Data-driven Quality Improvement in Primary Care - Trial

The trial described here is part of a programme which aimed to design a complex, primary care prescribing safety improvement intervention and test its effectiveness in a randomised controlled trial.

Non-steroidal anti-inflammatory drugs (NSAIDs) and antiplatelet drugs such as low dose aspirin and clopidogrel are responsible for a significant proportion of hospital admissions due to preventable adverse drug events (ADE), and are the drugs most commonly associated with fatal ADEs. Previous research has identified groups of patients and patterns of co-prescription in which use of these drugs is particularly high-risk , and national prescribing and safety guidance has embedded this research in clear recommendations to either avoid prescribing or to do so only when there is no alternative, and with caution. In previous epidemiological work, we have shown that high-risk use of NSAIDs, aspirin and clopidogrel is common, and pilot work in four practices has shown that focused review of prescribing by the practice reduced the targeted high-risk NSAID prescribing by approximately 40% after one round of feedback. This effect size is consistent with the PINCER trial where the intervention was a pharmacist facilitated review process.

We hypothesise that a multi-faceted intervention comprising of (1) educational outreach, (2) use of an informatics tool to monitor prescribing patterns at practice level and to prompt and facilitate the review of individual patients at risk of ADEs and (3) a small financial incentive to review patients will reduce rates of high-risk prescribing.

The specific research questions addressed by the trial are:

• Does the intervention reduce the specified primary outcome of a composite measure of high risk non-steroidal anti-inflammatory drug, aspirin and clopidogrel prescribing?
• Does the intervention reduce the specified secondary outcomes of: the nine individual measures constituting the composite; related admissions to hospital; repeat vs new prescribing?
• If found to be effective, then is the intervention cost-effective?

The trial will use a stepped-wedge design, which is particularly suited to a sequential roll-out of an intensive and informatics based intervention focusing on patient safety. In this design, all participating practices receive the intervention, but are randomised to a starting time. At the point of entering the intervention phase of the trial, all practices will receive an educational outreach visit which will include training in the use of the informatics tool.

The informatics tool will provide regular feedback of any change in rates of high-risk prescribing for each individual measure and the composite measure, with the ability to drill-down to individual patient level and review a summary of each patient's relevant conditions and prescribing.

Inclusion criteria

• General medical practices in NHS Fife and NHS Tayside with a compatible clinical IT system and agreeing to participate.
• Practices that agree to have relevant medication related data to be automatically extracted from their electronic clinical information systems from 1/10/10 to 30/9/13 (ie 12 months before first practice starts till 12 months after last practice starts).

Exclusion criteria

• Practices that use General Practice Administration System for Scotland (GPASS) or Egton Medicine Information System (EMIS) on the date of randomisation, since data extraction for the informatics requires Vision practice IT system.