The Cannabis Health Scare: How Flawed Studies Become Misleading Headlines

This is not an argument that cannabis is harmless. Frequent high-potency cannabis use is associated with increased psychosis risk. Smoking cannabis can cause respiratory symptoms, and THC has acute effects on heart rate and blood pressure. Those risks deserve serious research. But there is another problem worth examining: how cannabis research gets translated into public health warnings. Across several highly publicized studies, alarming statistics have been promoted without the context necessary to understand what was actually measured. A recurring pattern emerges: exposure is defined by a hospital billing code or a single survey question, relative numbers dominate the headlines, and limitations buried in the papers receive far less attention. The result is a growing collection of cannabis health scares that sometimes say considerably more than the underlying research can establish.

The Machine Behind the Headlines

Cannabis research has a translation problem.

A study identifies an association. A university or research institution publishes a press release. News outlets turn the most dramatic number into a headline. By the time the public encounters the finding, the distinction between association and causation can disappear.

Several of the studies examined here share a particularly important characteristic: they rely on observational data.

Observational studies can identify relationships between cannabis exposure and health outcomes. They can reveal patterns worth investigating and help researchers identify populations that may need additional attention.

But they cannot automatically establish that cannabis caused the outcome.

Other variables can influence the results, including tobacco, alcohol, other drugs, poverty, pre-existing conditions and the characteristics of the people whose records or survey responses are used.

One standard applies throughout this article: a study should show more than a statistical association before its findings are used to support claims about causation.

Case 1 — The Canadian Schizophrenia "Tripling"

The most widely reported cannabis study in this investigation is a Canadian study suggesting a legalization-related surge in schizophrenia. Daniel Myran and colleagues examined 13.6 million residents of Ontario in a JAMA Network Open study published in February 2025, covering 2006 through 2022. The related Institute for Clinical Evaluative Sciences (ICES) press release announced that the share of new schizophrenia cases associated with cannabis use disorder had nearly tripled since cannabis legalization, increasing from approximately 4% to 10%.

The study itself told a different story. An analysis of 13.6 million people from 2006 to 2022 found that age-standardized incidence rates remained essentially unchanged. The figures were 53.5 per 100,000 people before legalization, 52.8 at the time of legalization, and 53.3 afterward. One related category did rise: diagnoses coded as psychosis NOS — a nonspecific code used for uncertain presentations — increased from 30.0 to 55.1 per 100,000 over the same period. With the specific schizophrenia diagnosis flat, that rise is at least as consistent with changes in diagnostic coding as with new illness.

The study authors stated that the attributable fraction increased "with no accelerations after cannabis policy changes." The size of the "tripling" depends on the population attributable fraction. That fraction is an estimate of the proportion of total schizophrenia cases attributable to cannabis use disorder, taking rates of exposure into account. The fraction is modeled and does not establish whether cannabis use caused the schizophrenia.

The study relied on ICD diagnostic codes recorded during those visits and hospitalizations. The measured change was not an increase in schizophrenia itself. The exposure figures rose because the coding of "cannabis use disorder" could be identified in Ontario emergency department records. The underlying population's schizophrenia incidence remained flat, while cases that could be coded as involving cannabis use disorder increased.

That result raises a critical question about the exposure definition. The research did not measure cannabis use itself. It measured whether a person had an emergency department visit or hospitalization that included a cannabis use disorder code. Only 0.9% of Ontarians had such a code during the 17-year period, while about one in four adults uses cannabis at least once a year. The small group of people with a coded visit or hospital admission is not representative of all cannabis users.

Another critical issue is reverse causation. Pre-diagnosis schizophrenia can drive emergency department visits, cannabis use and other behaviours. Those effects may precede a formal diagnosis and can influence the association even before schizophrenia is documented.

Case 2 — "Triple the Death Risk"

In a second study published in JAMA Network Open in February 2025, Myran and colleagues reported that people who received hospital-based care coded for cannabis use disorder had a mortality risk 2.8 times higher than the general population over a five-year period. In absolute terms, 3.5% of the hospital-treated cannabis group died within five years, versus 0.6% of the matched general-population group.

The causes of death help explain the result. The primary causes were suicide, at 9.7 times the general population rate, opioid poisoning at 5 times the general population rate, and trauma. Those findings track the profile of the population that reaches hospital care for substance use — a group with high rates of other substance use and severe mental illness — rather than any established toxic effect of cannabis itself.

