The risk of drug-induced liver injury associated with flucloxacillin - a nationwide, entropy balanced cohort study

Objectives: In this nationwide cohort study in a Scandinavian setting, we aimed to investigate the magnitude of association between flucloxacillin use and drug-induced liver injury (DILI).

Methods: Nationwide cohort study among adults in Sweden, 2006-2018. Register data on filled prescriptions, patient characteristics, co-medications, and DILI were linked. All filled prescriptions for flucloxacillin and oral clindamycin, among Swedish adults aged 18-85 years were identified. Entropy balancing methods were used to control for confounding. Cox regression was used to estimate hazard ratios for a first diagnosis of DILI, defined as admission to hospital, emergency department or specialist care, or death due to DILI, within 45 days from start of treatment.

Results: Within the main 45 day risk period, there were 219 events of DILI among 1,443,622 flucloxacillin users (incidence rate [IR] 14/10,000 person-years) as compared to 9 events among 583,847 oral clindamycin group (IR 1.4/10,000 person-years). This corresponded to a hazard ratio (HR) of 7.32 (95% confidence interval [CI] 4.1 - 13.0). The absolute risk difference for users of flucloxacillin compared to clindamycin in the main period, was 11 cases of DILI (95% CI 5 - 20) per 100,000 courses. The risk diminished in the subsequent periods, 46-90 days (HR 4.17; 95% CI 1.44 - 12.10), 91-180 days (HR 0.72; 95% CI 0.36 - 1.44).

Conclusions: In this nationwide cohort study, the use of flucloxacillin was associated with a seven-fold increased risk of DILI, predominantly in the first 45 days of exposure.