EMPIRICAL COMPARISONS OF PROPORTIONAL HAZARDS, POISSON, AND LOGISTIC-REGRESSION MODELING OF OCCUPATIONAL COHORT DATA

This research was conducted to examine the effect of model choice on t he epidemiologic interpretation of occupational cohort data. Three mul tiplicative models commonly employed in the analysis of occupational c ohort studies-proportional hazards, Poisson, and logistic regression-w ere used to analyze data from an historical cohort study of workers ex posed to formaldehyde. Samples were taken from this dataset to create a number of predetermined scenarios for comparing the models, varying study size, outcome frequency, strength of risk factors, and follow-up length. The Poisson and proportional hazards models yielded nearly id entical relative risk estimates and confidence intervals in all situat ions except when confounding by age could not be closely controlled in the Poisson analysis. Logistic regression findings were more variable , with risk estimates differing most from the proportional hazards res ults when there was a common outcome or strong relative risk. The logi stic model also provided less precise estimates than the other two. Th us, although logistic was the easiest model to implement, it should be used only in occupational cohort studies when the outcome is rare (5% or less), and the relative risk is less than similar to 2. Even then, the proportional hazards and Poisson models are better choices. Selec ting between these two can be based on convenience in most circumstanc es. (C) 1998 Wiley-Liss, Inc.