key: cord-0989454-mrwbtaky
authors: Zhu, Yifan; Oishi, Wakana; Maruo, Chikako; Saito, Mayuko; Chen, Rong; Kitajima, Masaaki; Sano, Daisuke
title: Early warning of COVID-19 via wastewater-based epidemiology: potential and bottlenecks
date: 2021-01-21
journal: Sci Total Environ
DOI: 10.1016/j.scitotenv.2021.145124
sha: b2348dea34a02f06df638b17cb64b7b80a637eee
doc_id: 989454
cord_uid: mrwbtaky

An effective early warning tool is of great administrative and social significance to the containment and control of an epidemic. Facing the unprecedented global public health crisis caused by COVID-19, wastewater-based epidemiology (WBE) has been given high expectations as a promising surveillance complement to clinical testing which had been plagued by limited capacity and turnaround time. In particular, recent studies have highlighted the role WBE may play in being a part of the early warning system. In this study, we briefly discussed the basics of the concept, the benefits and critical points of such an application, the challenges faced by the scientific community, the progress made so far, and what awaits to be addressed by future studies to make the concept work. We identified that the shedding dynamics of infected individuals, especially in the form of a mathematical shedding model, and the back-calculation of the number of active shedders from observed viral load are the major bottlenecks of WBE application in the COVID-19 pandemic that deserve more attention, and the sampling strategy (location, timing, and interval) needs to be optimized to fit the purpose and scope of the WBE project.

The basic concept of WBE centers around this principle: certain chemical or biological agents (also referred to as 'biomarkers') excreted by human bodies can be collected by the sewage network and end up entering the wastewater, making it a rich source of these substances. Via physicochemical methods, biomarkers can be recovered from wastewater and the measured concentration can then be used to infer the size of the shedding population and provide community-level health information (Xagoraraki and O'Brien, 2020) . For SARS-CoV-2, although antigen testing is also emerging (Daughton, 2020) , the viral genome has been widely accepted as the biomarker. To date, a handful of studies have reported the detection of the SARS-CoV-2 viral genome in sewage networks D'Aoust et al., 2021; Hata et al., 2020; La Rosa et al., 2020b; Medema et al., 2020b; Randazzo et al., 2020; Westhaus et al., 2021; Wurtzer et al., 2020) .

However, some believe that the true standout of WBE is the early warning capability. The term "early warning" can be interpreted in two ways in the context of COVID-19 surveillance: (1) signaling an early stage of an outbreak. Presymptomatic/asymptomatic transmission of COVID-19 is considered a key factor behind its rapid spread (Gandhi et al., 2020) . Arons et al. (2020) recovered viable virus from 71% (17 in 24) of presymptomatic individuals 1 to 6 days prior to the symptom onset, and He et al. (2020) estimated that 44% (95% CI 25-69%) of secondary cases were infected when the index cases were in their presymptomatic stage. While asymptomatic and presymptomatic virus carriers can easily hide in the population due to the absence of appreciable symptoms such as fever and dry cough, by nature, WBE can indiscriminately detect their presence as long as they develop viral RNA shedding . Therefore, if a positive wastewater viral load is spotted in a region previously experiencing no or a low prevalence, it may indicate an unnoticed initial circulation of the virus J o u r n a l P r e -p r o o f in the community. This information can be made use by the local authority, who can take intervention by issuing warnings or administrative orders accordingly to inform the public of the potential threat and reduce the chance of invisible transmission. Also, as many countries and regions suffer from limited resources needed for a large-scale clinical testing program which greatly helps monitor the epidemic development and control the spread, getting a rough location of an initial circulation can help ease the burden and make the testing more efficient by guiding the valuable testing capacity to where it is most urgently needed; (2) foreshadowing an impending increase in infected individuals. The basic assumption behind this is: since infectiousness predates symptom onset, so can the viral shedding. Thus, if proper sampling tactics and quantification methods are adopted, an increased wastewater viral load may be observed and reported before the newly infected individuals develop symptoms and seek medical attention. Since there may be a correlation between wastewater viral load and the number of infected individuals, in addition to supporting the administrative and resource deployment measures previously described, from the perspective of disease treatment, the quantitative information also allows the healthcare facilities to take measures aimed at improving preparedness and coping with the anticipated new patients beforehand so that the facilities are less likely to be overwhelmed.

