The Overdose Decline Has a Hole in It, and Meth Is Standing in It

Kastalia Medrano used meth every day for three or four years. When she wrote about it in Filter this August, the line that stuck was not about the drug. It was about the services that were supposed to help her: harm reduction, she argued, was built for opioids and “retrofitted” for people like her. She called meth “the people’s drug,” the thing people reach for to stay awake through a double shift, to get through a night outside, to feel alive or numb, depending on the day.

This week the state of Minnesota put numbers on what she was describing, and they belong to our crisis-data coverage of the stimulants story we keep returning to.

America learned to save people from fentanyl and has not yet learned to keep them from meth, and the data has started to say so out loud.

The Minnesota Department of Health published its 2025 overdose report on Tuesday. Overdose deaths in the state rose 3 percent, from 994 to 1,025, in a year when the country as a whole kept falling. Opioid-involved deaths barely moved (678 to 671). Deaths involving methamphetamine and similar drugs climbed 11 percent. Cocaine deaths rose 6 percent. If you use, or love someone who does, you probably already felt this before the spreadsheet confirmed it.

The decline was real, and it was mostly one drug

Start with what is true and good. The CDC estimates the country lost 69,973 people to overdose in 2025, down from 81,313 in 2024, a drop of almost 14 percent and the third straight yearly decline. Deaths involving opioids fell from 55,296 to 44,564, which is about 19 percent. That is roughly ten thousand fewer opioid deaths than the year before. Nobody who has buried someone should be told to be less glad about that, and this piece is not telling you to be.

But look at what carried the decline. Naloxone, the nasal spray or injection that reverses an opioid overdose by knocking the drug off the receptors that are shutting down your breathing, got into pockets, glove compartments and library bathrooms. Medications like buprenorphine and methadone got easier to reach. Every one of those tools is built around a single mechanism: opioids make you stop breathing, and breathing can be restarted. The Minnesota report says it plainly: stimulants “do not yet have an antagonist that can prevent stimulant-use deaths as naloxone does for opioid overdoses.” A stimulant overdose usually looks nothing like a nod. The person stays conscious and may have chest pain, overheating, seizures, extreme agitation or stroke-like symptoms. There is no spray for that.

So when the opioid curve bends down and the stimulant curve doesn’t, the total stops falling. Minnesota’s own health department notes that fewer than 20 percent of states reported an increase in preliminary 2025 data, and every one of them is in the upper Midwest or the Southwest. which is the region Arizona sits in, and a reason to look hard at your own county’s numbers instead of the national headline. Minnesota’s commissioner, Dr. Brooke Cunningham, told reporters the state has to keep pushing its opioid work while it applies what it learned to stimulants. Dylan Ferguson, who runs the state’s Office of Emergency Medical Services, said “EMS providers are often the first to see when something is changing in a community,” and pointed to a new alert system, launched October 1 and used in 76 of Minnesota’s 87 counties, meant to let communities act before a spike becomes a funeral.

Kentucky shows what winning against fentanyl looks like from the inside

Kentucky is the cleaner experiment, because it did almost everything right and the result is still strange. The state’s 2025 overdose fatality report, prepared by the University of Kentucky’s injury research center for the state drug control office, counts 1,110 resident deaths, down 22.9 percent from 1,439 the year before. The state distributed 182,810 doses of naloxone and kept 82 syringe service sites open that served more than 25,000 people. Fentanyl was found in 504 of the deaths, a drop of 42.9 percent from 883.

The state distributed 182,810 doses of naloxone and kept 82 syringe service sites open that served more than 25,000 people.

Methamphetamine was found in 549 deaths, down 23.8 percent from 720. Cocaine was found in 227, down 20.4 percent. Read that again. Meth deaths fell by nearly a quarter, which is a good outcome by any normal standard, and meth still flipped from second place to first. In 2024 fentanyl showed up in more Kentucky deaths than meth did (883 to 720). In 2025 meth passed it (549 to 504), appearing in 49.5 percent of deaths against fentanyl’s 45.4 percent. One drug fell faster than the other, and the slower one is now the headline.

Two cautions belong here, because the people reading this have been lied to with statistics before. First, these toxicology percentages count a substance wherever it was found, and a single death can involve several drugs, so the shares add up to more than 100 percent and “found in” is not the same as “caused by.” Second, Kentucky’s data comes from a different system than Minnesota’s, and the two states are not interchangeable. What they share is the shape: the opioid line drops, the stimulant line drops slower or climbs, and the mix is where people die. Minnesota’s report says mixed-drug use accounts for 70 percent of its overdose deaths since 2020, and deaths involving both an opioid and a stimulant have more than doubled since 2019, from 154 to 371.

