Mobile Units Lowered Deaths but Not Treatment Uptake in Five-City Trial

A mobile medical unit that delivered HIV prevention and addiction treatment directly to people who inject drugs did not increase their use of either medication, according to trial results published Wednesday in JAMA Psychiatry. But four people died in the group assigned to the mobile units over 52 weeks, against 13 in the group referred to ordinary community services — a difference the National Institutes of Health described as a 71 percent reduction in all-cause mortality.

The trial, known as HPTN 094 or INTEGRA, enrolled 409 adults across Houston, Los Angeles, New York City, Philadelphia and Washington, D.C. All of them injected drugs, had opioid use disorder, were not taking medication for it at enrollment, and said they were willing to consider it. Every participant received 26 weeks of peer navigation. Half also got what the study called integrated PrEP, medication for opioid use disorder and limited clinical and supportive services delivered on a vehicle parked in their neighborhoods.

The primary outcomes came back flat

At 26 weeks, verified PrEP use was 5.4 percent in the mobile-unit arm and 2.9 percent among those referred to community services. Medication for opioid use disorder was used by 14.8 percent of the mobile-unit group and 10.7 percent of the comparison group. At 52 weeks the arms were essentially indistinguishable: 2.0 percent against 1.9 percent for PrEP, and 13.3 percent against 14.1 percent for addiction medication.

None of those gaps reached statistical significance, and the paper's conclusion states plainly that mobile delivery did not improve the primary outcomes beyond peer navigation alone. The finding cuts against a widely held assumption in harm reduction circles — that proximity is the binding constraint. In this trial, a van that brought the medications to the corner did not produce more people on them than a navigator who walked people to an existing clinic.

NIDA's director framed the population rather than the endpoint. "Injection drug use is an important contributor to HIV in the U.S. that also increases risk for other serious infectious diseases," Nora D. Volkow said in the agency's announcement. "Although we have effective interventions for this population, there are significant barriers to their accessibility and use. Nonetheless, these findings suggest that delivering low threshold services through mobile units could save lives."

A mortality difference nobody designed for

Death was a secondary outcome, not the thing the trial was built to measure. Four deaths occurred among the 203 participants assigned to mobile units and 13 among the 206 referred to community care, a hazard ratio of 0.29 with a confidence interval running from 0.1 to 0.9 and a P value of .03.

That is a wide interval. With 17 deaths in a study of 409 people, the estimate is compatible with a very large benefit and with a considerably smaller one, and the authors wrote that the result "bears replication" rather than offering an explanation. The trial also recorded three HIV acquisitions — one in the mobile-unit arm and two in the comparison arm — an incidence of 0.9 per 100 person-years.

The study ran from May 2021 through September 2024, a period in which the illicit supply shifted decisively toward fentanyl and, in some cities, toward stimulants mixed with it. Participants were recruited at a moment when the case for meeting people where they are had become an article of faith in federal drug policy, and the trial is one of the few randomized tests of that premise in the United States.

Why 14 percent counts as a result

Fourteen percent sounds like a failure to anyone reading it as a share of everyone who injects drugs. Within the trial it read differently, because no participant was on medication for opioid use disorder when they enrolled and the 26-week and 52-week numbers barely moved — 14.8 percent down to 13.3 percent in one arm, 10.7 percent up to 14.1 percent in the other. The people who started stayed, in a population that cycling in and out of treatment is the norm.

"People who inject drugs often face fragmented systems, unstable housing, stigma, and other barriers that make sustained care difficult," said Nabila El-Bassel, a professor of social work at Columbia University and a co-chair of the study, in the NIH release. "Our findings show why we must continue developing models that meet people where they are and connect HIV prevention, addiction treatment, and supportive care in ways that are accessible and responsive to their lives."

The PrEP numbers were the weakest part of the result. About one in nineteen participants in the mobile arm had verified use at 26 weeks, and the investigators wrote that the figure, while higher than self-reported PrEP use among people who inject drugs, points to a deeper problem of acceptance rather than of access. Their recommendation is to test long-acting injectable PrEP in this population, where a daily pill competes with housing instability, incarceration and the routines of drug use.

A qualitative paper from the same trial, published in the Journal of Acquired Immune Deficiency Syndromes in August, described what staff on the units ran into: keeping the vehicle stocked and staffed, coordinating with local health systems, and induction protocols for buprenorphine that were hard to run in a van. A separate analysis in Drug and Alcohol Review found fear of precipitated withdrawal — heightened by fentanyl's presence in the supply — shaping whether participants would accept buprenorphine at all.

What the trial changes for treatment planning

The practical takeaway is narrower than the headline the NIH put on its release. Mobile units are not a demonstrated way to raise medication uptake. They are a demonstrated way to reach people who otherwise would not have been reached at all, and in this trial that population had a lower death rate, for reasons the data cannot yet explain — whether from faster access to care after an overdose, more contact with staff, or something else entirely.

Health departments and nonprofits have spent the past several years buying and staffing mobile units under a 2021 federal rule that lets opioid treatment programs extend methadone dosing through them. Operators have stood them up in states from California to Florida, including a New Season unit that began dosing patients in Nassau County, Florida, in late October. The results published this week argue for measuring those units on retention and survival rather than on the share of a caseload newly started on medication. For a field that has leaned on medication-assisted treatment as its central answer to the overdose crisis, that is a different scorecard.

The trial was sponsored by NIH's National Institute of Allergy and Infectious Diseases and funded by NIDA, through awards UM1AI068619, UM1AI068617 and UM1AI068613. Steven Shoptaw of the UCLA Department of Family Medicine served as the other co-chair. Results from an earlier phase of the program, in which a mobile unit brought buprenorphine to patients in one California county, were covered here in October; the broader opioid use disorder picture behind this trial is one in which most people who need medication still do not receive it.