In Reply: Dr Fresco comments that the primary end point showed progression in the pioglitazone treatment group for patients not treated with statins. We disagree and recommend caution in interpretation of subgroup data. In the pioglitazone group, the 95% CI for the primary end point for the subgroup not receiving statins ranged from −0.45 to 2.29. The lower limit of the CI being less than 0 was consistent with the overall study results. In any clinical trial, results in subgroups are considered hypothesis-generating at best because when multiple groups are assessed, the likelihood of an apparent difference occurring by chance alone is substantial. In this case, we think that even more caution is required. There were only 29 patients in the glimepiride group who did not receive statins and only 25 in the pioglitazone group. We did not stratify for statin use because we expected nearly all patients to receive statins and knew that many different agents would be used, precluding any reasonable stratification approach.
Nissen SE, Nicholls SJ, Wolski K. Pioglitazone vs Glimepiride in the PERISCOPE Trial—Reply. JAMA. 2008;300(7):787-788. doi:10.1001/jama.300.7.788-a