Population Attributable Fraction: translating causal knowledge into prevention potential

Abstract As part of a Comparative Risk Assessment (CRA), the Population Attributable Fraction (PAF) is calculated to estimate the proportion of a synthetic population health indicator (mortality, morbidity, DALY, YLL, YLD) attributable to a given risk factor, and therefore theoretically preventable. This metric plays a critical role as public health shifts toward prevention and seeks to guide policy and resource allocation. This presentation will briefly review the definition of PAF, then show how its integration into the CRA framework has made it possible to overcome certain limitations historically associated with the causal attribution of health indicators. In particular, the CRA approach makes it possible to take account of the full distribution of exposure in the population, non-linear risk functions, and a counterfactual level of exposure that is more realistic than the simple “unexposed”. We will then illustrate and discuss how this measure makes it possible to quantify prevention potential intuitively, and to facilitate its use in support of public decision-making. Monitoring PAFs associated with different risk factors makes it possible to: • prioritise interventions in public health and research; • communicate effectively on the avoidable burden to decision-makers and the general public; • target interventions where their impact could potentially be greatest, particularly within certain population sub-groups. In this sense, the PAF is a central tool for linking epidemiological knowledge to policy action by quantifying what part of the burden of disease can be attributed to preventable behaviours or conditions.