key: cord-0960987-k4q6xx3b
authors: Fitzgerald, Stephen F.; Rossi, Gianluigi; Low, Alison S.; McAteer, Sean P.; O’Keefe, Brian; Findlay, David; Cameron, Graeme J.; Pollard, Peter; Singleton, Peter T. R.; Ponton, George; Singer, Andrew C.; Farkas, Kata; Jones, Davey; Graham, David W.; Quintela-Baluja, Marcos; Tait-Burkard, Christine; Gally, David L.; Kao, Rowland; Corbishley, Alexander
title: Site Specific Relationships between COVID-19 Cases and SARS-CoV-2 Viral Load in Wastewater Treatment Plant Influent
date: 2021-11-05
journal: Environ Sci Technol
DOI: 10.1021/acs.est.1c05029
sha: 168de2d28fac9038e4a69db2290536484955628d
doc_id: 960987
cord_uid: k4q6xx3b

[Image: see text] Wastewater based epidemiology (WBE) has become an important tool during the COVID-19 pandemic, however the relationship between SARS-CoV-2 RNA in wastewater treatment plant influent (WWTP) and cases in the community is not well-defined. We report here the development of a national WBE program across 28 WWTPs serving 50% of the population of Scotland, including large conurbations, as well as low-density rural and remote island communities. For each WWTP catchment area, we quantified spatial and temporal relationships between SARS-CoV-2 RNA in wastewater and COVID-19 cases. Daily WWTP SARS-CoV-2 influent viral RNA load, calculated using daily influent flow rates, had the strongest correlation (ρ > 0.9) with COVID-19 cases within a catchment. As the incidence of COVID-19 cases within a community increased, a linear relationship emerged between cases and influent viral RNA load. There were significant differences between WWTPs in their capacity to predict case numbers based on influent viral RNA load, with the limit of detection ranging from 25 cases for larger plants to a single case in smaller plants. SARS-CoV-2 viral RNA load can be used to predict the number of cases detected in the WWTP catchment area, with a clear statistically significant relationship observed above site-specific case thresholds.

The COVID-19 pandemic has necessitated the rapid implementation of surveillance programs worldwide to track and control the spread of SARS-CoV-2 (the coronavirus that causes the disease syndrome known as . Initially, such programs relied on syndromic surveillance, community testing, contact tracing, and the monitoring of morbidity and mortality rates. 1−3 Community testing relies on voluntary reporting of clinical signs and is only partially able to capture the presymptomatic, asymptomatic, and pauci-symptomatic cases of SARS-CoV-2 infection that can contribute significantly to community transmission, and are therefore subject to biases, which can influence estimates of disease burden. 1,2 Syndromic surveillance based on hospital admissions is less biased, but is subject to delays between infection and admission, 2 while implementing mass swab-testing on a nationally meaningful scale is not economically feasible for most countries. 2 Early studies identified SARS-CoV-2 RNA in the feces of infected individuals and COVID-19 has subsequently been associated with a range of gastrointestinal symptoms. 4 

CoV-2 has been detected in feces from both asymptomatic and symptomatic individuals, with prolonged shedding observed up to 33 days after the initial onset of symptoms or hospitalization. 1, 4, 5 Consequentially, wastewater-based epidemiology (WBE) has been explored as a tool to track the spread of SARS-CoV-2 by many countries. 1 Early in the pandemic, Medema et al. 6 detected SARS-CoV-2 RNA in the wastewater of three Dutch cities and a major airport up to 6 days before the first reported clinical cases. 6 Since then, WBE programs have been started by over 50 countries, 1, 7, 8 however a number of important questions remain relating to the implementation of these programs and the interpretation of WBE data. These include the impact of viral shedding dynamics in feces, viral persistence in wastewater and wastewater flow rates on viral detection in wastewater, whether differences exist between urban and rural wastewater systems and how viral levels in wastewater should be normalized with respect to population size. 2 Furthermore, there are a range of techniques available for detecting viruses in wastewater, while wastewater samples are diverse with respect to their physicochemical composition. There is therefore a need to determine which methodologies and process controls are appropriate when operationalizing WBE at a national scale. 2 This study describes the development and implementation of a national WBE SARS-CoV-2 surveillance program. We compared and optimized commonly used viral concentration techniques, validated Porcine Respiratory and Reproductive virus (PRRSv) as a suitable process control and optimized RT-qPCR assays for SARS-CoV-2 detection in wastewater. This methodology was adopted by the Scottish Environment Protection Agency (SEPA) and has been used to routinely monitor viral levels at 28 wastewater treatment plant (WWTP) sites across Scotland, serving 50% of the Scottish population (2.66 million people). These sites include large conurbations, as well as low-density rural and remote island communities.

