Estimating SARS-CoV-2 reinfection frequency and infection incidence: A modeling study

Confirmed coronavirus disease 2019 cases represent merely the tip of the iceberg of total infections. While seroepidemiological data help estimate undiagnosed infections, widespread reinfections, especially with the dominance of the Omicron variant, have complicated transmission dynamics interpretation. We developed a new method to simultaneously estimate severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) reinfection frequency and infection incidence, combining epidemiological surveillance data, seroepidemiological data, and a mathematical model. The model accounted for reinfection and waning of immunity with multiple viral antigenic variants. We estimated that the cumulative proportion of reinfections increased from 0.4% (95% credible interval [CrI]: 0.3%-0.5%) as of December of 2021 to 3.7% (95% CrI: 3.5%-3.9%) by June of 2022 in Japan, dominated by Omicron subtypes BA.1 and BA.2; this further increased to 28.1% (95% CrI: 27.1%-29.1%) by May 2023, dominated by BA.4 and BA.5. By May 8, 2023, the cumulative number of infections was estimated to be 4.7times (95% CrI: 4.3-5.0) the number of confirmed cases. The actual sizes of the SARS-CoV-2 epidemic and reinfection dynamics were reconstructed, potentially acting as a gold standard survey to directly infer infection volume. Our findings endorse the value of continued case-based and seroepidemiological surveillance.