Older Adults on Medical Cannabis Face Doubled Intoxication Risk When Co-Prescribed Narrow Therapeutic Index Drugs

A large Ontario cohort study finds that co-exposure to medical cannabis and narrow therapeutic index drugs significantly raises the risk of drug-related intoxication in older adults, while prescribing practices show no detectable adjustment after cannabis authorization, pointing to a clinically important gap in pharmacovigilance.

Why This Matters

Medical cannabis use among older adults is expanding rapidly across North America, yet this population carries the highest burden of polypharmacy and the greatest vulnerability to drug interactions. Cannabinoids are known inhibitors of cytochrome P450 enzymes responsible for metabolizing many common prescription drugs, making pharmacokinetic interactions biologically plausible and clinically consequential. Despite this, real-world evidence quantifying the actual adverse event burden of cannabis-drug co-exposure in older adults has been nearly absent until now, leaving clinicians without data to guide prescribing decisions at the point of care.

Clinical Summary

As medical cannabis authorization has grown among older Canadians, questions about its interaction with existing medications have outpaced the evidence base. A 2025 longitudinal cohort study published in Age and Ageing by Gomes and colleagues used linked clinical cannabis registry data and Ontario administrative health records to examine whether prescribers adjusted co-medications after cannabis authorization and whether co-exposure to interacting drugs produced measurable harm. The pharmacological rationale is well established: delta-9-tetrahydrocannabinol and cannabidiol inhibit several CYP450 isoenzymes, including CYP2C9 and CYP3A4, potentially elevating serum concentrations of narrow therapeutic index drugs (DNTI) such as warfarin, certain anticonvulsants, and immunosuppressants to toxic levels.

Using an interrupted time series design alongside propensity-score-matched cohort comparisons, the investigators found that dispensing patterns for drugs affected by cannabinoid-related CYP450 interactions (DARSCIC and DNTI) were essentially unchanged in the year following cannabis authorization compared to the year before, suggesting prescribers did not systematically account for potential interactions. Among 3,926 older adults co-exposed to cannabis and DNTI, the hazard ratio for drug-related intoxication was 2.61 (95% CI 1.42 to 4.79) compared to 12,223 matched controls on DNTI alone. The warfarin-bleeding analysis (378 exposed, 1,646 controls) yielded a non-significant hazard ratio of 1.19 (95% CI 0.71 to 1.98), though the small sample severely limited statistical power. An unexpected finding linked cannabis-levothyroxine co-exposure to increased heart failure-related emergency department visits. The authors note that residual confounding, reliance on dispensation rather than consumption data, and the inability to verify actual cannabis dosing or adherence are important constraints, and they call for prospective pharmacovigilance studies before firm clinical guidance can be issued.

Dr. Caplan’s Take

This study addresses something I encounter regularly: older patients already on complex medication regimens who ask about or are already using medical cannabis, often without their other prescribers being aware. The 2.6-fold intoxication risk with narrow therapeutic index drugs is not surprising given what we know about cannabinoid CYP450 inhibition, but seeing it quantified in a large real-world cohort makes the concern far more concrete. What is equally striking is the finding that prescribers did not adjust co-medications after authorizing cannabis. That gap between pharmacological knowledge and clinical behavior is where patient harm lives.

In my practice, every patient over 65 considering cannabinoid therapy gets a thorough medication reconciliation before we proceed. I review their drug list for CYP2C9 and CYP3A4 substrates, flag narrow therapeutic index agents explicitly, and coordinate with the prescribing physician when dose adjustments or enhanced monitoring may be warranted. This study does not change what I do, but it powerfully reinforces why I do it and why every clinician authorizing cannabis in this population should be doing the same.

Clinical Perspective

This study sits at an important inflection point in the cannabis-medicine evidence arc. Prior evidence for cannabinoid-drug interactions in older adults has consisted largely of case reports, in vitro pharmacokinetic studies, and small clinical series. Gomes and colleagues provide the first large-scale, population-level observational data confirming that the theoretical interaction risk translates into measurable clinical events. The intoxication finding with DNTI co-exposure is the study’s strongest signal and aligns with established pharmacology. The null warfarin result should not be interpreted as evidence of safety; the exposed cohort of 378 patients lacked statistical power to detect clinically meaningful effect sizes, and warfarin management in Ontario may involve closer INR monitoring that partially mitigates risk. The levothyroxine-heart failure association is provocative but mechanistically unclear and requires replication.

