A Perspective on the Use of Poisson Versus Logistic Regression in Exposure-Response Analysis: Insights and Considerations.

Abstract

This perspective evaluates the use of Poisson versus logistic regression in modeling binary exposure-response (ER) data. Through simulation studies across varying sample sizes, event rates, and ER slopes, we highlight the strengths and limitations of each method. Our findings show that Poisson regression is suitable under low event rates, while logistic regression provides consistent performance across broader scenarios. These insights help guide model selection and improve the robustness of ER analyses in drug development.