We demonstrate that daily SARS-CoV-2 viral RNA load can be used to predict the number of cases detected in the WWTP catchment area, with a clear statistically significant relationship observed between these two variables above site-specific case thresholds.

■ METHODS WWTP Site Selection. WWTP monitoring sites were selected by Scottish Water and SEPA to represent at least 50% of the population in each Scottish health board area (Table  S2 .1 of the Supporting Information, SI), using the minimum number of sites possible.

Wastewater Sample Collection. WWTP influent was collected at most sites using a refrigerated autosampler that obtained a fixed volume of influent every hour over each 24-h period (08:00 to 08:00). Refrigerated autosamplers at Dalbeattie, Allanfearn, Nigg, and Fort William obtained a fixed volume of influent, where the frequency of sampling over each 24-h period was dependent on the influent flow rate. Composite 24-h samples were mixed prior to analysis. Sites were typically sampled once a week, with increased frequency of sampling in response to increases in disease incidence in the community. There was no specific disease incidence threshold that was used to determine sampling frequency, however the local directors of public health were consulted, with sampling prioritized according to local needs. Due to resource limitations, any single site was not sampled more than four times a week. Samples were transported and stored at 4°C prior to analysis, typically within 24−48 h of collection.

Wastewater Concentration and Detection of SARS-CoV-2. Five viral concentration methods, Methods 1−5, based on filtration, precipitation, and adsorption were trialed (see the SI). Method 1 was further optimized by SEPA (Method 6) and used for routine wastewater monitoring. Viral RNA was extracted from concentrated wastewater samples using commercial kits (see SI). SARS-CoV2 was detected by RT-qPCR. During method development (April−May 2020), there was a national shortage of RT-qPCR reagents, with a number of suppliers providing contaminated oligonucleotides. Early experiments consequently relied on E gene detection, however once uncontaminated N1 gene reagents were available, performance of the E gene and N1 gene assays was compared using RNA extracted using multiple methods. Detection of the N1 gene was used during routine monitoring.

Data Collection. Two WWTP data sets were provided by SEPA via a publicly available portal. 9 The first data set reported sample date, location (WWTP name, coordinates, Health Board, and Local Authority), catchment area (CA) size (population band and population) and SARS-CoV-2 N1 and E gene average concentrations (gene copies/l). The second data set reported the daily WWTP influent flow (l/day) and three separate N1 gene technical replicates for each sample. All replicates (838 of 2967) not returning a detectable signal were marked as "negative" in the data set, and they were treated as zeros in the analyses. SEPA also provided the WWTP dry weather (i.e., licensed) flow (l/day) and Scottish Water provided the CA shapefiles for the 28 sites.

