Wastewater-based epidemiological investigation of SARS-CoV-2 in Porto Alegre, Southern Brazil

Abstract

Wastewater-based epidemiology (WBE) may be successfully used to comprehensively monitor and determine the scale and dynamics of some infections in the community. We monitored severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA in raw wastewater samples from Porto Alegre, Southern Brazil. The samples were collected and analyzed every week between May 2020 to May 2021. Meanwhile, different social restrictions were applied according to the number of hospitalized patients in the region. Weekly samples were obtained from two wastewater treatment plants (WWTP), named Navegantes and Serraria. To determine the SARS-CoV-2 RNA titers in wastewater, we performed RT-qPCR analysis targeting the N gene (N1). The highest titer of SARS-CoV-2 RNA was observed between epidemiological weeks (EWs) 33–37 (August), 42–43 (October), 45–46 (November), 49–51 (December) in 2020, and 1–3 (January), 7–13 (February to March) in 2021, with viral loads ranging from 1 × 10⁶–3 × 10⁶ genomic copies/Liter. An increase in positive confirmed cases followed such high viral loads. Depending on the sampling method used, positive cases increased in 6–7 days and 15 days after the rise of viral RNA titers in wastewater, with composite sampling methods showing a lower time lag and a higher resolution on the analyses. The results showed a direct relation between strict social restrictions and the loads of detected RNA reduction in wastewater, corroborating the number of confirmed cases. Differences in viral loads between different sampling points and methods were observed, as composite samples showed more stable results during the analyzed period. Besides, viral loads obtained from samples collected at Serraria WWTP were consistently higher than the ones obtained at Navegantes WWTP, indicating differences in local dynamics of SARS-CoV-2 spread in different regions of Porto Alegre. In conclusion, wastewater sampling to monitor SARS-CoV-2 is a robust tool to evaluate the viral loads contributing to hospitalized patients’ data and confirmed cases. In addition, SARS-CoV-2 detection in sewage may inform and alert the government when there are asymptomatic or non-tested patients.

Graphical abstract

Highlights

• •Wastewater-based epidemiology can be a useful tool for monitoring SARS-CoV-2 circulation in a community.
• •Social restriction measures influence viral titers on wastewaters over time in long-term wastewater-based epidemiology.
• •Multivariate regression models can predict the proportion of positive cases in the studied region.
• •The number of real active cases are estimated up to five to six times higher than the official confirmed cases.

1. Introduction

Human coronaviruses, including SARS-CoV and MERS-CoV, are known to cause gastrointestinal symptoms in addition to respiratory symptoms [1], and previous studies demonstrated that these viruses replicate in gastrointestinal tissues [2,3]. Also, the detection of coronaviruses (SARS-CoV, MERS-CoV, and SARS-CoV-2) was confirmed in stool samples in patients with or without symptoms or respiratory tract detection [4,5,6,7]. Consequently, multiple recent reports indicate the utility of wastewater-based surveillance (WBS) of SARS-CoV-2 to estimate population prevalence and set a trendline for the timing of pandemic while sanitary measures were applied [8,9,10].

Porto Alegre is the capital of Rio Grande do Sul state situated in Southern Brazil. The city area corresponds to 495,390 km² and estimates a population of 1,488,252 inhabitants. The economy is mainly based on services, followed by commerce and industries [11]. The city has been impacted by SARS-CoV-2 since April/May 2020, with the first confirmed cases of infection, and since then different restriction measures were taken to control the virus's spread in the community. The restrictions were applied according to indicators (capacity of health services to receive severely affected patients, and the rates of SARS-CoV-2 transmission, which were translated into hospital occupation rates). These indicators were useful to conduct the restrictions in order to avoid overcrowding in the health system [12].

