Science Wiki

Wastewater-based epidemiology (WBE), also called wastewater surveillance or sewage surveillance, is the measurement of chemical and biological markers in wastewater to obtain information about the health and behavior of the population that contributes to it. Samples are usually collected from the influent of municipal wastewater treatment plants (WWTPs), from points within the sewer network, or from institutions such as hospitals, prisons, and university dormitories. The excreta (feces and urine) of the contributing population, together with other inputs, are pooled in a single sample. The method is used in public health and epidemiology to track infectious agents, to estimate the use of drugs and other substances, to follow exposure to chemicals, and to monitor antimicrobial resistance (AMR).

Wastewater surveillance does not depend on people seeking care or on the availability of clinical testing, and it includes people with no symptoms. It is not subject to the biases of case reporting, and it requires no individual samples, so it can be done at low cost per person. It is also not perfectly unbiased. It reflects only the population connected to the sewer system, it depends on how much of a marker is excreted and how it is lost in the network, and it cannot show individual-level information. Wastewater data are therefore used to complement clinical surveillance, and not to replace it. Trends in wastewater signals can precede trends in reported cases, but the lead time depends on the pathogen, the testing practices, and the system, and it is not always present.

Polio Eradication, Illicit Drug Metabolites, and SARS-CoV-2 Networks

Wastewater has been used to monitor infections since the 1940s, when poliovirus was detected in sewage in the United States and elsewhere. Environmental surveillance of poliovirus became part of the Global Polio Eradication Initiative, and it has been used to detect silent circulation of wild and vaccine-derived polioviruses in countries such as Israel, Finland, and the United Kingdom, and, in 2022, in New York State. The concept of "sewage chemical-information mining" for drugs was proposed by Christian Daughton in 2001, and the first estimate of community cocaine consumption from the wastewater concentration of its metabolite was made by Zuccato and colleagues in 2005. The term "wastewater-based epidemiology" was applied to this approach in the following years. After the emergence of COVID-19, the detection of SARS-CoV-2 RNA in feces and in wastewater in early 2020 led to a rapid expansion of programs around the world, including national programs in the Netherlands, the United States (National Wastewater Surveillance System), Canada, Australia, and the European Union.

Wastewater contains a complex mixture of organic and inorganic matter, and it contains inhibitors of molecular assays. Obtaining a quantitative signal therefore requires a careful sequence of steps.

Sampling

A single grab sample is a point-in-time sample, and the composition of wastewater changes during the day, with strong peaks at times of high water use. Most programs therefore use 24-hour composite sampling, in which an automated sampler collects portions of the influent over the day. The portions can be taken at fixed intervals (time-proportional) or in proportion to flow (flow-proportional). For some uses, such as in small catchments or sewer sites, grab samples or passive samplers are used because they are cheaper and easier to deploy. A passive sampler, such as a cotton swab or a membrane suspended in the flow, can capture viral material over hours to days. Samples should be kept cold during transport, because the target may degrade, and they are usually processed within 24 to 48 hours or frozen.

Wastewater can be sampled at several scales:

• Treatment plant influent covers an entire sewershed, and for large cities serves hundreds of thousands to millions of people.
• Sewer network sites (manholes) allow the identification of neighborhood-level or building-level signals.
• Facility sampling covers hospitals, schools, long-term care homes, and airports, and can be used for targeted detection.
• Aircraft and ship wastewater has been used to monitor pathogens that travelers bring.
• Sludge and primary solids are settled material that concentrates viral material, which often provides a higher and more stable signal than the liquid fraction.

Concentration and extraction of viruses

Viral nucleic acids are often present at low concentrations. The sample may be concentrated before detection, and methods include:

• Polyethylene glycol (PEG) precipitation. PEG excludes water and viral particles from solution, so that the virus can be collected by centrifugation. It is often combined with sodium chloride.
• Electronegative membrane filtration. Many viral particles in wastewater have a negative surface charge at neutral pH. Cations such as Mg²⁺ or a lowered pH are added to improve binding to a negatively charged membrane, after which the viruses are eluted or the membrane is extracted directly. The details differ among methods, and recovery varies among viruses.
• Ultrafiltration and centrifugal concentrators retain particles above a molecular weight cutoff.
• Magnetic bead methods and other affinity-based approaches that can be automated for high throughput.
• Solids-based methods that precipitate or extract the nucleic acids from the settled solids directly.

Extraction of RNA or DNA is followed by purification to remove inhibitors, such as humic substances and other organic compounds, which can reduce the amplification efficiency of PCR. Recovery of the target through the whole process is tracked with a spiked control, such as a surrogate virus that is added to the sample (for example, bovine coronavirus, mengovirus, or a bacteriophage), and measured values can be corrected for it.