COVID-19 data in Scotland are collected by Public Health Scotland (PHS) and the data set used in this study reports the date and location of first COVID-19 tests and first positive tests (i.e., such that "positivity" is the proportion of individuals who test positive), with test results, and deaths, starting from March first, 2020. To protect patient anonymity, data were provided by PHS by "datazone", a small-scale geographic unit identified by the National Records of Scotland (NRS) containing approximately 500 to 1000 individuals. Each case was assigned to a datazone on the basis of the patients' reported address of residence, irrespective of where any treatment or testing was administered. Datazone size was set to avoid the need to mask any data to protect patient confidentiality, i.e., each datazone is large enough so that the identity of a case cannot be inferred from other publicly available information. Relevant shapefiles and population data were downloaded from the NRS portal, 10 facilitating a high resolution allocation of the number of tests, detected cases (i.e., positive tests), and COVID-19 related deaths for each of the CAs.

Data Analysis. The objective was to understand the association between the concentration and daily viral RNA load of SARS-CoV-2 RNA in WWTP influent and the number of detected cases in the corresponding CA. The daily WWTP influent viral RNA load was calculated by multiplying the wastewater sample viral RNA concentration with the total WWTP influent flow for the day of sampling. Since daily flow is not always available, SEPA included a flow estimate obtained with a linear regression model that considered ammonium concentration (provided by Scottish Water), catchment population, and site as independent variables (Roberts and Fang, private communication). Analyses were repeated using both reported flow rates and these estimates (see SI).

The number of detected cases and the positive test rate were calculated by counting the number of positive and total tests over the 7 days up to and including the day the sample was taken. We undertook a sensitivity analysis to test the effect of varying this time period from zero days, i.e., counting only the reported cases on the day of wastewater sample collection) to 28 days on our results (see SI).

First, a simple correlation between viral concentration or load and number of cases or positive test rate was calculated using Spearman's ρ rank correlation coefficient.

Further, to test the association between observed cases (Y i,j ) and daily WWTP viral RNA load (X i,j ), we fitted a basic linear mixed model 11 Environmental Science & Technology

where β 0 and β 1 represent the fixed intercept and coefficient of the daily WWTP viral RNA load X i,j . Parameters u j and b j are the random intercept and coefficient, associated with each group j (catchment), while ε i represents the error term. We used this model to allow both the intercept and the slope (i.e., the coefficient of the daily viral load) to be composed by a common and a group-specific part, therefore for each site j the final intercept and slope were, respectively, β 0 + u j and β 1 + b j . The addition of a random slope was verified with a χ 2 test. 12 Before the estimation, the dependent and independent variables were square root transformed, which was required to reduce the overdispersion of the distribution prior to linear mixed model analysis (the untransformed data are reported in Figure S2 .9, which shows the daily viral load average of the three sample replicates). We evaluated the model using the conditional pseudo-R 2 , which measures the variance explained by both fixed and random effects 12 and analyzed the resulting coefficients (intercept, β 0 + u j , and slope, β 1 + b j ) to assess the consistency of the signal and the potential causes of the differences between WWTPs. We first fitted a series of univariable linear regression models with the site's slope or intercept as the dependent variable and population, population density, number of wastewater samples, latitude, longitude, deprivation, and access indices 10 as independent variables. Deprivation and access indices measure the relative deprivation and the access to healthcare services respectively of a datazone. They were included as potential causes of bias in case detection. We then fitted a multivariable model to each coefficient, selecting as independent variables those returning a p-value below 0.2 in the univariable models. This threshold was chosen to allow the inclusion of variables not significant when considered in isolation, but potentially significant in a multivariable model. Variables were then further selected through a backward stepwise selection in order to eliminate the statistically insignificant ones, using the Akaike Information Criterion (AIC) for evaluation.

All data analyses were done in R 4.0.1, 13 using packages tidyverse 1.3.1, 14 scales 1.1.1, 15 and ggrepel 0.9.1 16 for data manipulation and representation, and packages lme4 1.1.27.1 17 and MuMIn 1.43.17 18 for the mixed model fit and evaluation.