Here we report our findings on SARS-CoV-2 assessment in wastewater samples collected in Porto Alegre from May 2020 to May 2021. We demonstrated the virus dynamics during this period through wastewater sampling, and how virus detection was impacted by the number of confirmed cases. Also, the results showed an efficiency of the different social restrictions applied during this period influencing the abundance of SARS-CoV-2 detected in Wastewater Treatment Plants (WWTPs) of Porto Alegre, reflecting the virus circulation.

2. Material and methods

2.1. Social restriction measures applied in the city during the first year of the pandemic

The restriction measures applied in Porto Alegre were based on a system called the “Flag system”. The system works in four different scenarios: Yellow flag, classified as low risk, translates low propagation rates and high capacity of the healthcare system; Orange flag, medium risk, meaning low propagation rates and medium capacity of the healthcare system or medium propagation rates and high capacity of the healthcare system; Red flag, high risk, medium propagation rates and low capacity or high capacity of the health care system and high propagation rates; Black flag, very high risk, high propagation rates and low capacities of the healthcare systems. Moving from one flag to the next is accompanied by an increase in measures of social restrictions. The definition of the risk level was revalidated weekly by the health authorities based on risk indicators of healthcare system capacity (rates of available intensive care units, testing level in the region, availability of personal protection equipment) and transmission indicators (e.g., confirmation of new cases, numbers of new hospital admissions). Both category indicators were analyzed based on data-informed from the previous week.

2.2. Wastewater samples

The wastewater samples were collected at five different points at Serraria WWTP and Navegantes WWTP in the municipality of Porto Alegre from May 2020 to May 2021. Wastewater samples were collected every week and identified by the epidemiological week (EW). In addition to these, three Wastewater Pumping Stations (WWPSs) (Baronesa do Gravataí, Ponta da Cadeia, and Sarandi) were monitored (the first two WWPSs pump the wastewater to Serraria WWTP) (Fig. 1). Twenty-four hour composite samples collected at the Serraria WWTP and grab samples were collected from the influent of Navegantes WWTP and from Baronesa do Gravataí, Ponta da Cadeia and Sarandi WWPS. Overall, we processed 136 samples, including 50 samples from the Serraria WWTP, 47 from the Navegantes WWTP, and 39 samples from the WWPSs.

Fig. 1.

Wastewater sampling points. Red dots indicate the locations of Wastewater Treatment Plants (WWTPs). Green dots indicate the locations of Wastewater Pumping Stations (WWPSs). Covered areas represent the locations served by each or both WWTPs in the city of Porto Alegre, Southern Brazil.

2.3. Sample concentration

The concentration process of the samples was based on ultracentrifugation [13] with adaptations [14]. After the concentration process, samples were stored at −80 °C until the RNA extraction process occured.

2.4. RNA extraction and detection

The RNA was extracted using the Maxwell® 16 Viral Total Nucleic Acid Purification Kit according to the manufacturer's instructions. The genomic region N1 of SARS-CoV-2 was targeted by the RT-qPCR. The amplification protocol, probes and primers followed the CDC_2019_N1N2 SARS-CoV-2 protocol [15]. The viral RNA loads were determined with serial dilutions (five points of 1:10 dilution factor) of a standard curve of SARS-CoV-2 positive control, previously quantified in digital droplet PCR (ddPCR) [16].

2.5. Calculation of cumulative cases and correlation analysis

The number of total cases was retrieved from the official COVID-19 website of the Rio Grande do Sul government [17]. The number of daily cases was calculated for the regions/neighborhoods contributing to each WWTP, including Serraria and Navegantes. Confirmed positive cases were considered from the notification of the first day of symptoms and only RT-PCR-confirmed cases were considered. Based on the census data from Porto Alegre in 2010 (the last time it was accounted for) and corrected to the 2021 estimated population (an increase of 5.9%), the relative prevalence of 100,000 was calculated for each region. Considering the potential underreporting solely for symptomatic cases, an underreporting factor was included based on the seroprevalence described in a SARS-CoV-2 survey undertaken in the State of Rio Grande do Sul [18]. The number of cumulative cases was obtained considering the day of disease onset and the dynamic viral shedding of SARS-CoV-2 in feces [19,20]. Additionally, data on social dynamics gathered from the official COVID-19 website of the Rio Grande do Sul government [17].