Quantitative detection

The most common methods for the quantification of viral RNA are reverse transcription quantitative PCR (RT-qPCR) and its digital counterpart, droplet digital PCR (ddPCR). RT-qPCR measures the number of cycles needed for a fluorescent signal to pass a threshold (the quantification cycle, C_q). Because C_q depends on the initial copy number, concentrations are obtained by comparison with a standard curve made from known quantities of the target. In ideal conditions, the amount of product doubles each cycle, so that:

where N₀ is the initial copy number, E is the amplification efficiency, and n is the number of cycles.

In ddPCR, the reaction mixture is divided into thousands of droplets (usually about 20,000), and each droplet is amplified separately. The fraction of positive droplets is then used to estimate the concentration by Poisson statistics. If p is the fraction of positive droplets, the mean number of target copies per droplet is:

and the concentration is λ divided by the droplet volume (about 0.85 nanoliters). For example, if 30 percent of droplets are positive, λ = −ln(0.7) ≈ 0.36 copies per droplet. Because ddPCR does not require a standard curve and counts positive and negative partitions, it is less affected by variation in amplification efficiency, and it is often more tolerant of inhibitors than qPCR. It is not completely immune to inhibition, and it is more costly and has a smaller dynamic range at high concentrations. Both methods are used in practice.

A raw measurement (copies per liter) does not show the infection rate directly. It depends on the dilution of the sample, the size of the population, how much virus each person sheds, and the fate of the virus in the sewer.

Flow and load

Rain, groundwater infiltration, and industrial discharges change the volume of wastewater, and heavy rain or a combined sewer overflow dilutes the sample. A concentration can therefore fall when infections are increasing. Concentration is often converted to a daily load:

where L is the daily load (copies per day), C is the concentration (copies per liter), and Q is the daily flow (liters per day). The load divided by the population served gives a per capita load. Flow-based normalization has limits. Flow is affected by factors that are not related to the human input, and the population served may be uncertain or may differ from the resident population (for example, because of commuters and tourists).

Normalization to human fecal markers

Concentrations are often normalized to a marker of the human fecal input. The most common is the plant virus pepper mild mottle virus (PMMoV), which is abundant in human feces because of the diet and is found in wastewater at fairly stable concentrations in many places. Other markers include crAssphage (a bacteriophage that infects gut Bacteroides), human-associated Bacteroides genes, and the concentration of ammonia, creatinine, or caffeine. Normalized values are obtained as a ratio:

Normalization can reduce the effects of dilution and, to some extent, of changes in the population. It does not remove all noise. PMMoV shedding differs among individuals and with diet, and PMMoV concentrations vary over time and among sewersheds. The ratio also adds the measurement error of the marker. Different programs use different normalizations, and results are not always comparable.

Decay in the sewer

Between the toilet and the sampler, targets are lost by degradation, adsorption to solids, and biological activity. Loss increases with temperature and with the time in the sewer. It is often described by first-order kinetics:

where C_t is the concentration at the sampling point, C₀ is the initial concentration, k is the decay constant, and t is the travel time. The half-life is t_1/2 = ln 2 / k. As an example, if k = 0.1 per hour and the travel time is 6 hours, then C_t/_C_₀ = e^−0.6 ≈ 0.55, so about 45 percent of the target is lost. Decay rates differ by orders of magnitude among targets. Enveloped viruses such as SARS-CoV-2 are generally less persistent than non-enveloped viruses, and decay rates depend on temperature and on the matrix. Large sewer networks therefore require models of hydraulics to relate a measured signal to the sources.

From load to prevalence

To estimate the number of infected people from a load, the average shedding rate per person is needed:

where s is the mean number of copies shed per infected person per day. This is difficult because shedding varies by orders of magnitude among people and over time, with a peak soon after the onset of infection, and it depends on the variant and on vaccination. Estimates of prevalence are therefore uncertain, and many programs report trends (increasing, decreasing, or stable) and relative levels and not absolute numbers. Statistical models, such as those that include time series and Bayesian methods, are used to relate wastewater signals to case numbers and hospitalizations. Because case reporting changes with the availability of tests, the relationship between wastewater and cases is not constant.

Respiratory viruses

SARS-CoV-2 is shed in feces by many infected people, including some who have no respiratory symptoms, at concentrations in the range of 10² to 10⁸ copies per gram, with large variation. The wastewater signal tracked clinical case numbers and hospitalizations in many places in the pandemic. Studies have reported that the wastewater signal can increase a few days before reported cases. The lead time varied by place and by period, and was longer when testing was limited or delayed, and some analyses found no consistent lead. The statement that shedding in feces begins before symptoms for all people is not established. Shedding of RNA in feces can also occur late in infection and for weeks after recovery, and RNA in wastewater does not show whether the virus is infectious.

