Measurement Error Correction for Survival Data Analysis with Covariates That Are Functions of Time-Varying Exposure Histories

• Activity Number: 115
• Type: Topic Contributed
• Date/Time: Monday, August 5, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Xiaomei Liao*+ and Donna Spiegelman
• Companies: Harvard School of Public Health and Harvard School of Public Health
• Keywords: Measurement error ; Time-varying exposure ; Risk set regression calibration

Abstract

Nutritional and environmental epidemiologists are often interested in estimating the prospective effect of functions of a time-varying exposure history such as the cumulatively updated average exposures or 12-month moving average exposures, in relation to chronic disease endpoints. By re-calibrating the measurement error model within each risk set, a risk set regression calibration (RRC) method has been proposed for Cox models in this setting. An algorithm for a bias-corrected point estimate of the relative risk using an RRC approach is developed, followed by the derivation of an estimator of its variance. Emphasis is on methods applicable to the main study/external validation study design, which is standard in nutritional and environmental epidemiology. Limitations of the current validation study designs are discussed and solutions to these limitations are proposed. Simulation studies under several realistic assumptions about the error model and correlation structures were conducted and the method is applied to a study of long-term exposure to fine particulate matter in air pollution in relation to all-cause mortality in the Nurses' Health Study.