Path-specific population attributable fractions

A population attributable fraction (PAF) represents the relative change in disease prevalence that one might expect if a particular exposure was absent from the population. Often, one might be interested in what percentage of this effect acts through particular pathways. For instance, the effect of excessive alcohol intake on stroke risk may be mediated by blood pressure, body mass index and several other intermediate risk factors. In this situation, attributable fractions for each mediating pathway of interest can be defined as the relative change in disease prevalence from disabling the effect of the exposure through that mediating pathway. This quantity is related to, but distinct from the recently proposed metrics of direct and indirect PAF by Sjolander. In particular, while differing pathway-specific PAF will each usually be less than total PAF, they may sum over differing mediating pathways to more than total PAF, whereas direct and indirect PAF must sum to total PAF. Here, we present definitions, identifiability conditions and estimation approaches for pathway-specific attributable fractions. We illustrate results, and comparisons to indirect PAF using INTERSTROKE, a case-control study designed to quantify disease burden attributable to a number of known causal risk factors.