Practice- and individual-level antibiotic prescribing associated with antibiotic treatment non-response in respiratory tract infections: a national retrospective observational study

Objectives: Antibiotic overuse results in adverse clinical outcomes. This study quantified the independent contributions of practice- and individual patient-Level antibiotic prescribing to antibiotic treatment non-response in respiratory tract infections (RTIs) in primary care. Methods: RTI episodes with antibiotic prescribed in 2018 were extracted from an Australian national general practice database. Practices were classified into tertiles by total antibiotic prescriptions per patient and ratios of broad- to narrow-spectrum antibiotic prescriptions. The association between practice- and individual patient-level antibiotic prescribing in the previous year and antibiotic treatment non-response (defined as prescription of a different antibiotic) <= 30 days after the initial RTI episode was quantified using generalized estimating equations. Results: Of 84597 RTI episodes with antibiotics prescribed in 558 practices, 5570 (6.6%) episodes of treatment non-response were identified. Patients with high individual-Level antibiotic prescribing (>= 4 prescriptions/year) had an increased risk of treatment non-response (versus no prescriptions/year: OR =1.64, 95% CI 1.52-1.77). At the practice Level, there was no significant association between total antibiotic prescriptions per patient and treatment non-response (high versus Low: OR 0.99, 95% CI 0.92-1.06). RTI episodes in practices with high broad- to narrow-spectrum antibiotic ratios had an increased risk of treatment non-response (versus Low-ratio practices: OR 1.14, 95% CI 1.05-1.23); this association was only observed among patients with <4 antibiotic prescriptions/year. Conclusions: The general practice-Level broad- to narrow-spectrum antibiotic ratio was a predictor of RTI antibiotic treatment non-response in patients with Lower individual-Level antibiotic use. The measure of practice-level antibiotic prescribing could potentially guide the improvement of antibiotic treatment.

Automated summary

This study found that higher individual-level antibiotic prescribing was associated with an increased risk of treatment non-response in respiratory tract infections in primary care, while there was no significant association between total antibiotic prescriptions per patient at the practice level. Practices with high broad- to narrow-spectrum antibiotic ratios had an increased risk of treatment non-response, particularly among patients with less than 4 antibiotic prescriptions per year. Practice-level antibiotic prescribing could potentially guide the improvement of antibiotic treatment.