The within-host reproductive number R0 is an important parameter to predict the minimum antiviral efficacy needed to suppress viral infection. However, this parameter has not been well quantified for SARS-COV-2. This is because accurate estimation of this quantity requires longitudinal viral load measurements during the initial phase of infection, when the virus population expands before the viral load peak; yet, most available measurements are made after the viral load peak.
We constructed viral dynamic models to describe a set of longitudinal viral load data from a study where individuals were tested frequently such that viral loads during the viral expansion phase were measured. We fit multiple models to data from a total of 42 infected individuals (14 symptomatic and 28 asymptomatic) to estimate R0 and used a model linking within-host viral load to the infectiousness of a person to evaluate the infectiousness of asymptomatic individuals compared to symptomatic individuals.
We estimated that the within-host R0 is between 8-16 across the 48 individuals. This suggests that antiviral efficacy has to be greater than 95% to suppress virus infection in a majority of individuals. The estimated R0 in asymptomatic individuals is lower than in symptomatic individuals (mean 10.0 vs. 13.8; p-value<0.0001). Our model suggests there exists large heterogeneity in infectiousness among individuals, and asymptomatic individuals may be on average 15% less infectious than symptomatic individuals (p-value=0.02), not considering isolation measures.
An antiviral efficacy of 95% or more is needed to suppress viral infection in most infected individuals. Asymptomatic individuals may be slightly less infectious than symptomatic individuals.