Expanding Bayesian Methods to Estimate Causal Effects of Environmental Exposures on Childhood Leukemia

• Activity Number: 337 - Environmental Epidemiology and Analysis of Large Database
• Type: Contributed
• Date/Time: Tuesday, August 9, 2022 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Sofia Vega* and Rachel Nethery
• Companies: Harvard T.H. Chan School of Public Health, Department of Biostatistics and Harvard T.H. Chan School of Public Health, Department of Biostatistics
• Keywords: Bayesian Statistics; Spatio-temporal Models; Matrix Completion; Childhood Leukemia; Environmental Exposures; Rare Outcomes

Abstract

The steady increase in childhood leukemia (CL) incidence in developed nations over the past 50 years suggests that exposures, such as mobile source benzene and 1,3-butadiene (MSBB), play a role in the development of CL. These chemicals are known to cause leukemia in adults with high exposure levels, but less is known about the relationship between MSBB and CL due to the epidemiologic challenges presented by the rarity of CL, difficulty in measuring exposure, and confounding. In the 1990s, the US EPA enacted regulations in select areas of the US that drastically reduced the amount of MSBB in gas, creating a quasi-experiment (QE) to study the effects of MSBB on CL. Existing methods for QE analysis, such as matrix completion, can fail when outcomes are rare and unstable, as with CL incidence. We develop a Bayesian space-time matrix completion model to robustly estimate causal effects of environmental exposures on rare outcomes. This model extends traditional methods to accommodate complex spatio-temporal correlation structures and enables stable estimation with rare outcomes. We test its precision through simulations and apply the model to estimate the causal effects of MSBB on CL.