key: cord-0765112-tsusmy3p
authors: Pillay, Leanne; Amoah, Isaac Dennis; Deepnarain, Nashia; Pillay, Kriveshin; Awolusi, Oluyemi Olatunji; Kumari, Sheena; Bux, Faizal
title: Monitoring changes in COVID-19 infection using wastewater-based epidemiology: A south African perspective
date: 2021-04-23
journal: Sci Total Environ
DOI: 10.1016/j.scitotenv.2021.147273
sha: ff47c3773512d6a108f8d66f3ebfec5e15d94b28
doc_id: 765112
cord_uid: tsusmy3p

Monitoring of COVID-19 infections within communities via wastewater-based epidemiology could provide a cost-effective alternative to clinical testing. This approach, however, still requires improvement for its efficient application. In this paper, we present the use of wastewater-based epidemiology in monitoring COVID-19 infection dynamics in the KwaZulu-Natal province of South Africa, focusing on four wastewater treatment plants for 14 weeks. The SARS-CoV-2 viral load in influent wastewater was determined using droplet digital PCR, and the number of people infected was estimated using published models as well as using a modified model to improve efficiency. On average, viral loads ranged between 0 and 2.73 × 105 copies/100 ml, 0–1.52 × 105 copies/100 ml, 3 × 104–7.32 × 105 copies/100 ml and 1.55 × 104–4.12 × 105 copies/100 ml in the four wastewater treatment plants studied. The peak in viral load corresponded to the reported COVID-19 infections within the districts where these catchments are located. In addition, we also observed that easing of lockdown restrictions by authorities corresponded with an increase in viral load in the untreated wastewater. Estimation of infection numbers based on the viral load showed that a higher number of people could potentially be infected, compared to the number of cases reported based on clinical testing. The findings reported in this paper contributes to the field of wastewater-based epidemiology for COVID-19 surveillance, whilst highlighting some of the challenges associated with this approach, especially in developing countries.

were selected, treating an average of 14 ML/d and 80 ML/d of wastewater respectively. The eThekwini district is largely made up of Durban, a commercial and industrial hub of the KZN province. It is a coastal city along the Indian Ocean and is a major tourist destination in South Africa. Pietermaritzburg, the capital of the KZN province, is also a major industrial hub, mainly producing aluminium, timber, and dairy products, and is the main city in the Umgungundlovu district. Grab samples (2 L) of raw sewage were collected at the head of works (post-primary screening) for each of the WWTPs weekly between the peak hours (7:00 -11:00 am). Sampling was carried out for approximately four months, from July to October 2020. Full personal protective equipment (PPE) (Face shield, FFP2 face mask, waterproof coveralls, and safety boots) was worn during each sampling event.

Samples were heat-treated at 60 °C for 90 min immediately upon arrival in the laboratory (within two hours of sampling) and were left to cool to room temperature before further analysis was done. Thereafter, samples were mixed thoroughly, and 250 ml aliquots were removed for processing and the rest of the samples were stored at -80°C. The 250 ml sample from each WWTP was then equally divided into 50 ml centrifuge tubes and clarified by centrifugation at 3500 x g for 10 min. The supernatants from each tube were then pooled and used for viral concentration while the pellets were stored at -80 °C for future analysis. The method of ultrafiltration was used to concentrate the virus particles as previously described by Medema et al. (2020) . Briefly, 60 ml of supernatant was filtered through a Centricon® Plus-70 centrifugal ultrafilter with a cut-off of 10 kDa at 3500 x g for 30 minutes. The volume of the resulting concentrate for each sample varied due to the composition of the sample matrix. In instances where the concentrate was less than 140 µl, (the minimum J o u r n a l P r e -p r o o f amount required for RNA extraction) the concentrate was topped up to 140 µl using PBS.

The volumes of the concentrate recovered as well as the volume of PBS added (if required) were recorded to account for the dilution effect.

