Disparities in Discontinuing Rosiglitazone Following the 2007 FDA Safety Alert

Responsiveness to the Food and Drug Administration (FDA) rosiglitazone safety alert, issued on May 21, 2007, has not been examined among vulnerable subpopulations of the elderly.

To compare time to discontinuation of rosiglitazone after the safety alert between black and white elderly persons, and across sociodemographic and economic subgroups.

A cohort study.

Medicare fee-for-service enrollees in 2007 who were established users of rosiglitazone identified from a 20% national sample of pharmacy claims.

Outcome of interest was time to discontinuation of rosiglitazone after the May alert. We modeled the number of days following the warning to the end of the days' supply for the last rosiglitazone claim during the study period (May 21, 2007-December 31, 2007) using multivariable proportional hazards models.

More than 67% of enrollees discontinued rosiglitazone within six months of the advisory. In adjusted analysis, white enrollees (hazard ratio=0.90; 95% confidence interval, 0.86-0.94) discontinued rosiglitazone later than the comparison group of black enrollees. Enrollees with a history of low personal income also discontinued later than their comparison group (hazard ratio=0.84; 95% confidence interval, 0.81-0.87). There were no observed differences across quintiles of area-level socioeconomic status.

White race and a history of low personal income modestly predicted later discontinuation of rosiglitazone after the FDA's safety advisory in 2007. The impact of FDA advisories can vary among sociodemographic groups. Policymakers should continue to monitor whether risk management policies reach their intended populations.