Doubly robust estimation of attributable fractions

The attributable fraction (AF) is a widely used measure to assess the impact of an exposure on a disease. It is commonly estimated through maximum likelihood, which requires a regression model for the outcome. Recently, it was demonstrated that the AF can also be estimated through inverse probability weighting, which requires a model for the exposure. In this paper, we derive doubly robust estimators for the AF. These estimators require one model for the outcome and one model for the exposure and are consistent if either model is correct, not necessarily both. We consider both cohort/cross-sectional studies and case-control studies.