COVID-19 wastewater epidemiology: a model to estimate infected populations

Abstract

Background: Wastewater-based epidemiology provides an opportunity for near real-time, cost-effective monitoring of community-level transmission of SARS-CoV-2. Detection of SARS-CoV-2 RNA in wastewater can identify the presence of COVID-19 in the community, but methods for estimating the numbers of infected individuals on the basis of wastewater RNA concentrations are inadequate.

Methods: This is a wastewater-based epidemiology study using wastewater samples that were collected weekly or twice a week from three sewersheds in South Carolina, USA, between either May 27 or June 16, 2020, and Aug 25, 2020, and tested for SARS-CoV-2 RNA. We developed a susceptible-exposed-infectious-recovered (SEIR) model based on the mass rate of SARS-CoV-2 RNA in the wastewater to predict the number of infected individuals, and have also provided a simplified equation to predict this. Model predictions were compared with the number of confirmed cases identified by the Department of Health and Environmental Control, South Carolina, USA, for the same time period and geographical area.

Findings: We plotted the model predictions for the relationship between mass rate of virus release and numbers of infected individuals, and we validated this prediction on the basis of estimated prevalence from individual testing. A simplified equation to estimate the number of infected individuals fell within the 95% confidence limits of the model. The rate of unreported COVID-19 cases, as estimated by the model, was approximately 11 times that of confirmed cases (ie, ratio of estimated infections for every confirmed case of 10·9, 95% CI 4·2-17·5). This rate aligned well with an independent estimate of 15 infections for every confirmed case in the US state of South Carolina.

Interpretation: The SEIR model provides a robust method to estimate the total number of infected individuals in a sewershed on the basis of the mass rate of RNA copies released per day. This approach overcomes some of the limitations associated with individual testing campaigns and thereby provides an additional tool that can be used to inform policy decisions.

Funding: Clemson University, USA.

Figure 1

Sewersheds under surveillance for SARS-CoV-2 in wastewater The 29631 ZIP code area overlaps mainly with the Cochran Road and Pendleton–Clemson sewersheds. The Clemson University sewershed encompasses the campus and a small residential area adjacent to the campus.

Figure 2

The susceptible-exposed-infectious-recovered model (A) Proportions of the population that are susceptible to SARS-CoV-2 infection, exposed, infectious, and recovered. (B) Model predictions for mass rate of SARS-CoV-2 RNA in wastewater over time. Individual black points represent each Monte Carlo simulation. (C) Predictions of the number of infections versus RNA mass rate. Individual grey points represent each simulation, with the median, 75% CI, and 95% CI shown. Coloured datapoints correspond to measured RNA mass rates (table 1) and estimates of infected individuals based on equation 10 and estimated positive cases (n=320), assuming that 2% of the population was infected. The green rectangle represents the average RNA mass rates for July 16, 2020, to Aug 18, 2020, (table 1) versus the 320 positive cases.

Figure 3

COVID-19 cases predicted by the SEIR model compared with SCDEHC cases after correction for under-reporting SEIR model predictions of active COVID-19 cases in the 29631 ZIP code area based on RNA mass rates in wastewater compared with the number of cases confirmed by SCDEHC and corrected for under-reporting using an estimated ratio of ten actual cases to every nine cases confirmed by testing. Individual grey points represent each simulation. The 1:1 ratio represents a perfect match between the model and active cases. SCDEHC=South Carolina Department of Health and Environmental Control. SEIR=susceptible-exposed-infectious-recovered.