From a safety standpoint, clinicians should be particularly vigilant with cannabis co-prescribed alongside anticonvulsants (clobazam, phenytoin, valproic acid), calcineurin inhibitors (tacrolimus, cyclosporine), and direct oral anticoagulants metabolized via CYP3A4. Cannabidiol-dominant products may pose greater interaction risk than THC-dominant formulations given CBD’s more potent CYP450 inhibition profile. The single most actionable step a clinician can implement now is to conduct a formal drug interaction screen using a tool such as the Lexicomp or Clinical Pharmacology database at the time of cannabis authorization for any patient aged 65 or older, and to document the results and any dosing modifications in the shared medical record.

Study at a Glance

Study Type

Interrupted time series combined with propensity-score-matched longitudinal cohort study

Population

Adults aged 66 and older authorized for medical cannabis in Ontario, Canada, covered by public drug insurance

Intervention

Medical cannabis authorization with concurrent dispensation of DARSCIC, DNTI, warfarin, or levothyroxine

Comparator

Propensity-score-matched patients dispensed the same interacting drugs without cannabis authorization (3:1 ratio)

Primary Outcomes

Bleeding (warfarin), drug-related intoxication (DNTI), thyrotoxicosis, heart failure, acute coronary syndrome, stroke (levothyroxine)

Sample Size

12,599 (ITS); 378 vs. 1,646 (warfarin); 3,926 vs. 12,223 (DNTI); 2,499 (levothyroxine)

Journal

Age and Ageing

Year

2025

DOI or PMID

Published in Age and Ageing, 2025 (specific DOI not available in extracted text)

Funding Source

Not specified in available text

What Kind of Evidence Is This

This is a peer-reviewed original observational study combining an interrupted time series analysis of prescribing behavior with propensity-score-matched cohort comparisons of clinical outcomes. It occupies a mid-level position in the evidence hierarchy, above case reports and cross-sectional analyses but below randomized controlled trials. The single most important inference constraint is that, as an observational design relying on administrative dispensation data, it cannot establish causation, and unmeasured confounding factors such as disease severity, recreational cannabis use, and actual medication adherence may account for part of the observed associations.

How This Fits With the Broader Literature

Prior evidence for cannabinoid-drug interactions has been largely mechanistic or anecdotal. In vitro studies and case reports have documented cannabinoid inhibition of CYP2C9 (relevant to warfarin) and CYP3A4 (relevant to many DNTI drugs), and a small number of clinical pharmacokinetic studies, notably those examining CBD-clobazam interactions in epilepsy populations, have confirmed elevated substrate drug levels during co-administration. The present study extends this literature substantially by demonstrating that these pharmacokinetic risks manifest as measurable adverse clinical events at the population level. It also adds a novel and concerning dimension: the finding that prescribers do not adjust co-medications after cannabis authorization, which echoes broader literature on the underrecognition of cannabis as a pharmacologically active agent in polypharmacy management.

Common Misreadings

The most likely overinterpretation is treating the 2.61 hazard ratio as proof that medical cannabis directly causes drug toxicity in older adults. While the association is biologically plausible and statistically significant, this is observational evidence subject to residual confounding. Patients authorized for medical cannabis may differ from controls in disease burden, health behaviors, or healthcare utilization patterns that independently affect intoxication risk. Equally important, the null warfarin finding should not be read as evidence that cannabis is safe to combine with anticoagulants. The exposed sample of 378 patients was too small to detect anything short of a very large effect, and absence of statistical significance is not the same as absence of risk.

Bottom Line

This study provides the strongest population-level evidence to date that co-prescribing medical cannabis with narrow therapeutic index drugs in older adults is associated with a clinically meaningful increase in drug-related intoxication risk. It also reveals a troubling gap in prescriber behavior. It does not establish causation and cannot substitute for prospective pharmacovigilance data. For now, the practical mandate is clear: any clinician authorizing cannabis for an older adult should conduct a formal drug interaction review and coordinate dosing adjustments with co-prescribers before initiating therapy.

References

1. Gomes T, et al. Cannabis-drug interactions and adverse outcomes among older adults authorized for medical cannabis: a population-based cohort study. Age and Ageing. 2025.
2. Nasrin S, Watson CJW, Perez-Paramo YX, et al. Cannabinoid metabolites as inhibitors of major hepatic CYP450 enzymes, with implications for cannabis-drug interactions. Clin Pharmacol Ther. 2021;109(6):1506-1517.
3. Gaston TE, Bebin EM, Cutter GR, et al. Interactions between cannabidiol and commonly used antiepileptic drugs. Epilepsia. 2017;58(9):1586-1592.
4. Antoniou T, Bodkin J, Ho JM. Drug interactions with cannabinoids. CMAJ. 2020;192(9):E206.