Monte Carlo Sensitivity Analysis of Smoking as an Unmeasured Confounder in an Investigation of the Association Between Occupation and Lung Cancer Mortality

• Activity Number: 209
• Type: Contributed
• Date/Time: Monday, July 30, 2007 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Jia Li*+ and Toni Alterman
• Companies: Constella Group, LLC and National Institute for Occupational Safety and Health
• Keywords: sensitivity analysis ; Monte Carlo simulation ; lung cancer mortality ; smoking ; occupation

Abstract

Traditional statistical methods focus on measured confounders and random errors. In observational epidemiology, systematic errors due to unmeasured confounders can limit the inferential power of the study, and hence need to be addressed and quantified. Data from the National Occupational Mortality Surveillance System was used to examine the association between death from lung cancer and occupation. After adjustment for gender, race, and age, the lung cancer mortality odds ratio (MOR) and 95% confidence interval (CI) for blue-collar vs. white-collar workers was 1.25 (1.19, 1.31). Smoking status was constructed repeatedly using Monte Carlo simulation based upon associations between lung cancer mortality, smoking, and occupation obtained from the literature and the U.S. National Health Interview Survey. With additional adjustment for smoking, the MOR and 95% CI became 0.99 (0.94, 1.04).