5c - Wastewater-based surveillance of SARS-CoV-2 has limited potential to predict the number of COVID-19 cases in Sweden

Background : During the SARS-CoV-2 pandemic, wastewater-based surveillance was used worldwide with varying degrees of success. We assessed its potential as an early-warning indicator for COVID-19 in Sweden.

Methods : We evaluated the levels of SARS-CoV-2 in wastewater as predictors of the number of reported COVID-19 cases in Sweden. We used weekly SARS-CoV-2 wastewater virus-levels collected by the Swedish University of Agricultural Sciences, including 18 wastewater treatment plants (WTP) in 14 of 21 regions, covering 11-52% of respective regional population. Regional weekly number of COVID-19 cases between July 24, 2023 and January 15, 2024 was obtained from the national surveillance system at the Public Health Agency of Sweden.For each WTP, we fitted four Poisson models to weekly COVID-19 cases. The models included virus-levels from either five or six weeks and lagged either one or two weeks with respect to cases. Each model predicted the number of cases one week ahead, with 95% prediction interval (PI) for a 20-week period. To assess the accuracy of the predictions, we calculated the proportion of weeks when the 95%PI included the observed number of COVID-19 cases (95%PI coverage) and mean square prediction error (MSPE).

Results : For 15 WTP, the best model for predicting COVID-19 cases included virus-levels from five weeks lagged two weeks, as they had highest 95%PI coverage or the lowest MSPE. For these models, 95%PI coverage varied between 15-50%, median 25%.

Conclusions : The low 95%PI coverage indicates a low predictive power, suggesting that wastewater virus-levels cannot accurately predict the number of COVID-19 cases one or two weeks ahead. Other approaches should be investigated when assessing the potential of wastewater virus-levels as an early warning indicator in real-time surveillance of COVID-19 in Sweden.