Credit: Will Ludwig/C&EN/Shutterstock
As the children’s book reminds us, everyone poops. They pee too. Then they flush all that solid and liquid waste down the toilet to the sewer system, where it joins everybody else’s waste and wends its way to a wastewater treatment plant.
Telltale signs of the things people ingest end up in sewage. That makes wastewater treatment plants rich sources of information about the health of the communities they serve. In a field called wastewater-based epidemiology, researchers are analyzing the wastewater flowing into sewage treatment plants to gather population-level data about communities. The biggest successes thus far have been in monitoring illicit drug use, but researchers are branching out to monitor other lifestyle chemicals, such as nicotine, and even biomarkers of health, such as proteins and lipids.
Because all that excrement contains chemical remnants of the food people ate, the cigarettes they smoked, and the medications they took, the wastewater treatment plant is a rich source of information about the health of the population it serves.
The sewage that flows into the facility can be thought of as a pooled urine and stool sample of every person whose sewage is treated there. Any tests that would be run on an individual’s sample—monitoring for illicit drug use, for instance—can conceivably be undertaken at a population level in a treatment plant. That’s the basis of the field of study researchers are calling wastewater-based epidemiology.
“If you are at the mouth of a wastewater treatment plant, you essentially can observe all the chemistry that is being used in a city,” says Rolf Halden, an environmental engineer and director of the Center for Environmental Health Engineering at Arizona State University’s Biodesign Institute. “You can measure the metabolites that have migrated through a human body. You can look at medications that are being taken. In essence, you’re at a place where you can observe human health in real time.”
Topping the list of chemicals that have been measured in wastewater are drugs—both legal and illicit. But researchers would like to move beyond drugs to measure markers of health and disease in populations—molecules such as DNA, proteins, lipids, or metabolites that are also present in excrement and that can indicate malfunction inside a person’s body. All these analyses combine to paint a picture of community health and serve as an early warning of epidemics.
The origins of wastewater-based epidemiology can be traced to the 1990s, when environmental scientists found pharmaceuticals and their metabolites in lakes and rivers. From there, it was a small step to begin measuring illicit drugs.
“After all, they’re drugs just like any other drugs,” says Christian G. Daughton, who in 2001 edited an American Chemical Society Symposium Series book about pharmaceuticals in sewage and the environment. The book’s focus was on measuring therapeutic drugs, but the last chapter, which Daughton wrote, proposed applying the same approach to illicit drugs (ACS Symp. Ser. 2001, DOI: 10.1021/bk-2001-0791.ch020).
When the chapter was peer reviewed, several reviewers called the idea “stupid,” remembers Daughton, who was then at the U.S. Environmental Protection Agency. “I shrugged my shoulders and thought maybe I was on to something.”
What Daughton proposed was measuring illicit drugs in wastewater samples and then using those data to calculate the amount of drug use in the geographical region served by the treatment plant where the samples were collected. That analysis could back up (or improve) estimates obtained through conventional epidemiological approaches, which are based on surveys and analyses of individuals. And unlike people, who may not be forthcoming when asked questions, the wastewater samples don’t lie.
But the topic of societal drug use wasn’t in EPA’s purview, so Daughton couldn’t initiate any of the measurements himself. The idea wouldn’t pan out until another scientist took it up and ran with it.
In 2005, Ettore Zuccato and his colleagues at the Mario Negri Institute for Pharmacological Research measured cocaine in Italy’s Po River and in the sewage coming into wastewater treatment plants there (Environ. Health: Global Access Sci. Source 2005, DOI: 10.1186/1476-069x-4-14). They estimated that cocaine use was considerably higher than official national figures indicated.
And so the race to measure illicit drugs in wastewater began. Environmental scientists and analytical chemists throughout Europe started making similar measurements in their own countries.
Since 2011, the SCORE network has analyzed drugs in wastewater samples from sewage treatment plants in participating cities and reported its findings to the European Monitoring Centre for Drugs & Drug Addiction.
EMCDDA makes the data publicly available. Shown here are representative data for cocaine.
The map shows the amount of cocaine per 1,000 people per day as measured in 2017. The graph shows trends for cocaine use from 2011 to 2017 in four cities.
Note: Some data are missing for certain cities between 2011 and 2017. Source: EMCDDA
If you’d like to play with the data yourself, including figures for the other three drugs that SCORE measures, check out EMCDDA’s interactive data tools at http://bit.ly/2g5MNm8.
A network is born
The next few years saw a flurry of papers reporting drugs in wastewater in cities across Europe. By 2010, researchers from seven organizations making such measurements banded together to form the SCORE network, which stands for Sewage Analysis CORe group Europe.