Pearson's correlation coefficient was calculated between wastewater data and clinical data. Higher correlations were seen when comparing wastewater data to relative cumulative cases than to daily new clinical confirmed cases (Fig. 3). Correlation analysis was done in R statistical software (R Core Team, 2021). P-values for the correlations between viral concentrations and cumulative clinical cases were not provided because of the likelihood of autocorrelations between the two datasets.

Fig. 3.

Relationship between virus load and incidence cases: Lag time between the viral loads and how many days viral loads lead to prior confirmed cases. In the Serraria WWTP (A),where composite samples were taken, the time lag varies from 6–10 days. In Navegantes WWTP (B), where simple grab samples were taken and the best time-lag was 15 days. Indicating that the sampling methods influence in the resolution of analysis.

2.6.1. Nonparametric fitting of viral load overtime

Generalized Additive Models (GAM) using a basis of cubic regression splines [21], and the Locally Estimated Scatterplot Smoothing (LOESS) [22] nonparametric regression models have been used to fit the weekly viral load, and as a function of time at SES-RS and WWTPs from Serraria and Navegantes served regions. The nonparametric LOESS allows modeling of complex nonlinear functions [23]. Otherwise, semiparametric GAM models allow the inclusion of linear parametric and flexible nonparametric effects of the predictors on the response. Here, GAM models were applied to define the relationship between the viral load and the date by using penalized regression splines [24] to estimate the smooth effect of the date on the viral load. A GAM model expression consisting of one linear, X₁, and two smooth predictors, T₁ and T₂, can be defined by:

$$Y={\beta }_0+{\beta }_1{X}_1+s_1\left(T_1\right)+s_2\left(T_2\right)+s_{12}\left(T_1,T_2\right)+\epsilon$$

whereby β₀ and β₁ are the parameters that define the linear effect of X₁ on the response Y, s₁(T₁) and s₂(T₂) are the smooth effects of T₁ and T₂ variables, S(T₁, T₂) accounts for the smooth effect of their interaction, and ε is the random error. Here, Y is assumed to be Gaussian distributed, but instead, it could be assumed Binomial or a Poisson distribution, in which case Y should be replaced by g(η). In the framework of this section, Y = Viral Load and X = Time, thus;

$$E\left(ViralLoad\right|Time={\beta }_0+s_{Time}\left(Time\right)\text{.}$$

s accounts for the smooth effects of the predictors (T) and are often estimated using penalized regression splines. Splines are piecewise polynomial functions of a specific degree that are continuous in a set of knots a < t1 < t2 < ⋯ < tL < b (up to L − 1°) defined in an interval [a,b].

Thus, the smooth effect of T can be expressed using a splines basis composed of L + degree elements,

$$sj\left(t\right)=\sum _{k=1}^{L+\mathrm{deg}ree}{\beta }_{jk}{\phi }_{jk}\left(t\right)={{\beta }^{\prime }}_{j\prime }{\phi }_{j,}$$

where φ_jk are the splines functions, and L accounts for the number of knots. In this work, degree = 3 (cubic regression splines) was chosen.

2.6.2. Viral load models

Flexible nonparametric and semiparametric models (e.g., GAM and LOESS) have been formulated to estimate symptomatic and asymptomatic cases. Diagnostic tests (Q–Q plots, residuals versus fitted values plots, and Cook's distance) were used for outlier detection. Statistical analyses were undertaken with R statistical software (R Core Team, 2021) with the following packages: mgcv [24] to fit GAM models and ggplot2 and GGally [25,26] to plot and fit LOESS models; and caret to fit and evaluate regression models [27]. Some RT-qPCR replicates could not be measured when there was small viral load due to the limitation of the detection technique (errors randomly occur when the number of copies/L is under 10,000).

