Tracking Opioids Beneath the Streets

For the past eight consecutive Monday mornings, public workers in Cary, North Carolina, have lowered what look like black suitcases down ten manholes across the city, where they remain for 24 hours.

Inside each black case, a suction system continuously collects sludge and shuttles it through a plastic-encased filtration system. By Wednesday of each week, the cases are shipped to Cambridge, Massachusetts, where the slurries of sewage are screened for 13 different opioid molecules, including heroin, fentanyl, and methadone. Based on the unique structure of each molecule, researchers can distinguish between illicit and prescription opioids as well as drugs that were flushed versus ingested.

Cary is not the town one might imagine would be at the forefront of the fight against opioid addiction. The city, a quiet suburb 20 minutes outside of Raleigh, routinely ranks as one of the “safest cities in America,” according to data from the Federal Bureau of Investigation. The median income, $94,000, is nearly double the state average. In the neighboring city of Durham, T-shirts bear the slogan: “I would rather get shot in Durham than die of boredom in Cary.”

While wastewater collection has frequently been used in other countries to track emerging public health threats, near real-time surveillance of illicit drug use is a first in the United States.

But, as national data show, the opioid epidemic transcends class, race, and partisan lines. Nearly every city in America has been affected, and Cary is no exception. In the fall of 2016, the city’s fire department received five calls reporting opioid overdoses, three of which were fatal.

“It didn’t make sense to me,” says Mike Bajorek, the deputy town manager. “Here we are, according to statistics, one of the safest places in the country, and we had these heroin overdoses.”

Shortly after, Cary town officials looked for a solution underground. The suitcases are part of a partnership between the city of Cary and Cambridge-based start-up BioBot Analytics to use sewage to reveal detailed patterns of opioid use in the community. The hope is that the near real-time information will allow local officials to respond better, and earlier, to spikes in drug use. Traditionally, they have relied on data from county-level hospital data and surveys, which can take months to collect and verify.

The project is the first of its kind in the United States, a test run of the method’s effectiveness. Wastewater collection has frequently been used in other countries to track emerging public health threats, but local, real-time surveillance of illicit drug use is a first in the United States. As such, the project raises serious questions about whether biological data can––or should––be collected that could potentially inform policing policies.

Mapping the Urban Microbiome

The labyrinth of pipes underneath our cities doesn’t just carry our waste—it also acts as a cache of biological data, tracking everything from our genetics, diets, illnesses, even hormonal cycles. Recently, scientists have used this trove of information to predict infectious disease outbreaks before the symptoms show up in hospitals.

In Israel, for example, in 2013, when the poliovirus reappeared in several towns, health officials began monitoring sewage to track the spread of the virus. The monitoring allowed for a swift and geographically-targeted vaccination campaign. Nearly 980,000 children were immunized, starting in the areas where the virus was found in the sewage. By the year’s end, the country’s health ministry concluded the crisis was over: no new cases of polio were reported.

Similar successes in wastewater epidemiology have been claimed across Europe and Australia, from tracking antibiotic resistance to salmonella outbreaks. But, until recently, the United States was noticeably absent from the list.

“We are absolutely behind,” says Kevin Bisceglia, an assistant professor of chemistry at Hofstra University who has extensive experience with wastewater epidemiology. “I actually moved away from this as an active area of research because there didn’t seem to be much interest at all in the U.S. But I would say in the last year or so it’s really, really spiked up.”

Much of the new attention in the field has focused on finding opioids in wastewater, Bisceglia says. Public health officials are calling the opioid epidemic the worst drug crisis in American history. The Centers for Disease Control and Prevention estimate that more than 70,000 Americans died from drug overdoses in 2017. Over half were from opioids.

But tracking opioids in wastewater presents a unique set of challenges. Unlike other drugs, including cocaine, opioid molecules break down quickly in water. For example, heroin’s only unique metabolite in urine, morphine-6-glucuronide, quickly breaks down to morphine itself. “So it can be difficult to differentiate morphine use, which may in some cases be legal, from heroin,” Bisceglia says.

Other potent drugs, like fentanyl, are used in such low concentrations that they can be difficult to detect accurately in sewage.

In Cary, officials found this out the hard way. Prior to the partnership with BioBot, some intrepid Cary utility workers unsuccessfully tried to detect opioid molecules in the city’s wastewater treatment plant, which has 10,000 gallons. “They tried and we failed miserably,” Bajorek says.

Going Upstream

Around the time Cary was testing its wastewater treatment plant, Mariana Matus, co-founder of BioBot, was wrapping up her doctoral dissertation in computational biology at MIT. Her research looked into the advantages of analyzing sewage closer to its source rather than after it converges and mixes downstream at treatment facilities. Collecting waste in the pipeline, she found, means molecules have less time to degrade, and drugs like fentanyl would be less diluted.

The problem was, no one was doing it, Matus says, in part because there was nowhere to send it afterwards. Commercial labs only accept urine or feces samples that come from a single person. “There are no commercial labs that will accept sewage,” Matus says. “We’re the first ones.”

Matus founded BioBot in 2017 with Newsha Ghaeli, a PhD student in architecture at MIT. Together, Matus and Ghaeli developed a device sophisticated enough to suction sewage for 24 hours, filter it for certain unique opioid molecules, and preserve them while the cases are shipped to the lab for analysis. How exactly this works is “proprietary,” Matus says.

Once the samples are back in the lab, the BioBot team uses mass spectrometry to analyze the molecules and population data to estimate the rate of drug use per 1,000 persons. In addition to the 13 different opioid molecules, they also look for naloxone, or Narcan, a medicine administered to reverse an opioid overdose. “That is useful because a municipality may know how much Narcan has been distributed, but not how much has been used,” Matus says.

