SARS-CoV-2 shedding sources in wastewater and implications for wastewater-based epidemiology

Abstract

Wastewater-based epidemiology (WBE) approach for COVID-19 surveillance is largely based on the assumption of SARS-CoV-2 RNA shedding into sewers by infected individuals. Recent studies found that SARS-CoV-2 RNA concentration in wastewater (C_RNA) could not be accounted by the fecal shedding alone. This study aimed to determine potential major shedding sources based on literature data of C_RNA, along with the COVID-19 prevalence in the catchment area through a systematic literature review. Theoretical C_RNA under a certain prevalence was estimated using Monte Carlo simulations, with eight scenarios accommodating feces alone, and both feces and sputum as shedding sources. With feces alone, none of the WBE data was in the confidence interval of theoretical C_RNA estimated with the mean feces shedding magnitude and probability, and 63% of C_RNA in WBE reports were higher than the maximum theoretical concentration. With both sputum and feces, 91% of the WBE data were below the simulated maximum C_RNA in wastewater. The inclusion of sputum as a major shedding source led to more comparable theoretical C_RNA to the literature WBE data. Sputum discharging behavior of patients also resulted in great fluctuations of C_RNA under a certain prevalence. Thus, sputum is a potential critical shedding source for COVID-19 WBE surveillance.

Graphical Abstract

1. Introduction

Infection with coronavirus disease (COVID-19) is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The current clinical diagnosis of COVID-19 relies on the testing of individuals by the detection of SARS-CoV-2 RNA using reverse transcription-quantitative polymerase chain reaction (RT-qPCR) (Chau et al., 2020). In contrast to clinical testing, which determines the prevalence by testing individuals, wastewater-based epidemiology (WBE) is a promising approach for population-wide surveillance (Hart and Halden, 2020a, Li et al., 2021a; Sherchan et al., 2020; Sherchan et al., 2021). By systematic collections of wastewater samples either at the inlet of wastewater treatment plants (WWTPs), in sewer pumping stations, or manholes, the WBE approach provides a prevalence estimation with a single wastewater sample for the connected catchment or buildings (Betancourt et al., 2021, Rusiñol et al., 2021, Wong et al., 2021). This requires much lower cost and effort compared with clinical testing for a community with a large population. Depending on the catchment size, the cost of WBE testing accounts for only 0.7–1% of a population-wide clinical testing (Weidhaas et al., 2021). To date, the presence of SARS-CoV-2 RNA in wastewater has been confirmed by numerous studies globally, the meta-analysis of which demonstrated the potential of applying the WBE approach for COVID-19 surveillance (Li et al., 2021b).

Two major approaches have been used for COVID-19 surveillance: 1) conventional prevalence back-estimation based on the concentration of SARS-CoV-2 RNA detected in wastewater samples (C _RNA) using WWTPs influent flow rate and inhabitant population (Eq.1) or using the average water consumption data (Eq. (2)), where feces are regarded as the major source (Ahmed et al., 2020a, Hasan et al., 2020, Saththasivam et al., 2021); 2) infection trend monitoring based on the correlation between the disease incidence (daily new cases, weekly new cases, etc.) and C _RNA (Graham et al., 2021, Wilder et al., 2021).

$$P_{\mathrm{COVID}}=\frac{C_{\mathrm{RNA}}\times F}{(1-D)\times P\times E}$$ (1)

$$P_{\mathrm{COVID}}=\frac{C_{\mathrm{RNA}}\times Q_w}{(1-D)\times E}\times {10}^5$$ (2)

Where P _COVID is the COVID-19 prevalence in the catchment area as the number of COVID-19 patients per 100,000 people; C _RNA is the concentration of SARS-CoV-2 RNA detected in wastewater samples (gene copies/ L); F is the daily wastewater flow during the sampling period (L); D is the decay ratio of SARS-CoV-2 RNA during in-sewer transportation (-); P is the population size in the catchment area (×100,000 people); E is the excretion rate of SARS-CoV-2 RNA from infected people (gene copies/day∙person); Q _w is the daily water usage that ends up in sewer systems (L/day∙person).

Through the conventional back-estimation approach, some studies have estimated the COVID-19 prevalence in the catchment area based on the C _RNA and fecal shedding parameters including load and probability (Li et al., 2021b, Wölfel et al., 2020). However, in clinical studies, the presence of SARS-CoV-2 RNA in various bodily fluids has been confirmed with feces, urine, blood, saliva, sputum, etc. ( Table 1). Since SARS-CoV-2 is a respiratory infectious virus, SARS-CoV-2 RNA is mostly detected in respiratory tract samples (typical range 70–100%), to a lesser extent in stool samples (typical range 30–60%), and rarely in urine (<5%) (Jones et al., 2020). Our previous meta-level analysis with more than 1500 patients also revealed that the shedding probability of SARS-CoV-2 RNA in feces and urine was 54.5% and 2.6%, respectively (Li et al., 2021b). The presence of SARS-CoV-2 in stool and urine samples could be related to the swallowing of respiratory secretions from the upper respiratory tract or residues of infected antigen-presenting immune cells, or, more likely, due to virus replication in gastrointestinal epithelial cells or tubular epithelium (Farkash et al., 2020, Moreira et al., 2020, Xiao et al., 2020). Thus, considering the shedding probability (Table 1) and the possibility of entering the sewer system, apart from feces, sputum is likely to be an additional major source. A recent study revealed that sputum shedding potentially contributed a great amount of SARS-CoV-2 RNA into wastewater through theoretical simulations (Crank et al., 2022). However, the changes in the prevalence estimation caused by the inclusion of sputum as another shedding source remain unclear.

Table 1.

Potential shedding sources of SARS-CoV-2 to wastewater.

Shedding source	Shedding magnitude (gene copies/g or mL)	Shedding probability	Reference
Feces	104.52 (95% CI: 104.26 to 104.78) copies/g	54.5% (95% CI: 37–73%)	(Li et al., 2021b)
		51.8% (95% CI: 43.8–59.7%)	(van Doorn et al., 2020)
	103.4 (95% CI: 100.24-106.5) copies/g		(Miura et al., 2021)
Urine	10–104 copies/mL	0.026 (95% CI: 6.3 ×10-4- 0.10)	(Li et al., 2021b)
		0.8%	(Kim et al., 2020)
Sputum	641 to 1.34 × 1011 copies/mL		(Pan et al., 2020)
	107.3-108.2 copies/mL		(Yoon et al., 2020)
		1/3 of the patients had sputum production	(Lai et al., 2020)
		98% sputum samples tested positive	(Mohammadi et al., 2020)
Saliva	105.58 (95% CI: 105.09 to 106.07) copies/mL	81% (95% CI: 71–96%)	(Wyllie et al., 2020)
Blood	8.04 copies/mL	11%	(Peng et al., 2020)

To avoid the uncertainties due to shedding sources, some WBE studies correlated the disease incidence (daily new cases, weekly new cases, etc.) with C _RNA and estimated the current or future incidence in the community based on such correlations (Graham et al., 2021, Wilder et al., 2021). The approach assumes that a constant or similar shedding behavior among the patients. Undoubtedly, nearly all human feces end up in sewers, but the sputum discharging behavior might vary based on the culture, or hygiene practice of patients, especially for patients with upper respiratory infections (a common symptom of COVID-19). To date, the instruction for sputum disposal of patients during self-isolation of COVID-19 is fairly limited. In many countries such as Australia and UK, domestic wastes contaminated with sputum from COVID-19 patients during home isolation are recommended to be double-bagged and stored at home for at least 72 h or until the clear of COVID-19 symptom before being collected (EPA, E.P.A.V., 2020, PICA, 2020, Anand et al., 2022). Infectious SARS-CoV-2 was detected in sputum excreted from humans for up to 48 h in room conditions (Matson et al., 2020), which leads to a potential risk caused by the leakage of sputum if being stored and transported in bags. Thus, due to the inconvenience of waste disposal and long-time storage at home, it is likely that COVID-19 patients are inclined to discharge sputum directly or through handkerchiefs into sewers. However, the impact of personal sputum discharging behavior on the COVID-19 surveillance through WBE approach is largely unknown.

This study aims to investigate the contribution of different shedding sources, primarily feces and sputum, to the SARS-CoV-2 RNA concentration detected in wastewater (C _RNA) through systematic literature review and Monte Carlo simulation. WBE data of C _RNA with clinically confirmed prevalence in various studies were collected through systematic literature review and compared with the theoretical range based on COVID-19 prevalence. The potential impacts of 1) asymptomatic infected individuals or uncaptured cases by clinical testing; 2) prolonged fecal viruses shedding of recovered cases and 3) recovery efficiency of analytical approach on viral concentration in wastewater were also assessed through simulations. The different simulation scenarios provide a comprehensive evaluation of the major virus shedding sources and their contributions to viral concentrations in wastewater. The knowledge would enhance the understanding of environmental circulation of SARS-CoV-2 in urban water systems and support the application of WBE in the environmental surveillance of COVID-19.

2. Methods

2.1. Systematic literature review of WBE data

The systematic literature search was conducted on August 4th, 2021 following PRISMA guidelines (Silverman and Boehm, 2020). The goal of the search was to collect a comprehensive set of WBE data regarding the C _RNA detected in wastewater and the prevalence of active cases (confirmed cases minus recovered cases) through clinical testing in the catchment area of the wastewater sampling. Databases (i.e., Web of Science core collection, Scopus, and PubMed) were searched using the term “SARS-CoV-2 AND wastewater AND prevalence”. A total of 602 unique papers were identified after removing duplicates using the EndNote Reference Manager software. Titles and abstracts of the retained articles were screened and assessed for eligibility using these criteria: 1) reported clear data regarding C _RNA and clinically confirmed prevalence in the catchment area; 2) the article is in English and is peer-reviewed. Relevant articles were further assessed by full-text reading and finally, 12 articles with a total of 206 data points were included in this study. Details of the review process are provided in the Supplementary Information (SI). Other WBE reports were not included in this study due to the lack of required information or inability to provide such information after communicating with their authors.

To date, 14 different RT-qPCR primer-probe sets, targeting various SARS-CoV-2 RNA regions, including nucleocapsid (N), envelope (E), and RNA-dependent RNA polymerase (RdRp), are recommended by WHO and have been applied worldwide (Pezzi et al., 2020, Zhang et al., 2022). In these 12 WBE papers, primer-probe sets targeting at N-gene (i.e. CDC_N1, CDC_N2, CDC_N3; NIID_N); RdRp-gene (i.e. RdRp_SARSr, RdRp_IP4), ORF1ab- gene (ORF1ab) and E-gene (E_Sarbeco) were applied, which are all in the list of WHO recommendations. However, previous studies found false positives caused by using CDC_N3, which was then excluded from the US CDC 2019-nCoV RT-qPCR diagnostic panel (CDC, 2020). E_Sarbeco was not specifically designed for SARS-CoV-2, which would detect other human pathogenic corona viral RNAs such as human coronavirus OC43 (HCoV-OC43), a common cause of mild respiratory tract infection (Park et al., 2021). Thus, considering the specificity, reliability, sensitivity, and WHO recommendations of different primer and probe sets, the results obtained through CDC_N1, CDC_N2, RdRp_SARSr, RdRp_IP4, and ORF1ab in these WBE papers were included in this study. The performance of these primer-probe sets was thoroughly compared in previous studies, where the Ct values or C _RNA detected in the sample were mostly independent of the primer-probe selection through in-laboratory and inter-laboratory comparisons (Fischer et al., 2021, Jung et al., 2020). Thus, the C _RNA detected in these WBE reports were included without differentiating the primer-probe sets in this study. The clinically testing ratio and testing practice of the country during the WBE surveillance period was collected through the database (https://ourworldindata.org/coronavirus-testing) and summarized in Table S1. The impact of sampling technique on the variations of C _RNA showed conflicting results from the literature. A 10-fold increase in C _RNA from 24 h composite sampling than that of corresponding grab sampling was observed in primary effluent samples, presumably highlighting diurnal variability in the SARS-CoV-2 signal (Gerrity et al., 2021). In contrast, another study found that the sampling technique showed negligible impacts on C _RNA, where a good agreement between most grab samples and their respective composite samples was observed (Curtis et al., 2021). Thus, to avoid unnecessary loss of data points, studies using either grab sampling or composite sampling were all included in this study. The sampling and analytical methods (including the RT-qPCR primer-probe sets) applied in these 12 articles were summarized in Table S2. The average daily water usage (Q _w, L/person∙day) in each article was collected through the governmental reports for the investigated regions at the year of the study if available; otherwise, the data from the most recent year was included. The wastewater temperature in each study was estimated from (Hart and Halden, 2020b) based on the country and sampling day.

2.2. Monte-Carlo simulation of the theoretical C_RNA range

The Monte Carlo simulation is commonly used when the exact value of results cannot be computed with deterministic algorithms. The principle behind the Monte Carlo methodology is the law of large numbers in probabilistic statistical theory, where the frequency of the random event is approximately equal to the probability of event occurrence after repeated trials (Zhao et al., 2017). Therefore, by taking a prescribed number of samples from defined distributions for model input parameters, a Monte Carlo simulation provides probabilities of different outcomes occurring in an estimated probability distribution (Gilks, 2005).

Our previous study found that estimating the prevalence using Eq. (2) showed lower uncertainty in comparison to Eq. (1), due to the lower uncertainty of average water consumption data (used in Eq. (2)) than that of WWTPs influent flow rate and inhabitant population (used in Eq. (1)) (Li et al., 2021b). Thus, Eq. (2) was applied to simulate the theoretical C _RNA range for a certain COVID-19 prevalence. Recent studies reported that SARS-CoV-2 RNA decay in wastewater followed the first-order kinetics (Eq. (3)), where the time in wastewater and decay rate constant (k, increasing with higher temperature) were critical factors (Ahmed et al., 2020d; Bivins et al., 2020). Depending on the wastewater temperature, the k values ranged from 0.067 to 0.286/day under different temperatures (Ahmed et al., 2020d; Bivins et al., 2020). In this study, to simplify the simulation, the k value observed at 20–25 ℃ (common wastewater temperature of the WBE studies included in this paper as described in Section 2.2) was applied as 0.1/day. Thus, Eqs. (4) and (5) were subsequently established for the scenario with feces as the only shedding source, and both feces and sputum as the major shedding sources, respectively.

$$C_{\mathrm{RNA},\mathrm{t}}=C_{\mathrm{RNA},0}\times e^{-\mathit{kt}}$$ (3)

$$C_{\mathrm{RNA}}=\frac{P_{\mathrm{COVID}}\times e^{-0.1\times \mathit{HRT}}\times P_s\times Q_s\times {10}^{\mathit{Rs}}}{Q_w}$$ (4)

$$C_{\mathit{RNA}}=\frac{P_{\mathrm{COVID}}\times e^{-0.1\times \mathit{HRT}}\times (P_s\times Q_s\times {10}^{\mathit{Rs}}+P_{\mathit{sp}}\times Q_{\mathit{sp}}\times {10}^{\mathit{Rsp}}\times C_f)}{Q_w}$$ (5)

Where C _RNA,_t and C _RNA,₀ are the concentrations of SARS-CoV-2 RNA (gene copies/L) in wastewater at time t and time 0, respectively, and k is the decay rate constant (1/day) (Ahmed et al., 2020a, Ahmed et al., 2020b, Ahmed et al., 2020c, Ahmed et al., 2020d). P _s is the shedding probability in feces from a COVID-19 patient (-); Q _s is the daily amount of feces of an individual (g/person∙day); R _s is the logarithmic shedding magnitude of SARS-CoV-2 RNA in feces (log₁₀, gene copies/g), HRT (hydraulic retention time) is the in-sewer transportation time (day). P _sp is the shedding probability in sputum samples from a COVID-19 patient (-); Q _sp is the daily shedding amount of sputum of an individual (mL/person∙day); R _sp is the logarithmic shedding magnitude of SARS-CoV-2 RNA in sputum samples (log₁₀, gene copies/mL), C _f is the ratio of sputum that enters sewers (-).

For WBE surveillance of COVID-19, the population-wide viral RNA shedding information is critical. SARS-CoV-2 RNA shedding magnitude, probability for each shedding source among patients are largely impacted by physiological factors such as gender, age, and pathological conditions (Novazzi et al., 2020, Wang et al., 2020). Our previous meta-analysis summarized the clinical results from around 1500 patients covering all the gender, age groups (children, adults<60 and adults over 60), and pathological conditions (severe, moderate, and mild), which revealed that the mean shedding magnitude was 10^4.523±0.133 gene copies/g, and the mean shedding probability (P _s) was 0.545 ± 0.093 (Li et al., 2021b). Similar values were reported in other meta-level analyses where the positive proportion of the fecal samples was found to be 51.8% (95% CI 43.8–59.7%) and the median shedding concentration was 10^3.4 (95%CI: 10^0.24-10^6.5) to 10^4.9 (95%CI: 10^3.9-10^6.8) (Crank et al., 2022, Miura et al., 2021, van Doorn et al., 2020). The detection efficiency of clinical protocol for SARS-CoV-2 RNA in stool samples was confirmed using standardized stool samples (stool samples with no SARS-CoV-2 RNA) spiked with synthetic SARS-CoV-2 RNA, where comparable results to the initial spike-in concentrations were achieved (Poon and Tee, 2021), suggesting that the detected SARS-CoV-2 RNA concentration can largely reflect the actual concentration in stool samples. Thus, the distribution for R _s and P _s was applied as Gaussian distributions in the form of Normal (μ, σ²): i.e., Normal (4.523, 0.017) and Normal (0.545, 0.009), respectively. The probability density of R _s and P _s during the simulations was detailed in Figs. S1 and S2. The median feces mass was around 150–250 g/person∙day in healthy individuals (Rose et al., 2015). Gastrointestinal conditions have been observed in 12% of COVID-19 patients although the feces amount generated has not been reported (Walsh et al., 2020). Thus, feces mass was applied as 200 and 300 g/person∙day (to accommodate the presence of fecal shedding of COVID-19 patients with gastrointestinal conditions in the community) in this study to simulate the minimum and maximum shedding conditions.

The contribution of daily water usage into wastewater flow varied regionally, based on the WBE data collected (described in Section 2.1), the range was applied as 150–400 L/day∙person. In-sewer transportation time, or more specifically hydraulic retention time (HRT) of sewers, could also affect the decay of SARS-CoV-2 (Eq.3). Due to the lack of or inability to obtain HRT information of WBE studies (described in Section 2.1) and the unknown distribution pattern of COVID-19 patients in the catchment area, two HRTs, i.e. 30 min and 10 h, were applied to reflect the conditions of a small and large scale of WWTPs (McCall et al., 2017). Thus, two scenarios were applied for the theoretical C _RNA simulation with feces shedding: F1 (lower boundary) with the shortest HRT, the highest Q _s, and the lowest Q _w, and F2 (upper boundary) with longest HRT, the lowest Q _s, and the highest Q _w ( Table 2). The maximum SARS-CoV-2 RNA concentration in clinical stool samples of COVID-19 patients ranged from 10⁸-10^8.5 copies/g in clinical reports and meta-analysis (Jones et al., 2020, Wölfel et al., 2020). Thus, the maximum C _RNA with feces as the only shedding source was simulated with the maximum concentration of SARS-CoV-2 RNA detected clinically in feces samples (10^8.5 copies/g), the shortest HRT, the highest Q _s, and the lowest Q _w in scenario FM (Table 2).

Table 2.

The range of input parameters of C_RNA simulation for seven scenarios.

Shedding source	Simulation scenario		Code	Parameters
Feces	Mean shedding magnitude and probability	Upper boundary	F1	HRT=30 min, Qs=300 g, Qw=150 L, Ps~ Normal (0.545, 0.009)a, Rs~ Normal (4.523, 0.017)
		Lower boundary	F2	HRT=10 h, Qs=200 g, Qw=400 L, Ps~ Normal (0.545, 0.009), Rs~ Normal (4.523, 0.017)
	Maximum shedding magnitude	Maximum CRNA concentration	FM	HRT=30 min, Qs=300 g, Qw=150 L, Ps~ Normal (0.545, 0.009), Rs= 8.5
Sputum and feces	Sputum shedding magnitude follows Normal (6, 0.6)	Lower boundary (20% sputum entering sewers)	SPS1–1	HRT=6 h, Qs=300 g, Qw=150 L, Ps~ Normal (0.545, 0.009), Rs~ Normal (4.523, 0.017), Psp = 0.33, Rsp~ Normal (6, 0.6), Qsp~Normal (95, 9.5), Cf= 0.2
		Upper boundary (80% sputum entering sewers)	SPS1–2	HRT=6 h, Qs=300 g, Qw=150 L, Ps~ Normal (0.545, 0.009), Rs~ Normal (4.523, 0.017), Psp = 0.33, Rsp~ Normal (6, 0.6), Qsp~Normal (95, 9.5), Cf= 0.8
	Sputum shedding magnitude follows Normal (8, 0.8)	Lower boundary (20% sputum entering sewers)	SPS2–1	HRT=6 h, Qs=300 g, Qw=150 L, Ps~ Normal (0.545, 0.009), Rs~ Normal (4.523, 0.017), Psp = 0.33, Rsp~ Normal (8, 0.8), Qsp~Normal (95, 9.5), Cf= 0.2
		Upper boundary (80% sputum entering sewers)	SPS2–2	HRT=6 h, Qs=300 g, Qw=150 L, Ps~ Normal (0.545, 0.009), Rs~ Normal (4.523, 0.017), Psp = 0.33, Rsp~ Normal (8, 0.8), Qsp~Normal (95, 9.5), Cf= 0.8
	Maximum shedding magnitude and amount	Maximum CRNA concentration	SPSM	HRT=6 h, Qs=300 g, Qw=150 L, Ps~ Normal (0.545, 0.009), Rs= 8.5, Psp = 0.33, Rsp= 11.1, Qsp~Normal (95, 9.5), Cf= 0.8

Note: ^a: Normal distribution is in the form of Normal (μ, σ²). The parameter μ is the mean or expectation of the distribution, while the parameter σ is its standard deviation.

To date, the understanding of SARS-CoV-2 RNA shedding in sputum samples is limited. A study assessed the shedding magnitude in sputum samples of 80 patients, where the viral loads varied from 641–1.34 × 10¹¹ gene copies/mL with a median value of 7.52 × 10⁵ gene copies/mL (Pan et al., 2020). A similar range of 10^7.3-10^8.2 gene copies/mL was observed in another study with two patients (Yoon et al., 2020). To reflect the population-wide shedding, the R _sp (logarithmic shedding magnitude of SARS-CoV-2 RNA in sputum) were assumed as normal distributions in the form of (μ, σ²). In these reports, the σ² of R _sp was observed as 5.4% of μ (μ = 7.78) (Yoon et al., 2020) and 9.3% of μ (μ = 5.88) (Pan et al., 2020). Thus, two shedding magnitudes were assumed as normal distributions in the form of (μ, σ²) with 10% of μ applied for σ², as Normal (6, 0.6) for SPS1–1 and SPS1–2, and Normal (8, 0.8) for SPS2–1 and SPS2–2, respectively (Table 2). SPS1–1, SPS1–2, SPS2–1, SPS2–2 represent the simulations when sputum is considered as a major shedding source in addition to feces.

A recent meta-analysis of 970 patients revealed that sputum production was observed in one third of the patients (Lai et al., 2020) and the SARS-CoV-2 positive ratio was 98% in sputum samples of patients at 0–7 days after the symptom onset (Mohammadi et al., 2020). Based on the limited data, P _sp was determined as 0.33 (98%×1/3) to reflect the shedding probability among patients. Healthy individuals generally do not produce a large amount of sputum (Balbi et al., 2007) but the daily amount of sputum produced in patients with respiratory diseases was 94.6 ± 21.6 mL (Lin et al., 1997). Currently, the amount of daily sputum production in SARS-CoV-2 patients has not been reported yet. Considering the respiratory symptom of COVID-19 patients, Q _sp was adopted as Normal (95, 9.5) (Table 2). The amount of sputum entering sewers has not been investigated to date. Due to the complexity of waste disposal as recommended by the health departments, it is more likely a large portion of sputum from COVID-19 patients is discharged into sewers. To explore the lower and higher possibilities, the percentage of sputum entering sewers (C _f) was set as 0.2 (SPS1–1, SPS2–1) and 0.8 (SPS1–2, SPS2–2), respectively. To simplify the input scenarios, the common HRT in sewers (i.e., 6 h), the highest Q _s, and the lowest Q _w were applied for the scenarios using both sputum and feces as virus shedding sources into wastewater (Table 2).

Using Eqs. (4) and (5), the theoretical C _RNA range with 8 input scenarios as described in Table 2 was simulated through Monte Carlo models using OpenBUGS (version 3.2.3). The details of the models are included in the SI. To provide stable distributions of results, an initial simulation of 5000 iterations was first performed as a burn-in, and a further 5000 iterations were used for calculating outputs of the model simulations.

2.3. Comparison between detected C_RNA from WBE studies and simulated C_RNA

The C _RNA detected in wastewater with corresponding prevalence (clinically confirmed active COVID-19 cases per 100,000 people in the catchment area) was compared with the theoretical range simulated in different scenarios as described in Section 2.2. The number of data fitted into each scenario was counted. The difference between detected C _RNA and simulated C _RNA with the same prevalence in each scenario was evaluated by the root mean squared error (RMSE).

2.4. Assess the contribution of sputum to the total SARS-CoV-2 RNA shedding in wastewater

To further assess the contribution of sputum to the overall SARS-CoV-2 RNA shedding (feces and sputum together), the contribution of sputum shedding (CR) was calculated as Eq. (6).

$$\mathit{CR}=\frac{P_{\mathit{sp}}\times Q_{\mathit{sp}}\times {10}^{\mathit{Rsp}}\times C_f}{P_s\times Q_s\times {10}^{\mathit{Rs}}+P_{\mathit{sp}}\times Q_{\mathit{sp}}\times {10}^{\mathit{Rsp}}\times C_f}$$ (6)

For fecal shedding, the maximum shedding quantity (Q _s, 300 g), the mean logarithmic shedding magnitude (R _s~ Normal (4.523, 0.017)), and the mean shedding probability (P _s~ Normal (0.545, 0.009)) were applied. For sputum shedding, the shedding probability (P _sp) and shedding quantity (Q _sp) was applied as 0.33 and Normal (95, 9.5), respectively. Seven different ratios of sputum that enters sewers (C _f) (i.e., 0.01, 0.05, 0.1, 0.2, 0.4, 0.6, 0.8) and two shedding magnitude (R _sp~ Normal (6, 0.6), R _sp~ Normal (8, 0.8)) were used to reflect the different shedding scenarios (14 scenarios in total). The mean of CR with standard errors under each shedding scenario was simulated through Monte Carlo models using OpenBUGS (version 3.2.3). The details of the models are included in the SI. To provide stable distributions of results, an initial simulation of 5000 iterations was first performed as a burn-in, and a further 5000 iterations were used for calculating outputs of the model simulations.

3. Results

3.1. Comparison between detected C_RNA from WBE studies and simulated C_RNA with and without sputum included as a shedding source

During the time span for each study, although these countries had different contacting tracing and clinical testing policies (text S1.1), the testing positive rates (daily positive cases/daily total tests) were all lower than 10% (Table S1), indicating a good coverage of the clinical testing (Saglietto et al., 2020). The WBE data were collected from eight different countries with the outbreak at the initial stage (within 1 detection window (i.e. 28 days) of SARS-CoV-2 for wastewater samples after the first COVID-19 patient) or later stage (after the initial stage) ( Fig. 1). At the initial stage, the number of active cases equals cumulative cases in the catchment area. At the later stage, the prevalence of COVID-19 in the catchment area was calculated based on the active cases (confirmed cases minus recovered cases).

Fig. 1.

Comparison of reported WBE data against the simulated C_RNA with 95% confidence interval under a certain prevalence for feces as the shedding source (A) and both feces and sputum as shedding sources (B). The C_RNA (copies/L) and prevalence (clinically confirmed COVID-19 cases/100,000 person) are illustrated on logarithmic scales. The color and shape of the points represent the country, and the outbreak stage of the WBE data, respectively. The initial stage of the outbreak is within 1 detection window (i.e. 28 days) of SARS-CoV-2 for wastewater samples after the first COVID-19 patient, and the later stage is after 1 detection window of the first patients in the catchment area (Ahmed et al., 2020a).

With feces as the only shedding source, the theoretical C _RNA range was estimated based on the mean shedding magnitude and probability in F1 (upper boundary) and F2 (lower boundary) (Fig. 1A). None of the WBE data points was in the range of F1 and F2, suggesting that the C _RNA detected were all higher than the theoretically estimated values. Furthermore, 63% of the reported WBE data points were above the confidence interval of FM (Fig. 1A). The data above the confidence interval of FM did not show a clear correlation with the outbreak stage or the country (Fig. 1A), suggesting consistent higher shedding loads than fecal shedding, regardless of the regional or temporal difference. FM was established based on the highest fecal shedding magnitude, the shortest HRT, maximum daily feces mass amount and the lowest daily water usage (Table 2). The feces shedding magnitude in FM was only observed in some clinical cases during a short period (2 days) as reported by (van Doorn et al., 2020). In addition, the limit of detection (LOD) of SARS-CoV-2 RNA was observed to be around 10³ copies/L (Ahmed et al., 2020b). With the mean shedding magnitude, only COVID-19 prevalence higher than 1000/100,000 can be captured by WBE (F1 and F2 in Fig. 1A), which conflicts with the successful detection of C _RNA in wastewater under much lower prevalence.

The maximum C _RNA with both feces and sputum as the shedding source was estimated with the highest sputum (10^11.1copies/mL) and fecal shedding magnitude (10^8.5 copies/g) observed in clinical cases (SPSM in Fig. 1B). The fixed shedding magnitude rather in SPSM than a Gaussian distribution in other scenarios (SPS1–1 to SPS 2–2) leads to a narrow confidence interval in SPSM. The region below SPSM reflects the greatest probability of the occurrence of WBE data. 91% of the WBE data fell into the range below SPSM, in contrast to that of 37% in FM (Fig. 1A). This suggests that sputum was likely another critical shedding source for SARS-CoV-2 into sewers, and a high proportion of sputum generated by COVID-19 patients was potentially discharged into sewers. Four scenarios were applied for estimating the C _RNA under a certain prevalence with both sputum and feces as major shedding sources (Table 2). For better readability, the highest (SPS2–2) and lowest (SPS1–1) theoretical C _RNA range were included in Fig. 1B, and two other scenarios (SPS1–2, and SPS2–1) are included in Fig. S3.

SPS1–1 and SPS1–2, and SPS2–1 and SPS2–2 were simulated under the sputum shedding magnitude of Normal (6, 0.6) and Normal (8, 0.8), respectively, where the upper (SPS1–2, SPS2–2) and lower boundary (SPS 1–1, SPS 2–1) under each shedding magnitude were estimated by assuming 20% and 80% of sputum entering sewers, respectively (Table 2). In these four scenarios, only three of the WBE data were in the confidence interval of SPS1–1, while about 49% of the WBE data was in the range of SPS2–2 (Fig. 1B). The RMSE of logarithmic C _RNA was 3.73, 3.20, 1.84, and 1.45 in SPS1–1, SPS1–2, SPS2–1, and SPS2–2, respectively. Thus, the range of C _RNA estimated under a higher sputum shedding magnitude and amount entering sewers was more comparable to the concentration detected in WBE studies. More importantly, when the sputum shedding load (magnitude× amount entering sewers) increased by about 100 times, the median C _RNA increased by about 50 times from SPS1–1 to SPS 2–1, but about 100 times from SPS1–2 to SPS 2–2 (Figs. 1B, S3). This implies a dominant role of sputum shedding at higher shedding loads for the prevalence estimation through WBE approach.

3.2. Contribution of sputum to the total SARS-CoV-2 RNA shedding in wastewater

The contribution of sputum discharging behaviors of patients to the overall SARS-CoV-2 RNA shedding (feces and sputum together) was further evaluated under different ratios of sputum entering sewers with two shedding magnitudes (Magnitude1 and Magnitude 2) ( Fig. 2). It is evident that sputum shedding played a critical role in the total virus shedding, especially under higher shedding magnitude and or higher ratio entering the sewers. Under the lower shedding magnitude (Magnitude 1), sputum may contribute from 13% to 73% of the total virus in wastewater, when the ratio of sputum entering sewers increase from 1% to 80%. With the higher shedding magnitude of sputum (Magnitude 2), even with 1% of sputum entering the sewers, the contribution of sputum reached 74% of the total virus RNA in wastewater. Furthermore, sputum became the dominant shedding source (>90% contribution) with > 10% of sputum entering sewers under the higher shedding load (Magnitude 2), where the fecal shedding became negligible.

Fig. 2.

Contribution of sputum to the total virus shedding under two sputum shedding magnitudes. Magnitude 1 and Magnitude 2 represent R_sp~ Normal (6, 0.6), and R_sp~ Normal (8, 0.8), respectively. The error bar represents the standard error of 5000 simulation results. For some data points, the error bars are shorter than the height of the symbol and thus are not visible.

The personal sputum discharging practice also greatly affected the range of theoretical C _RNA. With both feces and sputum as shedding sources, the increase of sputum entering sewer from 20% to 80% led to 2.5 times higher median C _RNA with the lowest shedding magnitude, but 4 times increase in C _RNA with the highest shedding magnitude (Fig. S3). More importantly, with the least shedding magnitude of sputum (Magnitude 1), 20% of sputum generated entering sewers would increase the C _RNA by about 2 times compared with only fecal shedding. With a higher shedding magnitude (Magnitude 2), even 1% of sputum entering sewers would increase the C _RNA by 5 times. Higher sputum shedding loads (magnitude × amount entering sewers) would lead to up to 70 times increase of the C _RNA in comparison to the results from fecal shedding alone.

4. Discussion

4.1. Feces contributes partly to detected SARS-CoV-2 RNA in wastewater

Through Monte-Carlo simulations, the C _RNA detected in WBE studies showed much higher values than their theoretical range under corresponding prevalence. More importantly, 63% of the reported WBE data points were above the confidence interval of the maximum shedding condition (FM). Previous studies also noticed that the C _RNA detected in wastewater was higher than the theoretical values estimated by the clinically confirmed prevalence using feces as the shedding source (Wu et al., 2020a, Wu et al., 2020b). This was attributed to a hypothetical surge of shedding (either from feces or other sources) before the symptom onset at several orders of magnitude greater than typical values. However, a recent study monitored the C _RNA in wastewater of a university campus with intense on-campus case surveillance, which revealed that individual shedding of RNA (sources unclear) into wastewater peaks on average six days after the symptom onset (95% UI: 4–8 days) (Cavany et al., 2021). Also, most of the virological assessments were carried out on patients with confirmed symptoms or clinical tests by oropharyngeal or nasopharyngeal swabs. Thus, the hypothetical surge of shedding before symptom onset is debatable.

Some other factors, such as the shedding from asymptomatic patients or uncaptured cases by clinical testing, analytical recovery efficiency, in-sewer RNA decay, and duration shedding could also affect the C _RNA detected in the wastewater although they were not included in the simulation. Thus, the differences between simulation results and the actual C _RNA detected in the wastewater, and the contributions of the above factors were further compared and discussed ( Table 3).

Table 3.

Potential contributions of additional factors to the simulated wastewater concentration of SARS-CoV-2 RNA.

Factors	Value	Changes to the simulated CRNA
		Increase (times)	Decrease (times)
Percentage of asymptomatic patients or uncaptured cases	15.6–76.7%	0.2–3.3	–
Analytical recovery efficiency	10–72%	–	0.4–9.0
In-sewer RNA decay	k = 0.084–0.286/day, HRT= 1 min to 10 h	0–1.0	0–1.0
Shedding duration	Up to four to ten weeks after the first symptom onset	–	–
Total		0.2–7.6	0.4–19.0

SARS-CoV-2 virus shedding has been found in feces of asymptomatic patients (Park et al., 2020, Tang et al., 2020, Zhang et al., 2020a, Zhang et al., 2020b). The COVID-19 prevalence was predominately confirmed through clinical testing of individuals. Symptom-onset can be a major trigger for the motivation of testing, in addition to some mandatory testing required for cross-border travelers or close contacts of infected patients (Table S1). A meta-analysis conducted in July 2020 (similar period as the WBE studies) estimated that the percentage of asymptomatic patients among COVID-19 patients was 15.6% (95% CI, 10.1%‐23.0%) (He et al., 2021). The Centre for Disease Control and Prevention (CDC) also estimated that 1 in 4.3 (95% UI 3.7–5.0) of total COVID-19 infections were reported (Reese et al., 2020). Therefore, considering the potential shedding from asymptomatic patients or uncaptured cases, the theoretical C _RNA would increase by 0.2–3.3 times.

The analytical approach (i.e. concentration, extraction, and detection) applied for wastewater could greatly affect the recovery efficiency of the SARS-CoV-2 RNA from wastewater, which thereby changes the C _RNA detected in wastewater (Li et al., 2021b, Rusiñol et al., 2020). The large variation of C _RNA is partly due to the limitations of wastewater analysis using RT-qPCR, as shown by the SARS-CoV-2 interlaboratory consortium report (Pecson et al., 2021). Generally, to further correct the C _RNA detected in wastewater, the recovery efficiency is quantified by spiking low-pathogenic surrogate viruses as external controls or using internal controls such as fecal load indicators (e.g. pepper mild mottle virus) (Ahmed et al., 2020c, Jafferali et al., 2020, Wu et al., 2020b). The recovery efficiency of the included articles varied from 10% to 72% (Table S2). Thus, the analytical approach could lead to 0.4–9 times decrease of theoretical C _RNA (Table 3).

The decay of SARS-CoV-2 RNA during in-sewer transportation was found to follow the first-order kinetics as Eq. (3), with k values increased from 0.084/day to 0.286/day from 4 ℃ to 37 ℃ (Ahmed et al., 2020d). Apart from the k value, the traveling time t (HRT of sewers) also impacts the decay of SARS-CoV-2 RNA in wastewater (Eq. (3)). For most of the WWTPs, the HRT of sewers ranged from several minutes to 6–10 h in small and large scale WWTPs, respectively (McCall et al., 2017). In this study, the k (0.1/day) value at 20 ℃ (common wastewater temperature in these WBE studies) was applied for simulation. The HRT of 30 min and 10 h was considered to simulate the upper (F1) and lower boundary (F2) for feces as the only shedding source. Thus, changes in wastewater temperature or HRT would potentially lead to a further 1-time increase for the upper boundary (k = 0.286/day, HRT=1–30 min), or a 1-time decrease for the lower boundary (k = 0.084/day, HRT=10 h) (Table 3). In addition, prolonged fecal shedding has been observed in patients for up to four to ten weeks after the first symptom onset and even after the patients' respiratory samples tested negative for SARS-CoV-2 RNA (Wu et al., 2020c, Xiao et al., 2020, Xu et al., 2020, Yang et al., 2020). However, a meta-analysis involving thousands of patients revealed that the shedding loads of recovered patients (after respiratory samples tested negative) was about 0.3–6.2% of the active patients (Jones et al., 2020). Thus, the contribution of prolonged shedding of recovered cases is negligible in comparison to active cases.

Compared with the reduction of C _RNA due to the analytical recovery efficiency, the increase of C _RNA caused by asymptomatic patients or uncaptured cases, and in-sewer decay is limited (Table 3). Even with 100% analytical recovery, the inclusion of all the factors would elevate the simulation results of C _RNA by 0.2–7.6 times (Table 3). With elevated maximum shedding condition (FM×7.6), 40% of the reported WBE data points were still above the confidence interval. Furthermore, the C _RNA detected in WBE studies was averagely 10^2.81 to 10^4.20 times higher than the upper (F1) and lower (F2) boundary of fecal shedding, respectively. Some C _RNA detected in wastewater were even about 10⁴ times higher than the theoretical maximum C _RNA with feces as the only shedding source (Fig. 1A). Thus, feces only contribute partly as a shedding source to SARS-CoV-2 RNA in wastewater.

4.2. Potential role of sputum as a major SARS-CoV-2 RNA shedding source

Through the simulation, it is evident that the range of C _RNA estimated under a higher sputum shedding magnitude was more comparable to the concentration detected in WBE studies. At higher shedding loads (magnitude × amount entering sewers), sputum became the dominant source, where the impact of fecal shedding becomes negligible. Thus, the hygiene behavior of patients in regards to the disposal of sputum becomes more important for the WBE surveillance of COVID-19.

As aforementioned, a clear shedding load or constant shedding behavior from patients is crucial, either for the WBE surveillance of COVID-19 through the conventional back-estimation approach or the correlation between C _RNA and COVID-19 prevalence or incidence. Our simulation results suggest that, with a higher shedding magnitude, even 1% of sputum entering sewers, the C _RNA would be increased by 5 times. Higher sputum shedding loads (magnitude × amount entering sewers) would lead to up to 70 times increase of the C _RNA in comparison to the results from fecal shedding alone. Even with both feces and sputum considered as shedding sources, the increase of sputum entering sewers from 20% to 80% led to 2.5 times higher median C _RNA with the lowest shedding magnitude, but 4 times increase in C _RNA with the highest shedding magnitude. Thus, the change of person sputum discharge behavior would introduce great variations into the WBE surveillance of COVID-19. Highly over-dispersed viral shedding was also observed in some WBE studies (Cavany et al., 2021), which is likely related to the personal sputum discharging behaviors. To date, the instructions regarding sputum disposal remain unclear yet for COVID-19 patients. Although discharging sputum into sewers is intuitively safer during self-isolation, the actual amount of sputum entering sewers is likely dispersed and varied greatly among the population. Thus, future research regarding the sputum discharging behavior among patients and clear instructions for sputum disposal are crucial for the WBE surveillance of COVID-19.

In addition, with both feces and sputum as the major shedding source, 9% of WBE data points were above SPSM (maximum fecal+sputum shedding) (Fig. 1B). The RMSE of logarithmic C _RNA with both sputum and feces as the shedding source ranged from 1.45 to 3.73 with the change of shedding loads of sputum (SPS1–1 to SPS2–2). As discussed in Section 4.1, other factors such as the shedding from asymptomatic patients or uncaptured cases by clinical testing, and in-sewer RNA decay as listed in Table 3 would also increase the C _RNA detected in the wastewater by up to 7.6 times. This could further reduce the RMSE of logarithmic C _RNA to 1.25–2.88, which is more comparable to the WBE data, resulting in 96% of WBE data falling below the elevated SPSM (SPSM×7.6). The remaining 4% of WBE data that was still above the elevated SPSM might be caused by the presence of other shedding sources. SARS-CoV-2 RNA was also detected in the water used for mouth/throat wash of COVID-19 patients at around 10²-10⁵ gene copies/mL (Liu et al., 2020). This could be caused by the presence of SARS-CoV-2 RNA in saliva, which might be another potential source of SARS-CoV-2 RNA in wastewater (Huang et al., 2021). The SARS-CoV-2 RNA load and detection rate in saliva were found comparable to that of sputum and respiratory tract samples (Zhu et al., 2020). However, due to the limited understanding of the concentration, detection probability, and daily discharge amount of mouth/throat wash or saliva into sewers, the contribution of mouth/throat wash and saliva was not included in the simulation, which requires future investigations.

The country and outbreak stage showed a negligible difference in the distribution of WBE data in comparison to the simulation results from fecal shedding (Fig. 1A). However, the WBE studies from the USA and France had more data points above the SPSM (maximum fecal+sputum shedding) (Fig. 1B). This might be related to the higher shedding load of SARS-CoV-2 RNA (from either feces, or sputum, or both) or sputum discharging practice from COVID-19 patients in these countries. To date, the impact of race, socioeconomic conditions, and country on shedding dynamics in COVID-19 patients remains unclear, which requires future investigations.

5. Implications, limitations, and future research recommendations

Our results suggest that feces is unlikely to be the only shedding source based on currently available WBE data. Sputum might play an important role as a shedding source to the sewer system. The theoretical ranges of SARS-CoV-2 RNA in wastewater estimated under higher sputum shedding magnitude (around 10⁸ copies/mL) were more comparable to the WBE data. Sputum shedding became the dominant source for WBE estimations under such scenarios. The discharge of sputum into sewers can lead to up to 70 times increase of SARS-CoV-2 RNA in wastewater while other factors such as asymptomatic or uncaptured cases by clinically testing, analytical recovery efficiency, in-sewer decay and prolonged shedding from recovered patients have comparably limited impacts on the change of C _RNA in wastewater in comparison to sputum shedding. However, sputum was not previously considered as a major source of SARS-CoV-2 shedding. WBE investigations were primarily focusing on feces detection and quantifications. Due to the lack of such information, the sputum shedding magnitude in this study was based on the data from two studies with limited numbers of patients. The data points of sputum shedding magnitude were comparatively fewer than that of fecal shedding (i.e. 1500 patients). The daily sputum amount generated by a COVID-19 patient was estimated based on patients with respiratory diseases. Furthermore, the amount of sputum entering the sewers will vary depending on the personal hygiene practices and recommendations provided by the relevant health agency. This study provided two scenarios with 20% and 80% of sputum entering sewers and found the percentage of sputum entering sewers became more important under higher sputum shedding loads. The contribution of sputum was further assessed with 1–80% entering sewers. Even 1% of sputum entering sewers under the higher shedding magnitude would increase the C _RNA in wastewater by 5 times, which would greatly impact WBE surveillance of COVID-19. Based on the currently available data on sputum shedding, the simulation results in this study suggest a critical role of sputum shedding on the COVID-19 WBE estimations. A comprehensive survey for the sputum discharge practice of COVID-19 patients and the amount of virus load in the sputum are recommended for future investigations. When such information becomes available, the contribution of sputum shedding would be further validated or evaluated based on stronger clinical evidence.

In addition, to date, there is a lack of assessment regarding the geological and or temporal differences in the contribution of daily water usage into wastewater flow. Since the WBE data included in this study was collected from eight countries (Australia, Canada, France, Germany, Japan, Netherlands, Spain, and the USA), the C _RNA was simulated based on their water usage. In other countries with different water usage cultures, the theoretical C _RNA is likely to be different and can be biased due to limited sanitation coverage.

CRediT authorship contribution statement

Xuan Li: Conceptualization, Methodology, Data curation, Writing – original draft. Jagadeeshkumar Kulandaivelu: Data curation, Writing – review & editing. Ying Guo: Data curation. Shuxin Zhang: Data curation. Jiahua Shi: Data curation. Jake O’Brien: Writing – review & editing. Sudipti Arora: Writing – review & editing. Manish Kumar: Writing – review & editing. Samendra P. Sherchan: Writing – review & editing. Ryo Honda: Writing – review & editing. Greg Jackson: Writing – review & editing. Stephen P. Luby: Writing – review & editing. Guangming Jiang: Conceptualization, Methodology, Writing – review & editing, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Editor: Danmeng Shuai

Appendix A. Supplementary material

Supplementary material

.