Since 2011, SCORE has coordinated annual measurements of four illicit drugs—cocaine, 3,4-methylenedioxymethamphetamine (MDMA, the synthetic drug commonly known as ecstasy), amphetamine, and methamphetamine—across Europe. The network reports those data to the European Monitoring Centre for Drugs & Drug Addiction (EMCDDA), which releases the findings to the public as part of its ongoing efforts to provide information about drug problems in Europe.
To determine which labs’ data can be included in the European annual report, network members participate in an annual interlaboratory exercise (TrAC, Trends Anal. Chem. 2018, DOI: 10.1016/j.trac.2018.03.009). For the exercise, which is part of SCORE’s quality control, participants receive a sample spiked with the four target drugs. All the samples are prepared by a single institution.
Labs analyze their samples and send data to Christoph Ort, an environmental engineer at Eawag, the Swiss Federal Institute of Aquatic Science & Technology, who crunches the numbers and determines whether each lab achieved results that match the spiked concentrations within an acceptable error range. Labs that pass can then submit their data for real samples in the drug-monitoring study.
The labs don’t all use the same instruments or methods to analyze samples, Ort notes. “They just have to use their established methods,” he says. “I don’t think it would be possible to require everybody to use the same method.” But he finds that the analytical techniques used among labs yield results that are accurate within 20%.
After Ort receives the data for the real wastewater samples from across Europe, he checks the calculations, which take into consideration parameters such as the flow rate of the wastewater when it was sampled. Then he submits the data to EMCDDA, which independently evaluates them.
Because the European data is collected annually, there are limits on what it can reveal. EMCDDA “wants to know the size of the drug market in Europe every year,” Kevin V. Thomas, current chair of SCORE, says. Thomas was formerly at the Norwegian Institute for Water Research and is now director of the Queensland Alliance for Environmental Health Sciences at the University of Queensland. “Ideally, we’d measure every single day of the year, but you can’t do that. What we have is a compromise to give the answers they ask for.”
Since establishing itself in Europe, SCORE has expanded to include labs in countries such as Australia, the U.S., Canada, and China.
Jochen F. Mueller, an environmental chemist also at the University of Queensland, got involved in wastewater-based epidemiology in 2007. At that time, Brisbane, Australia, was starting a water-recycling program in response to a water shortage.
“We were recruited by the government to find out what we have to remove from the water to make it safe for recycling,” Mueller says. “In the process of looking for what’s in the water, we ran into illicit drugs, methamphetamine in particular.”
Since then, Mueller’s program has grown into a large-scale drug-monitoring program funded by the Australian government. Mueller and coworkers analyze wastewater from approximately 50 wastewater treatment plants. Each plant is monitored for a week—every two months for those in the capitals of states and territories and every four months for those in other regional centers. The frequent analyses make it easier to establish a baseline and assess trends. The most recent report was published last month.
“Communities actually come to us and ask if we can help them find out how big the problem is,” Mueller says. His team also helps communities assess the success or failure of intervention strategies.
When searching wastewater for illicit or other compounds of interest, scientists need to agree on which forms of those compounds will give the most reliable results. SCORE helps them by establishing protocols for selecting the most appropriate marker molecules.
“We focus when possible on human metabolites,” says Sara Castiglioni, a SCORE network member at the Mario Negri Institute. By choosing metabolites—breakdown products of the original, parent compound—researchers can limit their measurements to substances that humans consumed, thus minimizing potential confusion over whether a substance ended up in wastewater through some other route, including being flushed down the toilet or being dumped after the manufacturing process. Ideally, the metabolites chosen by SCORE should be excreted in urine and stable in the sewer system.
For some types of drugs, though, researchers might need to detect the parent compound because not enough is known about its metabolites, Castiglioni says. This is true for psychoactive substances such as designer cathinones (also known as bath salts) being created by clandestine operations. She led a study in which European researchers investigated the presence of 17 amphetamine-like synthetic cathinones in eight cities in four countries (Environ. Sci. Technol.2016, DOI: 10.1021/acs.est.6b02644). They detected seven of the substances. No city had all of them, and one city had none of them.
In addition to establishing protocols for selecting marker compounds, SCORE has also established ethical guidelines for collecting and analyzing samples and interpreting the results. The guidelines, which were written with input from Australian researchers, identify potential ethical risks and propose strategies to mitigate them.
The guidelines recognize that because research focuses on the broader population and community, the risks to individuals are low. “The benefit of the wastewater approach is that it’s anonymous. That anonymity needs to be protected through the whole publication process,” Thomas says. The risks are greatest in studies that target specific sites, such as prisons, schools, hospitals, or workplaces.
“We wouldn’t go into a suburb or neighborhood and say this part of the city uses more drugs than that part of the city,” Thomas says. “There’s no benefit in terms of what it tells you about the overall drug use within a city, but it could end up stigmatizing an already-vulnerable group.”
