrated the overall odds of having at least one potentially inappropriate prescription was 0.87 (95% CI 0.81 to 0.93 I2 =0.0%) in the intervention compared to control group. Conclusion Interactive dashboards have the potential to support safe and effective prescribing in primary care. To support their implementation, it is essential to establish the necessary data infrastructure within primary cares systems. This encompasses electronic health records (EHR) systems, data integration tools, analytics platforms, and compliance with data privacy regulations, all working together to facilitate the efficient use of data for improving prescribing and ultimately patient care.