Comparative Risk Measures for Heterogenous Populations

Abstract

Comparing population risks is a crucial part of many policy-directed risk analyses. In estimating the public health risks from operation of a power plant, sale of a new pharmaceutical product, exposure of individuals to ionizing radiation, reductions in automotive safety standards, or other proposed changes in the status quo, the risk analyst typically compares the estimated population risk in a target population in the presence of the proposed change with the estimated population risk in the same population (or in a matched control population) in the absence of the proposed change. If the absolute population risk at any time is defined as the average incidence rate of an undesirable event or “response” (such as death, illness or injury) in the target population, then the population risk attributed to the proposed change is commonly expressed through epidemiological measures such as relative risk (the ratio of response incidence rates with and without the proposed change), attributable risk (the difference between the two incidence rates), or the odds ratio (the ratio of the odds of response among members of the exposed and unexposed groups.) 1,2