Reply to: Concerns about the Burden of Proof studies

Abstract

We, together with the Global Burden of Disease Study and Burden of Proof (BoP) collaborators, have published a BoP Capstone Methods paper and suite of associated meta-analyses that introduce and apply a new framework for synthesizing evidence to evaluate relationships between selected risk factors and health outcomes 1 , 2 , 3 , 4 , 5 . The BoP approach was designed to address key problems in current analytical frameworks. To provide helpful information to users making decisions around risk exposure, the approach systematically estimates a flexible mean risk–outcome function, avoiding making strong assumptions such as log-linearity of the relationship. Complementing the mean risk–outcome relationship, BoP analysis provides a BoP risk function (BoPRF) that represents the lowest estimate of excess harmful risk associated with a risk factor, incorporating unexplained between-study heterogeneity after accounting for known variation in study design characteristics. From the BoPRF, we calculate summary risk–outcome scores and star-rating measures that present conservative estimates of the risk–outcome relationship and enable comparisons across different risk–outcome pairs.