The study also compared cannabis use disorder hospital care with hospital care for alcohol, stimulant and opioid use disorders. In those comparisons, the alcohol, stimulant and opioid groups all had higher mortality rates than the cannabis group. Cannabis use disorder generated a headline for its threefold mortality ratio. The substance with the highest mortality was not cannabis, but the comparison received little coverage.

Myran himself reportedly said that the study did not establish that cannabis use caused the deaths.

Case 3 — A "475% Increase" in Cannabis Traffic Injuries

Myran and colleagues published a third widely reported JAMA Network Open study in 2023. Headlines warned of a 475% increase in cannabis involvement in traffic-injury emergency department visits after legalization. The 475% rise spans 2010 to 2021 — a period that begins eight years before legalization took effect.

The absolute numbers were far less dramatic. The study identified 426 emergency department visits that documented cannabis involvement out of a total of 947,604 traffic-injury visits. That is 0.04%.

"Cannabis involvement" depended on whether a clinician documented cannabis in the patient chart. Documentation, testing and clinical awareness increased after legalization, so changes in measurement coincided with the policy change.

The study did not mean that cannabis had a direct role in the traffic injuries. THC metabolites can persist for weeks, so the presence of THC does not indicate that the driver was impaired at the time of the crash or that the driver had cannabis in their own medical records.

Case 4 — The Sixfold Heart Attack

In 2025, a JACC: Advances study reported that cannabis use was associated with a sixfold increase in heart attack risk. The study involved 4.6 million patients, all aged 50 or younger, and "cannabis use" was defined using a diagnosis code in a person's medical chart.

The absolute risks were 0.558% among patients with a cannabis use diagnosis code and 0.09% among those without. That works out to about 5 additional events per 1,000 coded cannabis users over approximately three years.

The study found no evidence that cannabis use itself caused the heart attacks. It did not measure dose, administration route, potency, frequency or whether cannabis was used with other substances.

Kamel's own companion meta-analysis, presented at the same conference, pooled 12 studies and found a 1.5-fold increase, not a sixfold increase. Five of the twelve studies found no significant increase or a decrease. Two other meta-analyses found no statistically significant association between cannabis and heart attacks. The wider pooled literature is not one-directional, however: a 2025 meta-analysis in Heart pooling 24 real-world studies found cannabis use associated with higher risks of acute coronary syndrome (risk ratio 1.29), stroke (1.20) and cardiovascular death (2.10) — though it pools the same kind of observational designs examined here.

The weakest design received most attention because it produced the largest relative number. The pooled evidence did not.

Case 5 — The November Stroke Scare

A poster abstract presented at the American Heart Association's Scientific Sessions in 2026 introduced another set of alarming results. The American Heart Association press release, dated November 2, 2026, was issued four days before the poster was presented at the conference. The release acknowledged that the research was not peer-reviewed, although the headlines did not.

The study analyzed data from the Centers for Disease Control and Prevention's Behavioral Risk Factor Surveillance System telephone survey. It used self-reported cannabis use and self-reported stroke at a single time point.

The study found that people who reported using cannabis only 1–5 days a month had a 40% increase in odds of reporting ever having had a stroke, while those who used it more frequently had a 25% increase. Light use was associated with greater stroke odds than heavy use. The pattern did not show a clear dose-response relationship.

The study found no significant association between cannabis use and heart attack risk at any frequency. That finding did not appear in most headlines.

Case 6 — The 8-Point IQ Loss

Meier and colleagues published a study in the Proceedings of the National Academy of Sciences in 2012, based on a New Zealand cohort study of 1,037 participants followed from birth to age 38, which reported an eight-point decline in IQ among the heaviest users who began using cannabis in adolescence — specifically, the small subgroup who were dependent on cannabis at three or more of the study's assessment points.

The 8-point decline did not apply to all people who used cannabis. It was observed among a small subgroup of the heaviest users whose use began in adolescence.

In 2013, Ole Rogeberg published a critique in PNAS. Using simulation, Rogeberg demonstrated that socioeconomic confounding alone could reproduce the observed pattern without requiring any cannabis effect.

Meier and colleagues replied in PNAS the same year, arguing that adjusting for socioeconomic status and personality left their estimate unaltered. The exchange remains unresolved.

Two longitudinal twin studies by Jackson and colleagues, published in PNAS in 2016, used more rigorous designs. One study had 789 participants, the other 2,277. In both studies, users showed greater decline in raw comparisons than did non-users, but there was no dose–response relationship. A cannabis-using twin declined no more than the abstinent co-twin.

