People with a history of alcohol, opioid, or substance overdose who were later started on GLP-1 receptor agonist medications had a significantly lower risk for a repeat overdose, even if they discontinued it, according to an Epic Research study.

In a records review from 683,800 US adults with an emergency department or inpatient overdose encounter occurring between January 1, 2017, and March 9, 2026, active GLP-1 medication use was associated with a 39% lower risk for a repeat overdose between 30 days and 3 years after the overdose event compared with no GLP-1 use, the study found. Patients who started a GLP-1 and later discontinued the medication still had a 20% lower risk for a subsequent overdose than people who never used a GLP-1, “suggesting that the association with lower overdose risk might extend beyond the period of active treatment,” the authors wrote.

The pattern “was consistent across initial overdose type [alcohol, nonalcohol, or no addiction treatment],” they said:

• Patients whose first overdose involved alcohol had a 40% lower risk for a repeat overdose during active GLP-1 use and an 18% lower risk for a repeat if they discontinued use.
• Patients whose first overdose did not involve alcohol had a 37% lower risk for a subsequent overdose during active GLP-1 use and a 24% lower risk for a repeat if they discontinued use.
• Patients receiving no addiction-treatment medications had a 44% lower risk for a repeat overdose during active GLP-1 use and a 28% lower risk for a repeat if they discontinued use.

The study was completed by two clinician-research scientist teams working independently and coming to similar conclusions. Investigators reviewed records from Cosmos, a dataset created with over 350 organizations using Epic that incorporates records from nearly 2200 hospitals and almost 50,000 clinics. They excluded patients who were prescribed a GLP-1 before their overdose and accounted for demographics, social vulnerability, and rural status based on the patient’s most recent address, BMI category, diabetes or mental health diagnoses, and medications for opioid, alcohol, or nicotine use disorder.

Patients included in the study had a BMI of 25 or higher around the index date; 56% were male; 46%-48% had anxiety or depression. Most patients (86%) lived in metropolitan areas.

Investigators included patients who had an overdose of opioids, alcohol, stimulants, sedative/hypnotics, cannabinoids, cocaine, or hallucinogens and then compared those who had a GLP-1 prescribed after the overdose event to those who did not have a GLP-1 prescribed, Caleb Cox, head of research at Epic, said. Then they used the difference in repeat overdose frequency to calculate the risk ratio for subsequent overdose encounters. They repeated their analysis for three subgroups: patients whose first overdose involved alcohol; patients whose first overdose did not involve alcohol; and patients not receiving any addiction-treatment medications.

Due to the parameters of the study, investigators couldn’t determine the underlying reason for the association, explained Cox. “Some of the other related research suggests that it could be tied to the way it affects cravings and the reward system in the brain,” Cox told Medscape Medical News. “There is some interesting research, particularly in the substance abuse space, around what the mechanism of action might be.”

Still, the effect of a GLP-1 receptor agonist on reduced risk for a subsequent overdose was “bigger than I expected it to be,” he said. “It is just another piece of evidence that suggests it’s worth actually doing clinical trials and other biological research” to understand the underlying biology and for future development of drugs that could have a broad impact on addictive substances.

“I think it’s a decent study,” said Anne Peters, MD, professor of medicine at the University of Southern California (USC)’s Keck School of Medicine, Los Angeles, and director of the USC Clinical Diabetes Programs, who has experience prescribing GLP-1 medications for her patients. “There’s a large sample size, but it’s observational,” and there may be a selection bias as to which patients received a GLP-1, she noted. 

A recent Veterans Affairs study had similar findings, Peters said, so there is likely a strong association. “You can’t prove causality until you’ve done a randomized clinical trial…but it wouldn’t inhibit me from using one of these drugs if somebody had a substance abuse problem.”

There is preliminary evidence that the medications help people drink or use illicit substances less, she said. The drugs are good at calming down urges to drink or eat, at least in about a third of patients, she said. “It does lead to people, certainly in my experience, drinking less. You don’t know exactly why or how. It’s more than just satiety — it’s actually not wanting that substance. A lot of people will not want alcohol; they just lose their taste for it.”

However, some of her patients have expressed feeling worse, more depressed, or more anxious while on the medications, she cautioned. “As much as there’s benefits to these drugs, there’s also going to be downsides. As a physician, listening to how these drugs make [patients] feel is really important, and if something makes them feel not right, you can always taper down.”

It’s also wise to taper the drugs down instead of stopping suddenly, she added, as going “cold turkey” can make patients feel worse or allow cravings to resume.

Cox is employed by Epic Research. Peters reported having a relationship with Vertex.

Karen Blum is a freelance medical/science writer in the Baltimore area.