Do Various Glitazones Have the Same Risk of Acute Myocardial Infarction? Indirect Evidence from a Population-Based Norwegian Cohort Study

Abstract

Objective: Pioglitazone lowers triglycerides and is indifferent towards low-density lipoproteins (LDL), while rosiglitazone has no effect on triglycerides and increases LDL. Our purpose was to test the hypothesis that the risk of acute myocardial infarction (AMI) in thiazolidinediones differs. Methods: We followed a cohort from the Norwegian Prescription Database consisting of 4,009 and 740 first time users of pioglitazone and rosiglitazone respectively for three years. We estimated the propensity score for rosiglitazone vs pioglitazone based on age, gender, and the use of 13 drug classes. We used the initiation of platelet aggregation inhibitors, lipid lowering drugs, beta-adrenergic blockers, and renin-angiotensin inhibitors as a proxy for AMI after testing the validity of these endpoints in a separate cohort of patients suffering their first AMI. We estimated hazard ratios (HR, rosiglitazone vs pioglitazone) and their 95 percent confidence intervals (CI) for AMI using Cox proportional hazards models stratified by propensity score deciles. Results: During the first six months after initiation the incidences of the initiation of platelet aggregation inhibitors were the same with both glitazones (HR=1.0; 95 % CI: 0.65-1.52). More than six months after initiation, rosiglitazone was associated with an increased risk of initiating platelet aggregation inhibitors compared with pioglitazone (HR=1.68; 95 % CI: 1.09-2.61). We observed no difference between the glitazones and the initiation of any of the other drug classes assessed. Conclusions: Albeit indirectly, our cohort study supports the hypothesis that the two thiazolidinediones differ in their risk of AMI, based on monitoring over a period of three years the initiation of drug classes indicated after AMI.