Viral loads observed under competing strain dynamics

Abstract A plausible mechanism for the increased transmissibility of SARS-CoV-2 variants of concern (VOCs) results from VOC infections causing higher viral loads in infected hosts. However, investigating this hypothesis using routine RT-qPCR testing data is challenging because the population-distribution of viral loads changes depending on the epidemic growth rate; lower cycle threshold (Ct) values for a VOC lineage may simply reflect increasing incidence relative to preexisting lineages. To understand the extent to which viral loads observed under routine surveillance systems reflect viral kinetics or population dynamics, we used a mathematical model of competing strain dynamics and simulated Ct values for variants with different viral kinetics. We found that comparisons of Ct values obtained under random cross-sectional surveillance were highly biased unless samples were obtained at times when the variants had comparable growth rates. Conversely, comparing Ct values from symptom-based testing was largely unaffected by epidemic dynamics, and accounting for the time between symptom onset and sample collection date further reduced the risk of statistical errors. Finally, we show how a single cross-sectional sample of Ct values can be used to jointly estimate differences in viral kinetics and epidemic growth rates between variants. Epidemic dynamics should be accounted for when investigating strain-specific viral kinetics using virologic surveillance data, and findings should be corroborated with longitudinal viral kinetics studies.