Modeling Biases in SARS-CoV-2 infections Prediction using Genome Copies Concentration in Wastewater

Abstract

AbstractBackgroundSARS-CoV-2, the virus responsible for the COVID-19 pandemic, can be detected in stool samples and subsequently shed in the sewage system. The field of Wastewater-based epidemiology (WBE) aims to use this valuable source of data for epidemiological surveillance, as it has the potential to identify unreported infections and to anticipate the need for diagnostic tests.ObjectivesThe objectives of this study were to analyze the absolute concentration of genome copies of SARS-CoV-2 shed in Catalonia’s wastewater during the Omicron peak in January 2022, and to develop a mathematical model capable of using wastewater data to estimate the actual number of infections and the temporal relationship between reported and unreported infections.MethodsWe collected twenty-four-hour composite 1-liter samples of wastewater from 16 wastewater treatment plants (WWTPs) in Catalonia on a weekly basis. We incorporated this data into a compartmental epidemiological model that distinguishes between reported and unreported infections and uses a convolution process to estimate the genome copies shed in sewage.ResultsThe 16 WWTPs showed an average correlation of 0.88±0.08 (ranging from 0.96 to 0.71) and an average delay of 8.7±5.4 days (ranging from 0 to 20 days). Our model estimates that about 53% of the population in our study had been infected during the period under investigation, compared to the 19% of cases that were detected. This under-reporting was especially high between November and December 2021, with values up to 10. Our model also allowed us to estimate the maximum quantity of genome copies shed in a gram of feces by an infected individual, which ranged from 4.15×107gc/gto 1.33×108gc/g.DiscussionAlthough wastewater data can be affected by uncertainties and may be subject to fluctuations, it can provide useful insights into the current trend of an epidemic. As a complementary tool, WBE can help account for unreported infections and anticipate the need for diagnostic tests, particularly when testing rates are affected by human behavior-related biases.