Overdoses Rose 65% Within Blocks of a Minneapolis Sweep, an Unpublished Study Finds

When Minneapolis clears a homeless encampment, the block empties in an afternoon. The people who lived there disperse, usually within a few streets, and take their drug supply with them. According to research shared with Axios Twin Cities on Oct. 19, so does the overdose risk — condensed, rather than reduced.

Within a two- to four-block radius of a cleared site, the odds of a nonfatal overdose were 65 percent higher than expected during the following three days. The study, led by University of Minnesota epidemiologist Kumi Smith and not yet published, is one of the first to attach a hard number to a claim that outreach workers in American cities have made for years: that displacing people who use drugs does not scatter the danger, it concentrates it.

The finding lands in the middle of a live policy fight. Minneapolis City Council members advanced a proposal Tuesday that would let unsheltered residents sleep overnight in their cars in supervised lots, with staff on site, bathrooms, trash service and a police-reviewed safety plan. The full council could vote next week; the ordinance would take effect in January. St. Paul has said it intends to clear its remaining large encampments, without yet naming a date. Across the river from a study about the costs of clearing, two cities are sorting through what a non-clearance policy would actually look like.

How the Study Was Built

Smith's team did not start from city records. They worked with community members who had experienced homelessness to build their own database, documenting 87 camps cleared in Minneapolis between April 2024 and March 2025, along with the dates and locations of the 1,431 nonfatal overdose 911 calls dispatched in that span.

That pairing is the study's methodological core. Instead of comparing neighborhoods to each other, it compares a small area against itself: the overdose rate in the same blocks before a sweep and after one, with the expected rate estimated from surrounding space and time. The 65 percent excess is the difference between what was observed and what that baseline predicted.

The result held up under stress. When the researchers widened the ring to eight blocks, the increase shrank but remained statistically significant. When they extended the window to two weeks, it remained significant again — a decay curve rather than an on-off spike, consistent with a population that briefly loses its footing and then partially reassembles.

Smith's team used citizen science deliberately. "Probably captured the sweeps problem better than any other study that we know of," Joshua Barocas, a physician and public health researcher at the University of Colorado Anschutz, told Axios. Official logs tend to record sweeps as they were scheduled, not as they happened — the day crews arrived, the days they returned, the block that was cleared twice in a week.

What the Three-Day Window Implies

A 72-hour excess is short, and that is the point. Three days is roughly how long it takes for a displaced group to find a new place to sleep, a new route to a dealer, and a new set of people to use with. It is also how long it takes for tolerance to drift, for a naloxone kit to be lost in a hurried move, and for the one person who knew how to respond to a bad reaction to be sleeping somewhere else.

Read that way, the finding is less about geography than about the fragility of the arrangements that keep street drug use survivable. The tolerance drop after even a few days of reduced use is well documented; so is the protective effect of using alongside someone who can call 911. A sweep interrupts both at once, then adds a crowd of strangers to the same corner.

Not the First Signal

This is the fourth or fifth time the same pattern has surfaced, from four different methods.

In April 2023, a Barocas-led team publishing in JAMA modeled ten years of continual involuntary displacement across 23 U.S. cities and found no scenario in any city where the practice was neutral for health. Their projections included a 151 percent increase in overdose mortality, a 50 percent increase in hospitalizations, an 11 percent reduction in life expectancy among the unsheltered population, and a 38 percent decrease in initiations of medication for opioid use disorder. The National Health Care for the Homeless Council called it a seminal study; city officials called it a model.

The following year, the same research group tested a narrower claim — that sweeps make neighborhoods safer — using Denver's actual sweep records from November 2019 to July 2023. Published in the Journal of Urban Health, that analysis found no change in most crime categories within a quarter- to three-quarter-mile radius, negligible changes in others, and in some cases higher rates of violent crime after displacement. The encampments, the authors concluded, could not be tied to a rise in crime.

Minneapolis now supplies something neither of those could: measured, local, event-level overdose counts rather than projections. The difference matters for how the finding can be used. A simulation can be argued with on its assumptions. A count of 911 calls in specific blocks on specific days is harder to wave off, which is likely why the reception has been sharper.

