Can primary care research be conducted more efficiently using routinely reported practice-level data: a cluster randomised controlled trial conducted in England?
Blair PS., Ingram J., Clement C., Young G., Seume P., Taylor J., Cabral C., Lucas PJ., Beech E., Horwood J., Dixon P., Gulliford MC., Francis N., Creavin ST., Lane A., Bevan S., Hay AD.
Objectives Conducting randomised controlled trials (RCTs) in primary care is challenging; recruiting patients during time-limited or remote consultations can increase selection bias and physical access to patients' notes is costly and time-consuming. We investigated barriers and facilitators to running a more efficient design. Design An RCT aiming to reduce antibiotic prescribing among children presenting with acute cough and a respiratory tract infection (RTI) with a clinician-focused intervention, embedded at the practice level. By using aggregate level, routinely collected data for the coprimary outcomes, we removed the need to recruit individual participants. Setting Primary care. Participants Baseline data from general practitioner practices and interviews with individuals from Clinical Research Networks (CRNs) in England who helped recruit practices and Clinical Commission Groups (CCGs) who collected outcome data. Intervention The intervention included: (1) explicit elicitation of parental concerns, (2) a prognostic algorithm to identify children at low risk of hospitalisation and (3) provision of a printout for carers including safety-netting advice. Coprimary outcomes For 0-9 years old - (1) Dispensing data for amoxicillin and macrolide antibiotics and (2) hospital admission rate for RTI. Results We recruited 294 of the intended 310 practices (95%) representing 336 496 registered 0-9 years old (5% of all 0-9 years old children). Included practices were slightly larger, had slightly lower baseline prescribing rates and were located in more deprived areas reflecting the national distribution. Engagement with CCGs and their understanding of their role in this research was variable. Engagement with CRNs and installation of the intervention was straight-forward although the impact of updates to practice IT systems and lack of familiarity required extended support in some practices. Data on the coprimary outcomes were almost 100%. Conclusions The infrastructure for trials at the practice level using routinely collected data for primary outcomes is viable in England and should be promoted for primary care research where appropriate.