The COVID-19 pandemic, caused by SARS-CoV-2, highlighted the need for robust surveillance methods. Wastewater-based epidemiology (WBE) emerged as a valuable tool, tracking viral RNA in sewage to monitor community prevalence. However, RNA-based surveillance has limitations. This study proposes to enhance WBE by incorporating chemical biomarkers—specifically, COVID-19 treatment drugs and over-the-counter (OTC) medications—to gain a more comprehensive understanding of community response to the pandemic. The presence of SARS-CoV-2 RNA in wastewater, resulting from fecal shedding, provides a valuable signal for community-wide monitoring. WBE offers near real-time surveillance and respects individual privacy. While challenges exist, including the stability of certain biomarkers in wastewater and variations in population contribution, WBE serves as a valuable complement to existing surveillance techniques. Numerous studies have demonstrated the correlation between SARS-CoV-2 RNA levels in wastewater and reported COVID-19 cases, highlighting WBE's potential as an early warning system. However, most studies are qualitative, focusing on correlations and lead times rather than quantitative predictions of case numbers. Existing models for quantitative prediction either have short forecasting periods or large prediction errors. The inclusion of additional biomarkers, such as OTC drugs and COVID-19 treatment medications, is proposed to improve the accuracy and predictive capabilities of these models. Individuals experiencing COVID-19 symptoms often self-medicate with OTC drugs, while severe cases receive treatment drugs in hospitals. These pharmaceuticals are excreted and enter the wastewater system, providing additional data points for monitoring community health and disease spread. This study aims to identify complementary chemical biomarkers correlating with SARS-CoV-2 RNA levels, offering potentially more stable and easily measurable indicators compared to viral RNA itself.

Existing literature extensively documents the use of wastewater-based epidemiology (WBE) for tracking various substances, including illicit drugs and pathogens, to understand community-level health and behavior. WBE applications for COVID-19 surveillance have shown promising results, correlating viral RNA in wastewater with reported cases. However, these studies largely focus on qualitative correlations and lead times, lacking a formal quantitative link between biomarker concentrations and confirmed case counts. The challenge lies in the limited stability of some biochemical indicators in sewage, alongside uncertainties concerning wastewater flows and daily population variations. While some studies attempted quantitative prediction using WBE data and models like SEIR, the prediction accuracy and forecasting periods remain limited. To overcome these shortcomings, this study integrates data from COVID-19 treatment drugs and OTC drugs, aiming to enhance the accuracy of community-level prevalence prediction.

Wastewater samples (n=183) were collected from Suffolk County, NY's largest wastewater treatment plant, serving approximately 330,000 people. Samples were collected using an autosampler, creating 24-hour composite samples. Viral RNA analysis was performed immediately using a Qiagen QIAamp DSP viral RNA mini kit and digital PCR. For drug analysis, samples were filtered (1 µm glass fiber), aliquoted, and stored at -80°C. Solid-phase extraction (SPE) was used to concentrate COVID-19 treatment drugs before analysis by LC-MS/MS, while other pharmaceuticals were measured by direct injection after dilution. A stability experiment assessed the degradation of COVID-19 treatment and OTC drugs at different temperatures (4°C, 12°C, and 20°C) over 24 hours. Confirmed COVID-19 case data were obtained from the Suffolk County Department of Health, while hospitalization and drug prescription data came from Stony Brook University Hospital. Caffeine was used for population normalization, accounting for variations in wastewater flow. Bayesian models were developed to predict confirmed cases, incorporating log-transformed viral RNA data and rescaled pharmaceutical concentrations. The models used a binomial distribution for confirmed cases and logistic regression for the probability of infection. Model selection was based on the Watanabe-Akaike Information Criterion (WAIC).

SARS-CoV-2 RNA levels strongly correlated with confirmed COVID-19 cases (R=0.81, p<0.01), exhibiting a 3-4 day lead time. COVID-19 treatment drugs had varying detection frequencies due to wastewater instability; however, desethylhydroxychloroquine, a hydroxychloroquine metabolite, showed a significant correlation with SARS-CoV-2 genes (R=0.30, p<0.01). Remdesivir levels peaked immediately following FDA authorization. Among the 26 pharmaceuticals analyzed, acetaminophen exhibited a strong positive correlation with viral loads and confirmed cases, especially during the Omicron variant's dominance. This correlation showed a short-to-nonexistent lead time (0-2 days). The Bayesian models successfully reproduced the temporal trends of confirmed cases, with improved accuracy when including acetaminophen and desethylhydroxychloroquine. Model validation using an independent dataset of viral concentrations from January to May 2022 showed that the model accurately captured the confirmed case trend, except for slight discrepancies around the peak.

This study demonstrates the feasibility and value of integrating chemical biomarkers into wastewater surveillance for SARS-CoV-2. While COVID-19 treatment drugs proved less stable in wastewater, the inclusion of acetaminophen and desethylhydroxychloroquine greatly enhanced the prediction accuracy of COVID-19 cases. Acetaminophen's strong correlation with both viral loads and confirmed cases, especially during Omicron, suggests its potential as a robust and readily measurable indicator of COVID-19 prevalence in communities. The short lead time associated with acetaminophen aligns with the typical progression of COVID-19 symptoms, potentially providing a quicker indication of outbreaks than RNA surveillance alone. The use of Bayesian modeling offers a robust and adaptable framework for forecasting, continuously improving predictions with new data. The study findings suggest that combining traditional viral RNA monitoring with complementary chemical biomarkers provides a more comprehensive and accurate assessment of community-level COVID-19 dynamics.

This study successfully integrated chemical biomarkers, specifically acetaminophen, into existing wastewater-based epidemiology workflows for COVID-19 surveillance. The findings demonstrate the value of this approach for enhancing prediction accuracy and providing timely information on community-level disease prevalence. Acetaminophen's stability and correlation with COVID-19 cases highlight its potential as a supplementary marker for future outbreaks. Future research could explore the application of this combined approach to other viral diseases and investigate other relevant OTC drugs or metabolites for improved monitoring capabilities. The Bayesian modeling framework offers a robust foundation for continuous refinement and improvement of predictive models.

The study's main limitation is the potential instability of certain chemical biomarkers in wastewater, particularly COVID-19 treatment drugs. The analysis focused on filtered wastewater for viral RNA extraction, potentially underestimating the overall viral load. The data on hospitalization and drug prescriptions were sourced from a nearby hospital, which may not perfectly represent the entire catchment area. Future studies could address these issues by optimizing sample collection and analysis methods, and by expanding the data sources to ensure a more comprehensive representation of the community.