Both interpretations of early warning have been backed up by recent studies and events. Table 1 lists some selected recent wastewater surveillance studies that highlight the potential of WBE early warning of COVID-19. In terms of practical application, the University of Arizona made headlines in August 2020 when researchers there detected SARS-CoV-2 viral genome in the wastewater from a student dormitory, the university quickly took action and tested all 311 residents living and working in the said building and found two asymptomatic carriers among J o u r n a l P r e -p r o o f them, likely having prevented a potential outbreak and making it the first true application of WBE in COVID-19 early warning (Jaclyn Peiser, 2020).

However, despite the appealing side of WBE, a significant knowledge gap still exists regarding to what extent the promises regarding being an early warning system can be truly fulfilled, and potential bottlenecks are seldomly discussed in detail. To help readers fill the gap and draft a roadmap for further research, here in this study we would like to dive into the limiting factors in a real-world setting, discuss what has been done and made clear so far, what still awaits to be addressed, and where is the research frontier of these factors.

When used for detecting newly introduced virus carriers and initial virus circulation in a lowprevalence community, the viability of WBE and the confidence it offers largely lean on the lowest possible prevalence level that enables the detection of viral RNA in sewage. The practical value is governed by many factors including the sewage network layout and capacity, shedding profile of infected individuals, sewage characteristics, sampling strategy, the recovery efficiency of the concentration and quantification methods, and the detection limit of the instrument. A relatively reliable estimate requires the latest knowledge about the pathology of COVID-19, verified experimental method, as well as support from the local water agency. Some estimates have been given by previous studies, Hart and Halden (2020) performed a computational analysis with the City of Tempe, Arizona, USA being the studied region and estimated that a sensitivity of 1 in 144 to 2 million individuals can be achieved, depending on the assumptions used. Similarly, Ahmed et al. (2020a) reported an estimated prevalence level of 0.028% (95% CI 0.019-0.039%, 1 in 3571 individuals) based on viral RNA detection. However, these estimations may be too optimistic as some factors that can significantly affect the detection sensitivity are missing while others face significant uncertainty. For instance, neither of the two studies counted the recovery efficiency of the experimental method, the latter study also did not consider the natural degradation of the viral RNA. As for the example of the University of Arizona, despite a detection sensitivity of 0.64% (2 in 311) on paper, as further details (e.g., sampling strategy) remain undisclosed, it is unclear whether the same level of sensitivity can be expected under other conditions. Besides, in a quantitative sense, as an extension of calculating the lowest prevalence level that enables successful detection, a back-calculation model that projects the obtained wastewater viral load to the active shedding population is of foremost importance (Xagoraraki and O'Brien, 2020) , yet so far, very few studies have challenged this issue.

Another measure of the viability of WBE in COVID-19 early warning is how responsive it can be. Even if the detection sensitivity is adequate for low prevalence detection, the value of detection can be seriously undermined, even nullified, if the result cannot reach the correct hands in time. Also, in regions where prevalence level is high enough to enable consistent viral RNA detection, WBE can still shine from its quantitative side; if the wastewater viral load is closely monitored and there is a surge in infections, as fecal shedding may predate symptom onset, an increase in viral load may appear before the newly infected individuals develop symptoms, seek medical attention, and be admitted to healthcare facilities after diagnosis, and the number of them may be inferred from the viral load. Different from the low prevalence detection which only gives a qualitative result, the quantitative outcome gives local healthcare facilities and their supervising agencies a response window period and an anticipated capacity demand. Just as in the case of testing capacity, in a time when many regions are having logistic difficulty handling the rapid increase in infections with limited resources (Kamerow, 2020), being able to forecast J o u r n a l P r e -p r o o f the demand may help get an upper hand and improve the preparedness as local healthcare facilities can make use of this time to (1) prepare necessary medical supplies and equipment including beds, ventilators, protective clothing, and masks; (2) arrange human resources to make sure there would be adequate health workers for the increased workload. In addition, from a higher angle, this community demand forecast may enable regional reallocation of available resources which can come in handy if there is an overall shortage.

However, both measures of the feasibility of WBE early warning face considerable uncertainty.

In the previously mentioned studies regarding the lead of viral RNA in primary sewage sludge compared to local admissions, the analysis was performed in a retrospective way: the accumulated longitudinal SARS-CoV-2 quantification data were compared with the clinical reports during the same period. For WBE to be an active early warning tool, though, it needs to be performed in a timelier manner. One important index in the timeline of COVID-19 infection is the incubation period, which is the time gap between virus exposure to symptom onset. Some studies have reported very similar median or mean values of approximately 5 days (Lauer et al., 2020; McAloon et al., 2020) . After the symptom onset, there typically will be another period until the testing result comes out or the patient gets admitted to a hospital, Lauer et al. (2020) reported a mean value of 1.2 days, but it may vary greatly depending on the testing policy and the capacity of the hospital in question. Adding the two intervals up sets a reference for WBE; whether the workflow can be streamlined to beat this time largely affects its viability, although to what extent the outcome is useful also depends on how long is the response window period it leaves behind.

