Statistical Relationship Between Wastewater Data and Case Notifications for COVID-19 Surveillance in the United States, 2020-2023: A Bayesian Hierarchical Model

During the COVID-19 pandemic a number of jurisdictions in the United States began to regularly report levels of SARS-CoV-2 in wastewater for use as a proxy for SARS-CoV-2 incidence. Despite the promise of this approach for improving situational awareness, the degree to which viral levels in wastewater track with other outcome data has varied, and better evidence is needed to understand the situations in which wastewater surveillance tracks closely with traditional surveillance data. In this study, we quantified the relationship between wastewater data and traditional case-based surveillance data for multiple jurisdictions. To do so, we collated data on wastewater SARS-CoV-2 RNA levels and COVID-19 case reports from July 2020 to March 2023, and employed Bayesian hierarchical regression modeling to estimate the statistical relationship between wastewater data and reported cases, allowing for variation in this relationship across counties. We compared different model structural approaches and assessed how the strength of the estimated relationships varied between settings and over time. These analyses revealed a strong positive relationship between wastewater data and COVID-19 cases for the majority of locations, with a median correlation coefficient between observed and predicted cases of 0.904 (interquartile range 0.823 – 0.943). Across locations, the COVID-19 case rate associated with a given level of wastewater SARS-CoV-2 RNA concentration declined over the study period. Counties with higher population size and of higher levels of urbanicity had stronger concordance between wastewater data and COVID-19 cases. Ideally, use of wastewater data for decision-making should be based on an understanding of their local historical performance.