Sleep apnea is one of the few conditions where a cannabinoid went up against placebo with a sleep lab counting every pause in breathing, and sleep doctors read the results and said no. Researchers in Chicago gave prescription dronabinol, a synthetic THC capsule, to people with obstructive sleep apnea twice, in 2013 and again in the PACE trial published in Sleep in 2018. Both times the apnea count fell. Both times blood oxygen stayed where it was. In April 2018, a few months after PACE appeared, the American Academy of Sleep Medicine published a position statement saying medical cannabis and its synthetic extracts should not be used to treat the condition at all.
Our position sits with the sleep academy. An edible is a poor sleep apnea treatment, and nobody has measured whether it is even a neutral one, because no trial has put a gummy in front of a sleep lab. The drug that was studied is a capsule of pure THC in sesame oil, taken an hour before bed under a sleep physician's watch. If you snore, stop breathing at night, or wake up exhausted, the first step is a sleep study, and the second is whatever that study says to do.
Does THC help sleep apnea?
Prescription dronabinol lowered the apnea-hypopnea index in two small, short trials, and neither one showed better oxygen levels overnight. A 2024 systematic review in Pharmacotherapy found nine studies, seven of them on dronabinol, with a median treatment length of three weeks.
The idea came out of rat labs, where Carley's group at the University of Illinois showed dronabinol suppressing apneas in sleeping rats through the CB1 receptor. The 2013 pilot, led by Bharati Prasad, took that to 17 adults with an apnea-hypopnea index of at least 15 events an hour. Each came off CPAP for a week, then took dronabinol once a day, stepping from 2.5mg to 5mg to 10mg over three weeks. By night 21 the index had dropped by 14.1 events an hour on average, among the 15 who finished. Nadir oxygen saturation did not improve. Somnolence was the most common side effect, reported by half the participants at the 10mg step.
One more small trial belongs here, mostly for its ending. In 2023, an open-label study of 15 patients tested dronabinol paired with atomoxetine. The 10mg combination did not significantly reduce apnea severity, the 5mg one did, and a third of participants stopped treatment. The trial was funded by Apnimed, and Apnimed's drug now sitting at the FDA pairs atomoxetine with something other than THC.
What did the PACE trial actually find?
Against placebo, six weeks of dronabinol cut the apnea-hypopnea index by 10.7 events an hour at 2.5mg and 12.9 at 10mg. Part of that gap was the placebo group getting worse. Oxygen levels, sleep architecture and an objective wakefulness test did not change in any group, and 6 of 39 people on dronabinol met the trial's bar for a meaningful response.
PACE (Pharmacotherapy of Apnea by Cannabimimetic Enhancement) randomized 73 adults with moderate or severe apnea, an average age of 53.6 and an average BMI of 33.4, to placebo, 2.5mg or 10mg of dronabinol. Each participant took one capsule by mouth 60 minutes before bedtime. Baseline apnea index averaged 25.9, and on placebo it rose over the six weeks, something the authors could not explain. Self-rated daytime sleepiness improved by 2.3 points against placebo at 10mg, right on the edge of significance (p = .05).
The responder count is the number worth keeping. The investigators defined a meaningful result as an index of 15 or under plus a cut of at least half, and six people on the drug got there. Getting into the trial was its own filter: of 417 people enrolled, 78 met the criteria, and 60 were turned away for reported recreational drug use or a positive urine screen. The study that tested THC for sleep apnea excluded the people already using it.
What does the American Academy of Sleep Medicine say about cannabis for sleep apnea?
That it should not be used. The AASM's April 2018 position statement, written by its board of directors with Kannan Ramar of the Mayo Clinic as first author, says medical cannabis and its synthetic extracts should not be used for obstructive sleep apnea because of "unreliable delivery methods and insufficient evidence" on effectiveness, tolerability and safety.
The statement goes further than most society positions. It tells state legislators and health departments to keep sleep apnea off their medical cannabis lists, and it points patients to a licensed provider at an accredited sleep facility. It also notes that only dronabinol was tested, and that other cannabis products and delivery methods never were. Daytime sleepiness, the board added, carries a motor vehicle risk of its own.
Minnesota added obstructive sleep apnea to its qualifying list anyway, in August 2018, and it is still on the Office of Cannabis Management's list today. The state's own five-year report counted 3,102 people who enrolled with the condition through July 2023. It also shows what a qualifying list measures. Before each purchase, patients rate eight symptoms from 0 to 10, disturbed sleep among them, and 39.4% of those with moderate to severe disturbed sleep improved by 30% and held it for four months. No breathing measure is collected. Where a certifying practitioner reported it, the median gap between a patient's last sleep study and certification was three years. And 64.2% of the products bought were inhaled, which is a curious way to treat an airway.