Influenza A virus and respiratory syncytial virus (RSV) can be detected in wastewater, mainly because of shedding from the respiratory tract and the gut. Methods for these are used in programs for seasonal surveillance. Influenza A subtyping in wastewater has been used to detect the highly pathogenic avian influenza virus H5N1, which can come from human infection, but also from animal sources, such as raw milk or animal waste from dairy and meat processing. A positive signal for H5 therefore does not by itself show a human infection, and the source must be checked.

Enteric viruses and polio

Norovirus, rotavirus, hepatitis A virus, hepatitis E virus, astroviruses, and adenoviruses are shed in large amounts in feces, so wastewater is an effective means of monitoring them. For poliovirus, the detection of vaccine-derived or wild poliovirus in sewage can identify transmission among people who have no symptoms, because only a small fraction of infections cause paralysis. Environmental surveillance is a core part of polio eradication programs, and the detection of a vaccine-derived poliovirus in wastewater in the United Kingdom and in New York in 2022 led to public health responses.

Other pathogens

Wastewater methods have been used for mpox (the virus is shed in lesions, feces, and other material, and was detected in wastewater during the 2022 outbreak and the later clade I outbreaks), measles (RNA of wild-type and vaccine strains can be detected, and genotyping can distinguish them), hepatitis B, dengue, Zika, and bacterial pathogens such as Salmonella, Vibrio cholerae, and Candida auris. The methods and their sensitivity differ among agents. For viruses that are shed in low amounts or are spread by respiratory or vector-borne routes, detection in wastewater is less direct and less often informative.

Genomic surveillance

Sequencing of wastewater permits the identification of variants and the detection of lineages that have not been detected in clinical samples. Methods include amplicon sequencing with target-specific primers (for example, the ARTIC primers for SARS-CoV-2), hybrid-capture enrichment, and untargeted metagenomic sequencing, which can detect many viruses at once. Nanopore sequencing and short-read platforms are both used.

Wastewater contains a mixture of genomes from many people, and the RNA is fragmented because of degradation. As a result, it is hard to assemble complete genomes of each lineage, and linkage between mutations in separate fragments is usually unknown. Computational methods estimate the relative abundance of lineages from the frequencies of lineage-defining mutations, for example by regression or by deconvolution with a reference set of lineages. The relative abundance f_j of lineage j can be estimated by fitting the observed frequency p_m of each mutation m:

where x_mj is 1 if lineage j carries mutation m and 0 otherwise. Sequencing of wastewater has detected cryptic lineages, which are mutations that are rare in clinical sequences and are thought to originate from chronic infections in immunocompromised people or from animal reservoirs. Their origin is not always certain. Accuracy is limited at low abundance, and quality control, such as checking for contamination and sequencing errors, is needed.

Illicit drugs and other substances

Drug consumption can be estimated from the concentration of a drug or its metabolite in wastewater. The relation requires the excretion fraction of the compound as the target metabolite, which is known from pharmacokinetic studies, and a correction for the molecular mass of the parent drug. The mass of the substance consumed per day per 1,000 inhabitants is:

where C is the concentration of the metabolite in wastewater (mass per volume), Q is the daily flow, CF is a correction factor that includes the ratio of the molecular masses of the parent drug and the metabolite and the reciprocal of the excretion fraction, and P is the population. For example, benzoylecgonine is a main urinary metabolite of cocaine, and about 45 percent of a dose is excreted as benzoylecgonine, so CF includes a factor of about 2.3 (the reciprocal of the fraction) and the ratio of molecular masses (cocaine is 303 g/mol and benzoylecgonine is 289 g/mol, giving about 1.05).

The method has been used by the European Union Drugs Agency in annual multi-city studies of cocaine, amphetamine, methamphetamine, MDMA, and cannabis (as the metabolite THC-COOH). Wastewater data show weekly patterns, such as higher weekend use of cocaine and MDMA. Uncertainties include the estimate of the population, the stability of the biomarker, the excretion rates, and disposal of unused drug. Results are a measure of consumption at the community level, and they are not identification of individuals. The method has also been applied to alcohol, nicotine, caffeine, pharmaceuticals, and new psychoactive substances. Exposure to pesticides, plasticizers, and other chemicals, and markers of oxidative stress have been studied.