Getting a headcount
One of the challenges when interpreting wastewater epidemiological data is getting a handle on the number of people who correlate with measurements at a particular treatment facility at a certain time. A good estimate is needed because the values for the markers are reported in terms of amount per day per 1,000 people. To draw meaningful conclusions about trends over time or comparisons among communities, the population size is an essential piece of information.
“What we’ve always done is sample at a time that we know the population within a community is relatively static,” Thomas says. “We accept that there’s day-to-day variation” from people commuting in and out of the area being monitored. But, he adds, that variation is insignificant compared with times when people leave on holiday or when there’s a festival that draws crowds.
“Norwegians go on holiday en masse,” Thomas says. Easily 65% of Oslo’s population decamps in July, he says. “If you were to estimate drug use then based on your static population, you would come up with the wrong conclusion.”
For such reasons, SCORE’s annual drug-monitoring tests are done at times that are outside school and public holidays, typically during March.
Researchers have tried to find chemical markers of population size so that worrying about ideal monitoring periods wouldn’t be necessary.
Some proposed population markers have included metabolites of nicotine and caffeine (Water Res. 2015, DOI: 10.1016/j.watres.2015.02.002). But using a nicotine metabolite such as cotinine requires that populations in different areas have similar proportions of smokers. And caffeine comes from many sources.
Thomas and his colleagues have experimented with gathering mobile-phone data to get better population estimates, especially at dynamic times of year.
The researchers teamed up with a Norwegian cellular-service provider to extrapolate the population within a given area from signaling data detected by the cell phone network (Environ. Sci. Technol. 2017, DOI: 10.1021/acs.est.7b02538). They used those population estimates to calculate per-capita drug use during holiday time. If they hadn’t used the dynamic cell phone data and instead worked with population data estimated from normal, static times of year, the researchers would’ve calculated a 31% per-capita decline in drug use.
Thomas, however, sees mobile-phone data as a tool for validation rather than regular use in wastewater-based epidemiology. “It’s hard to get the data from mobile-phone operators,” he says.
In Australia, researchers are developing models for chemically estimating population by making measurements at the same time as Australia’s Census Day, the country’s population count that occurs every five years. Mueller and his team ran studies during the 2011 and 2016 censuses. They used the 2011 population numbers to develop a model for estimating population from the amounts of a variety of chemicals in pharmaceuticals and personal care products (Environ. Sci. Technol. 2014, DOI: 10.1021/es403251g). They found that the resulting model works better for large populations than for small ones.
Moving beyond illicit drugs
Although wastewater-based epidemiology has so far focused primarily on illicit drugs, researchers are branching out to look at other compounds. For example, members of SCORE have used the approach to monitor consumption of legal lifestyle compounds, including caffeine, nicotine, and alcohol (BMC Public Health 2016, DOI: 10.1186/s12889-016-3686-5).
“For nicotine, we measured the main human urinary metabolites—cotinine and hydroxycotinine—and then back calculated the nicotine consumption,” Castiglioni says. The nicotine values they measured did not correlate well with figures for cigarette sales, possibly because of the use of dipping tobacco.
They are also using wastewater measurements to determine human exposure to other chemicals, such as pesticides and plasticizers. For example, Castiglioni led an eight-city wastewater-based epidemiology study of pesticide exposure. The team analyzed wastewater for urinary metabolites of triazines, organophosphates, and pyrethroids (Water Res. 2017, DOI: 10.1016/j.watres.2017.05.044). The amounts detected per day per 1,000 inhabitants were highest for organophosphates and lowest for triazines.
A group in Spain, led by José Benito Quintana and Iria González-Mariño of the University of Santiago de Compostela, analyzed phthalate metabolites in wastewater (Environ. Sci. Technol. 2017, DOI: 10.1021/acs.est.6b05612). They used the detected concentrations to back calculate levels of exposure to six phthalate diesters. Four of the diesters were at levels that exceeded thresholds recommended by the U.S. EPA and the European Food Safety Authority.
So far, wastewater-based epidemiology has been most widely used in Europe and Australia. In the U.S., researchers are starting to wonder whether the approach could help combat the opioid epidemic.
Australia’s National Wastewater Drug Monitoring Program regularly analyzes use of fentanyl and other opioids, such as oxycodone, which are used as pain medications and are also drugs that are misused. But others remain unconvinced that meaningful interpretations can be drawn from such measurements.
The opioid heroin gets quickly metabolized to morphine, for example. “You basically can’t find heroin in wastewater. What you find is morphine,” says Caleb Banta-Green, a researcher at the University of Washington’s Alcohol & Drug Abuse Institute. Figuring out how much of that morphine can be attributed to misuse requires subtracting out the amount used as prescribed. “Wastewater testing just isn’t that precise” to be able to make that kind of estimation, he says.