2.6.3. Social dynamics influence

Social dynamics influence was investigated to account for the percent changes from the baseline of main social activities in Porto Alegre, namely presence in retail and recreation, grocery stores, pharmacies, parks, transit stations, workplaces and residence.

3. Results

3.1. Number of confirmed cases of COVID-19

The number of confirmed cases of COVID-19 between May 12 2020 and May 3 2021 is calculated as 66,695 in the region attended by Serraria WWTP and 13,828 in the region attended by Navegantes WWTP. Neighborhoods that send wastewater to both WWTPs showed increased numbers of accumulated cases during the first three weeks of March 2021 (see Supplementary data 1).

3.2. SARS-CoV-2 amplification efficiency

The slopes of the standard curves for the quantification of the N gene assays were −3.28 ± 0.12 for N1 and −3.42 ± 0.14 for N2. Amplification efficiencies were 101.5 ± 6.2%, 95.3 ± 5.7 and correlation coefficient of 0.992 ± 0.008, 0.993 ± 0.013 with Y-intercept values of 32.51 ± 1.02 and 31.99 ± 0.7 for N1 and N2 respectively. With few exceptions, detection of N1 occurred at lower CTs in comparison to N2, indicating a higher sensitivity of SARS-CoV-2 detection when N1 was targeted.

3.3. SARS-CoV-2 RNA in wastewater collected in Porto Alegre

A total of 136 wastewater samples were collected from May 2020 to May 2021 in the city of Porto Alegre. From the total, 97 samples were obtained at the WWTPs, and 39 were collected at WWPSs from Porto Alegre (see Table 1).

Table 1.

Number of wastewater samples obtained from each WWTP or WWPS between May 2020 and May 2021 in the city of Porto Alegre, Brazil.

		Samples (n)	N1 Positive samples (n)	N2 Positive samples (n)
Wastewater Treatment Plant	Serraria	50	42	42
	Navegantes	47	43	40
Wastewater Pumping Stations	Baronesa do Gravataí	10	6	9
	Ponta da Cadeia	15	12	12
	Sarandi	14	10	10

The results of SARS-CoV-2 quantification samples from Serraria WWTP are shown in Fig. 2, based on the results of N1 detection (Supplementary data). Fig. 2 also indicates the number of confirmed cases detected in the neighborhoods related to Serraria WWTP. In addition, distinct background colors in the figure indicate the different control measures implemented in the city (described as the “Flag” system) to restrict social mobility and control virus dissemination.

Fig. 2.

Relative cumulative cases of COVID-19 (white line) and viral loads detected in influent (blue bars), under the different flag systems of social movement restriction in the Serraria and Navegantes served regions (A and C). Viral loads detected in samples collected at WWTPs Serraria and Navegantes from May 2020 to May 2021 (B and D).

Wastewater samples obtained from Serraria WWTP were negative for SARS-CoV-2 RNA on EW 20 and EW 23 (May and June 2020) (Fig. 2 and Table 3 in Supplementary data). From EW 29 onwards, all samples analyzed were positive. Fig. 2 shows the average viral loads detected from this WWTP during different periods of social distancing (represented by the orange, red and black flags). Significant differences in the mean viral loads were observed when specific restriction measures were applied for Serraria WWTP (p < 0.05).

