Estimating daily antibiotic harms: an umbrella review with individual study meta-analysis

BACKGROUND: There is growing evidence supporting the efficacy of shorter courses of antibiotic therapy for common infections. However, the risks of prolonged antibiotic duration are underappreciated. OBJECTIVES: To estimate the incremental daily risk of antibiotic-associated harms. METHODS: We searched three major databases to retrieve systematic reviews from 2000 to 30 July 2020 in any language. ELIGIBILITY: Systematic reviews were required to evaluate shorter versus longer antibiotic therapy with fixed durations between 3 and 14 days. Randomized controlled trials included for meta-analysis were identified from the systematic reviews. PARTICIPANTS: Adult and paediatric patients from any setting. INTERVENTIONS: Primary outcomes were the proportion of patients experiencing adverse drug events, superinfections and antimicrobial resistance. RISK OF BIAS ASSESSMENT: Each randomized controlled trial was evaluated for quality by extracting the assessment reported by each systematic review. DATA SYNTHESIS: The daily odds ratio (OR) of antibiotic harm was estimated and pooled using random effects meta-analysis. RESULTS: Thirty-five systematic reviews encompassing 71 eligible randomized controlled trials were included. Studies most commonly evaluated duration of therapy for respiratory tract (n = 36, 51%) and urinary tract (n = 29, 41%) infections. Overall, 23 174 patients were evaluated for antibiotic-associated harms. Adverse events (n = 20 345), superinfections (n = 5776) and antimicrobial resistance (n = 2330) were identified in 19.9% (n = 4039), 4.8% (n = 280) and 10.6% (n = 246) of patients, respectively. Each day of antibiotic therapy was associated with 4% increased odds of experiencing an adverse event (OR 1.04, 95% CI 1.02-1.07). Daily odds of severe adverse effects also increased (OR 1.09, 95% CI 1.00-1.19). The daily incremental odds of superinfection and antimicrobial resistance were OR 0.98 (0.92-1.06) and OR 1.03 (0.98-1.07), respectively. CONCLUSION: Each additional day of antibiotic therapy is associated with measurable antibiotic harm, particularly adverse events. These data may provide additional context for clinicians when weighing benefits versus risks of prolonged antibiotic therapy.