Background: In a chronic disease setting such as cystic fibrosis (CF), antibiotics are often prescribed for emergent symptoms and it is unclear whether this affects endpoints in a clinical trial. Pulmonary exacerbations (PEs) are defined episodes of acute worsening and a key clinical efficacy measure in CF. Our hypothesis was that acute antibiotics given for illnesses not meeting the PE definition may alter estimates of treatment effect that do not account for this antibiotic use.
Methods: A randomized, placebo-controlled trial of azithromycin (AZ) including 260 participants with CF was utilized for this study. PEs were defined using a priori criteria. Physician initiated antibiotic therapy (PIT) not meeting the PE endpoint was characterized and its impact on treatment effect assessed.
Results: 40% (104/260) of participants were prescribed 188 courses of PIT in the absence of a PE; 19% (25/129) of placebo and 10% (13/131) of AZ participants received ≥2 courses of PIT and never fulfilled the PE definition (9% difference, 95% confidence interval: 1%, 18%, p = 0.04). Accounting for PIT through use of a composite endpoint including time to PE or need for repeated PIT altered treatment effect estimates (a 56% reduction in the event rate comparing AZ to placebo [p < 0.0001] as compared to a 50% reduction not accounting for PIT [p = 0.003]).
Conclusion: PIT is common in CF and may impact treatment effect estimates. Optimization of the PE endpoint to include meaningful events necessitating treatment may improve our ability to conduct efficient trials by reducing the sample size 30-50%, ultimately enabling rapid evaluation of new therapies.
Keywords: AZ; Antibiotic therapy; CF; CI; Cystic fibrosis; HR; IV; PE; PRO; Pulmonary exacerbation definitions; Sample size; Study design; azithromycin; confidence interval; cystic fibrosis; hazard ratio; intravenous; p; p-value; patient reported outcomes; pulmonary exacerbation.
Copyright © 2013 Elsevier Inc. All rights reserved.