OBJECTIVES: Recently, the FDA required that a warning be added to prescribing guidelines for fluoroquinolones. This decision was made, in part, due to RWE showing an elevated AA risk. The current analysis explored the association between fluoroquinolones and AA, compared to penicillin. A probabilistic sensitivity analysis (PSA) explored exposure misclassification.

METHODS: Patients were identified using an EMR network of over 26M US patients. The index event (IE) was the first fluoroquinolone or penicillin recorded. Patients were aged 60+, had no history of AA, no fluoroquinolone or penicillin use and no hospitalization in the 4-months before the IE, and no cardiovascular surgeries in the year prior to the IE. A propensity score analysis (1:1 greedy-nearest-neighbor matching) examined the occurrence of AA in the 60 days after the IE. Differences in treatment definition (class vs ingredient (ciprofloxacin, amoxicillin) and a PSA examining non-compliance with medications were explored.

RESULTS: Over 30,000 matched, treated patients were identified for each analysis, with fewer than 50 patients with AA in any cohort. No statistically significant results were identified. Comparing ciprofloxacin to amoxicillin alone found a RR of 1.2 (0.8-1.9). Comparing ciprofloxacin to penicillin found a RR of 1.1 (0.7-1.7). Fluoroquinolones compared to penicillin had a RR of 0.8 (0.5-1.2). The PSA assumed a small decrease in specificity (mode=0.975) and a moderate decrease of sensitivity (mode=0.85) and had a minimal impact on results.

CONCLUSIONS: Although not statistically significant, effect estimates comparing ciprofloxacin, but not fluoroquinolones as a class, to penicillin align with FDA-cited findings. Given the severity of AA, and the width of CIs in this analysis, results do not provide strong evidence to contradict the FDA’s decision. Whether all fluoroquinolones contribute equally to the risk of AA, should be further explored. To our knowledge, this is the first use of quantitative bias analysis examine fluoroquinolone use.