Content area
Full Text
Introduction
How much of the disease burden in a population could be eliminated if the effects of certain causal factors were eliminated from the population? To address this question, epidemiologists calculate the population attributable fraction. As noted in a recent editorial in the Journal, population attributable fraction estimates can help guide policymakers in planning public health interventions.1 Despite numerous articles on population attributable fraction estimation,2-7 errors in computation and interpretation persist. In addition, in certain settings, the value of a population attributable fraction estimate may be questionable. This commentary considers computational and conceptual issues relevant to population attributable fraction estimation that are infrequently discussed elsewhere, with illustrations from the breast cancer literature.
Background
In 1953, Levin8 first proposed the concept of population attributable fraction. Since then, the phrases "population attributable risk," "population attributable risk proportion," "excess fraction," and "etiologic fraction" have been used interchangeably to refer to the proportion of disease risk in a population that can be attributed to the causal effects of a risk factor or set of factors. Greenland and Robins4 distinguish between excess fraction (what epidemiologists usually estimate when they compute "population attributable risk" or "population attributable fraction") and etiologic fraction, which is not estimable without strong biologic assumptions. Our use of the term "population attributable fraction" corresponds to Greenland and Robins' (population) excess fraction.
The population attributable fraction is most commonly defined as the proportional reduction in average disease risk over a specified time interval that would be achieved by eliminating the exposure(s) of interest from the population while distributions of other risk factors in the population remain unchanged. This also can be interpreted as the proportion of disease cases over a specified time that would be prevented following elimination of the exposures, assuming the exposures are causal.
While population attributable fractions usually are estimated for single risk factors, they also can be estimated for groups of factors considered simultaneously. In this situation, a population attributable fraction estimates the proportional amount by which disease risk would be reduced if all of the factors were to be simultaneously eliminated from the population. The exposed group consists of those exposed to at least one of the factors. A population attributable fraction for a set of risk factors considered simultaneously is sometimes termed...