In the following sections, factors that may become bottlenecks, their significance, what previous and recent studies have revealed, and what awaits to be addressed and clarified by further studies J o u r n a l P r e -p r o o f are summarized and discussed to provide readers with a brief roadmap towards the final application of WBE as a part of the COVID-19 early warning system.

The shedding profile of infected individuals directly determines the wastewater viral load and is hence regarded as one of the most critical factors in WBE. The shedding profile consists of three parts: the shedding rate, the beginning of shedding, and the shedding duration. When the shedding profile is relatively predictable and stable while showing finite between-person variation, it will greatly simplify the modeling process, but on the other hand, if the shedding profile bears significant stochastic fluctuations and between-person discrepancy, substantial extra efforts would be needed to handle the data noise and uncertainty.

So far, reports regarding shedding rate have mainly focused on hospitalized symptomatic patients due to the availability. Walsh et al. (2020) summarized in their review that while some studies reported little to no difference in the viral loads of symptomatic and asymptomatic patients, there are also studies that found the severity of symptoms can affect viral load, indicating substantial heterogeneity. Generally speaking, fecal viral shedding shows significant uncertainty and the overall pattern is more erratic than respiratory shedding. One of the earliest assessments of the rate and duration of fecal shedding was conducted by Wölfel et al. (2020) , while the highest recorded viral load among 9 hospitalized patients reached 10 7 copies/gram of stool sample, the results also show significant variation between cases; the viral load of one patient had always stayed below 10 4 copies per gram of feces. In a review by Parasa et al. (2020) , the recorded viral load in stool samples also ranges from 550 to 1.21 × 10 5 copies per mL of feces. The viral shedding in the gastrointestinal tract seems more erratic than that in the respiratory tract (Walsh et al., 2020) . In addition to the variation in shedding rate, it has also been J o u r n a l P r e -p r o o f stated that not all infected individuals will develop fecal shedding. In the aforementioned study by Wölfel et al. (2020) , the stool specimen of one patient was negative during the entire testing course. A meta-analysis by van Doorn et al. (2020) reported that 51.8% (95% CI 43.8-59.7%) of patients have their stool specimens tested positive while another systematic review by Gupta et al. (2020) reported a similar percentage (53.9%), but very limited information is available about the shedding ratio among asymptomatic virus carriers. Not only is this uncertain fecal shedding a hindrance to the estimation of achievable detection sensitivity, it also means that when the number of infected individuals is low, statistically, there is a chance that none of them sheds viral RNA into wastewater, making their presence undetectable by WBE no matter how sensitive the assays are.

Also, the timing of fecal shedding has a decisive role in determining how "early" the shedding can be detected. As the routine testing of fecal shedding typically only focuses on symptomatic patients after their hospitalization, solid evidence remains scarce as to the actual starting point of fecal shedding, especially among asymptomatic virus carriers. Alternatively, the timing of infectiousness development (respiratory shedding) may be used as a proxy. He et al. (2020) estimated that the infectious period begins at 2.3 days prior to symptom onset and peaks at 0.7 days before it. Nevertheless, it is important to keep in mind that respiratory shedding does not perfectly represent fecal shedding and may not exactly parallel it, related information hence must be interpreted and used prudently until more medical evidence becomes available.

As another critical aspect of the shedding profile, the persistency of shedding should also be given some consideration. It has been revealed that the shedding of SARS-CoV-2 in fecal specimens can outlast that in respiratory specimens Wang et al., 2020; Xiao et al., 2020; Xu et al., 2020) . The long-tailed fecal shedding may cause a masking J o u r n a l P r e -p r o o f effect on newly infected individuals, making their presence indistinguishable from the patients in their post-infection phase, especially when an infection peak has recently ended and the shedding population remains large. Although according to previous medical reports, the intensity of shedding steadily declines during the infection course, further clinical evidence is still needed to confirm whether the long-tailed shedding will become a concern for WBE application.

