OBJECTIVES: To build a high-throughput signal detection algorithm to identify safety signals associated with roflumilast use in the United States.

METHODS: We used a sequence symmetry analysis (SSA) approach to discover potential adverse effects of roflumilast. We used Marketscan databases to build a cohort of new users of roflumilast between 2011 and 2021. The algorithm screened for safety signals against 211 different therapeutic classes within 1 year before and after roflumilast initiation. Stratified analyses were conducted by sex, age, and observation period to explore the robustness of the detected safety signals. Crude sequence ratios (cSR) and adjusted sequence ratios (aSR) were reported with their 95% confidence intervals.

RESULTS: A total of 11,091 patients used roflumilast for the first time in the study period (53% were 65 or older, and 52% were female). Our base-case analysis identified potential safety signals in 19 out of 211 distinct therapeutic classes. The strongest signals were found with anti-thyroid agents (aSR, 3.62, 95% CI 1.44–10.36), parathyroid hormones (aSR, 2.65, 95% CI: 1.33-5.51), and meglitinides (aSR, 2.43, 95% CI: 1.15-5.35). Most observed signals confirmed previous findings from clinical trials. However, we discovered novel signals for anti-thyroid and parathyroid agents.

CONCLUSIONS: We detected 19 safety signals for roflumilast, including notable associations with anti-thyroid agents and parathyroid hormones. Future investigations using more robust study designs are warranted to evaluate those signals.