Evaluating Risk from a 2 x 2 Table: Five Useful Measures

Abstract

Risk assessment from 2 x 2 tables can be very useful for both clinicians and researchers. The measures described go beyond demonstrating a statistical association between some factor and disease or death, and attempt to assess, in real world terms, just how important a particular factor may be. Five measures of risk are explained by means of a veterinary example. The odds ratio compares the odds of animals succumbing to disease or death if they are exposed to a factor, compared to the odds of succumbing if they are not exposed. Attributable rate and attributable proportion (risk) measure the amount of disease or death in the exposed group which can actually be attributed to the factor under study. Population attributable rate and population attributable proportion (risk) extend these measures to the real world: how much of the disease or death on the farm might we be able to prevent? Calculation of measures of risk from a data set requires that a number assumptions be met, and that the data used be of the highest quality.