A Hospital Sent Jean Descamps to a Bus Stop Instead of Treatment. He Died Hours Later.

Jean Descamps was 26, emaciated, and barely responsive when hospital staff at Providence Milwaukie Hospital, outside Portland, revived him with the overdose-reversal medication naloxone and decided he wasn’t a medical concern. Police officers on the scene disagreed. “He can’t be on his own,” one of them said, according to NPR’s investigation. Staff told them a bus station would do. Someone suggested he was faking his condition — malingering, in the clinical word one staffer chose. Nobody ran a toxicology test. Nobody offered him the medications that quiet opioid withdrawal and cut overdose risk roughly in half — buprenorphine and methadone — or even a referral to get either. He was wheeled into the parking lot. He was dead within hours.

That’s not a story about one bad shift at one hospital. It’s the default setting.

American medicine has the tools to stop opioid overdose deaths and routinely declines to use them — not because the evidence is thin, but because the people holding the prescription pad don’t trust the patient in front of them.

You already know what that room feels like if you’ve been in it. Maybe not that exact parking lot, but a version of it — the moment a nurse decides you’re a behavior problem instead of a patient, and the discharge paperwork moves faster than any conversation about what happens to you next. NPR’s investigation found that isn’t a fluke of one hospital’s culture. It’s a pattern documented across the country, and it’s the reason more than 40 million Americans living with a substance use disorder mostly navigate a healthcare system built to look past them.

The Numbers Behind the Parking Lot

Descamps’s death fits a shape NPR found repeated at scale: more than 250,000 Americans die every year from alcohol and drug disorders combined, and more than 80% of people who need medical treatment for addiction never receive it. Two out of three people who die of a fatal overdose had already crossed paths with the healthcare system — an ER visit, a hospital stay, a primary care appointment — where someone could have started treatment and didn’t. Descamps’s case isn’t the exception inside that two-thirds. It’s the textbook version of it: a hospital bed, a diagnosis, a discharge order, and silence where a prescription should have been.

The follow-up reporting is the part that should make you angry rather than sad. NPR found Providence Milwaukie was still discharging vulnerable patients without addiction treatment well after Descamps died, and no one — not the hospital, not an individual staff member — faced discipline. The system didn’t correct itself. It didn’t need to, because nothing inside it treats a missed opioid use disorder diagnosis the way it treats a missed heart attack. Dr. Judy Chertok, an addiction medicine physician at the University of Pennsylvania, makes the comparison directly: a hospital would never send a heart attack patient out the door without starting treatment for the heart attack. Opioid use disorder gets a different standard, and the difference isn’t clinical. Beth Meyerson, a researcher at the University of Arizona College of Nursing, told NPR that widespread treatment access could cut opioid overdose deaths by 60% to 80% — a number large enough that failing to act on it isn’t a gap in the system. It’s a choice the system keeps making.

The Medicine Works. The Doctors Don’t Trust It.

Here’s the plain version first: there are medications that quiet opioid withdrawal and cravings without producing the same high — buprenorphine and methadone — and a monthly injection that blocks opioids from working at all — naltrexone. Three different tools, all FDA-approved, all backed by decades of outcome data. None of them require a leap of faith. What they require is a physician willing to prescribe them, and NPR reports that only 1 in 5 physicians express any interest in treating opioid use disorder at all, a figure that traces back to a 2020 study in the Annals of Internal Medicine.

Three different tools, all FDA-approved, all backed by decades of outcome data.

Robert DeForde of the addiction nonprofit Shatterproof told NPR that 43% of healthcare professionals still believe addiction medications simply substitute one drug for another — the exact “trading one addiction for another” framing that treats methadone and buprenorphine as a moral compromise instead of medicine. Dr. Bobby Mukkamala of the American Medical Association didn’t dodge the pattern when NPR asked him about it; he described clinicians who say treating addiction is “not something I’m comfortable with,” which is a doctor admitting the barrier is personal, not scientific. Dr. Nora Volkow, who directs the National Institute on Drug Abuse, framed the fix as a training failure as much as a resource one — asking how medical and nursing schools even begin to “embed education and training” that most of them currently skip.

That last point matters more than it sounds. Addiction care isn’t a mandatory rotation in most medical schools. A physician can finish training having spent more hours on rare tropical diseases than on the medication most likely to save the next patient who overdoses in their ER. Stigma fills the vacuum where training should be — and stigma doesn’t require malice. It just requires a doctor who’s never been taught that buprenorphine works, meeting a patient who looks, to that doctor, like a problem instead of a diagnosis.

If you’re a provider or case manager reading this, here’s the one thing to change this week: before signing off on any discharge where opioid use is even plausibly part of the picture, ask the same question you’d ask for any other unstable condition — “what happens to this person in the next six hours without treatment?” — and don’t let the chart close until there’s an answer besides a bus schedule. Emergency departments that adopt a standing order letting any credentialed ER physician start a patient on buprenorphine on the spot, before an addiction specialist ever walks in — what’s often called a bridge order — cut that gap without requiring every doctor to become a believer overnight. It’s a policy fix for a trust problem, and it doesn’t wait for stigma to end first.

