By Dr. Benjamin Caplan, MD | Board-Certified Family Physician, CMO at CED Clinic | Evidence Watch
A large population-level study following over 11 million Ontario residents found that people with a cannabis-related hospital visit had more than five times the risk of deliberate self-harm and more than nine times the risk of dying by suicide compared to the general population. These findings suggest that cannabis-related hospital contacts should be treated as urgent opportunities to screen for suicide risk, even when patients have no documented psychiatric history.
Cannabis-Related Hospital Visits Linked to Sharply Elevated Risk of Self-Harm and Suicide
A large Ontario cohort study of over 11 million residents finds that individuals presenting to hospital for cannabis-related reasons carry a five-fold increased risk of deliberate self-harm and a nine-fold increased risk of death by suicide within three years, with elevated risk persisting even among those with no prior psychiatric history.
#82
Strong Clinical Relevance
Large population cohort with robust administrative linkage provides actionable risk-stratification data for clinicians encountering cannabis-related hospital presentations.
Suicide Risk
Self-Harm
Population Cohort
Mental Health Screening
Cannabis use disorder prevalence is climbing globally, and in several jurisdictions daily cannabis use now exceeds daily alcohol use. Yet population-level evidence linking cannabis-related healthcare encounters to downstream self-harm and suicide has remained limited, particularly compared to the well-established literature on alcohol. Emergency departments and hospital wards see thousands of cannabis-related presentations each year, and without clear data on what those encounters signal about future psychiatric risk, clinicians have lacked a strong evidence base to guide follow-up intensity. This study reframes cannabis-related hospital contact as a sentinel event deserving the same risk-assessment rigor traditionally reserved for alcohol and opioid presentations.
This retrospective population-level cohort study drew on ICES-linked administrative databases covering approximately 97% of Ontario residents, following 11.3 million adults from 2008 through 2021. The researchers identified 85,108 individuals whose first cannabis-related hospital contact (defined by validated ICD-10 codes for cannabis use disorder or poisoning) occurred during the study window. Each exposed individual was matched to up to ten general population controls on age, sex, and index date, producing a primary analysis sample of 861,291. The study leveraged hospital discharge abstracts, emergency department records, and vital statistics to track two primary outcomes: incident hospital contact for deliberate self-harm and death by suicide, using adjusted Cox proportional hazard models with cause-specific cumulative incidence functions.
Over a median follow-up of five years, individuals in the cannabis-contact group showed a 5.4-fold adjusted increase in deliberate self-harm risk (aHR 5.35, 95% CI 5.04 to 5.67) and a 9.2-fold adjusted increase in suicide death (aHR 9.22, 95% CI 5.24 to 16.23) at three years relative to controls. Notably, the wide confidence interval around the suicide estimate reflects smaller event counts for completed suicide. A critical sensitivity analysis excluding everyone with any prior mental health or substance use disorder diagnosis yielded an even stronger self-harm association (aHR 7.18, 95% CI 6.26 to 8.23), suggesting the signal is not fully explained by pre-existing psychiatric comorbidity. A secondary comparison against alcohol-related hospital contacts found that alcohol conferred modestly higher self-harm risk than cannabis (aHR 1.22, 95% CI 1.18 to 1.26), but both substances carried substantially elevated absolute risk. The authors emphasize that administrative data cannot capture cannabis use or self-harm events that never reach hospital, and that unmeasured confounders including trauma history, social adversity, and polysubstance use not captured in records represent important inference limitations.
This study reinforces something I have long believed in clinical practice: a cannabis-related hospital visit is not a benign event. The sheer magnitude of the hazard ratios here is striking, particularly the finding that even patients without any documented psychiatric history showed markedly elevated self-harm risk. Where the study adds real value is in quantifying what many of us have observed anecdotally. The gap between this evidence and clinical reality is that most emergency departments still do not treat a cannabis presentation as a trigger for structured suicide risk screening. This paper provides the population-level ammunition to change that practice pattern.
In my own practice, when a patient discloses problematic cannabis use or arrives with a cannabis-related concern, I treat it as a red flag for psychological distress, not simply a substance use issue. I routinely screen for suicidal ideation and self-harm, ensure safety planning is in place, and schedule close follow-up. I do not assume that cannabis is the cause of the psychiatric risk, but I take the presentation seriously as a marker that something deeper may be unaddressed. This study validates that approach and argues for broader adoption across clinical settings.
This study sits at an important inflection point in the research arc linking cannabis use to psychiatric harm. While prior work, largely cross-sectional or drawn from smaller clinical samples, has suggested associations between cannabis use disorder and suicidality, this is among the first population-level cohort studies to quantify the risk trajectory over multiple years using matched controls. Its design provides considerably stronger evidence than case reports or surveys, though it remains observational and cannot isolate a causal mechanism. Clinicians should understand these findings as describing a risk profile attached to a specific clinical presentation, the cannabis-related hospital contact, rather than to cannabis use broadly.
From a pharmacological standpoint, chronic heavy cannabis exposure can dysregulate the endocannabinoid system’s modulation of stress responses and mood, and high-THC products may amplify vulnerability to psychotic symptoms and impulsivity. However, this study does not capture product type, dose, frequency, or route of administration, so no dose-response inference is
Have thoughts on this? Share it:
