Reliable and scalable seroepidemiology methods are needed to estimate SARS-CoV-2 incidence and monitor the dynamics of population-level immunity as the pandemic evolves. We aimed to evaluate the reliability of SARS-CoV-2 normalized ELISA optical density (nOD) at a single dilution compared to titers derived from serial dilutions. We conducted serial serosurveys within a community-based cohort in Salvador, Brazil. Anti-S IgG ELISA (Euroimmun AG) was performed with five serial 3-fold dilutions of paired sera from 54 participants. Changes in nOD reliably predicted increases and decreases in titers (98.1% agreement, κ = 95.8%). Fitting the relationship between nOD and interpolated titers to a log-log curve yields highly accurate predictions of titers (r² = 0.995) and changes in titers (r² = 0.975), using only one to two dilutions. This approach can significantly reduce the time, labor and resources needed for large-scale serosurveys to ascertain population-level changes in exposure and immunity.

Data can be accessed using R.