The RNA was extracted from 140 µl of viral concentrate using the QiAmp Viral RNA MiniKit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. RNA was eluted in 80 µl of sterile nuclease-free water and then quantified using the Implen Nanophotometer®. The extracted RNA was then stored at -80 ºC for further analysis.

Detection of SARS-CoV-2 in the wastewater samples was performed as described by Amoah et al (2020: preprint) with some modifications. The current method involved the use of the One-Step RT-ddPCR Advanced Kit for Probes from Biorad (USA) together with primers and probes targeting the N2 region of the viral (SARS-CoV-2) genome. The N2 gene was chosen as it is the most widely used target gene in SARS-CoV-2 detection assays and it had the lowest limit of detection during preliminary tests in our laboratory. Other researchers have identified N2 as a good target for amplification compared to either N1 or N3 (Randazzo et al., 2020 , Shirato et al., 2020 . The 22 µl reaction mixture contained: 5 µl supermix, 2 µl reverse transcriptase, 1 µl dithiothreitol, 1.98 µl each of the forward and reverse primers (10 µM), 0.55 µl of 10 µM probe, 2.49 µl nuclease-free water, and 7 µl template RNA (1 ng). The 96well plates containing the reaction mixtures were then sealed, vortexed, and centrifuged.

Droplets were generated using the QXDx Automated Droplet Generator (Biorad, USA). The primer and probe sequences, together with the thermal cycling conditions can be found in Table 1 . The results after thermal cycling were read in the QX200 Droplet Reader, using the QuantaSoft 1.7 software (Biorad, USA) while further analysis was carried out using the QuantaSoft Analysis Pro 1.0 software (Biorad, USA).

To assess the efficiency of the concentration and extraction methods employed in this study, a 200 l suspension (corresponding to 3360 copies of N2 gene) of inactivated SARS-CoV-2 strain USA/WA1/2020 (Microbiologics, USA) was seeded into 400 ml of raw wastewater and mixed thoroughly. Thereafter, the samples were separated into 8 x 50ml centrifuge tubes (each containing approximately 420 copies of N2) and processed according to the methods described above and thereafter assessed for the presence of SARS-CoV-2. Additionally, unseeded wastewater was analysed in the same manner to determine the background concentrations of SARS-CoV-2. The ddPCR assay to determine the concentration of the N2 gene was conducted on the same day as RNA extraction to avoid any losses in RNA integrity that may result from storing and/or freeze thawing of the RNA. The entire process was also tested to determine the effect of possible inhibitors present in the wastewater matrix. This was done by spiking 120 ml of sterile Milli-Q water with 60 l of inactivated SARS-CoV-2 suspension and processing it under the same conditions. The recovery efficiency obtained using Milli-Q water was then compared with that of wastewater.

The recovery efficiency of SARS-CoV-2 for wastewater and Milli-Q water was calculated using the following equation:

Where C SW is the concentration of SARS-CoV-2 detected in spiked wastewater or Milli-Q water C UW is the concentration of SARS-CoV-2 detected in un-spiked wastewater or Milli-Q water.

Taking into consideration that 60 ml of a spiked sample contains 504 copies of N2, the recovery percentage for wastewater was calculated at 62.86 (±12.84) % as the concentration of N2 quantified via ddPCR was 144.32 (±14.93) copies/60 ml and 461.12 (±79.64) J o u r n a l P r e -p r o o f copies/60 ml in un-spiked and spiked samples respectively. For Milli-Q water, the recovery percentage was 78.62 (±1.79) % as there was 396.23 (±9.96) copies/60 ml of N2 gene detected in the spiked sample while no target genes were present in the un-spiked (raw) sample. The differences in the recovery efficiency between the wastewater and Milli-Q possibly indicate the presence of inhibitors in the wastewater sample.

Quality checks for the method used in this study were performed with the addition of positive, negative, and no-template controls, which were added to each plate/run during the ddPCR process. The positive controls contained synthetic RNA targeting 5 regions (E, N, ORF1ab, RdRP, and S genes) of the SARS-CoV-2 viral genome. Human genomic DNA and RNA contained in a synthetic matrix were used as negative control. These positive and negative controls were supplied by Exact Diagnostics (USA). No template controls were sterile nuclease-free water.

