Focusing a viral risk ranking tool on prediction

Abstract

Preparing to rapidly respond to emerging infectious diseases is critical. SpillOver: Viral Risk Ranking is an open-source tool developed to assess the risk of novel wildlife-origin viruses spilling over from animals to humans and spreading in human populations. Several risk factors used by the tool depend on evidence of previous zoonotic spillover itself or sustained transmission in humans. Therefore, we reanalyzed the Ranking Comparison after removing eight of the 31 risk factors that require postspillover knowledge and compared the adjusted risk rankings to the originals. The area under the receiver operating characteristic curve deteriorated from 0.94 for the original risk scores to 0.73 for the adjusted ones for predicting the classification as a human virus. We also compared the mean and SD of the risk scores for the human and non-human viruses at the risk factor level. Most excluded spillover-dependent risk factors had dissimilar means between the human and non-human virus classifications, but nonspillover-dependent risk factors frequently showed similar means between the two classifications. The original formulation of the tool depended on the inclusion of spillover-dependent risk factors to quantitatively assess the risk of zoonotic spillover for a novel virus. Future iterations of the tool should omit such risk factors and consider other nonspillover-dependent risk factors to ensure that the tool is fit for risk prediction of novel viruses.