The Pushback

Keith Humphreys, a Stanford psychiatrist and former Obama administration adviser, told Axios the research carries a blind spot: it tracks harms to people who use drugs and none of the harms that sweeps are meant to address. Fires in encampments, sanitation, human trafficking, the effect on nearby businesses, the loss of public parks — none of that enters a model built on overdose calls. "No one who isn't homeless and doesn't use drugs matters," he said of that framing. "People who use drugs matter. So does everybody else."

Humphreys also offered the argument that city officials make privately and rarely on the record: that a sweep can be the disruption that pushes someone toward housing or treatment. Barocas's response is that the evidence for that is thin. In the Denver work and in the modeling, displacement produced no measurable increase in treatment entry to set against the mortality.

Minneapolis and St. Paul officials have made a related case on different ground. Drug use, they argue, is one risk among several that encampment residents face, and the neighbors' complaints are not imaginary. The disagreement is less about whether those harms are real than about whether a policy that reliably increases overdose deaths is a reasonable trade for them.

Where the Study Stops

The findings are pending publication, which means no peer review, no methods section to inspect, and no chance yet for other researchers to test the specification. Three limits are visible from the summary alone.

The outcome is 911 calls for nonfatal overdose, which means fatal overdoses are excluded and reporting behavior is baked in. A block where residents have learned that an ambulance brings police may under-call; a block where outreach workers are present may over-call. Neither bias has an obvious direction relative to sweeps.

The design also cannot fully separate displacement from convergence. If people scatter two blocks east, the eastern ring absorbs both the original population and the newcomers, and the excess risk may reflect density rather than disruption. The eight-block robustness check helps with this but does not settle it.

And there is no individual-level linkage. The study can say that overdoses clustered near cleared sites; it cannot say that the people overdosing were the people displaced. That distinction is the difference between a mechanism and a correlation, and it is the first thing a journal's reviewers will press on.

The Clearings That Started This

Minneapolis did not arrive at safe parking by accident. Council members first raised the idea in 2024, after a run of high-profile clearings in south Minneapolis — among them Camp Nenookaasi, where residents packed up in January of that year as officials prepared to move them out. The safe parking proposal has been in circulation ever since, which means the ordinance now moving toward a vote is roughly two years old and has outlasted at least one round of encampment politics.

The study window, April 2024 through March 2025, sits inside that period. So does the legal shift that made aggressive clearing easier: the Supreme Court's 2024 decision in Grants Pass v. Johnson held that cities may enforce bans on sleeping outdoors, removing a constraint that had shaped municipal policy across the West. Barocas noted at the time that the ruling gave cities like Denver room to displace people, and argued the analysis should prompt them to ask whether sweeping was the best use of resources.

What Cities Are Doing Instead

The Minneapolis safe parking ordinance is, in effect, a test of the alternative the study implies. Rather than dispersing people across the grid, it concentrates them in managed lots with staff, sanitation and a curfew — the conditions that a sweep destroys. Council member Aurin Chowdhury framed it as using existing infrastructure to deliver wraparound services in a space that is managed, clean and can hold a community together. Maikao Vue, who manages the city's Homeless Response Team, told MPR News that the four-person unit has fielded rising 311 reports about occupied vehicles and that occupants are usually away when staff arrive, so the team leaves a pamphlet and often returns to an empty space.

Smith points to a different family of models, set out in the Benioff Homelessness and Housing Initiative's encampment resolution guide: gradual connection of residents to services, with closure as the end of a process rather than its first step. "Cities do not have to choose between only allowing encampments to persist or only displacing people," she told Axios. "Sweeps consume enormous public resources only to displace people down a few blocks, all while driving up the risk of overdose."

For treatment providers, the operational lesson is about timing. If the three days after a clear-out are the highest-risk window, then naloxone distribution, street medicine and outreach should be scheduled against the sweep calendar rather than the outreach team's. That is an argument for opioid addiction treatment that travels to where people land — mobile prescribing, same-day buprenorphine starts, and kits handed out during the clearing itself, when the population is briefly in one place and reachable.

Whether any of it happens depends on a vote next week, and on whether a study that has not yet cleared peer review is enough to change a practice that is easy to schedule and easy to defend. Smith's summary of the alternative is blunter than the data. Sweeps, she said, trap city governments in endless, expensive loops. The overdoses, on this evidence, are one of the costs of the loop.