When design is carefully controlled, the apparent decline is not found. That does not prove that cannabis is harmless; it suggests that the commonly cited eight-point IQ figure is not adequately supported.

Case 7 — "Damages the Brain" and Blood Vessels

In January 2025, JAMA Network Open published Gowin et al.'s study of 1,003 young adults using data collected from 2012 to 2015. The study measured seven brain tasks. Of these, only one small effect survived correction — a working-memory effect of d = -0.28, found among only 88 heavy lifetime users. Effects related to recent cannabis use did not survive correction.

Lower activation is not damage. Nevertheless, CNN headlined the study as evidence that cannabis "damages" the brain, and similar reporting framed the finding as evidence of lasting harm.

A UCSF study published in JAMA Cardiology in 2025 examined vascular function in 55 participants. The study made a single visit, used a surrogate ultrasound marker and recorded zero health outcomes. No tobacco-smoking comparison group was included in the study. Nevertheless, media coverage described cannabis as "comparable to tobacco" in harming blood vessels. The paper later required a formal correction — a July 2025 erratum for an error in its Abstract and Methods.

The headlines showed a consistent problem. Studies using surrogate markers, single visits and small samples made it into the news as if they provided evidence about long-term health outcomes.

What the Research Genuinely Suggests

The evidence for risk is stronger in other areas. There is established evidence that frequent use of high-potency cannabis increases psychosis risk. Smoking cannabis can cause respiratory symptoms, and THC can cause acute increases in heart rate and blood pressure.

Those risks require serious research. They should not be overpowered by inflated claims about other harms, which may also affect public health.

The Pattern

The implications of this investigation extend well beyond a list of seven studies. The articles follow a pattern that undermines how disputed research is synthesized and presented to the public.

The pattern includes:

1. Observational designs are presented as if they proved causation.
2. Exposure is defined by a billing code or a single self-report question.
3. No information is provided about dose, potency, route or the distinction between THC and CBD.
4. Potential confounding is not addressed from tobacco, alcohol, other drugs, poverty and pre-existing illness.
5. Relative ratios are headlined without their base rates.
6. Conference abstracts and press releases are treated as settled science.
7. Inconsistency and insufficient evidence disappear between the paper and press coverage.

What Readers Should Ask

Readers should ask this question whenever they encounter a headline about cannabis and health: Did the researchers demonstrate causation, or did they identify an association that can be explained in multiple ways?

A single study cannot establish whether cannabis caused a specific outcome. Studies using billing codes, self-reports and small samples all require critical evaluation before the results can be translated into public health warnings.

Public health messaging should follow the evidence, not the headlines.

Good science should survive scrutiny. Cannabis research should be no exception.

Read next

DEA Approves Its First 900+ Medical Marijuana Dispensaries — 6,083 Businesses Still Waitingarticles · Nov 2, 2026Cohen Bill Would Remove Marijuana From the Controlled Substances Act Entirelyarticles · Nov 2, 2026Edwin Rubis: 28 Years in Prison for Cannabis, A 40-Year Sentence That Could Keep Him Behind Bars Into His Sixtiesarticles · Nov 5, 2026

