Pennsylvania Halved Its Overdose Deaths. Its Hospitals Just Broke a Record.

Pennsylvania has spent four years reporting the best overdose news in a generation. Fatal overdoses peaked at 5,429 in 2017 and fell to 2,810 in 2025, a decline of nearly half, according to state surveillance data. Then this week a different state agency published the column nobody had been watching. Opioid use-related hospitalizations rose from 31,601 in 2024 to 37,287 last year, the highest total since the Pennsylvania Health Care Cost Containment Council began publishing the county tables, and the first time the figure has passed 32,000.

The council's reports are three pages of county rows, released with no press conference, and they split the question in two. Hospitalizations coded for opioid overdose went from 2,330 to 3,058 statewide, a 31 percent increase. Hospitalizations carrying an opioid use disorder diagnosis, a broader category that captures anyone admitted for withdrawal, infection, trauma or a cardiac event whose chart also records addiction, climbed 18 percent, with the statewide rate rising from 292.2 to 340.3 per 100,000 residents age 15 and older. Philadelphia accounted for much of the movement: overdose-coded admissions there rose 64 percent, from 917 to 1,504, and OUD-coded admissions 31 percent, from 9,650 to 12,685. No county in the commonwealth now records a higher OUD hospitalization rate than Philadelphia's 965.8 per 100,000.

Read together with the death trend, the numbers describe a state where the same drug supply is killing fewer people and hospitalizing more of them. Clinicians in Pennsylvania have a specific explanation for that, and it is not a happy one.

Two columns, two different patients

The distinction inside the PHC4 tables matters more than the headline total. A hospitalizations-for-overdose count is an event measure: someone arrived with respiratory depression and toxicology to match. A hospitalizations-with-OUD count is a population measure: anyone whose discharge record carries a code from the opioid use disorder block, whatever brought them through the door.

That second column is where the clinical story lives, and it is the one that has been climbing. It is also the one most sensitive to how much a hospital tests, codes and asks. The council limits the analysis to general acute care hospitals and suppresses counties with low volume, which means the state's smallest counties appear as blanks rather than zeros — a caveat that matters when the totals are read as a single number rather than as 67 local estimates.

A sedative that naloxone cannot reach

Medetomidine is a veterinary sedative, an α2-adrenergic and imidazoline receptor agonist, and it is between 100 and 200 times more potent at that receptor than xylazine, the tranquilizer that dominated the supply before it. The American Society of Addiction Medicine's new guidance traces the timeline: first detected in the drug supply in 2021 through seized samples, wastewater and paraphernalia, with the National Institute of Standards and Technology's rapid drug analysis program reporting nationwide presence in opioid-positive samples by December 2025. Prevalence is highest in the Northeast and Midwest, and adulteration reaches as high as 90 percent in some mid-Atlantic opioid supplies.

Philadelphia's public health laboratory puts a local number on it. In the second quarter of 2026, medetomidine appeared in roughly 81 percent of suspected dope samples submitted to the city's drug checking program. That is a supply transition, not a wave of new users: Pennsylvania scheduled xylazine as a controlled substance in 2024, and medetomidine filled the space, according to Carla Sofronski of the Pennsylvania Harm Reduction Network.

The clinical consequence shows up first in what naloxone does not do. A patient who is sedated on fentanyl and medetomidine can have their breathing restored and remain motionless, with a heart rate in the thirties. Emergency physicians in Philadelphia and Pittsburgh learned to titrate naloxone to respiration rather than to consciousness, because consciousness is not coming back on the schedule the drug usually keeps.

Four hours later

The withdrawal is the part that rearranged hospital capacity. Physicians in those same two cities first described the syndrome in 2024: anxiety, tremor, drenching sweats, vomiting, agitation and encephalopathy beginning within four to six hours of the last use, with heart rates observed above 170 beats per minute and blood pressure reaching hypertensive emergency ranges of 170 to 240 systolic over 120 to 140 diastolic.

What made it recognizable was what failed. Benzodiazepines and phenobarbital, which reliably quiet the sympathetic surge of alcohol and benzodiazepine withdrawal, did not touch it. Neither did full opioid agonists, which handle ordinary opioid withdrawal. What worked was an infusion of dexmedetomidine, the human-approved enantiomer of the same molecule.

"I've taken care of terrible overdoses and terrible withdrawal for two decades, and I've never seen a withdrawal syndrome like happens with medetomidine," Michael Lynch, medical director of the Pittsburgh Poison Center, told the Erie Times-News. Kory London, an emergency physician at Jefferson Methodist in Philadelphia, put the disposition problem plainly: "This isn't something that's safe for patients to kind of tough out at home."

