Use of wastewater metrics to track COVID-19 in the U.S.: a national time-series analysis over the first three quarters of 2022

Abstract

Background: Widespread use of at-home COVID-19 tests hampers determination of community COVID-19 incidence. Using nationwide data available through the US National Wastewater Surveillance System, we examined the performance of two wastewater metrics in predicting high case and hospitalizations rates both before and after widespread use of at-home tests.

Methods: We performed area under the receiver operating characteristic (ROC) curve analysis (AUC) for two wastewater metrics-viral concentration relative to the peak of January 2022 ("wastewater percentile") and 15-day percent change in SARS-CoV-2 ("percent change"). Dichotomized reported cases (≥ 200 or <200 cases per 100,000) and new hospitalizations (≥ 10 or <10 per 100,000) were our dependent variables, stratified by calendar quarter. Using logistic regression, we assessed the performance of combining wastewater metrics.

Results: Among 268 counties across 22 states, wastewater percentile detected high reported case and hospitalizations rates in the first quarter of 2022 (AUC 0.95 and 0.86 respectively) whereas the percent change did not (AUC 0.54 and 0.49 respectively). A wastewater percentile of 51% maximized sensitivity (0.93) and specificity (0.82) for detecting high case rates. A model inclusive of both metrics performed no better than using wastewater percentile alone. The predictive capability of wastewater percentile declined over time (AUC 0.84 and 0.72 for cases for second and third quarters of 2022).

Conclusion: Nationwide, county wastewater levels above 51% relative to the historic peak predicted high COVID rates and hospitalization in the first quarter of 2022, but performed less well in subsequent quarters. Decline over time in predictive performance of this metric likely reflects underreporting of cases, reduced testing, and possibly lower virulence of infection due to vaccines and treatments.

Figure 1.. U.S. counties submitting wastewater surveillance data to the NWSS between January 1 2022 and September 30 2022.

Regions mapped in dark blue show counties included in the analysis (n=268). Regions mapped in light blue show counties that submitted the analyzed metrics to NWSS, but did not meet inclusion criteria and were excluded from analysis (n=462).

Figure 2.. Time history of wastewater surveillance data and clinical case metrics from the most populous counties in each US Census region between January 2022 and September 2022.

Data are shown for the most populous counties in U.S. Census regions (A) West: Los Angeles County, California, (B) South: Harris County, Texas, (C) Midwest: Wayne County, Michigan, and (D) Northeast: Onondaga County, New York. Panel (i) shows smoothed spline-fit PCR concentrations of SARS-CoV-2 for each sampling location as reported by the CDC NWSS. When multiple sewersheds were sampled within a county, dashed grey lines in panel (i) represent individual sewersheds. The solid black lines in panels (i), (ii) wastewater SARS-CoV-2 percentile level, and (iii) wastewater SARS-CoV-2 15-day percent change show weighted mean values using each sewershed’s population served. Horizontal dashed lines in panels (iv) and (v) show thresholds for high COVID-19 community level (reported COVID-19 case rate equal to or greater than 200 per 100,000 population and reported hospitalization rate equal to or greater than 10 new inpatient admissions per 100,000 population, respectively). Panel (vi) shows state-level data (solid black lines show reported tests from the state of California (A), Texas (B), Michigan (C), and New York (D); dashed grey lines show estimates for all other U.S. states).

Figure 3.. Performance of wastewater percentile in reference to clinical case metrics stratified by calendar quartile of 2022.

AUC of wastewater percentile in reference to (A) Current reported COVID-19 cases (≥200 per 100,000 population), (B) New hospital admissions in two weeks (≥10 per 100,000 population).