Instituto Gulbenkian de Ciência

Testing for SARS-CoV-2 infection is a pillar of the strategy to contain this virus. However, we know that many infected people are missed by testing, either because they never take the test or because the test fails. Indeed, we have done a longitudinal serosurveillance study, with a stratified quota sample of the whole population in Portugal, and showed that 2.5 to 10 times more people had been infected than accounted for in the official statistics. To understand in more detail the relationship between a person’s viral load, its infectious profile and the probability of testing positive, we developed models of the within host SARS-CoV-2 viral load and linked them to the probability of detection by testing. These models described the viral load extremely well and allow quantification of key dynamical characteristics of the virus. Based on empirical data and models of the probability of detection for different tests, we define the profile of positivity for typical infected persons. These models link viral load dynamics with epidemiological transmission and intervention strategies to better define control strategies.