Changing the antibiotic prescribing of Australian general practice registrars' for acute respiratory tract infections: a non-randomized controlled trial.

Background: Inappropriate antibiotic prescription for self-limiting respiratory tract infections (RTIs) by general practitioner (GP) registrars (trainees) is less common than by established GPs but still exceeds evidence-based benchmarks. A 2014 face-to-face educational intervention for registrars and supervisors reduced registrars' acute bronchitis antibiotic prescription by 16% (absolute reduction). We aimed to establish the efficacy of an updated registrar/supervisor RTI-management intervention (delivered at distance) on antibiotic prescribing.

Methods: A non-randomized trial using a non-equivalent control-group nested within the ReCEnT cohort study. The intervention included online educational modules, registrar and supervisor webinars, and materials for registrar-supervisor in-practice educational sessions, and focussed on acute bronchitis as an exemplar RTI. The theoretical underpinning was the 'capability, opportunity, and motivation' (COM-B) framework. The intervention was delivered to registrars and supervisors of one large educational/training organization annually from mid-2021, with pre-intervention period from 2017, and with postintervention period ending 2023. Two other educational/training organizations served as controls. The primary outcome was antibiotics prescribed for acute bronchitis. Analyses used multivariable logistic regression with predictors of interest: time (before/after intervention), treatment group, and an interaction term for time-by-treatment group, adjusted for potential confounders. The interaction term P-value was used to infer statistical significance of the intervention effect.

Results: Of 4612 acute bronchitis presentations, 70% were prescribed antibiotics. There was a 6.9% absolute reduction (adjusted) of prescribing in the intervention-group compared with the control-group. This was not statistically significant (Pinteraction = .22).

Conclusions: Failure to find a significant effect on prescribing suggests difficulties with scalability of this (and similar educational) innovations.