The first sample obtained from Navegantes WWTP (EW20, in May 2020) was negative for SARS-CoV-2 RNA, and, from EW 23 onward, all were positive for virus RNA (Fig. 2 and Table 2 in the Supplementary data). Fig. 6 shows the average viral loads detected from Navegantes WWTP during different periods of social distancing. We tested the correlation of viral concentrations with relative cumulative cases, allowing for a variable time lag (Fig. 3). The relative cumulative cases were considered once SARS-CoV-2 can be shed in feces for more than 20 days, in which case the long tail of shedding may contribute significantly to the signal in wastewater [6,28,29]. There were higher correlations when comparing wastewater data to cumulative cases than daily new cases (Fig. 3). The maximum agreement between the time series was observed for a time offset of 9 days in the Serraria region (Pearson's r = 0.69, Fig. 3A), while in the Navegantes region it was observed for a time offset of 15 days (Pearson's r = 0.49, Fig. 3B). This time lag between the wastewater signal and the cumulative clinical confirmed cases is within the 20 days of fecal shedding of SARS-CoV-2. Thus, wastewater surveillance could potentially predict trends in new COVID-19 cases.

Table 2.

Final GAM models to explain the 20-day moving average as a function of ln (viral load) and percent change from baseline mobility of main places in Serraria and Navegantes served regions in Porto Alegre.

Serraria	Estimate	Standard error	95% CI	95% CI	P-value
Intercept	−1003.7	111.1	−1221.5	−785.9	<0.001
ln (viral load)	138.4	17.6	103.9	172.9	<0.001
Retail and recreationa	−9.6	1	−11.6	−7.6	<0.001
Grocery and pharmacya	5.9	0.7	4.5	7.3	<0.001
Transit stationsa	5	1.1	2.8	7.2	<0.001
Workplaces∗	−3.2	0.6	−4.4	−2.0	<0.001
Navegantes
Intercept	−452.04	67.2	−583.8	−320.3	<0.001
ln (viral load)	63.66	11.4	41.3	86.0	<0.001
Retail and recreationa	−8.2	0.9	−10.0	−6.4	<0.001
Grocery and pharmacya	4	0.7	2.6	5.4	<0.001
Transit stationsa	5.5	1	3.5	7.5	<0.001
Workplacesa	−2.6	0.6	−3.8	−1.4	<0.001

^aPercent change from baseline. CI Confidence interval.

Fig. 6.

Estimated number of COVID-19 active cases versus the viral load under the different flag systems of social movement restriction in the Serraria and Navegantes regions (A and B).

Wastewater samples collected in WWPSs in Porto Alegre were also processed and submitted to virus detection and quantification. Virus RNA was detected from EW 20 onward (May 2020) in wastewater samples obtained from different WWPS at variable viral loads (Supplementary data). The variation of the viral load through time was monitored to distinguish between the contribution from the two studied regions and analyzed by nonparametric LOESS models to avoid overfitting due to the small sample size. Both regions showed similar patterns (Fig. 5, Fig. 6, Fig. 7).

Fig. 5.

ICU patients versus estimated COVID–19 positive cases in the Navegantes served region using GAM regression model (A). Relative cumulative cases versus relative positive cases in Navegantes served region using GAM regression model (B). Estimated number of COVID-19 positive cases in the Navegantes served region and the LOESS fit (C). Estimated COVID-19 active cases in the Navegantes served region by the natural logarithm of the viral load (D).

Fig. 7.

Estimation of the number of real COVID-19 active cases relative to the exposed population in each Porto Alegre region (Censo 2010) based on prevalence estimate by [18]. These estimates are compared to the number of cases reported by the Health State Secretariat of Rio Grande do Sul relative to the exposed population per 100,000 persons (A and B). Estimated number of COVID-19 active cases in Porto Alegre vs the predicted number of COVID-19 real active cases at the Serraria served region (C) and the Navegantes served region (D).

The viral loads in the sewage were analyzed independently for each WWTP, using GAM and LOESS models to verify the efficacy of the flag system whilst controlling viral spread (Fig. 6). In both WWTPs served regions, the flag system efficiently controlled the viral spread, and more strict social restrictions translated into a viral load reduction in sewage.

