Letters Section Editor: Robert M. Golub, MD, Senior Editor.
In Reply: Dr Suissa raises a question about immortal time bias in our intent-to-treat analysis. This bias is a result of incorrectly classifying the person-time before the first exposure to a medication as being “on therapy.”1 However, our analysis was not performed in this manner. Follow-up started when each participant met the combination of criteria for heart failure and Adult Treatment Panel III criteria to receive cholesterol-lowering therapy, and this defined the participant's index date. After the index date, we structured our data in 2 ways for different analyses: intent-to-treat and time-varying exposure to statin therapy. In both data structures, person-time before the first-identified exposure to statins during follow-up was correctly classified as being “off therapy.”
Go AS, Yang J, Gurwitz JH. Statin Use and Risks for Death and Hospitalization in Chronic Heart Failure—Reply. JAMA. 2007;297(10):1057-1058. doi:10.1001/jama.297.10.1058-a