Conditional Poisson Regression with Random Effects for the Analysis of Multi-site Time Series Studies

Area of Science:

• Environmental Epidemiology
• Biostatistics
• Public Health Research

Background:

• Time series studies often link health indicators (e.g., mortality, crashes) with environmental factors (e.g., air pollution, temperature).
• Traditional two-stage meta-analysis approaches for multi-zone studies can be computationally intensive.
• Existing methods struggle with rapidly escalating strata in long time series with numerous zones.

Conclusions:

• The Bayesian conditional Poisson regression model with random slopes provides a flexible and efficient tool for environmental health time series analysis.
• This approach enhances the ability to control for complex trends and spatial variations across different zones.
• The study offers a reproducible framework for researchers in epidemiology and biostatistics.