Method Optimization and Detection of SARS-CoV-2 RNA in WWTP Influent. Reliable quantification of SARS-CoV-2 in wastewater requires consistent viral RNA extraction across a broad range of concentrations. To investigate this, aliquots of a single wastewater sample were spiked with a serial dilution of heat-inactivated SARS-CoV-2. There was no association between viral concentration and the efficiency of RNA recovery across 5 orders of magnitude of SARS-CoV-2 concentration (Figure S1.1.A). Significantly more SARS-CoV-2 (p > 0.0001) was recovered at a 10 −2 dilution, however there was no evidence that this anomaly was due to PCR inhibition, as no further increase in recovery was observed upon further sample dilution. As recovery across all other dilutions was comparable, we suggest this higher efficiency of recovery at 10 −2 was the result of handling error. We next validated PRRSv (a porcine enveloped nidovirus that can be cultured in vitro at Containment Level 2) as a suitable surrogate process control virus for SARS-CoV-2. The extraction efficiency of heat-inactivated SARS-CoV-2 was statistically significantly greater than either live PRRSv (p = 0.0348) or heat-inactivated PRRSv (p = 0.0056) (Figure S1.1.B) when spiked into a single wastewater sample, however it was within the same order of magnitude (approximately 1% vs 2%). Extraction efficiencies were also comparable between SARS-CoV-2 and heatinactivated PRRSv within wastewater samples from six individual WWTPs (p > 0.05) (Figure S1.1.C). Heatinactivated PRRSv was chosen as a process control for all subsequent testing.

Viral concentration methodologies based on filtration (Methods 1−3), PEG precipitation (Method 4), and adsorption (Method 5) were compared. The requirement to stir larger sample volumes for 8 h made the milk powder adsorption method insufficiently scalable and so it was excluded following initial pilot trials. PRRSv was recovered more efficiently by PEG precipitation than filtration from samples WWTP2 (p = 0.0162) and WWTP5 (p = 0.0382), however there was more variability between technical replicates observed using PEG precipitation (Figure S1.1.D). We compared liquid phase (influent and effluent) and solid phase (primary sludge and dewatered cake) samples for use in detection of SARS-CoV-2 RNA. Samples were taken weekly from a single plant, WWTP2, over a three-week period. The median recovery of PRRSv from influent was 20% across the 3week sample period (Figure S1.1.E), however SARS-CoV-2 RNA levels were below the limit of quantification (Figure S1.1.F).

SARS-CoV-2 RNA was detected in all primary sludge samples and 2/3 dewatered cake samples from WWTP2 despite poor recovery of PRRSv from both sludge (0.5−3.5%) and dewatered cake (0.2−0.8%). No SARS-CoV-2 RNA was detected in the effluent from WWTP2 (n = 3 technical replicates taken weekly over 3 consecutive weeks), however it should be noted that influent loading of SARS-CoV-2 RNA at WWTP2 during this time was close to the limit of detection and so the presence of SARS-CoV-2 in effluent at higher influent loads cannot be excluded. Although sludge and/or dewatered cake may be a more sensitive sample type for detection of SARS-CoV-2, 19 due to sampling difficulty and differences in sludge processing methods among WWTPs, influent samples were chosen for subsequent testing. Furthermore, some WWTPs treat sludge from other sites and hence sludge may not always be representative of the WWTP CA.

As Method 1 was both scalable and less variable for viral recovery efficiency than PEG precipitation, this method was selected to determine if SARS-CoV-2 RNA could be detected and quantified in wastewater collected from WWTPs in Scotland during the start of the pandemic. Influent samples from six wastewater treatment plants, WWTP1−WWTP6, were tested ( Figure S1 .2). Samples were taken on March 27 th , 2020, shortly before the first COVID-19 mortality peak in Scotland. A strong positive SARS-CoV-2 RNA signal of 18 000 genome equivalents per liter was detected in sample WWTP5 ( Figure S1.2.A) . SARS-CoV-2 RNA levels in each of the other five plants fell below our limit of quantification. Method 1 was further optimized by SEPA (Method 6; SI) and used for routine wastewater monitoring. Of note, detection of the N1gene by RT-qPCR was found to be more sensitive than the Egene ( Figure S1.2.B) and therefore N1-gene detection was adopted for the national program.

