Introduction Cluster randomised trials (CRTs) may have limited power because of the correlation of individual responses within clusters and the allocation of only small numbers of clusters. This ongoing study evaluates whether a CRT may be implemented utilising the electronic patient records of the general practice research database (GPRD) that includes large numbers of UK family practices.

Methods The trial aims to reduce antibiotic prescribing for respiratory illness in primary care, randomising practices between intervention of decision support or none. The primary outcome is the proportion of consultations for respiratory illness with antibiotics prescribed. In GPRD in 2006, the proportion was approximately 39% for all RTIs; 47 practices per group will be required to detect a 5% absolute difference in this measure.

Results Research governance approvals for all primary care organisations in England and Scotland were facilitated through the Central System for Permissions. All GPRD family practices in England and Scotland were invited to participate. In the first 4 months of the study, 71 family practices were consented. Random allocation is by minimisation controlling for region and practice size. Practices allocated to the active intervention have software installed remotely to provide decision support prompts for family doctors. The intervention will continue for 12 months. Outcomes will be analysed through analysis of data for diagnosis and prescriptions recorded into GPRD.

Conclusions Healthcare databases may be used to facilitate CRTs by providing data to inform trial design, a sampling frame for recruitment, as well as data for monitoring and outcome evaluation.