Time-dependent drivers explain correspondence between wastewater and clinical COVID-19 data

Abstract

Wastewater-based epidemiology has garnered increasing attention during the COVID-19 pandemic due to its potential for accurate and cost-effective population-level surveillance. In this study, we analyzed wastewater samples collected from six wastewater treatment plants in Tuscany, Italy, between April 2022 and March 2023. We compared SARS-CoV-2 RNA concentrations in wastewater with the number of positive COVID-19 tests provided by the Italian Ministry of Health and observed significant discrepancies between the two throughout the whole time window considered, with viral load ranging from 4 up to 8 orders of magnitude higher that clinical tests. These inconsistencies tend to increase with time by 1-2 orders of magnitude. To investigate the underlying causes of these discrepancies, we developed a Generalized Additive Mixed Model incorporating both clinical testing intensity (using the number of tests performed and the positivity ratio as proxies for testing accuracy) and viral subvariant prevalence. Our results indicate that variations in clinical testing intensity introduce changes in the relationship between their estimates and the wastewater-based time series, with an effect that is more than double the impact of Omicron subvariants. Shifts in viral subvariants produce systematic changes in the wastewater signal with an effect more than double the one of clinical tests. When not taken properly into account, they effectively act as a bias in the relationship between measured concentrations and case numbers.