Environmental Science & Technology pubs.acs.org/est Article Data Analysis. The weekly number of SARS-CoV-2 reported cases, deaths, and positivity are shown in Figure  1A . As of 1/29/2021, 989 wastewater samples, with three technical replicates each, have been analyzed across 28 WWTPs, with the earliest samples taken from late May 2020 ( Figure 1B) . The number of samples per WWTP ranged from 12 (Stornoway, Outer Hebrides) to 112 (Shieldhall, Greater Glasgow). The CAs are distributed across Scotland ( Figure   1C ) and despite covering only 1.2% of Scotland's land mass, they cover 50% of the population. Daily WWTP influent flow data was missing for 18% of the samples.

As evident in Figure 2 , wastewater RNA viral concentration (panels A, C, and E) and daily WWTP viral RNA load (panels B, D, and F) mimic the trends of the daily positive test rate (number of positive tests over the total) and the daily Environmental Science & Technology pubs.acs.org/est Article incidence curves, respectively. This was independent of the CA population size ( Figure S2 .1 to S2.5 for remaining WWTPs). Preliminary correlation analyses between the WWTP daily viral RNA concentration and the number of COVID-19 cases detected in the CA in the previous week resulted in a Spearman's ρ = 0.79, while the correlation between WWTP viral concentration and positive test rate resulted in ρ = 0.83. Using the viral load (i.e., multiplying the concentration by the WWTP daily flow rate), the correlation improved for the number of cases, ρ = 0.91, while it decreased for the positive test rate, ρ = 0.77 (all p ∼ 0). This result was robust to the choice of the period length considered to calculate the number of cases or the positive test rate (see Figure S2 .7). In this case, the correlations improve as the number of contributing days for case counts before sampling increases from zero to five, at which point it stabilizes. The full mixed model explained 78% of the variance in the number of cases in the CA (conditional R 2 = 0·78), while the Environmental Science & Technology pubs.acs.org/est Article daily viral RNA load as a fixed effect (i.e., the component of the slope constant across all sites) explained 45% of the variance (marginal R 2 = 0·45). The null hypothesis that the sites' random slope variance was zero, which can be interpreted as the absence of significant differences between the cases-viral load relationship strength across sites, was rejected with a χ 2 test (p ∼ 0). The normality assumption about the distribution of model residuals was verified graphically (Figure S2 .10). When the model was rerun using a different time period to calculate the number of detected cases, the conditional R 2 ranged from 0.71 to 0.89, with an average of 0.76 across the 29 periods considered ( Figure S2.11 ). The mixed model fit by site is reported in Figure 3 (and Figure S2.12) . While the daily WWTP viral RNA load coefficients, or slope, are an indicator of the strength of the relationship between viral RNA load and cases, the intercept provides an estimate of the limit of detected cases in each CA.

The median [interquartile] estimated slope across sites was 5.2 × 10 6 [4.50−5.37 × 10 6 ] and was positive in all sites, including the confidence interval ( Figure 4A) ) . This translates to a threshold of less than one recorded case from which the relationship between viral RNA load and cases is detectable in small catchments, five recorded cases in the medium-sized catchments and 25 cases in the large catchments. Among the latter group, Dalmuir and Meadowhead were outliers, with higher intercept and lower slope compared with similar-sized catchments ( Figure 4C ).

