Purpose: The need for drug-related safety warnings is undisputed, but their impact on prescribing behaviour is not always clear. Safety warnings usually do not contain therapeutic alternatives. Based on German outpatient routine healthcare data, our cohort study investigated the impact of three warnings for fluoroquinolones on prescribing behaviour.
Methods: Structural breaks were estimated in a time-series analysis (2005-2014) of 184,134 first antibiotic prescriptions for patients (≥ 18 years) diagnosed with community-acquired pneumonia (CAP), acute bacterial sinusitis (ABS), or acute exacerbation of chronic bronchitis (AECB). Subsequently, risk factors for patients' before/after safety warnings presented as risk ratios (RR) were estimated by Poisson regression.
Results: Following the 2008 warning for moxifloxacin, the RR of being prescribed moxifloxacin was reduced by 56% (95% CI 0.41-0.47; p < 0.001) for CAP, by 65% (95% CI 0.32-0.39; p < 0.001) for ABS, by 57% (95% CI 0.41-0.45; p < 0.001) for AECB. After the 2012 warning for levofloxacin, the RR of being prescribed levofloxacin was reduced by 31% (95% CI 0.64-0.74; p < 0.001) for CAP, by 14% (95% CI 0.77-0.96; p = 0.007) for ABS, by 27% (95% CI 0.69-0.77; p < 0.001) for AECB. We noticed a prescription-switch to other antibiotics which was not in line with the national guideline recommendations. The warning for moxifloxacin 2009 had no impact on prescribing behaviour.
Conclusion: This study observed an impact on prescribing behaviour in response to regulatory safety warnings for two out of three warnings. Information on therapeutic alternatives should be a part of any safety warning to encourage the intended changes in prescribing behaviour.
Keywords: Antibiotic use; Cohort studies; Fluoroquinolones; Patient safety; Pharmacovigilance; Practice patterns.