Urine drug test (UDT) data can generate timely estimates of overdose deaths, a new study suggests.
The study, published in JAMA Network Open by specialty lab Millennium Health and The Ohio State University, aimed to determine whether UDT data could provide near real-time indications of overdose trends. Effective responses to the overdose crisis must be prompt, the study noted, which requires a timely evaluation of current trends. However, current publicly available data on fatal overdoses in the U.S. can lag by at least six months.
“We were determined to close that gap,” Eric Dawson, vice president of clinical affairs at Millennium Health, told Fierce Healthcare. “We wanted to be able to tell people, here’s what’s happening today with overdoses—compared to here’s what you’re being told today happened six months ago.”
The study designed a model to estimate ODs for the most recent 6-month period, for which CDC provisional data is not yet available. To evaluate model performance, estimated OD death counts were compared to the observed OD death total. More than 3.1 million Millennium urine samples collected between January 2015 through January 2025 were analyzed. They were taken from adult patients in substance use disorder treatment across all 50 states and Washington, D.C. The study looked at monthly, national drug OD mortality counts from the Centers for Disease Control and Prevention (CDC) for the observed total.
The samples were analyzed for positivity rates of fentanyl as well as methamphetamine among fentanyl users, as well as their respective concentrations. These substances were chosen based on the researchers’ understanding that the fourth wave of the crisis is defined by fentanyl use and fentanyl-stimulant polysubstance use.
“This continuing evolution—it just paves the way for the next sudden change in overdoses,” Dawson said. “It could change tomorrow and [overdoses] remain a significant threat.”
Ultimately, the agreement between the estimated trends for fatal overdoses and what was actually observed was very close. The model is able to estimate as recently as the most recent complete month of UDT data, shaving the typical six-month lag window down to mere weeks. This is likely because UDT data serve as a proxy for local drug supply and prevalence, according to Dawson.
A specific spike in deaths the researchers hoped UDT data would capture was a substantial increase in early 2020 at the onset of COVID-19. More than 1 million people have died from a drug overdose in the U.S. since 2000, with about 40% occurring from 2020 to 2023. Indeed, the modeled projections captured the change in overdose trajectory during that time.
“Aggregated UDT data may function as a proxy measure for the presence of nonprescribed and/or illicit drug use at the population level, providing information on the prevalence of use of a variety of substances … across time,” the study said.
Having more timely data is a federal policy priority. In April, the Trump administration put out its drug policy priorities. They included innovation in research and data to support drug control strategies. “The Administration will prioritize the continuous collection and analysis of accurate, timely, and relevant data to inform policy,” an official document read (PDF).
“This is not a new problem. It’s been well characterized,” Dawson said. In a 2022 editorial, experts from the National Institute on Drug Abuse wrote that the lack of real-time data on drug overdose deaths means fighting the crisis “blindfolded.”
The editorial argued that resolving the crisis will require a multilayered data approach like the one the U.S. undertook during COVID-19. States and municipalities developed surveillance systems to track new cases and deaths daily, and these data included demographic information that enhanced interventions.
“It is time to realize the power of a parallel data system to address drug overdose deaths,” the authors wrote. “Localities could track real-time data on drug overdose events and deaths … Giving communities the ability to understand who needs prevention and treatment services will remove the blindfold and ensure efforts and resources are finally targeting those who are most at-risk.”
Millennium Health already works with the CDC today. As a next step, it hopes to explore additional data sources with its partners. Dawson isn’t sure whether any UDT data would suffice, given the need for validation of a model, but Millennium is open to a variety of other data sources including from EDs, EMS, poison control centers and workplace testing.
“Let’s combine these data sets and see if we can really strengthen these estimates,” Dawson said. “Instead of looking in the rears, let’s look six months forward.”
While Dawson has not yet had conversations with treatment providers about the study, “I hope, if nothing else, they feel empowered: that I have to stay in this fight, because there’s a lot of good work.”