Viral Interference and Coinfections: A Perspective From Hospital Surveillance of Respiratory Viruses

Viral interference may influence pathogen transmission at the population level, potentially affecting seasonal epidemics of respiratory infections. A frequently employed association measure purported to reflect interference effects is the prevalence ratio, the proportion of individuals coinfected with two viruses divided by the product of the proportions of individuals infected by each virus separately. However, the prevalence ratio neglects three important factors relevant to coinfection detection in epidemiological surveillance programs: undetected mono-infections, duration of viral excretion or detectability and circulation patterns of both viruses. We propose a generalization of the prevalence ratio that accounts for these factors to better assess the presence or absence of viral interactions in epidemiological surveillance data. We applied this association measure to influenza-respiratory syncytial virus (RSV) coinfection data from a hospital-based surveillance program of respiratory infections in the province of Québec, Canada, from 2012-2013 to 2018-2019 (HospiVir program). Our analysis suggests that influenza-RSV interference may be important in children but less in adults. However, our results are sensitive to population-level seasonal attack rates; coinfection data could be compatible with interference in adults if assumed attack rates increased from 3% to 5% to over 10%.