Positive Bias of the Combined Effect of Risk Factors Estimated by Marginal Aetiological Fractions

Publisher Website
Google Scholar
Abstract
The method of aetiological fractions is a commonly used epidemiological technique which attributes a certain fractionof each event of interest to a risk factor. An unrecognized property of this method is that when marginal analyses are done, the combined fraction attributed to the risk factors can be considerably overstated, particularly for prevalent risk factors such as alcohol and tobacco. In this paper a practical example of this problem is given. Theoretical bounds on the bias are derived and evaluated for a range of relative risks that are typical for alcohol and tobacco. It is shown that the overstatement can be quite substantial.

This publication has 0 references indexed in Scilit: