Potential risk signals for drug-related osteomyelitis: a comprehensive disproportionality analysis based on the FDA Adverse Event Reporting System (FAERS) database

Although current studies suggest that several drugs are associated with osteomyelitis, the risk of osteomyelitis associated with most medications remains unclear. This study investigated the potential risk signals for drug-induced osteomyelitis by analyzing a large dataset from FAERS. A comprehensive analysis of reports of drug-induced osteomyelitis from the first quarter of 2004 to the second quarter of 2025 was conducted using the reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network (BCPNN), and multi-item gamma Poisson shrinker (MGPS). Univariate and multivariate regression analyses were performed to identify associated potential risk signals, with Bonferroni correction applied for multiple comparisons. A total of 827 drugs were identified, corresponding to 18,819 reported cases. Forty-one drugs were consistently flagged by all four algorithms. Multivariate analysis revealed that age, gender, and 20 drugs-including canagliflozin, alendronic acid, denosumab, and methotrexate-were significant potential risk signals for drug-related osteomyelitis. The predictive accuracy of the model, as indicated by the ROC-AUC, was 0.757. The median time to onset of drug-induced osteomyelitis was 350 days (IQR 87-800.75 days), with a mean onset time of 624 days. Among the drugs, alendronic acid and pamidronic acid were associated with a relatively longer time to onset, whereas canagliflozin was linked to a shorter time to onset. These findings may assist clinicians in the early identification of drug-induced osteomyelitis and could provide insights for future research into the pathogenesis of drug-related osteomyelitis.