The variables that best explain differences in mixed model slopes across WWTPs were the population size and the number of samples taken, although geographical longitude (not significant) was retained after multivariable model stepwise selection (Table 1 ). The CA population size and deprivation index were significant in explaining the differences in the mixed model intercepts (see Figure S2 .13 for single variable plots). 25 France, 26 and Spain. 27 Importantly, our work uniquely describes the establishment of a WBE program covering 50% of a country's population across a wide range of WWTP sizes. We demonstrate how WBE can be adopted across a range of catchments, from densely populated urban areas (Edinburgh and Glasgow), to smaller towns, rural areas and islands. We have used granular geospatial data to determine accurate estimates of recorded COVID-19 cases within each CA and demonstrate the existence of a strong and measurable statistically significant relationship between the SARS-CoV-2 daily WWTP viral RNA load and the number of detected cases in the week preceding wastewater sample collection. While the importance of using viral load, rather than viral concentration, has been demonstrated by other authors, 24 we have gone further to validate the use of ammonium concentration to calculate viral load when daily influent flow data is missing. We have also used granular geospatial and longitudinal data to characterize, in detail, the relationship between viral load and community cases over the month preceding sample collection.

In keeping with work examining levels of SARS-CoV-2 RNA in WWTP settled solids, 22 we show that the precision of the relationship between influent viral load and community cases varies between sites, with differences in the slope mostly attributed to the size of the population being served. Our results identified a stronger relationship between cases and viral RNA load in the larger WWTPs. Uniquely, we also explored the impact of population density, longitude, latitude, and deprivation and healthcare access indexes on the relationship between influent viral load and community cases. The identified threshold for detection was typically under 25 cases, and for some smaller WWTPs, a single detected community case was sufficient to yield a positive wastewater result. Compared to similar-sized WWTPs, Meadowhead and Dalmuir were outliers ( Figure 4C ); given their size, the slopes imply a poorer relationship between detected cases and WWTP daily viral RNA load, and intercepts a poorer sensitivity than expected. These WWTPs are defined by fragmented and highly dispersed CAs compared to most WWTPs of this size. Thus network architecture may be important, and subcatchment sampling may be necessary for large, fragmented, and/or dispersed networks. Deprivation also had a significant impact on the intercept, possibly due to differences in case reporting and/or viral RNA load per case, or the impact of higher industrial discharge. Combined, these factors meant that the limit of detection of cases per 100 000 population was highly variable between WWTPs: median-[interquartile] 9.2 [5.6−16.9] for the smaller sites, 19.8 [9.4− 31.9] for the medium-sized sites, and 10.8 [6.2−24.6] for the larger sites.

In contrast to most previous studies, 21, [25] [26] [27] 30 we demonstrate the value of obtaining flow measurements from WWTPs to calculate daily viral RNA loads, which display a stronger correlation with detected community case numbers, compared with viral concentration data alone ( Figure S2.7) . The daily influent flow is mostly affected by the weather and the WWTP size and, because of the latter, the correlation between flow and population connected to the WWTP sewage system is very strong (see Figure S2 .6). The improvement of the correlations and model performance observed when using the daily viral load suggest that, not only can this substitute for scaling the cases by the total population, but that it might include other effects (i.e., dilution or weather) which would remain hidden otherwise. Our Spearman's rank correlation ρ = 0.79 when not normalizing using the influent flow rate is almost identical to ρ = 0.73 reported by other authors, 23 who obtained mixed results when attempting to normalize using other methods, and serves to further highlight the utility of normalizing using influent flow rate.

Our typically low limits of detection show that wastewater surveillance can be particularly valuable for areas reaching low prevalence and is therefore suitable as a logistically sustainable early warning system, making a targeted community testing strategy viable. For WWTPs collecting wastewater from cities, it is harder to isolate small clusters of infections. This hurdle can be overcome by sampling a site "upstream" to the WWTP (i.e., within the sewerage network) to improve spatial resolution. This is currently taking place in Scotland, with local health boards using subcatchment wastewater sampling to direct surge testing. For smaller catchments, the size and the spatial resolution is already fine enough to inform community interventions, however a potential issue here is the variability in the signal. Specifically, we observed sudden spikes in the viral RNA load or viral concentration in many small WWTPs (Figure 2 , E and F; Figures S2.4 and S2.5) . While smaller catchments might be more sensitive to individual variations in shedding, these spikes might also be caused by one or two households being infected in a short period of time. Given the sensitivity of these smaller WWTPs to a small number of cases, this may explain these sudden variations in the SARS-CoV-2 daily viral RNA load. This also raises important questions with respect to the frequency of sampling, where it may be necessary to sample smaller sites more frequently to ensure that brief intense signals are not missed.