The BioBot team has tested its method’s effectiveness at detecting bacteria in Cambridge, MA, as well as in Kuwait and Seoul, South Korea. Other projects in Australia, Belgium, Canada, and Croatia have had success “drug testing wastewater” for illicit drugs, including heroin and other opioids.

Bisceglia, who is not involved in the project, says the approach is “possible in theory,” assuming a number of factors are accounted for, including the presence of a high enough flow of water. He also notes that because of influxes in populations, for example from tourism or commuters, the results of wastewater sampling are not absolute numbers on drug use per capita.

“The science behind it is well-established,” he says. “It’s not the be-all-end-all in establishing concentrations, but it’s certainly a useful tool that, when combined with other kinds of information, can help provide a picture of opioid use in a community.”

Concerns Ahead

Right now, though the technology could theoretically be used in many different public health capacities, BioBot is only interested in measuring opioids and other drug compounds, according to Matus.

That limited scope isn’t without controversy, though. Wastewater epidemiology has long raised privacy and ethical issues.

The legality of sampling wastewater was first tested in 2004 when a court ruled in favor of the Environmental Protection Agency’s ability to test sewage outside of a steel mill that was illegally dumping chemicals. In effect, the ruling said that sampling wastewater via manholes was akin to searching trash left on the curb for pickup— any member of the public has access to it, and therefore there’s no expectation of privacy.

The approach has since been accepted by the public health community because large sewage samples also can’t be traced back to individuals. However, in the BioBot approach, the samples aren’t from a large catchment area, like a treatment plant. The surveillance is farther upstream, which offers more detailed data but also narrows the population being surveyed.

And while opioid use is perfectly legal in many instances, it isn’t in others. The information gathered has the potential to be used by law enforcement. “There is a bit of a Big Brother element to it,” Bisceglia says.

Medical ethicist Arthur Caplan cautions that this approach could indirectly harm residents. “If you’re going to sample neighborhoods, you start to get closer to identifying particular people or ethnic or racial groups,” says Caplan. “Then there’s a potential for stigma or stereotype. People will say, ‘Oh you live over there, you must be one of those addicts.’”

Bajorek, the deputy town manager, says Cary has addressed such concerns, specifically about whether the data will inform law enforcement practices. “I tell people all the time, we’re not benchmarking one community against another or even finding hot spots for where heroin is,” he says, “because I think our police department knows very well where the illegal drugs are going and coming from.”

Rather, Bajorek hopes that the pilot starts conversations previously hindered by the stigma of opioid use and lack of timely, location-specific data. In North Carolina, responsibility for public health resides at the county level, and data from toxicology reports and hospitalizations can take up to a year to reach city officials. “We want to use the data to be able to say, ‘Hey, this is happening here in Cary and we need to talk about it,’” he says.

However, because the information could also stigmatize people regardless of their drug use, Caplan says the data should be kept private so as not to discriminate against a particular neighborhood or section of the city. These problems can also be mitigated by getting more community buy-in and having a clear, coherent goal, says Caplan. Essentially, the public should see the inherent good in the project, like fluoride in tap water.

“Presumably when you do surveillance, you want to be ready to act depending on what you see,” he says. “So there’s the question of, what are you doing it for? Are you going to reallocate resources for drug education programs? Open up a drug rehab clinic? What’s going to happen?”

Preliminary Results

The city of Cary got its start with BioBot after town officials got wind of the start-up and, after entering the Bloomberg Business Challenge, won a $100,000 grant to implement a three-month pilot.

BioBot is now comprised of a team of five women who operate out of a start-up incubator on the outskirts of the MIT campus. The team has been traveling to Cary since August 2017. Early on, they had to make some adjustments to their robot’s design. The flow rate of the water in the Cary sewer system was slower, especially at night, than the one in more densely-populated Cambridge, where they’d been testing the robot.

Today, the robots are up and running and the pilot project is in its eighth week.

Bajorek admits that at the end of the three-month project, there may not be any sweeping policy reform. “Realistically, even after the pilot is done, I think it’s kind of hard sell to say that we will get some actionable information just from these ten sites,” he says. “I think what will really make a difference is going across the state, adding 100 to 200 samples sites to this database.”

It goes without saying that most counties in North Carolina don’t have the resources that the city of Cary does. However, that, says Bajorek, makes the city the perfect place to test-run the technology. “We’re kind of in a perfect position to make an impact because we’re not in a crisis mode,” Bajorek says. “We have the people and the resources.”
Even if other communities clear the resources hurdle, there still remains the major issue of whether other communities will want the secrets within their sewers to be revealed. Already, with the stigma attached to drug addiction, Bisceglia says he has to conceal the identities of some of the cities in his research because they’re reluctant to be connected with the opioid epidemic.

Caplan agrees that it may be a hard sell. “I can imagine blowback if people thought, ‘They’re stereotyping my neighborhood,’ or ‘I don’t want my property values to go down,’” he says.

However, that doesn’t mean the approach doesn’t have value, he says.

“I see value in getting signs of a problem before it gets out control. Just like we would watch for a flu outbreak, we want to watch to see if there’s an opioid outbreak,” says Caplan. “But, again, to me, it would be better if they also discussed this publicly, just because of the controversial nature of drug monitoring.”

Municipalities’ responses to these issues will determine whether the new approach catches on. Meanwhile, the urgency of the opioid crisis continues to spread. In North Carolina alone, nearly five residents die every day from opioid-related overdoses—which, during the three-month BioBot partnership, means nearly 450 lives will be lost.