Model Selection for Modeling Mesothelioma Mortality

• Activity Number: 269
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Leonid Kopylev* and Thomas Bateson
• Companies: EPA and EPA
• Keywords: cancer ; epidemiology ; exposure

Abstract

Mesothelioma is a rare cancer caused predominantly by exposure to asbestos fibers. Recently, models that include the clearance of fibers have been suggested for modeling mesothelioma mortality in several asbestos worker cohorts with long follow-up but relatively crude exposure data. It is seen that allowing for fiber clearance improves upon results obtained from traditional models of such data. We consider several mathematical models for parameterizing fiber exposures and use a cohort of amphibole asbestos exposed workers to evaluate these parameterizations for modeling mesothelioma mortality. The job exposure matrix of this cohort is uneven throughout the period of exposure; therefore, we discuss the impact of the quality of exposure information on model selection. We found that including a mathematically modeled fiber clearance term fit better than mathematically weighing earlier exposures higher. Although these results are based on specific individual-level exposure data, there are relatively low number of mesothelioma deaths in this analysis. Disclaimer: The views expressed are those of the authors, and do not necessarily represent the views or policies of the U.S. EPA.