Estimating the cumulative incidence of SARS-CoV-2 with imperfect serological tests: Exploiting cutoff-free approaches

As other pathogens, SARS-CoV-2 elicits antibody responses in infected people that can be detected in their blood serum as early as a week after the infection until long after recovery. The presence of SARS-CoV-2 specific antibodies can therefore be used as a marker of past infection, and the prevalence of seropositive people, i.e. people with specific antibodies, is a key measure to determine the extent of the SARS-CoV-2 pandemic. The serological tests, however, are usually not perfect, yielding false positive and false negative results. Here we exploit an approach that refrains from classifying people as seropositive or negative, but rather compares the antibody level of an individual to that of confirmed cases and controls. This approach leads to more reliable estimates of cumulative incidence, especially for the low prevalence and low test accuracies that we face during the current SARS-CoV-2 pandemic. We also show how this approach can be extended to infer the presence of specific types of cases that have not been used for validating the test, such as people that underwent a mild or asymptomatic infection.

Recommended citation: Bouman, JA, Riou, J, Bonhoeffer, S, Regoes, RR (2021). "Estimating the cumulative incidence of SARS-CoV-2 with imperfect serological tests: Exploiting cutoff-free approaches" PLOS Computational Biology. 17(2), e1008728. https://doi.org/10.1371/journal.pcbi.1008728