(260) Fluoroquinolones and the risk of aortic aneurysm or dissection: a cohort study in German health claims data

Background: Evidence for fluoroquinolones (FQ) increasing the risk of aortic aneurysm or dissection (AAD) was provided by several observational studies in recent years. Lately, Garg and colleagues reported an increased risk for FQ-associated AAD compared to macrolides in US health claims data.

Objectives: To replicate the study of Garg et al. 2023 in German health claims data previously used to assess FQ-associated adverse events, and to extend the results by contributing evidence for six further broad-spectrum antibiotic classes as active comparators (AC).

Methods: The German health claims data allowed replication of the design by Garg and colleagues with only minor modifications. The active comparator new user design comparing index episodes of FQ with macrolides was applied in a cohort of all adult patients from the German statutory health insurance AOK with ≥ 1 new prescription in the years 2014-2018. New users were defined as having a 365-day history before the index prescription without antibiotics dispensed or AAD diagnosed or hospitalisation, and were followed-up for 60 days. Nearest neighbour 1:1 propensity score matching for age, gender, drugs dispensed, Charlson comorbidity index, diabetes mellitus, and cardiovascular diseases was used. Adjusted hazard ratios for incident inpatient AAD were estimated by a Cox proportional hazard model. Sensitivity analyses on specific FQ agents, an extended risk window of 90 days, on dosage, and separately for aneurysm and dissection were conducted. All analyses were also conducted for FQ vs. tetracyclines, penicillins with extended spectrum, penicillins and beta-lactamase inhibitor combinations, 2nd and 3rd generation cephalosporins, sulphonamide and trimethoprim combinations, and lincosamides, respectively.

Results: FQ episodes were associated with an increased risk for AAD compared to macrolides (aHR = 1.56 [1.36;1.78]), which replicates the risk estimation by Garg et al. (aHR = 1.34 [1.17;1.54]). This association was robust in a 90-day risk window and for ciprofloxacin, levofloxacin and moxifloxacin. The latter comprised the greatest risk of AAD compared to macrolides (aHR=2.10 [1.61;2.73]).

Conclusions: The risk of FQ-associated AAD compared to macrolides can be replicated in German routine care data. However, the effect size may depend on the chosen AC. Further studies examining the extent of confounding by indication in antibiotic therapy are needed.