One patient comment in that report makes the best case anyone has for cannabis here: an "easier time tolerating CPAP." That is cannabis helping someone stay on the treatment that works.
Why is a dispensary edible different from dronabinol?
Dronabinol is a single synthetic molecule in a sealed sesame oil capsule, made in fixed 2.5mg, 5mg and 10mg strengths and prescribed for chemotherapy nausea and AIDS-related weight loss. A gummy carries plant-derived THC at whatever dose the state allows, often with CBD or CBN mixed in, and none of those products has been tested in a sleep lab.
The timing problem stacks on the formulation problem. PACE participants swallowed their capsule an hour before bed. Edible onset, per our guide to how long edibles last, runs 30 minutes to 2 hours, typically 30 to 90 minutes, with the peak 2 to 3 hours after eating and main effects lasting 4 to 8 hours. A sedating dose that peaks in the small hours lands right where apneas cluster, and nobody has measured what it does there. The Marinol label also tells patients not to drive until they know the drug does not impair them, which matters for a condition whose main daytime symptom is already sleepiness.
Can cannabis make sleep apnea worse?
No trial has shown that THC worsens airway collapse, and we found no human study measuring its effect on airway collapsibility or breathing drive directly. The observational studies disagree with each other. Alcohol is the sedative with real evidence against it, and it raised the apnea index in a meta-analysis of 14 controlled studies.
The alcohol numbers come from Kolla and colleagues at the Mayo Clinic, writing in Sleep Medicine Reviews in 2018: across 422 participants, drinking before sleep added 2.33 events an hour to the apnea index and deepened the lowest oxygen readings. A separate meta-analysis of 21 studies found higher drinking went with a 25% higher risk of sleep apnea. For THC the record is three snapshots:
- US survey, 2026. Among 14,485 adults in a federal telephone survey, using cannabis 16 to 30 days a month went with 1.73 times the odds of witnessed breathing pauses during sleep. The UCLA and USC authors say causality is unknown.
- Toronto sleep lab, 2024. Of 6,958 adults tested at Sunnybrook, 71 reported cannabis use. A link to lower apnea severity disappeared once medications and other illnesses were accounted for.
- Canadian sleep clinic, 2026. In 1,449 patients, most of them with moderate to severe apnea, the 151 who used cannabis daily for a year or more spent 21% more time awake after falling asleep.
That last finding is worth holding next to the reason most people reach for an edible at night, which is to stay asleep. The trial evidence for that use, apnea aside, is on our sleep maintenance page.
What is approved for sleep apnea in 2026?
Positive airway pressure remains the standard. The AASM's 2019 guideline recommends CPAP or auto-adjusting PAP for ongoing treatment. The first drug the FDA approved for it is Zepbound (tirzepatide), approved December 20, 2024 for moderate to severe obstructive sleep apnea in adults with obesity, alongside diet and exercise.
The Zepbound label sets the sleep apnea maintenance dose at 10mg or 15mg injected once a week. In the SURMOUNT-OSA trials, starting from an average index near 50, it cut about 25 to 29 events an hour over a year against roughly 5 on placebo, an effect the FDA ties to weight loss. A 2020 JAMA review also lists oral appliances that hold the jaw forward, airway surgery, and hypoglossal nerve stimulation for selected patients with a BMI under 32. Next in line is Apnimed's AD109, a bedtime pill combining aroxybutynin and atomoxetine, with an FDA action date of February 28, 2027. Set beside a year-long injection trial and two Phase 3 pill trials, dronabinol's record of three weeks in 17 people and six weeks in 73 looks like what it is, an early idea the field moved past.
What if I already use cannabis and think I have sleep apnea?
Get a sleep study first, keep any CPAP or prescribed therapy you already have, and tell the sleep physician about the cannabis. The AASM's diagnostic guideline says questionnaires alone should not be used to diagnose apnea; a lab study or a home sleep apnea test does that job.
The stakes are why. A JAMA review puts the added risk of cardiovascular and metabolic disease at two to three times, and the American Heart Association reports sleep apnea in 40% to 80% of patients with hypertension, heart failure, atrial fibrillation and stroke. A 2009 meta-analysis estimated that apnea raises the motor vehicle crash rate by a factor of 1.21 to 4.89 and called untreated apnea a significant contributor to crashes. An edible that makes the snoring feel less urgent does not change any of that. If the real complaint is falling asleep or staying asleep, the evidence on THC and CBN for those is on our sleep onset page and our chronic insomnia page, both written for people whose breathing has already been checked.