Banta-Green notes that in the early days of wastewater-based epidemiology, chemists didn’t properly report the error bounds of their data. Reporting the error of the analytical measurement is not enough, because it is often significantly lower than the overall error, he says. “As soon as you talk about flow measurement, as soon as you talk about population variability, the error is probably more like ±30%.”
In the case of illicit opioids, he says that the method might be useful for simply identifying whether they’re present but not for much more than that. Plus, he says, “When it comes to fentanyl, it’s killing so many people that you don’t really need to go looking in the wastewater for it. You can just look in the dead bodies.”
Looking for health markers
The future of wastewater-based epidemiology lies in health studies beyond the use of illicit or lifestyle drugs. But choosing endogenous biomarkers such as DNA, proteins, or metabolites that can be easily detected in sewage and easily interpreted is challenging.
The best biomarkers meet the same criteria as the ones SCORE established for drug metabolites. They should be excreted in urine, unique to humans, and stable in wastewater. Daughton, the now-retired EPA scientist whose book chapter launched the field, proposed isoprostanes as a prototype biomarker that meets those requirements (Sci. Total Environ. 2012, DOI: 10.1016/j.scitotenv.2012.02.038). Isoprostanes are generated in the body when phospholipids react with free radicals and serve as indicators of systemic oxidative stress.
“Since my initial search for a useful biomarker led to isoprostanes, I assumed there must be lots of others,” Daughton says. Instead, after painstakingly digging through clinical literature, he found only about 10 additional potential biomarkers for the field to consider as its next targets (Sci. Total Environ. 2018, DOI: 10.1016/j.scitotenv.2017.11.102). Examples include classes of compounds such as pterins and polyamines.
Daughton is interested in markers produced by regular metabolism. “Those have never been looked for in sewage. If you think about it, why should anybody be interested in them? It’s just stuff that’s excreted, and who cares? But now that I’ve proposed using them to gauge human health, they do have significance.”
Members of SCORE analyzed 8-iso-prostaglandin F2α, a member of the isoprostanes family initially proposed by Daughton, in the wastewater of 11 European cities (Sci. Rep. 2016, DOI: 10.1038/srep39055). In the same samples, they also measured ethyl sulfate (a metabolite of alcohol) and trans-3′-hydroxycotinine (a metabolite of tobacco). They found that the level of the oxidative stress biomarker correlated strongly with the tobacco metabolite level.
“We shouldn’t have been surprised, but it was a little surprising that the entire signal was explained by nicotine consumption,” Thomas says. “It does make sense, because if you look at the main drivers of disease in the developed world, smoking is quite high on the list.”
For population studies of health, establishing a baseline value is difficult, says Barbara Kasprzyk-Hordern, a SCORE member at the University of Bath. And a good baseline is essential for drawing meaningful conclusions about changes in that baseline.
Kasprzyk-Hordern’s team is working on a project, which they call ReNEW, in Stellenbosch, South Africa, to use wastewater profiling to develop a public health early-warning system. They’ll be monitoring wastewater and the surrounding environment for two years. They’ll need a year to establish a baseline to account for seasonally dependent conditions such as rainfall, which can affect flow conditions.
They’re targeting more than 200 biomarkers, including genes, proteins, and chemicals. “We’re hoping that long-term monitoring of selected biomarkers will allow for rapid evaluation of public health status, prediction of future crises, and development of mitigation strategies for rapid- or slow-onset hazards, even before they manifest characteristic end points, such as death in the case of pandemics,” Kasprzyk-Hordern says.
At Arizona State University, Halden and his colleagues are working at an even bigger scale. They’ve established the Human Health Observatory, a repository for samples from more than 300 wastewater treatment plants around the world (Sci. Rep. 2014, DOI: 10.1038/srep03731). “We aim to capture over a billion people,” Halden says.
To capture all the chemistry that flows through those people, the Arizona State researchers don’t restrict themselves to the liquid fraction of sewage. The chemicals that are partitioned into solid waste are ones that are more likely to accumulate in the body. Halden thinks the term “wastewater-based epidemiology” captures only a fraction of what his team is doing. So he’s coined what he thinks is a more-encompassing term: urban metabolism metrology.
Measuring the chemical profile at wastewater treatment plants “is a very economical way to identify chemical risk factors in urban environments and to estimate the toxic exposure of people,” Halden says. “It’s much easier to go to a wastewater treatment plant, take a sample, look at the toxin profile, and estimate the human exposure than to test 100, 1,000, or 10,000 people in that community to arrive at the same information.”
Compared with the earlier work on illicit drugs, wastewater-based epidemiology with endogenous health markers has gotten off to a slow start. “What we’re seeing is that fewer research groups are picking up the baton in the health space than did at the start of the illicit drug work,” Thomas says.
According to Daughton, “With illicit drug monitoring, within a space of five to 10 years after the concept was proposed, there was a useful approach that had been developed.” With the health-based markers, he says, “I can’t even guess” how long it will take to see similar progress.