Key Points

  • CDC provisional data show fentanyl overdose deaths fell 22% between 2024 and 2025, from 48,913 to 38,084, contributing to the lowest overall U.S. death rate ever recorded.1
  • Total drug overdose deaths dropped 14% in 2025, reaching 69,973 though access gaps in rural areas and for uninsured people have not closed alongside falling death counts.
  • A small study using real-time smartphone check-ins found that craving level, not mood, is what directly predicts whether someone uses a substance in the hours that follow, challenging treatment models that focus on emotional regulation alone.3
  • Dr. Sylvie Stacy explains why positive emotions can trigger cravings just as powerfully as negative ones, and why cravings require in-the-moment tools rather than weekly therapy appointments to address effectively.
  • An AI system deployed at the University of Washington cut urine drug test sign-off time by 23%, matching expert-level accuracy across 26 substances and enabling faster treatment decisions for people in addiction programs.4
  • A declining national overdose death toll is meaningful progress, but the tools already driving improvement, medication-assisted treatment, naloxone access, and expanded treatment capacity, remain just as essential for anyone seeking care today.

Addiction News Weekly Episode 1.9

In this episode:

Episode Transcript

Welcome to Addiction News Weekly by Rehab.com, where we break down the biggest stories in addiction, recovery, and public health. This week the news is about where and how quickly addiction treatment is heading. We have new federal data on fentanyl deaths, research that changes how clinicians think about what causes relapse, and an AI tool that is already making drug monitoring faster in real clinical settings.

Fentanyl Overdose Deaths Fall to a New Low

Let’s get into it. We start with the numbers. New provisional data from the Centers for Disease Control and Prevention show that fentanyl overdose deaths in the United States fell 22% between 2024 and 2025.1

Synthetic opioid deaths dropped from 48,913 to 38,084. Total drug overdose deaths fell 14%, from 81,313 to 69,973 over that same period. CDC officials have said the decline in overdose deaths was a major factor behind the lowest overall U.S. death rate ever recorded in 2025.1

The White House highlighted these figures this week as it announced a new federal summit on fentanyl enforcement and attributed the decline to a range of border, trade, and law enforcement actions taken over the past two years, including the HALT Fentanyl Act permanently classifying fentanyl analogs as Schedule I drugs and tariff pressure on precursor chemical suppliers. Those are the administration’s own causal claims rather than an independent analysis, and it is worth keeping that distinction in mind. One data point from DEA testing is harder to dispute.2

Just 29% of seized fentanyl pills now contain a potentially lethal dose, down from 76% two years earlier, which suggests real pressure on supply chain purity, regardless of which policy is most responsible.2

For anyone currently looking for addiction treatment, the practical question is whether a falling national death toll makes care easier to find. The honest answer is not necessarily.

Overdose mortality and treatment capacity move on different timelines, and access gaps, especially in rural areas and for people without insurance, have not disappeared alongside declining death counts. What the numbers do confirm is that the tools that were already working, like medication-assisted treatment, naloxone access, and expanded treatment capacity, are part of why the trend is moving in the right direction, and they remain just as valuable for someone in treatment today.

What Drives Relapse: The Craving-Mood Divide

Our second story is about what actually causes relapse, and why the answer matters for how treatment is designed.

A new study published in Drug and Alcohol Dependence Reports used a method called ecological momentary assessment, which tracks how people feel and behave in real time, rather than relying on memory of a difficult week.3 Researchers had 36 adults in outpatient addiction treatment complete brief smartphone check-ins three times daily for 14 days, rating their mood and craving each time, then reporting any substance use since the last check-in.

It is a small sample, so these findings are best understood as hypothesis-generating rather than definitive. But what they found was striking enough to get Dr. Sylvie Stacy’s attention. Here she is with the clinical takeaway.

The researchers in this study, they were trying to figure out how momentary emotions and cravings drive actual substance use in daily life. And they did this using smartphone check-ins with this group of adults in outpatient treatment. What they found was a separation between mood and craving. Positive emotions like happiness were linked to lower cravings, and negative feelings like sadness or anxiety were linked to higher cravings. But mood on its own didn’t directly predict whether someone used a substance hours later. What did predict use was the craving level itself. And that separation between mood and craving, it’s something I’ve definitely observed in my own patients.