Sources

1. Myran, Daniel T., et al. "Changes in Incident Schizophrenia Diagnoses Associated With Cannabis Use Disorder After Cannabis Legalization." JAMA Network Open, vol. 8, no. 2, 2025, e2457868. The Ontario cohort of 13.6 million people; schizophrenia incidence stayed flat while the modeled attributable fraction rose. https://doi.org/10.1001/jamanetworkopen.2024.57868
2. ICES. "Contribution of Cannabis Use Disorder to New Cases of Schizophrenia Has Almost Tripled Over the Past 17 Years." News release, February 4, 2025. The press release behind the tripling headline. https://www.ices.on.ca/news-releases/contribution-of-cannabis-use-disorder-to-new-cases-of-schizophrenia-has-almost-tripled-over-the-past-17-years/
3. Myran, Daniel T., et al. "Cannabis Use Disorder Emergency Department Visits and Hospitalizations and 5-Year Mortality." JAMA Network Open, vol. 8, no. 2, 2025, e2457852. The 2.8-fold mortality association, its causes of death, and the higher mortality of the alcohol, stimulant and opioid groups. https://doi.org/10.1001/jamanetworkopen.2024.57852
4. Edwards, India. "Study Links Heavy Cannabis Use to Higher Death Risk, Psychosis." MedicalXpress, February 7, 2025. Coverage of the mortality study. https://medicalxpress.com/news/2025-02-links-heavy-cannabis-higher-death.html
5. Myran, Daniel T., et al. "Cannabis-Involved Traffic Injury Emergency Department Visits After Cannabis Legalization and Commercialization." JAMA Network Open, vol. 6, no. 9, 2023, e2331551. The 475% figure rests on 426 documented visits out of 947,604. https://doi.org/10.1001/jamanetworkopen.2023.31551
6. Kamel, Ibrahim, et al. "Myocardial Infarction and Cardiovascular Risks Associated With Cannabis Use: A Multicenter Retrospective Study." JACC: Advances, vol. 4, no. 5, 2025, 101698. The sixfold relative risk on absolute risks of 0.558% versus 0.09%. https://doi.org/10.1016/j.jacadv.2025.101698
7. American College of Cardiology. "Cannabis Users Face Substantially Higher Risk of Heart Attack." Press release, March 18, 2025. The sixfold headline and the companion meta-analysis reporting a 50% increased risk. https://www.acc.org/about-acc/press-releases/2025/03/17/15/35/cannabis-users-face-substantially-higher-risk
8. Divya, Aabha, et al. Abstract Poster Mo3041, American Heart Association Scientific Sessions 2026. The BRFSS telephone-survey analysis with the reversed dose pattern and no significant heart-attack association.
9. American Heart Association. "Cannabis Use Linked to an Increased Risk of Cardiovascular Disease, Especially Strokes." News release, November 2, 2026. Issued before the conference; the release notes the abstract was not peer-reviewed. https://newsroom.heart.org/news/cannabis-use-linked-to-an-increased-risk-of-cardiovascular-disease-especially-strokes
10. Meier, Madeline H., et al. "Persistent Cannabis Users Show Neuropsychological Decline From Childhood to Midlife." Proceedings of the National Academy of Sciences, vol. 109, no. 40, 2012, E2657–E2664. The Dunedin cohort behind the eight-point IQ claim. https://doi.org/10.1073/pnas.1206820109
11. Rogeberg, Ole. "Correlations Between Cannabis Use and IQ Change in the Dunedin Cohort Are Consistent With Confounding From Socioeconomic Status." Proceedings of the National Academy of Sciences, vol. 110, no. 11, 2013, pp. 4251–4254. The simulation reproducing the pattern with no cannabis effect. https://doi.org/10.1073/pnas.1215678110
12. Jackson, Nicholas J., et al. "Impact of Adolescent Marijuana Use on Intelligence: Results From Two Longitudinal Twin Studies." Proceedings of the National Academy of Sciences, vol. 113, no. 5, 2016, E500–E508. Using twins declined no more than their abstinent co-twins, with no dose–response. https://doi.org/10.1073/pnas.1516648113
13. Gowin, Joshua L., et al. "Brain Function Outcomes of Recent and Lifetime Cannabis Use." JAMA Network Open, vol. 8, no. 1, 2025, e2457069. One small working-memory effect among 88 heavy lifetime users; recent-use effects did not survive correction. https://doi.org/10.1001/jamanetworkopen.2024.57069
14. LaMotte, Sandee. "More Frequent Marijuana Use May Damage an Important Memory Skill." CNN, January 28, 2025. Coverage framing the activation finding as damage. https://www.cnn.com/2025/01/28/health/marijuana-working-memory-study-wellness
15. Mohammadi, Leila, et al. "Association of Endothelial Dysfunction With Chronic Marijuana Smoking and THC-Edible Use." JAMA Cardiology, vol. 10, no. 8, 2025, pp. 851–855. The 55-person UCSF study; formally corrected in July 2025. https://doi.org/10.1001/jamacardio.2025.1399
16. Colliver, Victoria. "Whether It's Smoking or Edibles, Marijuana Is Bad for Your Heart." UCSF News, May 27, 2025. The university release on the vascular study. https://www.ucsf.edu/news/2025/05/430051/whether-its-smoking-or-edibles-marijuana-bad-your-heart
17. Storck, Wilhelm, et al. "Cardiovascular Risk Associated With the Use of Cannabis and Cannabinoids: A Systematic Review and Meta-Analysis." Heart, 2025. Pooled 24 real-world studies; found significant associations with acute coronary syndrome, stroke and cardiovascular death. https://doi.org/10.1136/heartjnl-2024-325429
18. Meier, Madeline H., et al. "Reply to Rogeberg and Daly." Proceedings of the National Academy of Sciences, 2013. The Dunedin authors' response to the socioeconomic-confounding critique of the IQ study.