Spatial Time-to-Event Analysis of Air Pollution and Preterm Birth

• Activity Number: 335
• Type: Topic Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Howard Chang*+ and Brian Reich and Marie Lynn Miranda
• Companies: Statistical and Applied Mathematical Sciences Institute and North Carolina State University and Duke University
• Keywords: Air pollution ; Preterm birth ; Spatial epidemiology ; Reproductive epidemiology

Abstract

Most studies of ambient air pollution and preterm birth rely on birth certificate data where the challenge of confounding due to unmeasured spatial and temporal risk factors is well recognized. We describe a flexible spatial discrete time survival model to estimate the effect of environmental exposure on the risk of preterm birth. Specifically, cases and controls are compared only during a gestational age window at-risk of preterm birth. This design is similar to a time series analysis that examines the risk of preterm birth among ongoing pregnancies between days. However unlike a time series analysis, our model also allows the inclusion of individual-level risk factors. The approach is applied to a dataset of geocoded births during 2002 in North Carolina and we report an increased risk of preterm birth associated with short-term exposure to fine particulate matter.