Because the shedding rate and duration determined from clinical case reports may be subjected to stochastic error and person-to-person variation, if possible, packing available data and biological explanation into a mathematical model for better generalization and easier extrapolation of the shedding dynamics is preferable. Currently, available information about this approach is very limited and further study is needed. Recently, Miura et al. (2020) fitted a shedding dynamics model (Eq. 1) originally developed by Teunis et al. (2015) for norovirus fecal shedding. shedding of SARS-CoV-2 may have its own distinct characteristics and follow a different biological mechanism, it is unclear whether the same pathologic assumptions and consequently these mathematical models can be applied to SARS-CoV-2.

In conclusion, though much information has been made available, the current knowledge is still far from enough to support successful WBE application in absolute calculations. A heavy workload still lies ahead until the uncertainty in the fecal viral shedding can be properly addressed. However, it should be clearly stated that there is no guarantee that such a goal will finally be achieved therefore the worst scenario also needs to be considered: if the shedding profile is eventually found to be too erratic and unpredictable, as some existing literature suggests, to be clearly described and properly modeled, the prospect of WBE will be critically impaired as it lacks the ability to be a tool for absolute quantitative analysis. But to proceed from where we are now, a more comprehensive and holistic image of the shedding profile, including the rate, starting time, and duration is needed, which will benefit from further clinical evidence.

Stable and efficient recovery and detection of the viral RNA is a decisive factor in wastewater surveillance. The recovery efficiency and instrument detection limit provide a critical reference when estimating the threshold prevalence level and back-calculating the shedding population from the viral load. For primary concentration and RNA extraction, several research articles and reviews have looked into this technical issue, focusing on either surrogates or other coronavirus strains (Ahmed et al., 2020d; La Rosa et al., 2020a; Rusiñol et al., 2020; Torii et al., 2020) . is not considered, in other words assuming a 100% recovery, the estimated detection limit would be lower than the actual value, which may lead to a falsely high sense of security. However, even though both MHV and φ6 are enveloped viruses and may better resemble the behavior of SARS-CoV-2 than nonenveloped surrogates, discrepancy may still exist and the measured recovery efficiency should be used discreetly and only as a reference.

It is also worth mentioning that many established primary concentration methods were originally that wastewater solids may support more sensitive SARS-CoV-2 detection. Therefore, an extra step that helps release viral RNA from the solids (e.g., heat treatment and adsorption-elution) J o u r n a l P r e -p r o o f may improve recovery efficiency (Corpuz et al., 2020; Schwab et al., 1997) , but additional research needs to be conducted to verify the efficacy for SARS-CoV-2.

The last barrier of the quantification assay is the detection and quantification limit. For RT-qPCR, a standard curve is necessary for converting the cycle threshold (Ct) value into virus titers, but if the signal intensity is below a certain Ct value, it would be indistinguishable from the potential noise. In practice, this Ct value limit is usually translated to gene copies per unit volume by referring to the standard curve. However, if the dilution series is not well configured, there could be a difference between the limit of detection (LoD) and the limit of quantification (LoQ).

Attention should be paid to reduce or eliminate the gap between LoD and LoQ. PCR reaction inhibition is also a concern in wastewater surveillance, the introduction of process control, whether applied to the whole process, before RNA extraction and/or before RT-qPCR, has been proposed to help evaluate the extent of inhibition . In addition, the design of RT-qPCR assay Hamouda et al., 2020; Hata et al., 2020; Kitamura et al., 2021) and nucleic acid extraction kit (Sidhu et al., 2013) can also affect the detection sensitivity.

As in the case of the primary concentration method, at the current stage, a consensus of optimal recovery-detection assay has not been reached, researchers may need to conduct their experiments to determine the assay suitable for the lab condition and wastewater characteristics.

CoV-2 RNA in clinical samples (Dang et al., 2020; Falzone et al., 2020; Suo et al., 2020; Yu et al., 2020) and suggested that ddPCR is a superior choice for clinical diagnosis for its higher sensitivity and other benefits such as not needing a standard curve for quantification. However, D'Aoust et al. (2021) compared RT-ddPCR and RT-qPCR using wastewater sludge samples and the results did not support the statement that RT-ddPCR performs better than RT-qPCR. It is J o u r n a l P r e -p r o o f possible that the low detection limit offered by ddPCR can enhance the performance of the WBE approach, but related research needs to be further extended to investigate the effect of factors such as inhibition and optimize the assay.

