Cannabis Use Was Linked to Higher Stroke Odds in a Large U.S. Adult Cohort
Adults who reported cannabis use had higher odds of a past cardiovascular disease diagnosis, with the clearest signal appearing for stroke. The finding comes from an analysis of 442,906 U.S. adults that will be presented at the American Heart Association Scientific Sessions 2026 in Chicago.
The pattern was not a simple dose-response curve. People who used cannabis one to five days a month had 40% higher odds of reported stroke and 23% higher odds of reported cardiovascular disease compared with nonusers. Among those reporting use 16 to 30 days a month, the corresponding odds were 25% and 21%. Heart attack did not show a statistically significant association across use frequency, which makes the stroke signal stand out even more sharply.
A Signal Centered on Stroke Rather Than Myocardial Infarction
Researchers drew on Behavioral Risk Factor Surveillance System telephone survey data collected from 2020 through 2024. The dataset is large enough to detect patterns that smaller single-center studies would miss, and it reflects a broad cross-section of U.S. adults rather than a narrowly selected specialty population. Even so, the analysis remains observational and cross-sectional, which means it can identify associations but cannot establish causation.
The uneven risk pattern matters clinically. If cannabis were producing a uniform vascular effect, one might expect more parallel movement across stroke and myocardial infarction. Instead, the strongest association emerged around stroke, a result that raises questions about how cannabis may influence cerebral blood flow, thrombosis, arrhythmia burden, endothelial function, or other stroke-relevant pathways more than coronary outcomes.
One sentence captures the key point: the heaviest users did not carry the biggest stroke odds.
That finding complicates the assumption that risk should rise steadily with increasing exposure. It suggests either biologic heterogeneity in the way cannabis affects the cerebrovascular system, or residual confounding from unmeasured factors that cluster differently across frequency groups. Tobacco co-use, alcohol use, hypertension, diabetes, socioeconomic status, and access to medical care all remain plausible contributors in a self-reported survey design.
Why the Cerebrovascular Signal May Be More Visible Than the Cardiac One
Cannabis has long sat at the center of cardiovascular debate because it may affect multiple systems at once. Acute changes in heart rate, blood pressure, sympathetic tone, and vascular reactivity could all influence cerebral perfusion and embolic or ischemic risk. Stroke also tends to be more sensitive than coronary disease to short-term hemodynamic swings and vascular instability, which may help explain why the stroke association appeared more consistent than the heart attack association in this analysis.
The study author, Aabha Divya, M.B.B.S., M.Ch., an assistant professor in the division of cardiothoracic surgery at Tulane University School of Medicine, said the relationship between cannabis use and cardiovascular disease was not straightforward. That framing fits the data well. This was not a monotonic pattern in which more use simply meant more risk at every endpoint; instead, the signal varied by outcome and by use frequency.
For practicing clinicians, the most relevant bedside message is not that cannabis proves causality, but that it deserves routine cardiovascular history-taking. Patients often volunteer tobacco use and omit cannabis unless asked directly. If a patient has a prior stroke, transient ischemic attack, atrial fibrillation, uncontrolled hypertension, or multiple vascular risk factors, a cannabis history belongs in the same conversation as cigarette exposure, stimulant use, and medication adherence.
Self-Reported Exposure and Diagnosis Leave Important Gaps
The abstract depends on self-reported cannabis use and self-reported cardiovascular diagnoses. Both introduce room for misclassification. People may underreport use, especially if they use cannabis irregularly or live in places where legal status has changed quickly. Others may misunderstand or misremember a diagnosis, particularly in a telephone survey format.
Cross-sectional data also flatten time. A person who reported cannabis use in 2024 and a stroke years earlier may or may not have had meaningful exposure before the event. Without temporal sequencing, the analysis cannot separate cause from consequence or assess whether cannabis preceded the vascular event.
A single point in time can still be clinically provocative, though. Large survey data are often where risk signals first surface before prospective studies or mechanistic work can test them more rigorously. In this case, the result adds to the reason many cardiovascular specialists already discuss cannabis with the same seriousness they reserve for tobacco and other inhaled substances.
Why Clinicians Are Being Told to Ask About Cannabis Directly
Robert L. Page II, Pharm.D., M.S.P.H., who chaired the American Heart Association’s 2020 scientific statement on cannabis and was not involved in the study, said clinicians should be addressing cannabis use with patients as they would tobacco use. That advice is pragmatic rather than alarmist. It reflects the reality that cannabis exposure may intersect with vascular risk, symptom reporting, and treatment adherence in ways that are easy to miss unless the clinician asks.
The finding does not justify overstatement. It does not prove that cannabis causes stroke, and it does not show that every user faces the same level of danger. It does, however, add weight to the argument that cannabis should no longer be treated as a neutral background exposure in cardiovascular care.
For now, the most defensible approach is straightforward: screen for cannabis use, document frequency, ask about route of administration, and place the result in the context of stroke risk factors. The abstract leaves open the larger question of which biologic pathway drives the association, and whether route, potency, or co-exposures modify risk in ways this survey design could not capture.