医薬品誘発性甲状腺機能障害リスクのFDA有害事象報告システム（FAERS）を用いた評価

To quantify drug-specific associations with thyroid dysfunction (TD) using data from FDA Adverse Event Reporting System (FAERS) and to identify agents with significant disproportionality signals.

Using data from FAERS (Q1 2004 to Q1 2025), disproportionality analysis (Reporting Odds Ratio, ROR; Bayesian Confidence Propagation Neural Network, BCPNN) was employed to assess signals of drug-induced TD. Univariate analysis, Least Absolute Shrinkage and Selection Operator (LASSO) regression, and multivariate logistic regression were applied to identify independent risk factors. Time-to-onset (TTO) analysis was used to evaluate the timing of TD following drug initiation.

A total of 44,317 reports associated with TD were identified, with a higher proportion involving female patients (60.3%). The median patient age was 59 years (IQR: 45-69 years). Signal detection identified 172 suspect drugs significantly associated with TD. Following disproportionality screening, drug candidates with a ROR whose 95% confidence-interval lower bound exceeded 1, cell count a > 100, and P-adjust < 0.01 were advanced to LASSO regression. LASSO identified 43 drugs associated with TD. In the multivariable logistic regression model, female sex, the 45-59-year age group, and the 43 drugs were independent risk factors for drug-associated TD. Furthermore, the median TTO for drug-induced TD was 34 days (IQR: 7-105 days), with approximately 75% of cases occurring within the first 100 days of drug initiation.

This study identified multiple drugs significantly associated with TD and highlighted sex, age, and specific medications as independent risk factors. These findings support improved pharmacovigilance and clinical awareness of drug-induced thyroid dysfunction.