The risk of acute liver injury among users of antibiotic medications: a comparison of case‐only studies.

PURPOSE: The aims of this study were two‐fold: (i) to investigate the effect of exposure to antibiotic agents on the risk of acute liver injury using a self‐controlled case series and case‐crossover study and (ii) to compare the results between the case‐only studies. METHODS: For the self‐controlled case series study relative incidence ratios (IRR) were calculated by dividing the rate of acute liver injury experienced during patients' periods of exposure to antibiotics to patients' rate of events during non‐exposed time using conditional Poisson regression. For the case‐crossover analysis we calculated Odds Ratios (OR) using conditional logistic regression by comparing exposure during 14‐ and 30‐day risk windows with exposure during control moments. RESULTS: Using the self‐controlled case series approach, the IRR was highest during the first 7 days after receipt of a prescription (10.01, 95% CI 6.59–15.18). Omitting post‐exposure washout periods lowered the IRR to 7.2. The highest estimate in the case‐crossover analysis was found when two 30‐day control periods 1 year prior to the 30‐day ALI risk period were retained in the analysis: OR = 6.5 (95% CI, 3.95–10.71). The lowest estimate was found when exposure in the 14‐day risk period was compared to exposure in four consecutive 14‐day control periods immediately prior to the risk period (OR = 3.05, 95% CI, 2.06–4.53). CONCLUSION: An increased relative risk of acute liver injury was consistently observed using both self‐controlled case series and case‐crossover designs. Case‐only designs can be used as a viable alternative study design to study the risk of acute liver injury, albeit with some limitations. © 2015 The Authors Pharmacoepidemiology and Drug Safety Published by John Wiley & Sons Ltd.