Purpose: This study examined the potential risk of cardiovascular (CV) events associated with modafinil and the consistency of the risk estimates across databases.
Methods: A retrospective, inception cohort design of patients who initiated treatment with modafinil between 2006 and 2008 was used in three US health care claims databases. Modafinil users were matched with nonusers. Patients were further divided into two cohorts of obstructive sleep apnea (OSA) and non-OSA (NOSA) cohorts. Endpoints of interest, including myocardial infarction (MI), stroke, CV hospitalizations, and all-cause death, were assessed using incidence rates and Cox proportional hazard ratios (HRs), adjusted for potential confounding factors.
Results: The cohorts included a total of 175 524 patients in MarketScan CM; 77 266-in IMS LifeLink; and 8174-in MarketScan Medicaid. No increased risk for MI in the OSA and NOSA cohorts was observed across all three databases. The risks of CV hospitalization in the OSA and NOSA cohorts were not different between the modafinil users and nonusers, except for IMS LifeLink database where the HR was lower than one in the modafinil users compared with the nonusers (HR, 0.69; 95% confidence interval [CI], 0.54 to 0.87). For OSA patients with prior stroke, an adjusted HR of 1.96 (95% CI, 1.02 to 3.76) was observed for stroke among modafinil users compared with nonusers. Among the NOSA, the HRs for all-cause death in the OSA were inconsistent across databases.
Conclusions: Except for few CV outcomes, applying one common protocol generated consistent risk estimates of CV events following modafinil use across cohorts and databases.
Keywords: administrative claims databases; cardiovascular events; modafinil; pharmacoepidemiology; safety.
© 2018 John Wiley & Sons, Ltd.