Arizona's Cocaine Deaths Rose Nearly 70%. The Treatment That Works Is a Gift Card.

Bernard Groves had spent five years trying to quit methamphetamine when he started peeing in a cup for gift cards. He was 35. He had lost his job and his car and come close to losing his apartment, and he had been through rehab programs in San Diego and San Francisco where, as he described it, the therapy felt like people talking at him. He kept going back to meth. Then, one June, he began weekly sessions with a counselor named Andrew Dertien, where each clean urine test earned a gift card and each consecutive clean test earned a bigger one. “Like, how could you say you’re excited to pee in a cup?” Groves told the reporter Ryan Levi. “But I was, every week.”

By the first week of July he had stopped using, one of the longest stretches he’d managed in five years. He spent the cards on weights, on food for his pet bird, London, on donuts, and on a movie night with his mom, his sister and his grandmother. He said it brought “some joy back in my life.”

Arizona’s cocaine deaths rose nearly 70 percent last year, and the one treatment with real proof behind it for cocaine and meth is a gift card that, as far as we can find, Arizona’s Medicaid program doesn’t pay for.

Arizona is the place where Groves’s story matters most, and the place least likely to be able to repeat it.

The national numbers are good news, and Arizona is the exception hiding inside them

Start with what is going right, because it is real. The CDC’s latest provisional count, announced August 31, predicts 67,798 overdose deaths for the 12 months ending March 2026, a 12.3 percent drop from the year before. Calendar-year 2025 came in at about 70,000 deaths, down roughly 14 percent and the third straight annual decline, according to the Associated Press’s reading of the CDC data. Naloxone in every pharmacy, test strips in every outreach bag, and a shrinking supply of the people most likely to die are all on the list of reasons researchers give.

Arizona went the other way. The Arizona Public Health Association’s analysis of the same CDC data estimates deaths rose 18 percent, from 2,531 in 2024 to 2,988 in 2025. Deaths involving fentanyl and other synthetic opioids rose nearly 33 percent. Deaths involving methamphetamine and other stimulants rose more than 15 percent. Cocaine-related deaths rose almost 70 percent. The association’s write-up doesn’t name a cause, and neither will we, because nobody has proved one yet.

What researchers do have is a strong guess. Brandon Marshall, a Brown University researcher who studies overdose trends, told the AP that the increases in Arizona, Colorado and New Mexico might come from people increasingly combining fentanyl and methamphetamine. If you use, or love someone who does, you already know what that means in practice: the supply stopped arriving one drug at a time. A person who thinks of themselves as a meth person can be taking in fentanyl they never chose, and naloxone, which reverses fentanyl, does nothing for the stimulant on top of it. Marshall’s warning is worth reading twice: “If deaths are going down rapidly, that means they can increase just as rapidly if we take our foot off the gas.”

We covered the first wave of this argument in our look at HHS’s claim that overdoses were falling. The new piece of the puzzle is what Arizona has to offer the person on the stimulant side of the ledger. Nearly everything the state built in the last decade was built for opioids. Methadone, buprenorphine and naloxone are medicines with decades of evidence and a reimbursement code. For meth and cocaine, there is no equivalent. No medication has been approved by the FDA for either one. That is the hole the 70 percent fell through.

Nearly everything the state built in the last decade was built for opioids.

The field spent twenty years calling the one thing that works a bribe

Contingency management, the clinical name for what Groves did, is simple to describe: a person earns a small, immediate reward each time they test negative for stimulants, and the reward grows as the streak does. Compared with traditional counseling, people are about twice as likely to stop using meth or cocaine when they get the gift cards, according to the Tradeoffs reporting on the research, and studies find they are more likely to stay off stimulants for up to a year than after counseling or 12-step programs.

You would think a treatment like that would be on every intake form in the country. It is not, and the reason is a fight about whether paying someone to stay off drugs is treatment at all. We have watched a field decide a cheap thing that works is morally suspect before. Contingency management dates to the 1980s. The VA has offered it since 2011. In 2005, a California Medicaid pilot was stopped after federal officials raised concerns about billing rules, The state’s current Medicaid program launched in 2023 under a federal waiver, with gift cards starting at $10 and rising to $26.50 per negative test, capped near $599 over six months. Rick Rawson, the UCLA psychology professor emeritus who helped build the evidence, remembers the objection exactly: “You would hear things like, this isn’t treatment, this is just paying people not to use drugs.” The cap tells you who won that argument. Most studies find the approach works best when patients can earn more than $1,000, and California kept its numbers low, in part to avoid tax problems and jeopardizing food assistance and other benefits.