While we have shown that daily viral RNA load has the best correlation with detected cases ( Figure S2.7) , daily WWTP flow measurements are not always available. This may be more of a problem in smaller WWTPs, where flow rates regularly exceed the working range of the flow meter or in low resource settings, however our model retained substantial detection power when daily flow was estimated using easily obtained ammonium concentrations, with the conditional R 2 dropping by only 2% (R 2 = 0.76).

To better understand the relationship between WWTP viral RNA load and infected individuals, we need to consider the level of viral shedding in feces and how this varies over time. While SARS-CoV-2 RNA can be detected in the feces of hospitalized patients for over 4 weeks, 28, 29 our work and that of others 30 implies a relatively short period of time over which infected individuals substantially contribute to the wastewater signal. This was observed in two distinct sensitivity analyses, one on correlations and the other on mixed model performance (see SI). Specifically, the correlation between cases and viral RNA load (and between positive test rate and viral concentration) stabilizes once detected cases are included up to and including the 5 days prior to wastewater sampling. Furthermore, even with declining incidence, when the cumulative effect of older infections would be expected to have a greater contribution to the overall signal if shedding duration was long, the conditional R 2 of the mixed models did not deteriorate significantly (0.76 compared to 0.78 when incidence was increasing), and was consistent with a short period of peak viral shedding. Unfortunately, there is currently very limited data on fecal shedding of SARS-CoV-2 RNA in nonhospitalized individuals. Our understanding of the relationship between the WWTP viral RNA load and infected individuals is further complicated by the biases in community testing and movement (although restricted during lockdowns) of individuals between CAs. Specifically, testing of symptomatic individuals is unlikely to fully reflect the population incidence, with an analysis of English data suggesting that approximately 1 in 4 cases were being reported via community testing up to November 2020. 31 It is therefore likely that the model in this study underestimates the true prevalence of infection within the community. It is also possible that factors that have not been considered in this study, such as the degree of movement in and out of a CA, complicate the relationship between WWTP viral load and reported cases attributed to residents within the CA.

The value of our results extends beyond the first year of the COVID-19 pandemic. We have demonstrated how COVID-19 WBE can be implemented at a national scale across a diverse range of urban and remote communities. At the time of writing, this program has been expanded to cover 75% of the population of Scotland and is being used by local health boards to direct surge testing within the community. This program will continue to be important during the rollout of COVID-19 vaccinations, particularly with respect to disclosing areas of ongoing disease transmission and surveillance for novel SARS-CoV-2 variants. 32, 33 There is currently no data comparing the fecal shedding of SARS-CoV-2 RNA between different variants, however the lower Ct values observed in respiratory swabs from patients infected with variant B.1.617.2 (Delta) 34 imply that fecal shedding may also vary between some variants. It is possible that models that relate influent viral load to cases within the community may need to be adjusted in the future to account for the prevalence of specific variants within the population served by the WWTP CA. It also provides public health authorities with an unbiased surveillance network for other viral and bacterial infections, including antimicrobial resistance genes, shed in feces. Until the COVID-19 pandemic, WBE was predominantly limited to the surveillance of a narrow range of viruses (e.g., polio, norovirus, Hepatitis A/E) in low resource, sewered settings. 35−38 This study demonstrates the rapid inception, development, validation, and operationalization of a national COVID-19 WBE program to provide community surveillance during the pandemic.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.1c05029.