Antimicrobial resistance

Sewage contains bacteria from many people and from the environment, and it is a place where resistance genes can be exchanged by horizontal gene transfer. Wastewater metagenomics and qPCR of resistance genes, such as bla_NDM and bla_KPC carbapenemase genes, mcr-1 (colistin resistance), mecA (methicillin resistance), and vanA, combined with mobile elements, can characterize the resistome of a community. Global studies of sewage found regional differences in the abundance and diversity of resistance genes that correlate with factors such as sanitation and antibiotic use. Detection of a resistance gene in wastewater does not show how many people carry the gene, because it may be in environmental bacteria, and it does not show the infection rate. The method is best used to compare places and to follow changes over time.

Early Warning, Resource Allocation, and Coverage Limitations

Wastewater data are used in several ways:

• Early detection of introduction or re-emergence of pathogens, and confirmation that an area is free of a pathogen, such as poliovirus.
• Trends for planning of health services, such as when respiratory virus levels increase.
• Targeted action, such as testing campaigns, in neighborhoods or facilities with high signals.
• Monitoring of immunity, vaccine-derived strains, and variants.
• Evaluation of public health interventions.
• Monitoring of travelers' pathogens at airports.

Wastewater programs also face limitations. Coverage is incomplete, because many people are not on sewers (in many countries, a large fraction of the population uses septic tanks or latrines, and in some places the sewer system covers less than half of the population). In such places, other sampling sites, such as open drains and fecal sludge, are used. Equity of coverage is a concern. Results can be affected by changes in shedding with the variant and with vaccination, by differences in laboratory methods, and by hydrological conditions.

Spatial Granularity, Anonymity, and Community Consent

Wastewater from a large sewershed does not identify individuals. As the sampling area becomes smaller (a building, a dormitory, or a street), the data can reveal information about small groups, and could be used in ways that stigmatize those groups or that have legal consequences. The ethical principles for sampling at small scales include informing and consulting the community, limiting data access, and using the results for health purposes. Wastewater data are generally published at the level of aggregated regions and at set intervals, and some jurisdictions regulate the data. Consent is not usually required for sampling at the plant level, but it may be in places where specific groups are tested.

Matrix Recovery Controls, Inhibition Testing, and Metadata Reporting

Comparison of data among laboratories requires standard methods and quality control. Common practices include:

• Spiking and recovery controls for each sample, and extraction and amplification controls to monitor inhibition.
• Negative controls for each batch, to detect contamination.
• Defined limits of detection and quantification, and a rule for reporting non-detects (for example, the use of a fraction of the limit of detection).
• Replicate measurements, since variability among replicates is large at low concentrations.
• Participation in interlaboratory comparisons and the use of reference materials.
• Documentation of metadata, such as sample type, flow, temperature, and rainfall.

Why is droplet digital PCR (ddPCR) often preferred over RT-qPCR for wastewater matrices?

Municipal wastewater contains high concentrations of humic acids, fulvic acids, polyphenols, and divalent metal ions that partially inhibit Taq polymerase and reverse transcriptase, delaying the amplification cycle threshold (C_q) in conventional RT-qPCR and causing systematic underestimation. Because ddPCR partitions the reaction into ~20,000 nanoliter droplets and relies on binary end-point fluorescence counting with Poisson statistics (λ = −ln(1 − p)) rather than real-time amplification kinetics, it exhibits superior tolerance to partial PCR inhibition and performs absolute quantification without external standard curves.

Why is Pepper Mild Mottle Virus (PMMoV) used to normalize human pathogen concentrations in sewage?

Pepper mild mottle virus (PMMoV) is a non-enveloped, rod-shaped plant tobamovirus ingested universally through processed pepper products and excreted in human feces at consistently high concentrations (10⁶ to 10⁹ RNA copies per gram of dry stool) without seasonal pathogenic fluctuations. Dividing target pathogen RNA concentrations by PMMoV RNA levels (R = C_target / C_PMMoV) corrects for stormwater dilution in combined sewer systems, variable fecal solid recovery, and transient commuter population shifts.

How do bioinformatics pipelines deconvolute mixed viral variants from fragmented sewage RNA?

Unlike clinical swabs that contain a single monoclonal viral strain, municipal sewage pools fragmented RNA from thousands of infected individuals across multiple co-circulating lineages. Tools such as Freyja and LolliPop align tiled-amplicon sequencing reads against a reference genome to calculate single-nucleotide variant (SNV) allele frequencies at each position, then apply constrained weighted least-squares regression or convex optimization against a barcode matrix of known lineage-defining mutations to estimate the relative population fraction (f_j) of each variant.

• Epidemiology
• Public health surveillance
• Sewage
• Reverse transcription polymerase chain reaction
• Antimicrobial resistance
• Poliovirus