The economics have caught up with the evidence. A 2026 model in the journal Addiction, by G. L. Qian, Rawson, Keith Humphreys and colleagues, found a 24-week program cost roughly $12,000 per year of healthy life gained, far under the $50,000 benchmark health economists use, and estimated it would prevent about 153 deaths per 10,000 people in the first year, mostly by preventing overdoses. It is a simulation, not a trial, and its authors say it can’t measure how long the benefit lasts or capture real-world dropout and housing chaos. Even so, the primary paper concludes that the federal incentive cap, raised to $750 a year in early 2025, is consistent with good value.

Here is where Arizona comes in. Medicaid coverage of contingency management has so far come through special federal waivers. KFF’s tracker, last updated in January 2025, lists approved waivers in California, Delaware, Hawaii, Montana and Washington, with Michigan and Rhode Island pending. Arizona does not appear. We searched for any AHCCCS coverage of the service and found none, though we did not review the agency’s internal service guides, and a state that had quietly added it would be worth hearing from. The gap between what the evidence supports and what a person on AHCCCS can get is the story.

Denver just showed what a gift card can’t do on its own

If contingency management were a cure-all, the obvious move would be to bolt it onto every program and walk away. A randomized trial published September 2 in JAMA Network Open shows why that would be a mistake.

The BEAT Meth trial, led by Alia Al-Tayyib of Denver Health, enrolled 192 adults who had come through Denver’s acute-care system after a methamphetamine-related emergency. Half got usual care, which meant discharge planning with a withdrawal counselor. Half got three months of a care navigator focused on social needs, with gift cards and a prize-wheel spin as rewards for showing up. These were not easy patients. About 85 percent were unstably housed, 82 percent were unemployed, and 49 percent didn’t have regular access to a phone.

Half got three months of a care navigator focused on social needs, with gift cards and a prize-wheel spin as rewards for showing up.

The rewards worked on the thing they were aimed at. In the navigation group, 62.5 percent had at least two sessions with their navigator, and 49.0 percent completed the 90-day study visit, compared with 38.5 percent of the usual-care group. People came. What the rewards did not do, at least within the study’s window, was get significantly more of them into treatment. At 30 days, 25 percent were linked to treatment in both groups. At 90 days it was 33.3 percent in the navigation group and 25 percent in usual care, a gap whose statistical range runs from a 15 percent loss to a more than doubling, meaning the trial can’t tell us whether the difference is real. Overdose diagnoses, emergency visits and deaths did not differ in any way the authors could detect.

Read that result carefully, because it cuts two ways. It is one hospital system, the researchers and patients could not be blinded to who was getting what, and three months may simply be too short for someone sleeping outside to work their way into a treatment bed. The authors say as much and suggest longer follow-up and multiple points of access. This is not proof that the approach failed. It is proof that a gift card buys attention, not a door. When almost half your patients can’t reliably receive a phone call, the reward for answering the call is not the bottleneck.

What a provider in Arizona can do this week, before Medicaid catches up

None of this requires waiting for a waiver. Some of it requires a Monday meeting.

If you run intake at a treatment program, add stimulant use to what you screen for, and test for it, because the person who comes in for opioids may be a person who is also using meth, and nothing in your current protocol may say so. If you are a case manager, treat the Denver trial as a design spec: build in a second way to reach the person, because a phone number is a guess for half your caseload, and schedule the next contact before they leave the room, since the people in that trial who kept coming were the ones given a reason to. If you hold or influence settlement dollars, which in Arizona means counties and cities deciding where the state’s roughly $1.2 billion in opioid settlement money goes, ask the plain question of every proposal: what are we buying for the person who uses stimulants? Then ask whether a small, fast reward with a streak attached is in the budget, because the research finds the amounts have to be large enough to matter. And if you can’t give a gift card, give the other half of what Groves described. Be glad to see the person. Say it out loud.

If you are the person reading this, and you use, here is what is still yours tonight. Naloxone reverses the fentanyl part of a meth-and-fentanyl overdose even though it does nothing for the meth, so carry it, and carry it even if you don’t think you use opioids. The SAMHSA helpline, 1-800-662-4357, is free and open around the clock for treatment referrals, and Rize’s find-help tool can filter Arizona programs by insurance. Our earlier reporting on where contingency management stands has the longer version of the evidence.

Groves told the reporter the point of staying off meth was “to be that Bernard I used to be for the people that I love.” Nothing in that sentence requires a medication. It requires somebody to be glad you showed up, week after week, long enough that you start to believe you might too. Arizona has 2,988 deaths’ worth of reasons to build that. It has not yet built it.

It requires somebody to be glad you showed up, week after week, long enough that you start to believe you might too.