Additional experimental details, materials, methods and results, including the relationships between SARS-CoV-2 viral RNA concentration or load and test positivity or reported cases for each wastewater treatment plant included in the study (PDF) Alison S. Low − The Roslin Institute and Royal (Dick) School

Wastewater-Based Epidemiology (WBE) and Viral Detection in Polluted Surface Water: A Valuable Tool for COVID-19 Surveillance-A Brief Review

Implementation of environmental surveillance for SARS-CoV-2 virus to support public health decisions: Opportunities and challenges

SARS-CoV-2 in wastewater: State of the knowledge and research needs

Faecal shedding models for SARS-CoV-2 RNA amongst hospitalised patients and implications for wastewaterbased epidemiology

Presence of SARS-Coronavirus-2 RNA in Sewage and Correlation with Reported COVID-19 Prevalence in the Early Stage of the Epidemic in The Netherlands

Show us the Data: Global COVID-19 Wastewater Monitoring Efforts, Equity, and Gaps. medRxiv 2021

HERA Incubator: Anticipating together the threat of COVID-19 variants

National Records of Scotland Population Estimates by Scottish Index of Multiple Deprivation(SIMD)

Extension of Nakagawa & Schielzeth'sR2GLMMto random slopes models

R: A language and environment for statistical computing

Welcome to the Tidyverse

Scale Functions for Visualization

Automatically Position Non-Overlapping Text Labels with 'ggplot2

Fitting Linear Mixed-Effects Models Using lme4

MuMIn: Multi-Model Inference

SARS-CoV-2 RNA in Wastewater Settled Solids Is Associated with COVID-19 Cases in a Large Urban Sewershed

Nonparametric Analysis of Covariance

Measurement of SARS-CoV-2 RNA in wastewater tracks community infection dynamics

Scaling of SARS-CoV-2 RNA in Settled Solids from Multiple Wastewater Treatment Plants to Compare Incidence Rates of Laboratory-Confirmed COVID-19 in Their Sewersheds

Evaluation of Sampling, Analysis, and Normalization Methods for SARS-CoV-2 Concentrations in Wastewater to Assess COVID-19 Burdens in Wisconsin Communities

Wastewater surveillance of SARS-CoV-2 across 40 U.S. states from February to

First confirmed detection of SARS-CoV-2 in untreated wastewater in Australia: A proof of concept for the wastewater surveillance of COVID-19 in the community

Evaluation of lockdown effect on SARS-CoV-2 dynamics through viral genome quantification in waste water

Characteristics of pediatric SARS-CoV-2 infection and potential evidence for persistent fecal viral shedding

Prolonged viral shedding in feces of pediatric patients with coronavirus disease 2019

SARS-CoV-2 titers in wastewater are higher than expected from clinically confirmed cases

Estimating the proportion of SARS-CoV-2 infections reported through diagnostic testing

Genome Sequencing of Sewage Detects Regionally Prevalent SARS-CoV-2 Variants

An integrated national scale SARS-CoV-2 genomic surveillance network

SARS-CoV-2 testing and sequencing for international arrivals reveals significant cross border transmission of high risk variants into the United Kingdom

Environmental surveillance system to track wild poliovirus transmission

Seasonal and spatial dynamics of enteric viruses in wastewater and in riverine and estuarine receiving waters

Detection of Hepatitis E Virus in Sewage After an Outbreak on a French Island

The authors would like to thank all the sampling and courier staff at Scottish Water and the team of scientists and other staff at the Scottish Environment Protection Agency for their efforts in acquiring, transporting and analyzing the wastewater samples throughout the pandemic, as well as the electronic Data Research and Innovation Service (eDRIS) who provided the COVID-19 test and death data. We would also like to acknowledge the support of Andrew Millar, the Scottish Government's Chief Scientific Adviser for Environment, Natural Resources, and Agriculture.