3.4. Estimated number of real active cases of COVID-19 in neighborhoods served for two WWTP in Porto Alegre

The number of real estimated cases were five to six times higher than the official confirmed cases (Fig. 7), green line. Once data did not assume a normal distribution, parametric regression was disregarded, and further non-parametric models were further undertaken based on the viral load. The logarithmic of the viral load was significant using the GAM model, with a small adjusted R² of 0.08 in the Navegantes served region and 0.06 in the Serraria served region. The ICU patients did not seem to be a good predictor of relative positive cases in both regions (Fig. 3, Fig. 4). In the Serraria region, an exponential followed by a linear trend was observed (Fig. 4), while in the Navegantes region, a linear trend was fitted (Fig. 5).

Fig. 4.

ICU patients versus estimated COVID–19 positive cases in the Serraria served region using GAM regression model (A). Relative cumulative cases versus relative positive cases in Serraria served region using GAM regression model (B). Estimated number of COVID-19 positive cases in the Serraria served region and the LOESS fit (C). Estimated COVID-19 active cases in the Serraria served region by the natural logarithm of the viral load (D).

Multivariate GAM models with a 20-day moving average as a function of ln (viral load) and percent variance from baseline of mobility in retail and recreation, grocery and pharmacy, transit stations and workplaces achieved the best adjusted R² for both regions (0.40 for Serraria and 0.34 for Navegantes) compared to the other two moving average modeled.

3.5. Wastewater epidemiological models based on viral load for COVID-19 real active cases prediction

A scatterplot of the estimated number of COVID-19 active cases in the studied regions versus the logarithm of the viral load, along with the LOESS fitted curve shows a change from a quadratic shape from 4.5 to 6 logarithms of the viral load to an exponential shape above that level in the Serraria served region (Fig. 4). In the Navegantes, the effect was smoother and increased to a logarithmic shape above the 5.5 level with a wider confidence band than in Serraria (Fig. 5). The estimated and predicted values of the real number of COVID-19 active cases showed a relative dispersion from the diagonal line, representing a moderate model prediction (Fig. 7).

GAM models were fitted to predict the real number of COVID-19 active cases based on the viral load and the most relevant social mobility variables. The best model included the viral load and the percent change from baseline in workplaces, retail and recreation settings, groceries and pharmacies and transit stations (Table 2) for both regions (adjusted R² = 0.17 for Navegantes and 0.35 for Serraria).

4. Discussion

We determined the presence and abundance of virus RNA in samples from two WWTPs and six WWPSs in Southern Brazil in the city of Porto Alegre, Southern Brazil for twelve months, when different social-restriction measures were applied by authorities and COVID-19 vaccination was initiated. Until the end of May 2021, the vaccination schedule in Brazil was in its initial phase and restricted to priority groups and health professionals. The Serraria WWTP is located in the south region of Porto Alegre and receives wastewater from circa 760,000 people (50% of the total population, with a flow rate of 4115 L/s). Navegantes WWTP is responsible for about 13% of the wastewater treatment and receives wastewater from circa 180,000 people (13% of the total population, with a flow rate of 444 L/s). Together, these WWTPs receive the sewage generated by 63% of the total population.

Our first SARS-CoV-2 positive sample was identified at EW 23 (2020) from Navegantes WWTP. Since then, all samples from this WWTP were positive for SARS-CoV-2, although at lower viral loads, when compared to Serraria WWTP. On the other hand, samples from Serraria WWTP were positive from EW 29 onwards at higher and more homogeneous viral loads throughout the period. Variations of viral loads from environmental samples are well described, as chemicals or amplification inhibitors may influence the final amplification results [30]. In addition to intrinsic aspects of the sample, samples from Serraria WWTP were collected by an automatic composite sampler for 24 hours, while samples from Navegantes WWTP were grab samples. We reinforce the differences between sampling methods as a variable in this study. After viral titers increase in the sewage, positive cases also increase but with a different time lag for each WWTP. In Serraria WWTP, the time lag was smaller than Navegantes WWTP, 6–10 and 15 days, respectively (Fig. 3).