That overlap is the part that should change what we hand people. Hospitals in the state’s surveillance program found amphetamines in 65 percent of the opioid-positive overdose cases they tested. If you only use meth and believe fentanyl is someone else’s problem, the supply has not been consulted on your belief. The department’s own advice to clinicians is blunt: tell patients that fentanyl may be in stimulants, and give them naloxone.

We have watched this before: the drug we decided to punish instead of treat

We have watched a government build its whole response to a drug around punishment, and then keep it long after the evidence turned. In 1986 Congress passed the Anti-Drug Abuse Act, and it set mandatory minimums that treated crack cocaine as if it were a hundred times worse than powder: five grams of crack drew the same five-year sentence as 500 grams of powder. The gap stayed on the books until President Obama signed the Fair Sentencing Act on August 3, 2010, and even then the ratio only fell to 18-to-1, as the White House drug policy office’s own announcement described. That was twenty-four years of a sentencing rule built on a number, with a person attached to every year of it.

The pattern is back with a different stimulant. Federal meth sentencing is driven by purity rather than weight, and the criminal defense bar’s national association told the Sentencing Commission last year that meth is now punished more severely than any other drug. Medrano’s Filter piece traces what that does on the ground, including a death in a Wyoming jail restraint chair that a pathologist attributed in part to “excited delirium,” a label that explains a death by pointing at the drug. In the case Medrano describes, a Black man in a Wyoming county jail spent about forty minutes in a restraint chair with straps across his chest, saying he could not breathe, and the pathologist’s cause of death named “methamphetamine overdose and the cardiac effects of excited delirium.” If the drug is the whole explanation, nobody has to ask about the straps. That is what a drug panic does to a cause of death: it hands the system an answer before the question gets asked. The pattern is the one from 1986: a stimulant gets handled as a crime first and a health problem later, and the delay is paid for by the people using it. The opioid response got its rescue tools. The stimulant response is still waiting on its own.

That is what a drug panic does to a cause of death: it hands the system an answer before the question gets asked.

The tool that works exists, and almost nobody can get it

Here is the part that is not bleak. There is a treatment for stimulant use disorder with a strong evidence base, and it is not a pill. The ASAM/AAAP clinical guideline on stimulant use disorder, published in the Journal of Addiction Medicine in 2024, recommends contingency management, a structured program that pays people small rewards for tests that show they have not used, as a primary part of treatment, and rates that recommendation strong with high certainty. Medications like bupropion, mirtazapine and naltrexone are used off-label, and none is FDA-approved for stimulant use disorder. A Stanford review of 157 cocaine-treatment trials, as described by the psychiatrists Sally Satel and David Farabee in the Los Angeles Times, found contingency management the notable exception to a field of mostly failed treatments, more than doubling the odds of a negative cocaine test. They do not name the Stanford study, so treat the headline number as reported and not independently confirmed here.

The best real-world signal they cite comes with an honest limit. Patients in the VA’s program who earned vouchers between 2018 and 2020 were, per Satel and Farabee, 41 percent less likely to die within a year than a matched comparison group. Matched is not randomized, and people who stay in a rewards program may differ from people who don’t. Still, it is a death-rate result, not a lab result, and it is the kind of thing naloxone had to prove too.

The gate is access. As of early 2025, Satel and Farabee report, only Delaware, Hawaii, Montana and Washington had federal approval to pay for contingency management through Medicaid, and the federal grant cap on rewards had risen from $75 to $750 per person per year, in gift cards or services rather than cash. California has run a 24-week program since 2023. Maricopa County is among the places funding one from opioid settlement money. We covered the Medicaid side when the first states signed on. That list may be longer now. Check it for your state before you promise a patient anything.

If you work in a clinic, here is a week’s worth of moves. Add one question to every stimulant visit: “what else could be in this?” Put naloxone and fentanyl test strips into the same conversation, in the same breath, without making it a lecture, because the Minnesota health department tells its clinicians to do exactly this. Ask your team on Monday whether your referral logic still sends every meth patient to a residential bed by default, since the best-supported treatment is outpatient and rewards-based. Find out which payer, grant or settlement line in your county could pay for even a small incentive program, and put one name next to that task.

If you use, this is still yours: naloxone works on mixed overdoses, and the Minnesota report says to give it when you cannot tell what someone took. Call 911 if someone is in trouble, and know that Minnesota’s Good Samaritan law is built to protect the person who calls. If you want to talk to a person tonight about alcohol or drugs, 988 takes calls and texts.

The country can keep counting what it stopped and call the rest unfinished business. Or it can notice that Medrano told us what the retrofit would cost before the spreadsheet did. Somewhere tonight someone is on a night shift, awake on purpose, and the question for the rest of us is whether anything we built is standing there when they need it.