Wastewater Surveillance Research Group
Background
The measured amount of SARS-CoV-2 (the virus causing COVID-19) in wastewater can change a lot from day to day, similar to traditional surveillance clinical metrics such as case counts and hospital admissions, which makes the data challenging for public health officials to interpret. We tested a method called Bayesian smoothing and forecasting, which has been used before to track and predict COVID-19 cases, hospitalizations, and deaths.
Methods
We used data from the municipal wastewater treatment plant in Ottawa, Canada, collected from July 1, 2020, to February 15, 2022. We checked how well this method could average out the measurements and predict future amounts of the virus. We also tested this method with data from 15 other wastewater treatment plants in Ontario.
Results
When we plotted the virus levels in wastewater over time using Bayesian smoothing, we could clearly see the different waves of COVID-19 that matched the number of cases and hospitalizations. This was true both for Ottawa and the other 15 Ontario communities. The method’s predictions for virus levels one week ahead were close to what was actually observed from December 23, 2020, to August 8, 2022. Initially, the model underpredicted the virus levels during fast increases and overpredicted during fast decreases. After making some adjustments, the predictions closely matched the observed data.
Conclusion
Using Bayesian smoothing on wastewater data gives accurate estimates of COVID-19 growth rates and can forecast virus levels one and two weeks ahead for 16 communities in Ontario, Canada. This method should be tested further in other areas with different sewage systems and environmental conditions.