Estimating proportionate changes in rates.

Abstract

Quantitative measures are defined for use in health policy formulation. Consider a dichotomized risk factor for a specific disease or a protective factor known to promote health. A health policy that alters the prevalence of exposure to the factor will create corresponding proportionate changes in the disease rate and in the health rate. Statistical formulas are presented for estimating such proportionate changes. Formulas are given for cohort, case-control, and some hybrid study designs, but only for those studies where it is appropriate to present the data in a 2 X 2 table. It is assumed that the population is stable and that all subjects are observed over the same fixed time interval. Formulas for estimating the standard error of the estimate are also provided.