Separating Signal from Noise in Wastewater Data: An Algorithm to Identify Community-Level COVID-19 Surges

Abstract

Wastewater monitoring has shown promise in providing an early warning for new COVID-19 outbreaks, but to date, no approach has been validated to reliably distinguish signal from noise in wastewater data and thereby alert officials to when the data show a need for heightened public health response. We analyzed 62 weeks of data from 19 sites participating in the North Carolina Wastewater Monitoring Network to characterize wastewater metrics before and around the Delta and Omicron surges. We found that, on average, wastewater data identified new outbreaks four to five days before case data (reported based on the earlier of the symptom start date or test collection date). At most sites, correlations between wastewater and case data were similar regardless of how wastewater concentrations were normalized, and correlations were slightly stronger with county-level cases than sewershed-level cases, suggesting that officials may not need to geospatially align case data with sewershed boundaries to gain insights into disease transmission. Wastewater trend lines showed clear differences in the Delta versus Omicron surge trajectories, but no single wastewater metric (detectability, percent change, or flow-population normalized viral concentrations) adequately indicated when these surges started. After iteratively examining different combinations of these three metrics, we developed a simple algorithm that identifies unprecedented signals in the wastewater to help clarify changes in communities' burden. Our novel algorithm accurately identified the start of both the Delta and Omicron surges in 84% of sites, potentially providing public health officials with an automated way to flag community-level COVID-19 surges.

Identifying trends in SARS-CoV-2 RNA in wastewater to infer changing COVID-19 incidence: Effect of sampling frequency

“…Specifically, the CDC classifies trends into five categories based on the magnitude of the PC estimate over the last 15 days: large decrease (−100% or less), decrease (−99% to - 10%), stable (−9% to 9%), increase (10% to 99%), and large increase (100% or more) [52]. However, categorizing PC trends based on magnitude alone can be misleading when wastewater SARS-CoV-2 concentrations are low or around the limit of detection [17]. If public health departments want to report both the significance and magnitude of trends, the criteria in Table 1 could be modified such that the value of the PC or tau estimate must be both significant and above or below a certain threshold.…”

“…A recent study suggests that wastewater concentrations of SARS-CoV-2 RNA are also well-correlated with COVID-19 prevalence, which was estimated from randomized nasal swab sampling, more so than incident clinical cases because case data are prone to reporting biases such as underreporting of asymptomatic cases [9]. In addition, trends in wastewater concentrations of SARS-CoV-2 RNA were found to precede trends in incident clinical cases in communities [10][11][12][13][14][15][16][17]. WBE may therefore be a more reliable and objective tool than incident clinical case data for continued monitoring of COVID-19 because wastewater captures both symptomatic and asymptomatic individuals and does not depend on test seeking behavior or testing availability.…”

Identifying trends in SARS-CoV-2 RNA in wastewater to infer changing COVID-19 incidence: Effect of sampling frequency

“…Specifically, the CDC classifies trends into five categories based on the magnitude of the PC estimate over the last 15 days: large decrease (-100% or less), decrease (-99% to -10%), stable (-9% to 9%), increase (10% to 99%), and large increase (100% or more) [58]. However, categorizing PC trends based on magnitude alone can be misleading when wastewater SARS-CoV-2 concentrations are low or around the limit of detection [17]. If public health departments want to report both the significance and magnitude of trends, the criteria in Table 1 could be modified such that the value of the PC or tau estimate must be both significant and above or below a certain threshold.…”

“…A recent study suggests that wastewater concentrations of SARS-CoV-2 RNA are also well-correlated with COVID-19 prevalence, which was estimated from randomized nasal swab sampling, more so than incident clinical cases because case data are prone to reporting biases such as underreporting of asymptomatic cases [9]. In addition, trends in wastewater concentrations of SARS-CoV-2 RNA were found to precede trends in incident clinical cases in communities [10][11][12][13][14][15][16][17]. WBE may therefore be a more reliable and objective tool than incident clinical case data for continued monitoring of COVID-19 because wastewater captures both symptomatic and asymptomatic individuals and does not depend on test seeking behavior or testing availability.…”

“…Yet there has been limited work on how to actively monitor trends in epidemiology [21]. Common trend metrics (e.g., simple moving averages, rates of change) provide little attention to the statistical significance and stability of trends and can be misleading or confusing [17,21,22]. Standardized trend analysis methods that are robust and easily interpretable are needed to appropriately identify trends.…”