Abstract
A SAS Program for the Stratified Analysis of Follow-Up Data:Yasushi HONDA, et al. Environmental Health Sciences Division, National Institute for Environmental Studies-Mantel-Haenszel (M-H) estimators are robust measures of association for the stratified analysis of epidemiologic studies. This paper describes a SAS macro developed to compute the M-H rate ratio (RR) for a binary exposure and its confidence interval from follow-up data. The program accommodates a virtually unlimited number of stratification factors, allows setting the confidence level, and can easily be applied to the analysis of multi-level categorical exposure variables. Printed output includes stratum-specific rates and RRS and summary M-H RRS and confidence intervals. Summary measures of association are meaningful only if effects are homogeneous across strata. The printed output of the program allows inspection of stratum-specific rate ratios to check this important assumption.