Once the viral RNA is released from shedding individuals, it will enter the sewage network and get mixed with the rest of the wastewater. In its simplest form, the dilution factor can be period is set to 24 h and a flow-proportional sampler is used (Medema et al., 2020a; Michael-Kordatou et al., 2020; Thompson et al., 2020) . However, considering the biological rhythm and living habits of human beings, the variance may, in turn, be beneficial and the grab sample may offer a higher chance of detection if the sampling time is optimized to capture the peak hours of the toilet flushing. In the aforementioned study of Hata et al. (2020) in which a positive signal was detected when the catchment area had a low prevalence level (less than one confirmed case per 100,000 people), grab samples were used. However, due to the unevenly distributed in-sewer travel time in a large sewage network, this specific method may be more applicable to confined environments (e.g., dormitories and nursing homes). An early study on defecation concluded that J o u r n a l P r e -p r o o f defecations are more likely to occur in the early morning (Heaton et al., 1992) , and Campisano and Modica (2015) reported that there are three toilet flushing peaks during a day, although it is necessary to point out that the result is merely based on a case study of a household and whether it also applies to a larger community needs further verification.

Previous studies have employed different sampling frequencies, from taking samples on discrete dates (Kumar et al., 2020; La Rosa et al., 2020b) , to routine sampling with a relatively stable interval (D'Aoust et al., 2021; Hata et al., 2020; Randazzo et al., 2020) and daily sampling . For retrospective analysis, frequent sampling over a long period (e.g., several months) may unnecessarily increase the total workload required for sample processing, therefore lower frequency is acceptable and more realistic. But under the premise of using the measured viral load for early warning, especially considering the rapid progression of the COVID-19 epidemic and the potential social significance of the measures to be taken, daily or similarly frequent sampling is highly recommended as long as the laboratory capacity allows.

Apart from direct measuring, human fecal indicators may also help normalize the flowrate as well as help identify the peak flushing hours in a day if there are any. Some previous studies have opted for the usage of pepper mild mottle virus (PMMoV) as an internal control because of its universal and stable presence in the wastewater matrix (Kitajima et al., 2014; Symonds et al., 2018) , but more options including crAssphage, HAdV, JCPyV, human microbiome-specific HF183 Bacteroides 16S ribosomal rRNA, and eukaryotic 18S rRNA may also be used (Ballesté et al., 2019; Bofill-Mas et al., 2006; D'Aoust et al., 2021; Medema et al., 2020a) , although it is worth mentioning that because their nature may be very distinct from that of the target biomarkers, these internal control targets cannot be used for signal normalization.

Once the viral RNA is released from infected individuals and enters the wastewater, it will spend some time traveling in the sewer pipes until it reaches the designated sampling site. The in-sewer travel time is a function of many characteristics of the sewage system in question, such as the spatial configuration of the sewer network and wastewater flow rate in a given time, hence its value may vary greatly among sewage networks and is highly recommended to be determined for each WBE project individually if needed. Although its importance in WBE has been underlined , the in-sewer travel time for a given sewage network has rarely been reported, presumably due to the difficulty of performing an experiment or establishing a hydrological model. But according to limited existing estimates made for multiple purposes including WBE application, the mean value of in-sewer travel time, or wastewater residence time, typically falls within several hours. For instance, it has been estimated that the national median in-sewer travel time for the U.S. is 3.3 h (Kapo et al., 2017) , and the approximate sewer transit times of the UK and Rome have also been estimated to be ~2 h and 3-5 h, respectively (D'Ascenzo et al., 2003; Holt et al., 1998) . Also, in the case of grab sampling, the mean or median value does not consider the population heterogeneity, the wastewater produced by those living close to the sampling site will have significantly shorter in-sewer travel time compared to that from people living in the upstream area. To offset the potential impact, the demographic and geographic distributions of the population may also be considered a factor.

Wastewater is a complex matrix with a high concentration of microorganisms and substances that are either organic and inorganic, all of which may contribute to the natural degradation of viral RNA. Previous studies have investigated and reviewed the degradation of human coronaviruses including SARS-CoV and SARS-CoV-2 and their surrogates such as murine hepatitis virus (MHV) in the wastewater matrix (Ahmed et al., 2020c; Bivins et al., 2020a;  J o u r n a l P r e -p r o o f Journal Pre-proof Mandal et al., 2020) , as the results suggest, the viral RNA of SARS-CoV-2 is more persistent than viable virus particles and can stay in wastewater for a relatively long period, and the decay rate increases as the wastewater temperature goes up. Bivins et al. (2020a) recorded a 26.2 days T 90 value in untreated wastewater at 20 °C, which is comparable to the study by Ahmed et al. (2020c) in which the T 90 values in untreated wastewater at 15 and 25 °C were 20.4 and 12.6 days, respectively. Because the typical residence time of wastewater is several hours, the effect of degradation may not be as pronounced as other factors, but it is still recommended to take the degradation kinetics into consideration for better accuracy. For instance, in a long-term monitoring project, the seasonal change in wastewater temperature may bring change to the prevalence estimation as the degradation during summer would be more significant and may reduce the measured viral load (Hart and Halden, 2020) .

