Can We Predict Viral Outbreaks Using Wastewater Surveillance?

Yes, it is possible—but the approach needs to go above and beyond simple surveillance of wastewater. Recently, a lot of attention has been paid in wastewater testing for monitoring COVID-19. Effective detection of SARS-CoV-2 in wastewater is a critical first step in the development of prediction models. However, detection in wastewater by itself is not adequate for prediction purposes. For accurate identification and prediction, many more measurements, data, and processes need to be included in the analysis.

The immense global burden of human and zoonotic viral infections is widely recognized. Infectious outbreaks can cause devastating and uncontrollable negative effects especially in dense urban areas. The current COVID-19 pandemic is a striking example. Traditional disease detection and management systems are based on diagnostic analysis of clinical samples, but this approach assumes that patients are examined in a clinical setting after symptoms have developed and after the outbreak has been recognized. Furthermore, testing each individual for active infections or immunity is practically impossible. Traditional systems fail to detect early warnings of public health threats at a population level and fail to predict outbreaks in a timely manner. Environmental-based epidemiology may provide a means of obtaining early warnings of potential upcoming outbreaks as well as predictions of fluctuations of established outbreaks (Xagoraraki and O’Brien 2020).

The first step in this research and management approach is the identification of critical environmental (natural and built) reservoirs and critical environmental exposure pathways of human and zoonotic viruses. Identification of these reservoirs and pathways will inform locations and times for community composite sampling (CCS) (Fig. 1) and the design of environmental surveillance systems at appropriate locations and times for early detection of threats at the population level (O’Brien and Xagoraraki 2019a, b). A convenient CCS location for urban areas is untreated municipal wastewater. In rural or urban locations without centralized sewage collection, though, identification of sampling CCS locations is more complex. Watershed modeling, including microbial pollution source tracking, needs to be conducted to identify appropriate locations for CCS. Locations may include surface water, soils, or sediments in selected subwatersheds. In addition to identifying appropriate CCS locations, appropriate sampling times should be identified based on flow fluctuations and other parameters.

The second step in this research and management approach is appropriate sampling of wastewater or other environmental samples, pretreatment, and subsequent analysis. Genetic testing of environmental samples is much more complex than testing of clinical samples. Environmental samples such as wastewater are dilute, and they contain a multitude of substances that may cause inhibition of molecular assays that are usually used (Xagoraraki et al. 2014). Selected sampling for viruses, appropriate sample concentration, and sample cleanup are needed. Sampling automation is another important issue for large-scale application of this approach (Miyani et al. 2020). Following sample concentration, DNA and RNA are extracted and viral signals are tested with molecular methods such as quantitative polymerase chain reaction (qPCR). When the virus of concern is known, such as SARS-CoV-2, primers and probes can be designed to target that particular virus. When screening for the emergence of potential unknown or novel viruses is desired, methods such as next-generation sequencing and subsequent metagenomic analysis using bioinformatics tools can be used to identify virus-related genes (McCall et al. 2020; McCall and Xagoraraki 2019). These then need to be confirmed with more conventional analysis.

The third and most complex step in this research and management approach is the determination of the relationship between measured viral concentrations in wastewater and viral disease incidence in the community. To determine this relationship and predict viral disease fluctuations in the community over space and time, multiple measurements and data need to be analyzed (Fig. 2). In a system with centralized wastewater collection, it is critical to estimate dilution and detention times in the sewer collection network. To achieve that, hydrological and other network data need to be analyzed and modeled. Precipitation is an important parameter, especially in systems with combined sewer overflows. Furthermore, sorption and desorption of viruses to wastewater solids need to be taken in to account. It is also important to estimate the contributing population, which can be achieved by measurement and quantification of biomarkers and excreted metabolites in the wastewater samples, in addition to measurements of wastewater strength. Disease characteristics such as incubation times, shedding duration, and shedding rates need to be incorporated in the model to estimate delays between measured viral concentrations in wastewater and the appearance of disease symptoms. Furthermore, anthropometric data (age, race, gender distribution) of the service population may be included because they will affect the average disease characteristics of the particular population. Mechanistic/deterministic and statistical models can be built to relate all these parameters. In large systems these models can be complex; data science and machine learning methods may be the best approaches. Public health records of clinical cases for incidence and prevalence should be used to test and calibrate the models. Such models may be used to identify endemic disease, novel viruses, hot spots (spatial disease variations), and critical moments (upcoming disease peaks) for potential outbreaks.

Sampling municipal wastewater at critical locations and times, followed by DNA and RNA testing and multiscale deterministic, statistical, and data-driven modeling of multiple data and measurements, is an example of an environment-based surveillance system. Such a system has the potential to describe patterns of endemic disease, identify potential novel viruses, and forecast hot spots and critical moments for the onset or spread of outbreaks prior to the full-blown demonstration of disease in clinical settings. Such a system has the potential to be faster, more comprehensive, and more economical than traditional systems focused on clinical diagnostics, rapid or not, that are inherently limited to the analysis after of an outbreak. However, for such a system to be valuable, it needs to go above and beyond simple surveillance of wastewater or other community composite samples.