And traditional treatment models often lumped them together. I’ve had many patients who’ve been told that if they could just fix their depression or get their anxiety under control, that urge to use substances would suddenly vanish. But it’s really not that simple.

I treated a patient for alcohol use disorder not long ago, and he was doing really well. He was abstaining from drinking for several months. He felt stable. His mood was improved. And he met with me shortly after he got a promotion at work. And he was totally thrilled with this promotion. But literally the next day, he had a lapse to alcohol use. And it wasn’t sadness or depression or grief that triggered that lapse. It was that intense burst of excitement and happiness that came along with his promotion that really caught him off guard and set off this massive craving for him.

And that anecdote kind of matches this study pretty well. Negative emotions like sadness and anxiety, they certainly can drive cravings up. And positive emotions can sometimes buffer against them. But the emotion itself isn’t what pushes someone over the edge to drinking or using drugs. It often is the craving itself. And a challenge is that cravings can change over time. They rise and fall, sometimes on an hour-by-hour basis. So if you have a therapy appointment, say once a week or every couple of weeks, that time frame is simply too long to help you address cravings that fluctuate so rapidly. So we have to equip people with the tools that they can use in the moment to address cravings.

Modalities like certain behavioral therapies, they teach patients concrete tolerance skills that kind of break the chain between an emotion and a craving.

AI Speeds Up Drug Test Interpretation in Clinical Settings

Our third story is about what happens behind the scenes in addiction treatment and how artificial intelligence is starting to speed it up.

Researchers at the University of Washington School of Medicine, publishing in JAMA Network Open in June 2026, developed an AI system to generate preliminary interpretations of urine drug tests, the very lab work clinicians use to monitor patients in addiction treatment, chronic pain management, and emergency care.4

The study analyzed more than 83,000 urine tests from over 26,000 patients, collected between 2014 and 2024. The AI matched expert interpretation with more than 99% accuracy across nearly all monitored substances, including opioids, benzodiazepines, and stimulants. After the tool was deployed into the University of Washington’s live clinical workflow in August 2025, it cut the average sign-off time by 28.5 seconds per case, a 23% efficiency gain. When paired with a second workflow improvement, the combined time savings reached 51%.4

That may sound like small numbers, but in an outpatient addiction program running dozens of tests a week, faster results mean faster care decisions. For someone on buprenorphine or methadone, quicker lab turnaround can mean a provider adjusts a dose or catches a warning sign days sooner than they otherwise would.

The system was designed to support, not replace, the clinicians who still sign off on every result. The research team noted that it is built differently from generative AI tools, using a structured prediction model rather than one that composes new text, which is intended to reduce the risk of fabricated output in a clinical setting.

Conclusion

Put these stories together, and the direction is clear. The national death toll from fentanyl is gradually falling. New research is refining what treatment needs to do, targeting craving directly rather than just the emotions that drive it. And technology is starting to close the gap between a test result and the care decision that follows.

Progress in addiction treatment rarely moves in a straight line, but this week is different. If you or someone you love is looking for treatment, Rehab.com lists thousands of verified centers across the country, and free, confidential support is available at any time through the SAMHSA National Helpline at 1-800-662-4357. And if you are in crisis, you can call or text 988 at any time.

Those are the top stories in the news. For more, visit Rehab.com. We will be back next week. I’m Kay, and thank you for listening.

Sources in This Episode

  1. U.S. Overdose Deaths Decrease for Third Consecutive Year in 2025. Centers for Disease Control and Prevention, National Center for Health Statistics. Published May 13, 2026. https://www.cdc.gov/nchs/pressroom/releases/20260513.html
  2. Fentanyl Deaths Fell 22% Nationwide, CDC Data Shows. Rehab.com. https://www.rehab.com/fentanyl-deaths-fell-22-nationwide-cdc-data-shows
  3. Laha N, Keebaugh M, Liao H, et al. Development and Implementation of an AI System for Generating Clinical Urine Drug Test Sign-Outs. JAMA Netw Open. 2026;9(6):e2619816. https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2850599