Abstract
INTRODUCTION: Roflumilast, a selective phosphodiesterase-4 inhibitor, is prescribed to reduce exacerbations in severe COPD, but its real-world safety profile remains insufficiently characterized. METHODS: We conducted a retrospective pharmacovigilance study using the FDA Adverse Event Reporting System (FAERS). Reports from 2004 to 2025Q1 listing roflumilast as the primary suspect were extracted, deduplicated, and analyzed. Disproportionality analysis employed four algorithms-Reporting Odds Ratio (ROR), Proportional Reporting Ratio (PRR), Bayesian Confidence Propagation Neural Network (BCPNN), and Multi-item Gamma Poisson Shrinker (MGPS). Safety signals were defined by established thresholds. Time-to-onset and Weibull modeling were applied to assess temporal patterns. RESULTS: A total of 3,140 reports were identified, primarily involving older COPD patients. The median time-to-onset was 4 days (IQR 0-34), with 72% occurring within 30 days. Thirty Preferred Terms met signal criteria. Frequent signals included diarrhoea, weight decreased, nausea, dyspnoea, insomnia, and headache. Psychiatric events such as depression and suicidal ideation were notable. The strongest disproportionality was observed for gastroduodenal ulcer (ROR 47.1) and COPD (ROR 25.7). System Organ Class enrichment was most evident in gastrointestinal, psychiatric, and respiratory disorders. DISCUSSION/CONCLUSION: This real-world analysis confirms roflumilast's established adverse effects (gastrointestinal upset, weight loss, insomnia) and highlights concerning psychiatric signals. Most events occurred early, underscoring the need for close monitoring during treatment initiation. The use of multiple disproportionality methods enhances signal detection robustness and supports ongoing pharmacovigilance in clinical practice.