Prediction of confirmed, hospitalized, and severe COVID-19 cases and mechanistic insights from viral concentrations and variant dynamics in wastewater

Background Following the end of a public health emergency of international concern, divergence emerged between reported coronavirus disease 2019 (COVID-19) cases and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA concentrations in wastewater. Exploring viral, clinical, patient, and surveillance-related factors underlying this divergence, we developed models to predict clinically confirmed infections, hospitalizations, and severe cases. Methods In this observational study, we analyzed ~2 years of data from January 2022 in Kanagawa Prefecture, Japan, assessing associations between wastewater SARS-CoV-2 RNA concentrations and confirmed, hospitalized, and severe cases, adjusting for wave and variant effects. Findings Our models based on wastewater viral RNA concentrations showed high predictive accuracy (R^2 = 0.8199-0.9961), closely tracking confirmed, hospitalized, and severe cases. Models derived from earlier waves were applied to subsequent waves with residual correction based on prior prediction errors and maintained good predictive performance (root mean square error = 0.0665-0.2065). Divergence between wastewater viral RNA concentrations and reported cases was not explained by changes in viral shedding. Declines in patients' healthcare-seeking behavior and testing were associated with trends in confirmed cases, whereas milder clinical presentation was associated with severe case trends. The lineages XBB.1.9.2 and BA.2.86 were identified as candidates associated with reduced virulence. Interpretation By incorporating understanding of viral, clinical, and surveillance-related mechanisms, wastewater surveillance may enable prediction of case trends approximately one week earlier than official reporting and inform healthcare capacity planning.