Risk of aortic aneurysm or dissection following exposure to systemic fluoroquinolone antibiotics – a large-scale distributed network analysis

Background: Medicine regulators have issued safety warnings regarding the risk of aortic aneurysm or dissection (AA/AD) with fluoroquinolones (FQ) antibiotics. However, risk remains uncertain as prior studies that contributed to these safety warnings were limited due to inconsistencies in study design, approaches to identifying exposures and outcomes and methods to address confounding.

Objectives: To estimate the risk of AA/AD following FQ exposure compared to other standard treatments for urinary tract infection.

Methods: A distributed network analysis was conducted across 14 databases standardised to the OMOP Common Data Model. Patients were included if they were aged 35 years or older and initiated an antibiotic for treatment of urinary tract infection (UTI) between 2010 and 2019. Patients initiating systemic fluoroquinolones (FQ) were compared to trimethoprim (TMP) and cephalosporin (CPH). The primary outcome was hospitalisation for aortic aneurysm or dissection (AA/AD). For the cohort study, a large-scale propensity score (PS) model using covariates measured at baseline was generated using regularised regression. Target and comparator cohorts were matched using a 1:1 ratio. Covariate balance was assessed using standardised difference of means between the matched cohorts, clinical equipoise was assessed by determining overlap in preference score distributions, and systematic error was assessed by evaluating the expected absolute systematic error across 50 negative controls. Cox proportional hazards models were used to model outcome risk for each time at risk window (30, 60, 90 or 365 days). Hazard ratios were pooled across databases using Bayesian meta-analysis among those databases that passed study diagnostics. A self-controlled case-series design was also used and included only those who initiated FQ and experienced an AA/AD during the study period.

Results: Seven databases contributed data to the meta-analysis of the association between FQ and AA/AD compared with TMP; pooled calibrated hazard ratio (HR) for the 60-day risk window was 0.91 (95% Confidence Interval (CI) 0.73-1.10). Nine databases contributed to meta-analysis of association between FQ and AA/AD compared with CPH; pooled calibrated HR for the 60-day risk window was 1.01 (95% CI 0.82-1.25). Despite considerable variation in patient characteristics and treatment patterns for comparator antibiotics across different jurisdictions, there was limited heterogeneity in treatment effect estimates.

Conclusions: This is the largest distributed network study conducted assessing the risk of AA/AD following treatment with FQ for UTI. We found no association between use of FQ and risk of AA/AD compared to other broad spectrum antibiotics in patients with UTI.