Accounting for Time‐Varying Confounding in a Self‐Controlled Case Series of Fluoroquinolone Treatment for Uncomplicated Urinary Tract Infections and Risk of Collagen‐Related Events

ABSTRACT

We report findings from three SCCS conducted to quantify the association between fluoroquinolone (FQ) use for the treatment of uncomplicated urinary tract infection (uUTI) and tendon rupture, retinal detachment, and uveitis. Female patients aged ≥ 12 years old in the Optum Clinformatics Data Mart database with an outcome of interest and exposure to either oral FQ or trimethoprim/sulfamethoxazole (SXT) in the 5 days following a newly reported uUTI during the study period (01/01/2011–02/10/2019) were included. We considered a 90‐day risk window for each outcome following drug exposure. Incidence rate ratios (IRR) and 95% confidence intervals (CI) were estimated using conditional Poisson regression, adjusting for age and calendar time, incorporating SXT as an active comparator. We found little evidence of an association between FQ use compared to SXT and any of the outcomes of interest (tendon rupture: IRR = 1.01, 95% CI = 0.91–1.21; retinal detachment: IRR = 0.99, 95% CI = 0.86–1.14; and uveitis: IRR = 1.09, 95% CI = 0.97–1.22). Incorporating the active comparator using two different methods did not change conclusions. The lack of evidence for an association between short‐term use of FQ and the comparator antibiotic (SXT) for the treatment of uUTI and collagen‐related events also implies that there was no marked association between uUTI and these outcomes. However, power was limited, and a small increased risk cannot be ruled out. Using active comparators in SCCS can improve the robustness of studies of antibiotics and adverse events.