One bag, one hundred ICU loading doses

The dose arithmetic explains why. ASAM's guidance states that in the parts of Pennsylvania with the highest measured medetomidine concentration in the supply, a single bag of unregulated fentanyl-medetomidine can contain an amount of dexmedetomidine roughly 100 times a typical intensive care loading dose. That is not a potency comparison to other adulterants. It is a comparison to the dose hospitals give under continuous monitoring.

The bill arrived before the protocol did

London and colleagues measured what that did to critical care in a study published in the Journal of Addiction Medicine. Across two Philadelphia ICUs and 102 months of discharge data, the median quarterly charges for admissions coded for opioid withdrawal rose from $5,585,340 in the fentanyl-only era to $9,627,485 once xylazine dominated, then to $15,283,391 after medetomidine arrived. Median quarterly ICU admissions for the same diagnosis went 18, then 34, then 64.

The detail that makes the series worth citing is the one nobody quotes. Charges per patient did not rise. They fell, from $193,492 to $176,225 to $121,402. The spending increase came entirely from more people needing an ICU bed, not from each stay becoming more expensive — which is exactly the shape of a supply problem rather than a billing one.

Two ICUs in one city over eight and a half years is a narrow base for a national claim, and the authors label it a longitudinal retrospective analysis at a single center's two units. ICD-10 code F11.23 catches what physicians documented, not what patients took.

ASAM's first guide, and the limits it states

On Nov. 10, ASAM released the first national clinical guidance for medetomidine intoxication and withdrawal, written by Durkin, Lynch and Perrone and approved by the society's Quality Improvement Council and board in October. Its literature search ran through April 21, 2026, screened 529 studies and included 46, and the document says outright that "there is insufficient evidence available on this topic to support a clinical practice guideline." What follows is expert consensus, and the society labels it that way.

The recommendations, in order of what a patient needs first: treat it as a polysubstance overdose, give naloxone incrementally until breathing is adequate, and stop titrating toward wakefulness. Do not reach for an α2 antagonist to reverse the sedation; none is approved for human use, and in a patient with chronic exposure it can precipitate the withdrawal the clinician is trying to prevent. Expect three to six hours of observation for the intoxication itself.

For withdrawal, the guidance points to α2 agonists at doses far above the usual, intravenous dexmedetomidine for severe tachycardia and hypertension, and disposition outside the emergency department: mild cases to ASAM Criteria Level 2.7 or 3.7 programs with nurse monitoring, moderate and severe cases to hospital and often ICU care, then to medically managed residential treatment once stabilized. It asks payers to cover that level of care, and it says something that will be quoted in courtrooms and budget hearings alike: existing reports cannot support a definitively safe observation period. Nobody can yet say which patient will deteriorate after six stable hours.

The guidance is equally clear that none of this replaces opioid use disorder treatment. "Medications for opioid use disorder remain a key treatment, even in the setting of this novel adulterant," Durkin wrote.

What the numbers cannot tell you

No point-of-care test exists for medetomidine. Toxicology panels do not include it, and liquid chromatography with quadrupole time-of-flight mass spectrometry is a reference laboratory method, not something an emergency department runs at 3 a.m. Diagnosis is clinical, based on bradycardia, sedation and a withdrawal course that behaves like nothing else — which means every count in this article is a count of what clinicians recognized.

That cuts in both directions. Rising hospitalizations could reflect more illness, better recognition of it, or simply more people surviving long enough to be admitted, since naloxone saturation converts what would have been a death into a transport. The PHC4 tables do not adjudicate among those explanations, and no state agency is currently positioned to, because deaths are counted in one system and discharges in another.

What treatment has to absorb

The practical consequence is a capacity question that the overdose death decline has allowed everyone to postpone. A withdrawal syndrome that lands people in intensive care for days, in a state where the alternative to hospitalization is a medically managed residential bed that may not exist nearby, turns a pharmacology problem into a bed problem.

The system's two anchors still work. Naloxone restores breathing, and medication for opioid use disorder — methadone, buprenorphine, extended-release naltrexone — keeps people out of withdrawal and out of the supply. Neither addresses the sedative, which is why the guidance's most portable recommendation may be the least clinical one: watch the local supply, and let surveillance rather than the individual chart tell you what is in the bag.

For families trying to make sense of a hospitalization that followed an overdose someone survived, the useful frame is that opioid addiction treatment now has two problems to solve at once — the dependence, which has three medications, and the adulterant, which has a protocol that was ten days old this week.