Although the majority of existing studies took samples from wastewater treatment plants (WWTPs) for better coverage and convenience, they are not the only option. If a better spatial resolution or a shorter response time is required, the strategy of 'upstream sampling' can also be employed, which means samples are taken from locations closer to the origins, such as sewer pumping stations and maintenance holes, to narrow down the coverage and shorten the in-sewer travel time (Medema et al., 2020a; Thompson et al., 2020) .

The turnaround time for sample treatment consists of sample transportation, virus concentration, quantification, and data analysis and organizing. Primary concentration methods can take anywhere from about an hour (ultrafiltration, electronegative membrane vortex) to overnight (PEG precipitation) (Ahmed et al., 2020d; Torii et al., 2020) , the time required for subsequent steps varies depending on the reagent kits and instruments used but is also typically in the range J o u r n a l P r e -p r o o f of several hours if all steps are done consecutively. The time needed for sample transportation and data analysis depends heavily on real-life factors including the transportation method, the distance between sampling site and laboratory, and the way of data processing, so far, very limited information about these two steps is available from existing literature. Due to the varied conditions, although it is possible for the WBE approach to predate symptom onset and hospital admission, the entire workflow is subject to significant uncertainty and different settings hence an estimated time cannot be given. We encourage future studies to include the information of the time required for each step to give a better image of the total turnaround time.

Worrying that the standard off-site RT-qPCR method may not fulfill the demand for rapid detection, recent studies have discussed alternative methods. Mao et al. (2020) and Bhalla et al.

(2020) discussed the potential of paper-based analytical devices, which are easy-to-carry tools that can be deployed for rapid on-site nucleic acid testing. Yang et al. (2017) reported a paperbased "sample-to-answer" platform for the detection of human genomic DNA in untreated wastewater based on the loop-mediated isothermal amplification (LAMP), through which the result can be yielded within 45 min. Nguyen et al. (2020) shared similar optimism about LAMP, specifically stating that LAMP can be a potential candidate for COVID-19 early detection. While alluring on paper, compared to the conventional laboratory apparatus, the reliability of new methods and devices remains unverified, and very limited information is available regarding the practical application despite the strong interest. Although the development of novel devices and methods that can enable rapid and reliable detection of SARS-CoV-2 genome RNA is highly encouraged, considering that the derived information will be used to help make crucial decisions, the application must be proceeded with caution.

J o u r n a l P r e -p r o o f

After obtaining the quantification result, there is a final step of data analysis. If the SARS-CoV-2 signal in a low prevalence region is positive for the first time, it indicates a high possibility that the predetermined threshold prevalence level has been exceeded, but for better dependability, especially considering the information will be used to support critical decisions that may leave a permanent impact on the society, an additional validation step may be employed at the cost of longer response time. As for long-term monitoring in a middle-to-high prevalence region where the viral load is consistently higher than LoQ, the concentration of viral RNA, or the active shedding population if a back-calculation model can be successfully established and validated, should be combined with previous results to assemble a longitudinal pattern.

Although theoretically, qualitative and semi-quantitative analysis can be performed without a back-calculation model that connects the viral load to the active shedding population (Daughton, 2020) , the lack of quantitative projection would certainly impair the usefulness of the result.

