Abstract

Brazil has become one of the epicentres of the COVID-19 pandemic, with cases heavily concentrated in large cities. Testing data is extremely limited and unreliable, which restricts health authorities’ ability to deal with the pandemic. Given the stark demographic, social and economic heterogeneities within Brazilian cities, it is important to identify hotspots so that the limited resources available can have the greatest impact. This study shows that decentralised monitoring of SARS-CoV-2 RNA in sewage can be used to assess the distribution of COVID-19 prevalence in the city. The methodology developed in this study allowed the identification of hotspots by comprehensively monitoring sewers distributed through Belo Horizonte, Brazil's third largest city. Our results show that the most vulnerable neighbourhoods in the city were the hardest hit by the pandemic, indicating that, for many Brazilians, the situation is much worse than reported by official figures.

1. Introduction

On March 11, 2020, the World Health Organization (WHO) declared COVID-19 a pandemic. COVID-19 is a respiratory syndrome caused by the SARS-CoV-2 virus, with over 130 million confirmed cases and nearly 3 million confirmed deaths worldwide as of April 12, 2021. Official data suggest that over 13 million Brazilians have been infected, with a death toll second only to the USA (Worldometers, 2021). Cases have been heavily concentrated in big cities but have spread to the countryside, affecting isolated indigenous communities. These statistics are fraught with uncertainties, because the tests are notoriously unreliable, and the sample size (less than 0.1% of the population) is very small (Worldometers, 2021). In large Brazilian cities, testing is mainly occurring among people who suffer from symptoms severe enough to seek health care or have private health insurance. Brazil is one of the most unequal countries in the world (OECD et a. 2019), and its large cities are starkly heterogeneous in demographic, social and economic terms. During the pandemic, it is imperative that limited available resources are directed to neighbourhoods where they will have the greatest impact, which requires the spatial distribution of infections to be characterized.

Following the detection of SARS-CoV-2 RNA in faeces of COVID-19 patients (Wang et al. 2020) and later in sewage (Medema et al. 2020) and sewage sludge (Peccia et al. 2020), an increasing number of studies have reported the use of sewage monitoring as a tool for epidemiological surveillance. Sewage-based surveillance of other viruses has been well documented in the literature (Wigginton et al., 2015; Sims and Kasprzyk-Hordern 2020) and has been successfully applied to assess the spread of polio in populations since 1980s (Hovi et al. 2012). By sampling raw sewage from the inlet of treatment plants and determining SARS-CoV-2 RNA concentrations, it is possible to extract valuable information on infection levels for the entire population that contributes sewage to the treatment plant. This approach has shown that SARS-CoV-2 RNA concentrations in sewage correlated with the local numbers of infected individuals (Ahmed et al. 2020; Randazzo et al. 2020). Extending this principle by sampling wastewater directly from the sewers at multiple points upstream of the treatment plant could reveal the spatial distribution in prevalence of the virus in a large city. This can be even more critical in cities where not all collected sewage reaches the treatment plant but is instead discharged in water bodies, a common limitation in sanitation infrastructure in the developing world. In places with severely scarce testing data during the current pandemic, such as Brazil, information obtained from decentralised sewage monitoring could be vital in devising potentially life-saving public health interventions.

We have carried out a comprehensive sewage monitoring programme for SARS-CoV-2 RNA in Belo Horizonte, Minas Gerais, Brazil since April 2020, early on its epidemic curve. Our objective was to assess temporal and spatial information on the virus load in different neighbourhoods to assist local health authorities in decision-making through weekly bulletins. Here, we present the results for the period between epidemiological weeks 20 and 32 (from May 11 to August 07). To the best of our knowledge, this is the first study to use decentralised sewage monitoring to assess spatial distribution of COVID-19 prevalence in a large city.

Bilal et al. (2021) evaluated the correlation between the Social Vulnerability Index (SVI) and the number of COVID-19 tests done, confirmed COVID-19 cases, and COVID-19 deaths for hundreds of zones in Chicago, Philadelphia, and New York City. The SVI used included 15 variables from the American Community Survey, comprising 4 domains: socioeconomic status, household composition, race and language barriers, and housing and transportation. Sectors with higher SVI (more vulnerable) had lower testing rates, higher positivity rates, more confirmed cases, and greater COVID-19–related mortality.

Vulnerable communities are more affected by the pandemic likely due to a number of factors or a combination of them, including higher risk of exposure to the virus (e.g., due to their stronger reliance on public transport, which can be overcrowded, even during the pandemic), lower access to safe water and sanitation, higher number of people per household, cohabitation with relatives who work in essential services, higher comorbidity burden, decreased access to health care, barriers to testing, and other factors.

Our data on SARS-CoV-2 RNA concentrations in sewage (Fig. 5) indicate that the monitored regions had distinct epidemiologic dynamics, with clear differences in peak duration, week of peak occurrence and number of peaks among regions. The factors that govern these differences have yet to be determined.

Conclusion

Our data suggested that local restrictions had a strong effect on hospital bed occupation, as well as suspected COVID-19 cases and SARS-CoV-2 RNA loads in sewage. The monitoring of wastewater closer to the target population, in several neighbourhoods and not only at the treatment plant, revealed the spatial distribution in prevalence of the virus in a large, heterogeneous Brazilian city. The relative prevalence index (RPI) was proposed to help identify hotspots during different periods in the pandemic. Since the RPI assesses the regional prevalence relative to the prevalence in the entire sewershed, several parameters are normalized in its calculation and are therefore not needed, including shedding rates per infected person and sewage flow rates. This can be greatly advantageous, as shedding rates per infected person can vary by several orders of magnitude among individuals, and accurately measuring sewage flowrates during sampling can be very challenging. In-sewer SARS-CoV-2 RNA decay was modelled, and the results showed negligible decay estimates in the local sewer network.

The novelty of the current work is that hotspots could be identified in the city based on data generated by decentralized sewage monitoring, rather than based on clinical data. This could help major cities deal with the current pandemic, especially in places with utterly inadequate clinical testing data, such as Brazil and most developing countries.

Our results highlight the importance of planning and carrying out control measures at targeted areas. This is especially true for favelas, which tend to be densely populated, with limited sanitation infrastructure, and where it may be virtually impossible for residents to effectively isolate and follow basic prevention measures such as hand washing. Our results show that the most vulnerable neighbourhoods in the city were the hardest hit by the pandemic, indicating that, for many Brazilians, the situation is a lot worse than reported by official figures. Therefore, specific pandemic control actions should be prioritised in the identified hotspots, including carrying out targeted epidemiologic surveys, intensification of contact tracing of patients, implementation of full, mandatory local lockdowns, and possibly prioritizing vaccination of people living under such conditions.