Estimation of the number of people infected within the communities connected to the wastewater treatment plants sampled was done using the prediction model published by Ahmed et al., (2020) 

We used the same input data as published by Ahmed et al., (2020) , except for the RNA copies per liter of wastewater, which was taken from the data generated from our study using the ddPCR protocol described above. Furthermore, we used different input data for the daily stool mass per person. This was because Ahmed et al., (2020) used daily stool mass per person representing high-income countries. The daily stool mass produced per person specific for South Africa was therefore used, and this was taken from Burkitt et al., (1972) . These were modeled as a normal distribution with a mean of 2.07 log 10 and a standard deviation of 1.08 log 10 . The amount of SARS-CoV-2 shed per gram of feces was modeled as a loguniform distribution, with minimum and maximum values of 2.56 and 7.67respectively (Wölfel et al., 2020) . To reduce variability, and improve on the outputs of the models, Monte

Carlo Simulations using 10 000 iterations were performed. All models were built with @Risk (Palisade Corp, USA) addon to Excel (Microsoft Corp, USA).

To improve the accuracy of the prediction model in estimating the number of people infected, we added the viral load shed per mL of urine in infected persons, as well as the recovery percentage of viral particles in the wastewater. The viral load shed per mL of urine was taken to be 2.50 Log 10 (Peng et al., 2020) . The volume of urine produced per person per day was modelled as a log-uniform distribution with a minimum of 2.78 and a maximum of 3.76 (Lemann et al., 1996) .

In this study we measured the recovery efficiency to be 62.86 (±12.84) % of inactivated SARS-CoV-2 spiked into untreated wastewater. We, therefore, modelled the recovery efficiency as a uniform distribution with a minimum recovery of 50.02% and a maximum of 0.2 copies/µl (See Figure S1 in Appendix I). The observed changes in the viral loads in the wastewater throughout the study, as described above, were statistically significant at a 95% confidence interval (p value ≤0.05). Viral loads recorded for the majority of this study are significantly higher than those reported by Ahmed et al. (2020) , Serchan et al. (2020) , Wu et al. (2020) , and Randazzo et al. (2020) amongst many others. However, the discrepancies in results could be attributed to differences in disease prevalence and the efficiency of the various processes involved in the detection and quantification of SARS-CoV-2 from wastewater. It must also be noted that the higher viral loads observed in this study may also be partly attributed to the use of the ddPCR platform, which has been reported to have lower limits of detection (0.2 copies/µl) and is more sensitive and accurate than the RT-qPCR methods that are currently being used in majority of WBE studies (Lu et al., 2020; Zhou et al., 2020) . 

The current study began one month after the implementation of the level 3 lockdown. When stay-at-home regulations were eased on 17 August 2020 (level 3 to level 2) and 21 September 2020 (level 2 to level 1) people were now allowed to move freely across the country for recreational purposes and many returned to the city for work-related purposes. As social and economic activity resumed within the country, it was expected that this would reflect in the wastewater as many studies have already demonstrated how analysis of population pooled wastewater can represent the activity and lifestyle of a given community (Sims and Kasprzyk-Hordem, 2020). The results presented in Figure 2 demonstrates this effect. Using the Central WWTP as an example, viral loads increased from 2.25 x 10 5 copies/100 ml (11 August 2020) to 7.32 x 10 5 copies/100 ml after the move from level 3 to level 2. Wastewater

Journal Pre-proof samples collected from the Central, Isipingo, and Howick WWTPs on 29 September 2020 (after the start of lockdown level 1) was indicative of this trend as well. Where viral loads of 2.43 x 10 5 copies/100 ml, 1.09 x 10 5 copies/100 ml and 1.26 x 10 5 copies/100 ml for Central, Isipingo, and Howick WWTPs respectively were recorded. The findings of our study are following that of Wurtzer et al. (2020) who evaluated the effect of lockdown regulations on SARS-CoV-2 dynamics in wastewater in Paris. Wurtzer et al. (2020) observed that the WBE approach was not necessarily an early warning system and concluded that quantitative monitoring of wastewater is a time-related indicator of the health status of the community.

