Previous Drug Exposure in Patients Hospitalised for Acute Liver Injury: A Case-Population Study in the French National Healthcare Data System.

Introduction: Acute liver injury (ALI) is a major reason for stopping drug development or removing drugs from the market. Hospitalisation for ALI is relatively rare for marketed drugs, justifying studies in large-scale databases such as the nationwide Système National des Données de Santé (SNDS), which covers 99% of the French population.

Methods: SNDS was queried over 2010-2014 for all hospital admissions for acute toxic liver injuries not associated with a possible other cause, using a case-population approach. Exposures of interest were drugs dispensed from 7 to 60 days before date of admission. Individual drugs were analysed by their frequency (if five or more cases) and by the ratio of exposed cases to the number of exposed subjects and to exposed patient-time in the general population over the same timeframe.

Results: Over 5 years, 4807 cases of ALI were identified, mean age 54.5, 59% women, 76% exposed to at least one of 249 different drugs. Drugs most commonly identified were non-overdose paracetamol (31% of cases), esomeprazole or omeprazole (18%), phloroglucinol, domperidone, co-amoxiclav, furosemide, and atorvastatin (more than 250 cases each). When compared to population exposures, the highest per-person risks were observed with antimycobacterial antibiotics, with one case for 1000 or fewer users, followed by colestyramine and erythromycin (around 1/5300), antiepileptic drugs, anticoagulants, and anti-Alzheimer drugs (1/6000-1/10,000 users). When a person-time approach was considered, the drugs with the highest per-tablet risk were still the antituberculosis drugs, followed by a number of other antibiotics.

Conclusions: This nationwide study describes drugs associated with ALI, according to absolute population burden and per-patient and per-tablet risk. Some of these associations may be spurious, others causal, and others yet were unexpected. Systematic analysis of drug classes will look for outliers within each class that could raise signals of unexpected hepatic toxicity.