In a recent study published in PNAS, researchers develop and test a novel algorithm to detect coronavirus disease 2019 (COVID-19) surges before they occur. This model, which is otherwise known as Covid-SURGE, was associated with a true positive rate of over 80%.
Study: Separating signal from noise in wastewater data: An algorithm to identify community-level COVID-19 surges in real time. Image Credit: ImageFlow / Shutterstock.com
COVID-19 and wastewater
Throughout the COVID-19 pandemic, researchers have utilized wastewater sampling and analysis to monitor the transmission of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). This approach overcomes some of the limitations associated with clinical evaluation, including differential access to diagnostic tests and healthcare availability. Wastewater sampling is also a cost-effective method to detect community outbreaks and identify emergent SARS-CoV-2 variants.
Wastewater sampling allows for an entire community to be assessed, including asymptomatic individuals and those who test for COVID-19 at home without reporting their results to local authorities. However, certain factors, including laboratory analysis methods and sewer design, have resulted in a latency ranging from two days to three weeks between the initiation and completion of a process.
About the study
In the present study, researchers collected and analyzed data from the North Carolina Wastewater Monitoring Network (NCWMN) between January 2021 and March 2022. This data, which comprised 1,783 wastewater samples, was used to screen and train the Signaling Unprecedented Rises in Groupwide Exposure (Covid-SURGE) model.
The 1,783 water samples collected in this study were obtained from 19 wastewater treatment plants (WWTP) in 16 counties. Data on county- and sewer shed-level COVID-19 prevalence was also obtained during the study period. Wastewater viral concentrations were correlated with county- and sewer-level case counts, the latter of which were found to be more reliable indicators of viral surges.
Differences in the WWTP characteristics, including water flow rate and contributing population size, were used to standardize findings and evaluate the model’s sensitivity in identifying surges at different population scales. Spatial and temporal wastewater metrics preceding and during the Delta and Omicron SARS-CoV-2 surges were also assessed.
Study findings
Comparisons between wastewater and clinical data indicated that wastewater data is a leading indicator of COVID-19 surges. The lag period varied from WWTP to WWTP, with an average of seven days to a maximum of 12 days across sites. When restricting the analyses to data obtained during Delta and Omicron surges, correlations between wastewater viral load and case data were observed at a rate of 84% and 85% across sites, respectively.
The effect of sample size was an important predictor of correlation strength, with large WWTPs servicing a larger population size depicting higher correlations than their smaller counterparts.
Comparing wastewater metrics with clinical data during Delta and Omicron surges showed that viral concentrations in wastewater rapidly increased during the viral surge. This correlation was more robust during the Omicron surge than the Delta surge; however, both were susceptible to small sites being lower than the limit of detection (LOD).
Ten days before the start of the Delta surge, viral loads of 66%, which reflected 1.4 million viral copies for each person, were observed. At the start of the Delta surge, viral loads increased to 507% or 5.9 million viral copies for each person at the start. Likewise, Omicron viral loads increased from 96% to 237% after the start of that COVID-19 wave.
The Covid-SURGE algorithm was trained and validated using case-confirmed surge start dates. Out of the 19 counties samples, Covid-SURGE successfully identified the start of the Delta surge in 16 counties and Omicron in 15 counties.
The true positive rate of Covid-SURGE was 82%, with a false positive rate of only 7%. In large WWTPs, the model’s accuracy rose to almost 100%, thus highlighting the benefits of larger datasets in model sensitivity.
To generalize the algorithm’s predictive ability, the researchers tested Delta surge data from samples obtained in seven other states by the United States Centers for Disease Control and Prevention (CDC) National Wastewater Surveillance System (NWSS). To this end, Covid-SURGE accurately predicted the start of the Delta surge with 71-75% accuracy.
Conclusions
In the present study, researchers develop and evaluate a logical algorithm to distinguish between SARS-CoV-2 signal and noise in viral load data collected from wastewater samples. Covid-SURGE successfully predicted COVID-19 surges with an accuracy between 71-82% as early as four to five days before the start of the surge.
The findings from our analysis of data from North Carolina’s statewide wastewater monitoring program validate the use of wastewater monitoring to assess patterns in community transmission of COVID-19. Temporal trends in wastewater viral concentrations aligned well with trends in reported case counts, and wastewater data captured the differing trajectories of the Delta and Omicron surges.”
Analyzing the viral loads of over 1,700 wastewater samples from different sites allowed the researchers to determine that wastewater viral concentrations are leading indicators of impending COVID-19 surges. Given that this tool can identify high risk counties before clinical testing indicates rising cases, Covid-SURGE can be implemented as an early warning tool to mitigate the spread of COVID-19.
Similar models developed for other diseases like flu-SURGE can also provide clinicians and future policy-makers with important data needed to limit the spread of virulent infections and prevent future epidemics.
One in four healthy young Marines report long-term physical, cognitive, or psychiatric effects after mild COVID-19