What's the poop? Wastewater data predicts overdoses

One person’s bathroom business could be key to a public health worker tackling the opioid crisis. In the millions of gallons of water that Americans flush down their toilets every day, lie valuable insights on what’s happening in communities. Wastewater data can reveal, for example, the presence of viruses like COVID-19 or contaminants like PFAS. And now, sewage could play a critical role in combating the opioid crisis.

A new study from the research and data analytics firm Mathematica has found that insights from wastewater analysis can not only help local leaders better understand drug use in their communities, but also potentially prevent overdoses. 

In the study, researchers showed how analysis of wastewater data can identify an emerging drug threat, predict suspected overdoses and evaluate the effectiveness of harm reduction policies such as the decriminalization of fentanyl test strips. They say the study’s findings can help drive more effective solutions to save lives, particularly as gaps in substance use data persist. 

“With billions of dollars in opioid settlement funds coming out, officials need evidence on which programs work and should be expanded, and which programs are less effective,” said Aparna Keshaviah, principal researcher and director of wastewater research at Mathematica. With detailed data on what drugs are in their wastewater and what harm reduction strategies work, communities can better advocate for funding that meets their needs. 

The study evaluated data collected by the wastewater analytics company Biobot Analytics and the Washington/Baltimore High Intensity Drug Trafficking Area’s Overdose Detection Mapping Application Program. Starting in early 2023 and running up to 39 weeks, Biobot sampled untreated wastewater from five rural, urban and suburban counties across California, Kentucky, New Jersey and West Virginia. Fentanyl, methamphetamine, cocaine and xylazine were found, according to an analysis of the samples.

But the data analysis didn’t just identify what substances were present in the water, it also proved its worth in detecting emerging drug threats. Researchers found, for example, wastewater data showing the presence of the animal tranquilizer xylazine, which is increasingly being found mixed in with illicit fentanyl. They identified it at least four weeks before the first reported xylazine-related overdose in each county. In New Jersey, after initial detection of xylazine in one urban county, the wastewater concentration of xylazine spiked by 38% three days before the first recorded overdose, according to the study.

That early detection can help local officials get ahead of potential overdoses with outreach that  raises awareness of dangerous substances entering the community, said Alex Buben, principal epidemiologist at the wastewater monitoring analysis company Biobot Analytics. 

Typically, overdose data is compiled from hospital records, law enforcement reports or self-reported behaviors, Keshaviah said. But that information can present an incomplete picture of the risk because it doesn’t account for all overdose events. Many overdoses go unreported to officials out of fear of arrest or other punitive consequences, and it’s not uncommon for substances contributing to a person’s death to be excluded from death records. 

Wastewater, though, can provide a more comprehensive view of real-time, community-level drug use because it can include anonymous data based on hundreds or thousands of individuals. That’s particularly useful for rural communities, where fewer overdose events likely occur and are reported by police or hospitals. 

Plus, researchers say wastewater data gives officials a more cost effective and timely way to measure the efficacy of harm reduction policy and programming. In March 2023, for instance, Kentucky decriminalized fentanyl test strips, a move experts say can help prevent overdoses. Researchers studied wastewater in two Kentucky counties one month before the policy was implemented and about five months after it went into effect. In both counties, fentanyl concentrations decreased following the test strips’ decriminalization. The decline in the urban county was statistically significant, according to the study.

Researchers also developed the predictive algorithm Drug-SURGE to flag unprecedented rises in wastewater concentrations of the four substances evaluated in the study. Across the five counties, the algorithm successfully predicted fentanyl overdoses with an 86% positivity rate, according to the study. On average, alerts occurred eight to 11 days before overdoses were recorded. 

An eight- to 11-day window is particularly valuable for identifying new drug threats like xylazine, Keshaviah said. That early detection can alert first responders in the area to where data indicates the drug is in circulation to prepare targeted overdose treatments. 

At the regional level, public health officials can better predict potential overdoses along drug trafficking routes if, for example, drug-use data from neighboring communities indicates an uptick in local supply, said Allison Burrell, program manager at the Washington/Baltimore High Intensity Drug Trafficking Area. 

Buben argues that the return on investment for agencies leveraging wastewater data is a worthwhile investment. Collecting and analyzing that kind of data is “substantially cheaper” than alternative methods to obtain drug-use data such as surveys, which require considerable resource capacity for agencies to develop and administer, he said. Plus, surveys can only scrape data that reflects a certain time frame and depend on user participation, while wastewater data can capture a larger sample with fewer restrictions on when or where data was collected. 

That said, the approach is new and can create a challenge for officials trying to, for example, interpret what’s driving changes in local wastewater data so that they can confront the root cause. 

“There’s still more work to be done in terms of education and interpretation … to extract the full value of [wastewater] data,” Keshaviah said.

But, she adds, the approach is promising. Wastewater offers “a different type of data source that has really unique advantages,” like community-level, nonidentifying information.