A 2025 survey of 322 older Canadians who use cannabis for mental health symptoms found that most perceive it as helpful and safe, yet nearly two-thirds also reported experiencing at least one adverse effect in the past year. This gap between perceived safety and experienced side effects, particularly dizziness, cognitive fog, and balance impairment, raises important concerns for a population already vulnerable to falls and drug interactions.

Older adults represent the fastest-growing demographic of cannabis users in Canada, and many are turning to cannabis for mental health symptoms such as insomnia, anxiety, and depression without any clinical guidance. This population is also among the most vulnerable to adverse drug events given age-related pharmacokinetic changes, polypharmacy, and fall risk. Understanding what older adults believe about cannabis, what they experience when they use it, and whether they disclose that use to their healthcare providers is essential for clinicians who want to protect this group proactively rather than reactively.

Cannabis use among older adults in Canada has been rising steadily since legalization in 2018, and mental health symptom management, particularly insomnia, anxiety, and depression, is one of the most commonly cited motivations. Despite this growing trend, most existing evidence on cannabis for mental health comes from younger or clinical populations, leaving a substantial gap in our understanding of how community-dwelling older adults use cannabis, what they think about it, and what they experience when they do. This survey, administered online through the Qualtrics platform between February and September 2022, recruited 1,615 Canadians aged 50 and older through social media, websites, and organizational email lists. The investigators analyzed a subsample of 322 respondents who reported using cannabis at least in part for mental health symptoms, comparing them to non-mental-health cannabis users and non-users across demographic, health, and perception variables.

Among the 322 mental health cannabis users, insomnia was the most commonly cited indication (84.8%), followed by anxiety (43.8%) and depression (22.4%). A total of 70.8% rated cannabis as somewhat or extremely helpful, 57.1% perceived it as more effective than pharmaceuticals, and 73.8% rated it as safe or very safe compared to medications. However, 62.4% reported at least one adverse effect in the past year, with dry mouth (34.5%), balance impairment (22%), intoxication or hallucination (20.8%), and impaired mental alertness (20.2%) being the most common. The mental health user group was significantly more likely to be female, younger within the 50-plus range, and living with multimorbidity and polypharmacy compared to non-users. The authors acknowledge that the high adverse effect rate is likely inflated by the prompted-checklist methodology and that the convenience sample limits generalizability. They call for controlled longitudinal studies to determine whether the perceived benefits translate into measurable clinical outcomes.

The Perception-Safety Gap: What a Large Canadian Survey Reveals, and Cannot Resolve, About Cannabis for Mental Health in Older Adults

Two out of three older Canadians using cannabis for their mental health reported side effects last year, yet nearly three-quarters called it safe. That gap between experience and perception may be the most important thing this survey has to teach us. The study itself is well constructed for what it is: a descriptive snapshot of a community-dwelling population that is largely invisible to clinical research. It gives us something genuinely useful by documenting who these patients are (more often female, carrying multiple chronic conditions, taking multiple medications) and by revealing that more than one-third never told their healthcare provider about their cannabis use. Those descriptive observations are actionable today. But the study does not, and structurally cannot, tell us whether cannabis is actually working for these patients’ insomnia, anxiety, or depression. What it measures is perception, not outcome. Surveying only people who continue to use cannabis and finding that most consider it helpful is like surveying only people still dining at a restaurant and concluding the food must be excellent. You never hear from those who tried it, found it unhelpful or harmful, and stopped. This is survival bias, and it inflates every perceived-effectiveness figure in the paper.

The adverse effect data, paradoxically, may be the most clinically important finding, even though it too requires careful interpretation. The 62.4% adverse effect rate looks alarming, but the methodology partly explains it. When you hand someone a checklist of possible symptoms and ask them to check off anything they have experienced, you will always get higher rates than if you first ask an open-ended question. This is a well-known phenomenon in survey design: the shape of the question determines the shape of the answer. Prior surveys that used two-stage ascertainment (open recall first, then prompts) reported rates closer to 15%. The real number likely sits somewhere between these extremes. But even if we halve the reported rate, 11% of older adults experiencing balance impairment from a self-administered substance is clinically significant in a population where a single fall can be catastrophic. Similarly, 10% reporting cognitive dulling in a group already managing polypharmacy raises legitimate concerns about medication adherence, driving safety, and decision-making capacity.

What I find myself telling patients, colleagues, and policymakers from this data comes down to three points. To patients: I am glad you told me about your cannabis use, and I want to work with you on it rather than around it. To colleagues: screen proactively, because one in three of these users will never volunteer the information. To policymakers: adoption is running ahead of evidence, and we need investment in the controlled trials that will tell us whether this practice is genuinely beneficial or merely perceived as such. This study makes a real contribution by illuminating a growing, under-supervised clinical phenomenon, but the widely cited 70% helpfulness figure should be understood as a measure of patient attitude, not clinical outcome. Self-reported perceived effectiveness in a convenience sample of current users is one of the weakest forms of evidence for treatment efficacy, and it is simultaneously the most common type of evidence people find persuasive. When interpreting any survey showing that patients “found it helpful,” the first question must always be: helpful compared to what, as measured by whom, in whom, and accounting for which biases?

This survey sits early in the research arc for cannabis and mental health in older populations. It provides useful hypothesis-generating data and demographic characterization, but it occupies a rung on the evidence ladder well below the cohort studies, pragmatic trials, and randomized controlled trials that would be needed to inform treatment decisions. The absence of validated mental health instruments such as the PHQ-9, GAD-7, or Insomnia Severity Index means the study cannot speak to clinical improvement at all, only to what respondents believe about their own improvement. For clinicians, the study should be understood as a signal about patient behavior and attitudes, not a signal about treatment effectiveness.