In this study, we demonstrate that after implementation of more strict social restrictions, viral loads found in both WWTPs decreased, as well as the number of confirmed cases (Fig. 2, Fig. 3, Fig. 4, Fig. 5). Interestingly, we observed a significant increase in viral loads by the end of 2020 and, especially, the beginning of 2021. During this period, social distancing rules were strict but with some flexibility, and viral loads in both WWTPs were elevated; highest viral loads observed were detected at EW 8 and 9 in 2021. These results are correlated to the introduction and dissemination of the P1 variant of SARS-CoV-2 in this region [31] and the increase in the number of clinical cases observed (See Fig. 2, Fig. 4; Table 1 of the Supplementary data). In addition, if we carefully analyze the results obtained by the end of February, before the highest peak of virus infections in Porto Alegre, we observed an increasing trend of viral loads in both WWTPs.

The real number of COVID-19 cases is known to be higher than the number of confirmed cases of the disease. This occurs due to the high numbers of asymptomatic people and the lack of mass testing in Brazil during the days of the pandemic. Such data support the WBE as a useful tool of surveillance, once it includes all people served by a WWTP. Estimating the real number of positive cases provides crucial data on the extent of the viral spread and the pressure it will exert on the healthcare system. Though, how the estimated real cases and hospitalizations correlate remains an issue. When we consider viral loads in sewage, in addition to the number of people served for each WWTP and the estimated seroprevalence for the Rio Grande do Sul state [18], we conclude that the underreporting rate is about five to six times and even higher in the peak of infections (Fig. 7), reinforcing the importance of WBE as a source for complementary data on the circulation of the virus in the community.

The applied flag system, despite limitations, has influenced the levels of virus detection in sewage. When we evaluate the impact of social mobility data on confirmed cases, we observe three key indicators that directly impact viral spread: the increased movement of people in grocery stores, pharmacies and public transport (Supplementary data). On the other hand, the lower mobility of people reflects a negative impact on viral propagation as expected. As Porto Alegre's economy is primarily based on services and commerce, people from socially vulnerable neighborhoods who make up the majority of workers in these roles cannot stay at home for long periods of time. Based on this context, we accentuate the importance of a coordinated response by the authorities, which includes the expansion of public transport in regions with greater circulation, as well as the disinfection of spaces and the use of masks.

This work has a series of uncertainties, such as the difference between the sampling methods. Automated composite samples are collected on the Serraria WWTP, samples were collected hourly over a period of 24 hrs, being representative of the inflow of sewage in the WWTP and Navegantes WWTP, the sample collection method are simple to samples. How to properly correlate this data is a challenge. In our concentration processes, we do not verify the recuperation rate of SARS-CoV-2 RNA, as it can be made by adding a spike quantity with other viruses of a known amount and evaluating the recuperation of this virus, which is highly recommended.

These findings emphasize the importance and applicability of WBE authorities’ strategies to predict and follow the spread of SARS-CoV-2. In addition, our results indicate that the focus of WBE should lie on evaluating the trend of viral loads over time in a chosen WWTP, as such an approach raises more information than the data obtained by comparison of different WWTPs at certain time points.

5. Conclusions

The abundance of SARS-CoV-2 RNA in wastewater has been monitored over the first year of pandemics and viral RNA was present for almost the whole period of monitoring. Viral titers on wastewater reflect the virus circulation in the community and the social restrictions used to control virus spread rapidly influence viral loads in wastewater. Multivariate GAM regression models were fitted to predict the proportion of positive cases based on the cumulative.

Author contributions

ACF, AASC, FSC, and CR conceived the experiments and contributed to the paper preparation, which was carried out by BAP, ATM, WSN and LVL, who also carried out data interpretation and drafted the manuscript. BAP and ATM contributed equally to this work. AJ, ERCC, CBA, LCBT, MSR, RL, ASL, and PROB were responsible for sample collections and provided support for the sample concentrations.

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.

Appendix A. Supplementary data

The following are the Supplementary data to this article.