The Regulation Stigma Built, and Never Tore Down

This isn’t new. In the early 1970s, methadone maintenance was already showing it could keep people alive and functional — and it terrified Congress anyway. Lawmakers and the DEA didn’t see legitimate medicine; they saw the government handing out a second addiction to treat the first one. So they built a cage around it: methadone for addiction can only be dispensed through federally licensed clinics, with patients required to show up in person, often daily, to swallow a dose under supervision, for months or years, regardless of how stable they are. That regulatory structure passed in 1974 and has survived, largely unchanged, for more than fifty years — one of the most restrictive dispensing regimes attached to any medication in the country, built entirely on the belief that people in recovery couldn’t be trusted with their own treatment. The ER doctor in 2026 who hesitates to prescribe buprenorphine to a patient who “looks like a user” isn’t inventing a new instinct. He’s carrying the same one Congress legislated into concrete in 1974 — just without the paperwork to make it official this time.

Recovery Priced Like a Luxury Nobody Budgeted For

Discharge rooms aren’t the only place the system fails people already doing everything right. STAT News reported on a patient who had spent years stable on buprenorphine, passing every urine screen, holding a job, and raising kids — until an insurance premium hike moved their monthly cost from $40 to $138, a 245% increase, with no clinical reason and no warning. That’s not a rounding error to someone living paycheck to paycheck. It’s the difference between filling a prescription and not.

John Fomeche, an addiction medicine fellow at Yale who wrote the piece, put it as bluntly as the numbers deserve: “That number may not register as catastrophic to policymakers or insurers. But to someone living paycheck to paycheck, it is not an ‘adjustment.’ It is a threat.” His larger point is the one this whole story keeps circling back to — that stability isn’t a nice-to-have wrapped around treatment, it’s part of the treatment itself. “Insurance stability is not ancillary to treatment,” he wrote. “It is treatment.” A person whose medication access depends on a premium they can’t predict from one renewal to the next isn’t in stable recovery. They’re in recovery until the bill changes.

Multiply that patient by every state Medicaid program absorbing federal cuts to addiction services, and the picture gets worse before it gets better. Arizona’s Medicaid agency, AHCCCS, is still working through the fallout after federal substance use treatment and mental health block grant dollars were abruptly terminated, money that funded exactly the kind of outpatient and prevention infrastructure that keeps a $40 premium a $40 premium. And the newest CDC data shows why the state can’t absorb that kind of hit quietly: while national overdose deaths fell for a third straight year — 69,973 in 2025, down from 81,313 in 2024, with opioid deaths dropping from 55,296 to 44,564 — Arizona was one of only three states, alongside New Mexico and Colorado, where overdose deaths rose by more than 10% instead.

Multiply that patient by every state Medicaid program absorbing federal cuts to addiction services, and the picture gets worse before it gets better.

What Craving Actually Does to Your Brain

There’s a piece of this that isn’t about hospitals or insurance at all — it’s about what’s happening inside your head when you want to use, and new research says it’s more mechanical than most people, including plenty of clinicians, assume. Researchers at Yale studied 132 people with moderate to heavy alcohol or cannabis use, having them play a simple slot-machine game where the payouts alternated between cash and substance-related images — a beer, a bong. Then they built a mathematical model of how each person learned from wins and losses, a method researchers call computational modeling, to see whether craving changed the learning itself, not just the choice in the moment.

It did. People with stronger alcohol cravings picked up winning strategies for alcohol-linked rewards faster than everyone else — their brains got quicker at chasing the thing they craved. People with stronger cannabis cravings showed the opposite: slower learning around cannabis-linked rewards, a different kind of distortion pointed at the same target. “Addiction is very hard to treat, and one of the reasons is that we don’t fully understand the craving linked with it,” said Xiaosi Gu, the study’s senior author and director of Yale’s Computational Psychiatry Unit. “This could explain why breaking the addictive cycle feels so difficult, as the brain is adapting constantly.”

That’s worth sitting with if you’ve ever been told, directly or by implication, that relapse is a willpower failure. The Yale team’s data says craving isn’t a mood you should be able to argue yourself out of — it’s actively rewiring how your brain values rewards and learns from what happens next, in real time, every time you’re exposed to the thing you’re trying not to want. That’s a biological account of why “just white-knuckle it” was never a real treatment plan, and it’s the same argument, from a different angle, for why a hospital that skips medication at discharge isn’t offering a neutral choice. It’s leaving a rewired system to fend for itself against the exact mechanism the medication was built to interrupt.

None of this erases the fact that the trend line is moving in the right direction — the CDC’s third straight year of falling overdose deaths is real, and naloxone remains federally funded and available without a prescription in every state, which means the tool that could have bought Jean Descamps more time than a parking lot was never the scarce part of this story. What’s scarce is a system willing to hand him what came next.

Descamps didn’t die because medicine didn’t have an answer for him. He died in a parking lot with the answer sitting on a formulary somewhere inside the building he’d just left.