Background: Rosiglitazone was approved by the U.S. Food and Drug Administration (FDA) for type 2 diabetes in 1999. The unique mechanism of action and low risk of hypoglycemia contributed to rapid market uptake of rosiglitazone, but safety concerns became more prominent in 2007. There were 5 major events on 4 calendar days in 2007 regarding safety concerns related to rosiglitazone in certain patients: (1) the May 21, 2007, online release of the rosiglitazone meta-analysis performed by Nissen and Wolski and the FDA safety warning on the same day; (2) the July 30, 2007, conclusion of an FDA advisory committee meeting that rosiglitazone increased cardiac ischemic risk; (3) the August 14, 2007, update of thiazolidinedione (TZD) labels with a black-box warning for heart failure; and (4) the November 14, 2007, update to the warnings and precautions section of the rosiglitazone label for coadministration of nitrate or insulin.
Objectives: To (1) describe TZD (rosiglitazone and pioglitazone) utilization trends from January 1, 2007, continuing through May 2008 amid public announcements of safety concerns and (2) determine the percentage of TZD users who had medical claims indicating increased cardiovascular (CV) risk before and after release (May 21, 2007) of the FDA safety warning and online release of the meta-analysis performed by Nissen and Wolski.
Methods: A retrospective analysis of pharmacy claims was performed from 9 commercial plans with a combined 9 million eligible members, including a 1.4 million-member cohort from 1 of the plans for which medical claims data were available. We evaluated trends in TZD use for each month for the 17-month period from January 1, 2007, through May 31, 2008, including the percentage of TZD users at increased CV risk. In the trend analysis, for each calendar month of 2007, we calculated mean pharmacy claim counts per day per million members for each of the 2 TZD drugs and for a comparison drug, sitagliptin, a new oral hypoglycemic agent in a different class (dipeptidyl-peptidase-IV inhibitors). For the CV risk analysis, we used the database of integrated medical and pharmacy claims for the 1.4 million-member cohort to identify patients with a current days supply of a TZD on May 20, 2007, December 7, 2007, or May 20, 2008. The medical claims for all identified patients were queried back 2 years from May 20, 2007, December 7, 2007, or May 20, 2008, respectively. Rosiglitazone users at increased CV rsk were defined as those with a medical claim with a primary diagnosis for congestive heart failure (CHF; International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 428.xx or 398.91), those with a current supply of nitrate or insulin therapy, or those with ischemic heart disease, including myocardial infarction (MI; ICD-9-CM codes 410.xx through 414.xx, or surgical procedure codes [36.0x through 36.3x for removal of obstruction and insertion of stents, bypass surgery, and revascularization] in the primary diagnosis field). Pioglitazone users at increased risk were identified from medical claims with a CHF diagnosis code.
Results: The average number of claims per day per million members in January 2007 was 97.3 for rosiglitazone and 107.2 for pioglitazone. The average number of claims for rosiglitazone per day per million members began to decrease in May 2007, falling to 41.0 in December 2007, for a total decrease of 58.6% from the February 2007 peak (99.1), and fell further to 31.8 in May 2008. Pioglitazone use increased 8.0% from January to June 2007 (107.2 to 115.8) and remained relatively flat through December 2007 (114.6) and through May 2008 (108.9). Sitagliptin claims increased 5-fold, at a consistent rate, from an average of 8.6 claims per day per million members in January 2007 to 43.4 in December 2007, and continued to increase to 48.7, in May 2008. Of the 5,117 rosiglitazone users on May 20, 2007, 1,296 (25.3%) were identified at increased CV risk versus 590 (22.5%) of 2,621 users on December 7, 2007 (P = 0.006), and 336 (21.8%) of 1,541 users in May 2008 (P = 0.005). Of 6,056 pioglitazone users on May 20, 2007, 170 (2.8%) had a CHF diagnosis versus 160 (2.5%) of 6,275 users on December 7, 2007 (P = 0.376), and 122 of 5,998 users in May 2008 (P = 0.006).
Conclusions: Although rosiglitazone utilization per million members declined by more than half in 2007, when CV safety concerns started to emerge, about 1 in 5 rosiglitazone users had elevated CV risk at year-end 2007 and in May 2008. About 3% of pioglitazone users in May 2007 had a diagnosis of CHF in claims history, which declined to 2% in May 2008. Insurers should consider the impact of persistent utilization of TZDs among members with CV risk factors when making formulary decisions.