Practice-level variation in potentially inappropriate prescribing in middle-aged adults: A repeated cross-sectional...

To quantify between-practice variation in potentially inappropriate prescribing (PIP) among middle-aged adults (45-64 years) in UK general practice, and to identify patient- and practice-level characteristics associated with PIP.

Repeated cross-sectional study using linked data from the Lambeth DataNet and Quality Outcomes Framework across 41 GP practices in South London (2014-2019). The study included patients aged 45-64 years prescribed ≥1 medicines in each year from 2014 to 2019 (n = 46 633-52 582). PIP was identified using the 22 PROMPT (PRescribing Optimally in Middle-aged People's Treatments) criteria. Multilevel negative binomial regression estimated associations with patient- and practice-level characteristics, expressed as rate ratios (RR) with 95% confidence intervals.

PIP prevalence was 18%-20%, stable across the study period. Practice-level prevalence ranged from 14% to 24%. The SCV was 1.72 (low) and the ICC was 0.08, with approximately 8% of variance attributable to practice-level differences. Polypharmacy was the strongest predictor (adjusted RR 4.04, 95% CI 3.96-4.11), followed by multimorbidity (RR 2.44, 95% CI 2.40-2.50). Black/African/Caribbean/Black British ethnicity was associated with lower PIP (RR 0.77, 95% CI 0.75-0.78). Practice-based pharmacist presence and teaching practice status were associated with modestly lower PIP, while APMS contract type was associated with higher PIP (RR 1.20, 95% CI 1.10-1.32).

Most variation occurred at the patient level, but practice organizational characteristics were also associated with PIP and warrant further investigation in an age group for which direct medicines-optimization evidence remains limited.