Speaker
Description
Understanding which types of contact drive pathogen transmission is critical for outbreak control. Traditionally, high-risk contacts are inferred with contact tracing or network data. Including genetic data may better elucidate the contribution of different types of contact to transmission. Vice versa, contact data may improve transmission inference (who infected whom) of outbreaks.
We formulated a transmission likelihood for who infected whom, from data about different contact types observed between cases of an outbreak, based on \cite{Campbell2019}. We implemented this in the phylodynamic model phybreak, to analyse outbreaks using pathogen sequences and sampling times. The resulting model estimates the fractions of transmission events attributable to the types of contact. We evaluated performance with simulations across various scenarios of contact frequencies, transmission risks by contact type, and correlations between contact types.
Results show that sparse contact types with high transmission risk can be accurately identified. Performance declines when most transmission occurs via common contact types, or when contact types are positively correlated. Use of contact data improves inference of who infected whom, particularly when genetic data alone lack resolution. When applying the method to the 2020 SARS-CoV-2 outbreak in Dutch mink farms, shared personnel was identified as a high-risk contact, with a limited role of veterinary service providers and feed suppliers.
Bibliography
@article{Campbell2019,
author = {Campbell, Finlay and Cori, Anne and Ferguson, Neil and Jombart, Thibaut},
title = {Bayesian inference of transmission chains using timing of symptoms, pathogen genomes and contact data},
journal = {PLOS Computational Biology},
year = {2019},
volume = {15},
number = {3},
pages = {e1006930},
doi = {10.1371/journal.pcbi.1006930},
publisher = {Public Library of Science}
}