The findings of our study agree with this observation by Wurtzer et al. (2020) .

In addition to its utility as a surveillance tool, the potential of WBE as a predictive tool has been reported. For instance, it was used to assess the circulation of polio in populations WWTPs followed a similar trend as the SARS-CoV-2 viral loads measured, corresponding to peaks in reported clinical active cases as well. However, our estimated numbers were J o u r n a l P r e -p r o o f significantly higher than the total number of active cases for the two districts within which these WWTPs are located. This could be attributed to two factors, firstly under clinical testing, which may result in a lower number of active cases compared to the actual number of infected persons within the district. For instance, South Africa has a testing rate of 54, 224

tests per million population (Chitungo et al., 2020) , which is lower than the testing rate in the UK, USA, and many other developed countries. However, comparatively, South Africa has a better testing rate than any other African country and most countries in Asia (Chitungo et al., 2020) . It is estimated that for every 10 diagnosed infections, there are 7 undiagnosed infections . Therefore, with lower testing the number of infections reported by health authorities could be much lower; therefore, our estimates could be a true reflection of the actual number of people infected within the catchments. Secondly, the higher estimated number of infected individuals could be attributed to overestimation by the models employed. For instance, sensitivity analysis by Ahmed et al., (2020) (Campioli et al., 2020; Xu et al., 2020; Long et al., 2020; Li et al., 2020) . It is worth noting there is no information on the viral load shed in stool by infected individuals in South Africa. This creates an information gap and could potentially account for the disparity in the estimated infection number using the predictive models and the reported actives cases.

Weight of stool produced per person per day: In addition to information on the concentration of SARS-CoV-2 shed by each infected person, an accurate estimation of the number of infected people will be affected by the weight of stool produced per person per day. This information varies from region to region and could be impacted by several factors (Rose et al., 2015) . For instance, Ahmed et al (2020) used stool production per person per day from high-income countries as reported by Rose et al., (2015) . This data cannot be used for a middle-income country like South Africa.

Therefore, the local supporting information is critical. In this study, we had to rely on J o u r n a l P r e -p r o o f old data from 1972 on the weight/quantity of stool produced per person per day in South Africa. These values could have changed over the years, therefore update information might have improved the accuracy of the predictive model.

Considering all of the above, correlating epidemiological data with viral loads in wastewater is currently difficult as not all COVID-19 carriers are included in epidemiological data, making the implementation of the WBE surveillance strategy complicated for many parts of the world (Polo et al., 2020) . Based on this we will like to make the following recommendations:

1) There is an urgent need for a univocal testing framework, which takes into consideration the different analytical sensitivities of each step in the testing process (especially with PCR assays and platforms) (Michael-Kordatou et al., 2020) . This framework should include a sampling approach that will ensure an accurate representation of viral load within the catchment. This could involve a more frequent sampling regime or the use of autosamplers to make composite samples.

2) Changes in environmental conditions and the unique signature of each WWTP are probably the most important contributing factors to the variability of the WBE approach. In the environment, rainfall events and temperature play an important role and affect the dilution and stability of the virus in water (La Rosa et al., 2020) . While in WWTPs, it is imperative to know hydraulic retention times, peak flow rates, as well as the size and configuration of sewer networks in each community. In addition to macropollutant loads, which are used to calculate the population served by the WWTP, it is also important to consider the contribution of stormwater incursions, greywater input, septic tank discharge at the plant as J o u r n a l P r e -p r o o f well as the presence of industrial waste which are common challenges in South African

WWTPs.