Although lagged correlation has been found in wastewater viral load and reported patient number (Nemudryi et al., 2020; Peccia et al., 2020) , there is no model currently in existence that can infer the size total infected or shedding population, which could be much larger than reported cases due to undertesting and asymptomatic virus carriers. Previous WBE projects have used excretion-dilution-recovery mass balance models for the back-calculation of chemical biomarkers (Feng et al., 2018; Gracia-Lor et al., 2017; Lai et al., 2011) , but in the case of COVID-19 surveillance, the same model may suffer greatly from the limited understanding and the variance of some parameters (mainly the shedding profile and dilution factor) as well as the potential data noise. One reason is that due to the persistent fecal shedding, the wastewater viral load is likely to be contributed by patients in different infection stages, this population may even include those who have the virus cleared in the respiratory tract. Thereby, not only modeling J o u r n a l P r e -p r o o f tools that help reduce the uncertainty (e.g., Bayesian inference and maximum likelihood estimation) are highly recommended, other mathematical models featuring different structures are also worth looking into as long as they offer a decent capability of capturing the correlation between viral load in wastewater and the shedding/infected population and dealing with the noisy data. Here, we not only encourage researchers who are already working on COVID-19

wastewater surveillance to reach out and look for potentially suitable modeling techniques but also call on experts from other disciplines, such as epidemiology and statistics, to join in and tackle the challenge together.

Knowing the de facto population size in the catchment area also helps reduce the uncertainty when doing quantitative analysis (Choi et al., 2018; Daughton, 2020; Medema et al., 2020a; O'Brien et al., 2014) . Census data or estimation based on facility capacity can be used as an approximation, but they may deviate from reality. Another option is to use certain population biomarkers including exogenous markers such as nicotine and caffeine, and endogenous markers 5-hydroxyindoleacetic acid (5-HIAA) and ammonia (Choi et al., 2018) , but it should be kept in mind that significance discrepancy may exist between regions and countries due to various lifestyles. Apart from the de facto population, in regions with high mobility (e.g., tourist attractions and transportation hubs), the frequent movements of people not only increase the risk of virus introduction but also introduce a new source of uncertainty for quantitative analysis (Lai et al., 2011) . Further studies will need to employ appropriate methods to estimate and validate the population covered by the studied sewage network and consider how to incorporate the dynamics of the population into the result interpretation process.

As mentioned previously, WBE should not be considered as a standalone solution, but rather as a complementary data source in public health management (Ahmed et al., 2020b; D'Aoust et al., 

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

Detection of SARS-CoV-2 RNA in commercial passenger aircraft and cruise ship wastewater: a surveillance tool for assessing the presence of COVID-19 infected travellers

Decay of SARS-CoV-2 and surrogate murine hepatitis virus RNA in untreated wastewater to inform application in wastewater-based epidemiology

Comparison of virus concentration methods for the RT-qPCR-based recovery of murine hepatitis virus, a surrogate for SARS-CoV-2 from untreated wastewater

Presymptomatic SARS-CoV-2 infections and transmission in a skilled nursing facility

Dynamics of crAssphage as a human source tracking marker in potentially faecally polluted environments

Opportunities and challenges for biosensors and nanoscale analytical tools for pandemics: COVID-19

Persistence of SARS-CoV-2 in water and wastewater

Wastewater-based epidemiology: Global collaborative to maximize contributions in the fight against COVID-19

Quantification and stability of human adenoviruses and polyomavirus JCPyV in wastewater matrices

Appropriate resolution timescale to evaluate water saving and retention potential of rainwater harvesting for toilet flushing in single houses

Wastewater-based epidemiology biomarkers: Past, present and future

Viruses in wastewater: occurrence, abundance and detection methods

Quantitative analysis of SARS-CoV-2 RNA

Fate of natural estrogen conjugates in municipal sewage transport and treatment J o u r n a l P r e -p r o o f Journal Pre-proof facilities

Comparison of qualitative and quantitative analyses of COVID-19 clinical samples

Wastewater surveillance for population-wide Covid-19: The present and future

Sensitivity assessment of droplet digital PCR for SARS-CoV-2 detection

Monitoring of regional drug abuse through wastewater-based epidemiology-A critical review

Asymptomatic transmission, the achilles' heel of current strategies to control Covid-19

Persistent viral shedding of SARS-CoV-2 in faeces -a rapid review

Wastewater surveillance for SARS-CoV-2: Lessons learnt from recent studies to define future applications

Computational analysis of SARS-CoV-2/COVID-19 surveillance by wastewater-based epidemiology locally and globally: Feasibility, economy, opportunities and challenges

Detection of SARS-CoV-2 in wastewater in Japan during a COVID-19 outbreak

Temporal dynamics in viral shedding and transmissibility of COVID-19

Defecation frequency and timing, and stool form in the general population: A prospective study

UK monitoring study on the removal of linear alkylbenzene sulphonate in trickling filter type sewage treatment plants. Contribution to GREAT-ER project 2

Modeling the dynamics of virus shedding into the saliva of Epstein-Barr virus positive individuals

University of Arizona used wastewater testing to detect cases of coronavirus in a dorm

