Legal Cannabis' 3.7% Yearly Heart Failure Gap Is About 8 Per 100,000
The headline figure is a yearly 3.7% gap in heart failure admissions. The paper's own tables turn it into single digits per 100,000 residents, across a whole region rather than among cannabis users.
In 2012 and 2013, hospitals in the Western United States recorded about 206 acute heart failure admissions per 100,000 residents each year. A study in BMJ Open found that each year of legal recreational cannabis sales was then associated with that rate rising 3.7% faster than in the rest of the country. Worked through the paper's own tables, the model implies about 7.6 extra admissions per 100,000 residents in the first year, call it 8. That is a gap measured across a whole region, and the authors say an observational study cannot prove cause.
The paper, published in BMJ Open on Thursday, Oct. 22, 2026 (DOI 10.1136/bmjopen-2025-111050), comes from Dr. Samuel D. Slavin of Boston Medical Center and Boston University and colleagues at Harvard-affiliated institutions. Using the National Inpatient Sample, a U.S. hospital-discharge database covering 2012 to 2018, they compared the West, where Colorado began legal sales in Jan. 2014, with regions that had none until Massachusetts in late Nov. 2018. Per year of legal sales, admissions for atrial fibrillation or flutter (an irregular heart rhythm) rose 2.3% faster in the West, acute heart failure 3.7% and heart attack 3.1%. Stroke, blood clots and a dangerous rhythm called ventricular tachycardia did not reach statistical significance, so they are not findings.
The 3.7% is a yearly gap, and it compounds
A percentage on a rate sounds large until you put it on the rate. Acute heart failure admissions in the West ran at 206.46 per 100,000 residents a year before legal sales, roughly 1 admission per 480 residents. A 3.7% change on that base is about 7.6 more admissions per 100,000, or roughly 1 in 13,000 residents, and under 4% of the baseline. The table sets out all four outcomes the same way.
| Diagnosis | West baseline, 2012-13 (per 100,000) | Yearly change (95% CI) | Extra per 100,000 (95% CI) | If it compounds 4 to 5 years (model-implied) |
|---|---|---|---|---|
| Acute heart failure | 206.46 | +3.7% (2.9% to 4.6%) | 7.6 (6.0 to 9.5) | 32 to 41 |
| Heart attack | 147.20 | +3.1% (1.4% to 4.8%) | 4.6 (2.1 to 7.1) | 19 to 24 |
| Atrial fibrillation/flutter | 98.86 | +2.3% (1.0% to 3.5%) | 2.3 (1.0 to 3.5) | 9 to 12 |
| Ischemic stroke (not significant) | 114.19 | +1.1% (-0.1% to 2.4%) | 1.3 (-0.1 to 2.7) | 5 to 6 |
Source: Slavin et al., BMJ Open 2026, Tables 2 and 3 (fully adjusted model). Extra-per-100,000 columns are Daybreak Wire calculations: baseline x (yearly ratio - 1), compounded for the last column. Association, not proof of cause.
The last column is where "nearly 4% a year" gets bigger. The paper's model reports an annual relative change, so the gap multiplies year over year. If it held for four to five years of sales, the model implies roughly 32 to 41 extra heart failure admissions per 100,000 by the end of the study, a gap of 16% to 20% against the path without legalization. That is Daybreak Wire's arithmetic, not a reported result, and the paper does not say whether 2018 counts as year four or five, since Western states joined in stages. Even the top figure is 0.04% of residents.
The first-year numbers are modest, the end-of-study numbers are larger, and neither says how risky anything is for one person. Lead author Slavin, a general cardiologist, told Gizmodo in a story published Oct. 22:
"This is a modest relative increase. For any one person, the associated risk is likely small. But across millions of people, even small shifts can mean a large number of additional hospital visits."
Samuel Slavin, lead author and general cardiologist, Boston Medical Center
A regional hospital count cannot see who used cannabis
The design is a difference-in-differences: it measures how the West changed over time and subtracts how everywhere else changed, so national trends cancel out. The unit is the census region, because, the authors write, the survey design "only allows for regional comparisons". The data hold no information on who used cannabis, in what form, or whether any admitted patient did. The extra admissions are spread across every resident of 13 states, not assigned to users, and the outcome is hospital admissions, not deaths.
The West also started lower. Its heart failure rate was 206.46 against 333.29 elsewhere, and its patients were younger, with a median age of 51 against 55. Adjustment for age, sex, race and ethnicity and ZIP-code income shaved the estimates a little; the unadjusted heart failure figure was 4.6%.
Two checks give the design some credibility. Diagnoses that should not respond to cannabis policy did not move significantly: urinary stones (1.007) and opioid-related admissions (0.966). Cannabis-related diagnoses, which should respond, rose by about 14%. The authors also report that pre-2014 trends in the West and elsewhere were not significantly different.
Still, they list what could undo it. "It is impossible to exclude an unmeasured factor disproportionately affecting the West," they write, though CDC data show stable traditional cardiovascular risk-factor trends across regions. The U.S. switched from ICD-9 to ICD-10 diagnosis codes on Oct. 1, 2015; the authors inspected quarterly data but acknowledge a bias if Western coding practices responded differently. And this is effectively one treated region, with Colorado's start in Jan. 2014 and Washington's in July 2014 followed by one more Western state each year. By Jan. 2018, 79.3% of Western residents lived in a state with legal sales, so some states counted as exposed had none. Of that, the authors write: "Misspecification would be expected to bias our results towards the null." That points toward understating any effect, and they add that the data stop before markets matured.
Their summary: While an observational study cannot prove causality, our findings point to potential adverse effects of recreation legalisation on population health.
Cardiology groups already advise caution for heart patients
None of this is the first signal. The American Heart Association's 2020 scientific statement, published Aug. 5, 2020, had already noted that in states where cannabis was legalized, hospitalizations for heart attacks had increased. It also said most data are weak. Robert L. Page II, Pharm.D., of the University of Colorado, who chaired the writing group, put it this way:
"most of the available data are short-term, observational and retrospective studies, which identify trends but do not prove cause and effect"
Robert L. Page II, Pharm.D., University of Colorado, chair of the AHA statement writing group
Individual-level studies point both ways. The CARDIA study, which followed 5,115 people for 30 years, found no link between cannabis use and major cardiovascular events. A 2024 Journal of the American Heart Association survey analysis of 430,000 adults, led by Abra Jeffers of Massachusetts General Hospital, found daily users had 25% higher odds of heart attack than non-users. Both are observational, and survey answers are imperfect. A video by Dr. Vin Gupta on this individual-level research sits below.
No outside cardiologist or epidemiologist had commented on the paper in the coverage the desk could read as of Sunday morning, Oct. 25. The only voices on record are the authors'. That is normal three days after publication, but it means the read on the study's weak points above comes from the paper itself.
For a reader with heart disease, the AHA's 2020 advice is more useful than a headline:
"For people diagnosed with heart disease, cannabis should be used with extreme caution because cannabis increases the heart's need for oxygen at the same time as it decreases available oxygen supply, which could cause angina (chest pain)."
American Heart Association news release on its scientific statement, Aug. 5, 2020
The CDC says cannabis can speed the heartbeat and raise blood pressure right after use, that most studies tying it to heart attacks and strokes rest on reports from people who smoked it, and that cannabis smoke contains many of the same substances as tobacco smoke. The AHA's Rose Marie Robertson, M.D., said the group "recommends that people not smoke or vape any substance, including cannabis products, because of the potential harm to the heart, lungs and blood vessels". Page added that oral and topical forms, for which doses can be measured, "may reduce some of the potential harms", and that people should use only legal products, since street products have no quality controls. This study cannot say whether smoked or edible cannabis mattered, and it does not show that a given joint or edible raised anyone's risk.
In practice, someone with heart disease, a rhythm problem or heart failure who uses cannabis in any form can tell their cardiologist or GP, and bring the product name. It matters for medicines: the AHA lists blood thinners, antidepressants, antipsychotics, rhythm medicines and statins as possible interactions. Nobody should stop prescribed heart medicines, or start or stop anything, on the strength of a headline. Chest pain, fainting or a racing, irregular heartbeat is a call to emergency services, as with any heart symptom. That is general medical advice, not a quotation from the AHA.
Slavin made the same distinction between audiences. "I think these findings matter more for policymakers than for someone deciding whether to visit a local dispensary," he told Gizmodo, adding that People who use cannabis—especially those with existing heart conditions or risk factors—may want to talk with their doctor.
The authors argue their results support a cautious approach to legalization with data collection on cardiovascular outcomes, and the proposed federal move of cannabis from Schedule I to Schedule III to allow research. A rise of single digits per 100,000 is a reason to ask a question, not to panic. Daybreak Wire applied the same arithmetic to a melanoma vaccine trial, and absolute risks matter just as much in the menopause hormone therapy guideline.
The authors write that sales data might one day let researchers separate smoked from edible products. For now, the hospital counts stop at 2018, and the smallest place they can name is a region of 13 states.