Our findings show that WBE can be used to give an indication of infection levels in connected communities. This is due to the correlation observed between high viral loads in the untreated wastewater and peak in active clinical cases within the province. We also observed that the transition between the levels of lockdown (from higher to lower levels of restriction) resulted in an increase in viral loads in the untreated wastewater. Additionally, we also showed that mathematical models estimated a higher number of infected people compared to data from clinical testing. However, this is not conclusive due to the scarcity of active clinical cases specific to the catchments of the WWTPs studied. Despite the challenges faced (highlighted above), we can conclude that WBE can be used to detect possible surges in COVID-19 infections in communities serviced by WWTPs. Additionally, with an improved predictive model, WBE will be useful in forecasting the potential number of people that could be infected, an approach that is important for risk reduction interventions. Barra et al. (2020) 

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

Asymptomatic coronavirus infection: MERS-CoV and SARS-CoV-2 (COVID-19)

Analytical Sensitivity and Specificity of Two RT-qPCR Protocols for SARS-CoV-2 Detection Performed in an Automated Workflow

Statistical methods for engineers and scientists

Clinical predictors and timing of cessation of viral RNA

COVID-19: Unpacking the low number of cases in Africa

Wastewater-based epidemiology biomarkers: Past, present and future

Estimating alcohol consumption by using wastewater-based epidemiology in Adana Province

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

Interlaboratory assessment of droplet digital PCR for quantification of BRAF V600E mutation using a novel DNA reference material

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

Review of analytical performance of COVID-19 detection methods

COVID-19 surveillance in Southeastern Virginia using wastewater-based epidemiology

Sequential analysis of viral load in a neonate and her mother infected with SARS-CoV-2

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

Detection of pathogenic viruses in sewage provided early warnings of hepatitis A virus and norovirus outbreaks

SARS-CoV-2 known and unknowns, implications for the water sector and wastewater-based epidemiology to support national responses worldwide: early review of global experiences with the COVID-19 pandemic

Role of environmental poliovirus surveillance in global polio eradication and beyond

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

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

Severe Acute Respiratory Syndrome Coronavirus 2 Shedding by Travelers

Urinary oxalate excretion increases with body size and decreases with increasing dietary calcium intake among healthy adults

Clinical and virological data of the first cases of COVID-19 in Europe: a case series

Virus shedding dynamics in asymptomatic and mildly symptomatic patients infected with SARS-CoV-2

SARS-CoV-2 in wastewater: potential health risk, but also data source

Clinical and immunological assessment of asymptomatic SARS-CoV-2 infections

SARS-CoV-2 detection using digital PCR for COVID-19 diagnosis, treatment monitoring and criteria for discharge

The potential of wastewater-based epidemiology as surveillance and early warning of infectious disease outbreaks

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

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

Environmental surveillance of poliovirus and non-polio enterovirus in urban sewage in Dakar

Covid-19 news: UK government under pressure to impose England lockdown

SARS-CoV-2 can be detected in urine, blood, anal swabs, and oropharyngeal swabs specimens

Making waves Wastewater-based epidemiology for COVID-19 approaches and challenges for surveillance and prediction

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

The Characterization of Feces and Urine: A Review of the Literature to Inform Advanced Treatment Technology

First detection of SARS-CoV-2 RNA in wastewater in North America: a study in Louisiana

Development of genetic diagnostic methods for novel coronavirus 2019 (nCoV-2019) in Japan

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

Highly predictive regression model of active cases of COVID-19 in a population by screening wastewater viral load

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

Europe Imposes New Covid-19 Restrictions as Second Wave Accelerates

Correlation of SARS-CoV-2 RNA in wastewater with COVID-19 disease burden in sewersheds

Virological assessment of hospitalized patients with COVID-2019

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

Time course quantitative detection of SARS-CoV-2 in Parisian wastewaters correlates with COVID-19 confirmed cases

Evidence for gastrointestinal infection of SARS-CoV-2

Duration of respiratory and gastrointestinal viral shedding in children with SARS-CoV-2: a systematic review and synthesis of data

Comparative study on virus shedding patterns in nasopharyngeal and fecal specimens of COVID-19 patients

Potential transmission risk of SARS-CoV-2 through medical wastewater in COVID-19 outbreak cities