Shedding of SARS-CoV-2 in feces and urine and its potential role in person-to-person transmission and the environment-based spread of COVID-19

Aligning SARS-CoV-2 indicators via an epidemic model: application to hospital admissions and RNA detection in sewage sludge

Estimation of U.S. sewer residence time distributions for national-scale risk assessment of down-the-drain

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

Relative abundance and treatment reduction of viruses during wastewater treatment processes -Identification of potential viral indicators

Efficient detection of SARS-CoV-2 RNA in the solid fraction ofwastewater

Level of underreporting including underdiagnosis before the first peak of COVID-19 in various countries: Preliminary retrospective results based on wavelets and deterministic modeling

First proof of the capability of wastewater surveillance for COVID-19 in India through detection of genetic material of SARS-CoV-2

Coronavirus in water environments: Occurrence, persistence and concentration methods -A scoping review

First detection of SARS-CoV-2 in untreated wastewaters in Italy

Refining the estimation of illicit drug consumptions from wastewater analysis: Co-analysis of prescription pharmaceuticals and uncertainty assessment

A review on presence, survival, disinfection/removal methods of coronavirus in wastewater and progress of wastewaterbased epidemiology

Can a paper-based device trace COVID-19 sources with wastewater-based epidemiology?

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

Presence of SARS-Coronavirus-2 RNA in sewage and correlation with reported COVID-19 prevalence in the of the epidemic in the Netherlands

Sewage analysis as a tool for the COVID-19 pandemic response and management: the urgent need for optimised protocols for SARS-CoV-2 detection and quantification

Duration of SARS-CoV-2 viral shedding in faeces as a parameter for wastewater-based epidemiology: Re-analysis of patient data using a shedding dynamics model

Temporal detection and phylogenetic assessment of SARS-CoV-2 in municipal wastewater

Novel coronavirus disease (COVID-19): Paving the road for rapid detection and point-of-care diagnostics

A model to estimate the population contributing to the wastewater using samples collected on census day

Zika viral dynamics and shedding in rhesus and cynomolgus macaques

Prevalence of gastrointestinal symptoms and fecal viral shedding in patients with coronavirus disease 2019: A systematic review and meta-analysis

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

SARS-CoV-2 RNA in wastewater anticipated COVID-19 occurrence in a low prevalence area

Concentration methods for the quantification of coronavirus and other potentially pandemic enveloped virus from wastewater

Use of heat release and an internal RNA standard control in reverse transcription-PCR detection of Norwalk virus from stool

Sensitive detection of human adenovirus from small volume of primary wastewater samples by quantitative PCR

Future perspectives of wastewater-based epidemiology: Monitoring infectious disease spread and resistance to the community level

ddPCR: a more accurate tool for SARS-CoV-2 detection in low viral load specimens

Pepper mild mottle virus: A plant pathogen with a greater purpose in (waste)water treatment development and public health management

Shedding of norovirus in symptomatic and asymptomatic infections

Effects of sewer conditions on the degradation of selected illicit drug residues in wastewater

Making waves: Wastewater surveillance of SARS-CoV-2 for population-based health management

Applicability of polyethylene glycol precipitation followed by acid guanidinium thiocyanate-phenol-chloroform extraction for the detection of SARS-CoV-2 RNA from municipal wastewater

Novel wastewater surveillance strategy for early detection of coronavirus disease 2019 hotspots

SARS-CoV-2 detection, viral load and infectivity over the course of an infection

Fecal viral shedding in COVID-19 patients: Clinical significance, viral load dynamics and survival analysis

Detection of SARS-CoV-2 in raw J o u r n a l P r e -p r o o f Journal Pre-proof and treated wastewater in Germany -Suitability for COVID-19 surveillance and potential transmission risks

WHO Coronavirus Disease (COVID-19) Dashboard [WWW Document

Virological assessment of hospitalized patients with COVID-2019

Prolonged presence of SARS-CoV-2 viral RNA in faecal samples

Evaluation of lockdown impact on SARS-CoV-2 dynamics through viral genome quantification in Paris wastewaters

Wastewater-based epidemiology for early detection of viral outbreaks

Evidence for gastrointestinal infection of SARS-CoV-2

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

Monitoring genetic population biomarkers for wastewater-based epidemiology

Survivability, Partitioning, and Recovery of Enveloped Viruses in Untreated Municipal Wastewater

Quantitative detection and viral load analysis of SARS-CoV-2 in infected patients