Predicting hospital admissions due to COVID-19 in Denmark using wastewater-based surveillance

Abstract

Wastewater surveillance has become a fundamental tool to monitor the circulation of SARS-CoV-2 in order to prepare timely public health responses. In this study we integrate available clinical data on hospital admissions with wastewater surveillance data and investigate if predictions of the number of hospital admissions due to COVID-19 in Danish hospitals are improved by including wastewater concentrations of SARS-CoV-2. We implement state space models to describe the relationship between the number of hospital admissions due to COVID-19, available with a three-week classification delay, and more recent numbers of total hospital admissions with COVID-19. Including wastewater concentrations of SARS-CoV-2, we consider five-week predictions of the number of hospital admissions due to COVID-19. As a result of the three-week classification delay, the predictions translate into two hindcasts, one nowcast and two forecasts. The predicted values for all time frames follow the observed numbers well. We find that log likelihood values are higher when including wastewater concentrations (across all horizons) and that lagging the wastewater observations to investigate whether changes in wastewater concentrations occur before changes in hospital admissions does not result in further improvements. Our study shows that including wastewater concentrations improve estimates of the number of hospital admissions due to COVID-19, implying that wastewater concentrations add valuable information about the underlying transmission and that the imminent development of the near-future disease burden from COVID-19 is better informed when carefully including wastewater concentrations.