From a pharmacological standpoint, the overlap of cannabis use with polypharmacy in this population warrants serious attention. Both CBD and THC are known inhibitors of several CYP450 enzymes, which means they have the potential to alter the metabolism of commonly prescribed medications including anticoagulants, benzodiazepines, and certain antidepressants. The balance impairment reported by 22% of respondents and the cognitive effects in over 20% compound these concerns, particularly for older patients on sedating medications. The single most actionable recommendation from this study is straightforward: clinicians should routinely ask older patients about cannabis use, especially those with multimorbidity or polypharmacy, and document cannabis products and patterns with the same rigor applied to prescription medications.

This is an original cross-sectional survey using an online convenience sample, placing it among the lower tiers of the clinical evidence hierarchy. Cross-sectional surveys can describe prevalence, attitudes, and associations at a single point in time, but they cannot establish causal relationships, measure clinical effectiveness, or produce population-level prevalence estimates when the sample is not probability-based. The single most important constraint on inference is that all findings reflect what respondents believe and report, not objectively verified clinical outcomes.

This study aligns with prior surveys by Reynolds et al. (2018), Yang et al. (2021), and Kaufmann et al. (2023) in confirming that a substantial minority of older adults use cannabis for mental health purposes and generally view it favorably. It extends those findings by providing more granular adverse effect data and by comparing mental health cannabis users to other user groups and non-users within the same sample. However, the much higher adverse effect rate reported here (62.4% versus roughly 15% in prior work) likely reflects methodological differences rather than a genuinely more hazardous cannabis landscape. The study’s findings are broadly consistent with Wolfe et al.’s (2023) review, which noted limited and mixed evidence for cannabis effectiveness in mental health conditions, and with Vaillancourt et al.’s (2022) observations about pharmaceutical substitution patterns, though this study did not directly measure substitution.

The most consequential analytic choice was the use of a prompted adverse effect checklist without a prior open-ended recall stage. Had the investigators used a two-stage ascertainment method, asking respondents first whether they had experienced any problems and then following up with specific prompts, the adverse effect rate would almost certainly have been substantially lower and more consistent with the 10 to 20% range reported in prior surveys. This single methodological decision is responsible for the most eye-catching finding in the paper. Additionally, the absence of any multivariate modeling means that demographic and health differences between subgroups (multimorbidity, polypharmacy, age) were not adjusted for, leaving open whether group-level differences in perceptions or adverse effects are driven by confounders rather than cannabis use per se.

The most likely overinterpretation is treating the finding that “70.8% found cannabis helpful” as evidence that cannabis is clinically effective for mental health conditions in older adults. This conflates a self-reported attitude measure with a clinical outcome measure, two fundamentally different constructs. No validated mental health instruments were used, no baseline symptom severity was recorded, and the current-user sample inherently excludes people who tried cannabis and stopped because it did not work. Separately, the 62.4% adverse effect rate should not be cited as a reliable population-level estimate of harm without noting that the prompted-checklist methodology is known to substantially inflate reporting. Readers should also avoid interpreting the 19.9% prevalence figure as representative of all older Canadians; it reflects a convenience sample skewed toward cannabis engagement.

This survey contributes valuable descriptive data on a rapidly growing and under-supervised clinical phenomenon: older adults self-managing mental health with cannabis in the community. It identifies actionable gaps in clinician-patient communication and raises legitimate concerns about adverse effects relevant to fall risk and cognition. It does not establish that cannabis is effective or safe for mental health in older adults. Clinicians should use these findings to prompt proactive screening conversations, not to guide prescribing decisions.

Does this study prove that cannabis works for anxiety, depression, or insomnia in older adults?

No. The study measured what people believe about cannabis, not whether it produced measurable clinical improvement. Without validated outcome instruments, a control group, or follow-up assessments, no conclusion about effectiveness can be drawn from this data. The 70.8% who “found it helpful” are expressing a personal perception, which is meaningful but categorically different from demonstrated clinical benefit.

Should I be worried about the 62% adverse effect rate?

That number is likely inflated by the way the survey asked the question. When researchers present a checklist of possible symptoms, people naturally recognize and endorse more of them than when asked an open-ended question. Prior studies using a different approach found rates closer to 15%. However, even if the true rate is substantially lower, the specific types of adverse effects reported, particularly dizziness, balance problems, and mental fogginess, are genuinely concerning for older adults who may be at risk of falls or who are managing multiple medications.

I use cannabis for sleep and feel it helps. Should I stop?

This study does not provide a basis for stopping or continuing cannabis. What it does strongly suggest is that you should discuss your cannabis use with your healthcare provider, including the specific products, doses, and frequency. This is especially important if you take other medications, as cannabis can interact with many common prescriptions. Over one-third of older adults in this survey had never disclosed their cannabis use to a clinician, and that gap in communication represents a real safety concern.

Is cannabis safer than prescription medications for mental health?

This study cannot answer that question. While 73.8% of respondents perceived cannabis as safe compared to pharmaceuticals, perception of safety and actual safety are different things. Comparative safety would need to be assessed in controlled studies measuring objective outcomes, adverse events, and drug interactions in matched populations. For now, the evidence is insufficient to support or refute this claim.

Have thoughts on this? Share it: