Interpractice variability in antibiotic prescribing for acute respiratory tract infections: a cross-sectional study of Australian early-career general practitioners.

Abstract

Frequency of general practitioners' (GPs') antibiotic prescribing for acute, self-limiting respiratory tract infections (aRTIs) is high. The practice environment and culture influence the clinical behaviour, including prescribing behaviour, of GP specialist vocational trainees (registrars). We aimed to assess inter-practice variability in registrars' antibiotic prescribing. This was a cross-sectional analysis from the Registrar Clinical Encounters in Training (ReCEnT) cohort study, from 2010 to 2020. ReCEnT documents registrars' clinical experiences and behaviours. Before 2016, 5 of 17 Australian training regions participated in ReCEnT. From 2016, three of nine regions (~40% of Australian registrars) participated. 3210 registrars (response rate 91.8%) from 1286 training practices contributed to the analysis. The outcomes were prescription of an antibiotic for new diagnoses of (1) all aRTIs and (2) acute bronchitis diagnoses specifically. Prescribing percentages were calculated at the training practice level. Intraclass correlation coefficients (ICCs) were used to measure the ratio of interpractice variation to total variance. Median ORs (MORs) were also estimated to quantify interpractice variability. Practice-level antibiotic prescribing percentages ranged from 0% to 100% for both aRTIs and acute bronchitis diagnoses in the primary analysis. ICCs for aRTI prescribing were 0.08 (unadjusted) and 0.02 (adjusted). For acute bronchitis, ICCs were 0.10 (unadjusted) and 0.05 (adjusted). MORs were 1.66 (unadjusted) and 1.32 (adjusted) for aRTIs. MORs for acute bronchitis were 1.80 (unadjusted) and 1.53 (adjusted). This indicates a marked variation in the odds of a patient receiving antibiotics for an aRTI if randomly attending different practices. There was considerable interpractice variation in registrars' antibiotic prescribing frequencies. Further research is required to examine the factors accounting for this variation and to develop practice-level interventions to reduce antibiotic prescribing in high-prescribing practices.