Using Bayesian Logistic Regression to Estimate the Risk or Prevalence Ratio

• Activity Number: 257
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Charles E. Rose, Jr.*+ and Andrew Lewis Baughman
• Companies: CDC and CDC
• Keywords: Risk Ratio ; Prevalence Ratio ; Bayesian Logistic Regression ; Log-Binomial Model ; Poisson Model

Abstract

In cohort and cross-sectional studies or when the outcome is common, the risk ratio (RR) is the preferred measure of effect rather than an odds ratio (OR). The logistic regression OR is often used to approximate the RR when the outcome is rare. However, whether the outcome is rare or common, logistic regression predicted exposed and non-exposed risks can be used to form an appropriate RR. We developed a Bayesian logistic regression model to estimate the RR, with associated credible interval, and applied the model to published data. We compared our results to four commonly used RR modeling techniques: stratified Mantel-Haenszel, logistic regression, log-Binomial, and log-Poisson. Our Bayesian logistic regression provides a flexible framework for investigating confounding and effect modification on the risk scale and compares favorably with existing RR modeling methods.