Before Canada legalized recreational cannabis, about 1 in 8 Canadians aged 15 and older reported past-year use. A nationally representative 2015 survey reveals that those who called their use “medical” tended to be older, in worse health, lower income, and far more frequent users than recreational-only consumers. However, the cross-sectional design and absence of medical authorization verification mean these associations cannot be taken as causal or used to characterize formally authorized patients.

When Canada legalized recreational cannabis in 2018, the country needed a clear baseline against which to measure change. This 2015 national survey provides that baseline, but it also raises pressing clinical questions. People who describe their cannabis use as medical are reporting more pain, poorer health, lower incomes, and dramatically higher frequency of use than those who use recreationally. Whether those patterns reflect genuine therapeutic need, inadequately managed chronic disease, or polysubstance vulnerability has immediate implications for how clinicians screen, counsel, and monitor cannabis-using patients in a post-legalization landscape.

Cannabis use in Canada prior to legalization existed along a spectrum from purely recreational to self-described medical, but the boundaries between these categories were porous and largely unverified by the healthcare system. Using data from the 2015 Canadian Tobacco, Alcohol and Drugs Survey (CTADS), this study examined approximately 15,000 respondents to estimate the prevalence of non-medical-only (NMO) and self-defined medical and non-medical (SDMNM) cannabis use among Canadians aged 15 and older, and to characterize the health, sociodemographic, and substance-use profiles that distinguished these groups. The CTADS employed a stratified random-digit-dialing design, included cellphone-only households, and used population weights and bootstrap variance estimation to produce nationally representative estimates.

The study found that 9.5% of Canadians reported NMO use and 2.8% reported SDMNM use in the past year, with British Columbia showing the highest prevalence for both types. Pain was the leading reason for medical use, cited by approximately half of SDMNM users. SDMNM users were nearly twice as likely as NMO users to report daily or near-daily cannabis use (47.2% versus 26.4%) and were significantly more likely to report fair or poor general and mental health, lower household income, and non-therapeutic use of psychoactive pharmaceuticals. A critical limitation is that medical use was self-defined rather than verified against Health Canada authorization records, meaning the SDMNM group is a heterogeneous category. The authors emphasize that no causal inferences can be drawn from these cross-sectional data and that post-legalization longitudinal studies are needed.

Medical or Recreational? What a 2015 National Survey Reveals About Canada’s Cannabis Users

Before Canada became one of the first countries to fully legalize recreational cannabis, roughly 1 in 8 Canadians were already using it, and a surprising number called it medicine. Who were they, and what set them apart from those who used it purely for pleasure? This nationally representative survey from 2015 offers the clearest pre-legalization picture we have, along with a set of important caveats that matter just as much as the numbers. I want to give the authors credit for something that is harder than it looks: they stayed in their lane. The paper does not claim that cannabis causes worse health, does not pretend to evaluate Canada’s formal medical access program, and explicitly tells readers that “self-defined medical” is not the same as “authorized medical.” That kind of intellectual honesty in government-sponsored research is a model worth noting. What the study genuinely contributes is a population-level snapshot of who was using cannabis and what else was true about them at that moment. The finding that SDMNM users were older, sicker, poorer, and using cannabis nearly daily should get any clinician’s attention, not because it proves cannabis is harmful, but because it identifies a cluster of vulnerability that deserves a clinical response.

The central methodological challenge here is straightforward but consequential. Asking someone whether they use cannabis “for a medical reason” without checking whether a doctor authorized it is like measuring prescription drug use by asking people if they take pills for their health. It captures behavior accurately, but tells you nothing about clinical appropriateness. The SDMNM category therefore conflates formally authorized patients, unauthorized self-medicators, and people who may simply frame recreational use as medical. This heterogeneity means that the health differences observed between groups could reflect characteristics of genuine patients with documented conditions, characteristics of people rationalizing heavy use, or some inseparable blend of both. The survey’s skip-logic also introduced a subtle but important artifact: one-time users could not be classified as SDMNM, which artificially inflated the frequency gap between the two groups. The cross-sectional design compounds the problem. Taking a photograph of a parking lot tells you which cars are there right now, but cannot tell you whether the red car arrived first or last, or whether it caused traffic or got caught in it. We know these users had worse health, but we cannot know whether cannabis was treating, exacerbating, or simply coexisting with their conditions.

As a clinician, what I take from this study is practical rather than prescriptive. If a patient tells me they use cannabis for pain, anxiety, or sleep, I now have national data confirming that this person is statistically more likely to be in worse overall health, more likely to be using other substances non-therapeutically, and more likely to be using cannabis daily. That profile warrants engagement, not judgment. To a colleague, I would say: screen for cannabis use in patients with chronic pain, mood disorders, and insomnia, and treat the disclosure as a clinical opening rather than a dead end. To a policymaker, I would say: this study gives you a baseline, not a verdict. The clustering of worse health and lower income in self-defined medical users suggests that legalization without equity-sensitive access provisions risks leaving a vulnerable population poorly served. Population-level surveillance can tell us who is doing something and what else is true of them at the same moment in time, but it cannot tell us why, in what sequence, or with what consequence. When policy windows are short and the need for evidence is urgent, these snapshots are invaluable; they should be read as the start of an evidence conversation, not its conclusion.

This study sits at the descriptive, foundational level of the research arc. It does not test hypotheses about cannabis efficacy or harm. It does not follow patients over time. What it provides is the population-level denominator and demographic profile against which future post-legalization studies can be compared. For clinicians working in jurisdictions where cannabis is now legal, these 2015 data offer a pre-legalization reference that can help contextualize changes in patient demographics, use patterns, and clinical presentations.

The co-occurrence of SDMNM cannabis use with non-therapeutic psychoactive pharmaceutical use is a finding that warrants particular clinical vigilance. Polysubstance use patterns, especially those involving cannabis, opioids, benzodiazepines, or sedative-hypnotics, carry compounded risks including respiratory depression, cognitive impairment, and functional decline. The near-doubling of daily-use prevalence in SDMNM users (47.2% versus 26.4%) also aligns with dose-response evidence linking high-frequency cannabis use to greater dependence risk. Clinicians should consider routinely asking about cannabis use, its self-reported purpose, and its frequency in patients presenting with chronic pain, insomnia, anxiety, or depression, and should treat affirmative answers as invitations for clinical partnership rather than confrontation.

This is a cross-sectional observational study using nationally representative population survey data. In the evidence hierarchy, cross-sectional studies sit below cohort studies and randomized trials because they capture a single point in time and cannot establish temporal sequence or causation. The single most important inference constraint is that all findings are associational: the study can describe who uses cannabis and what characteristics they share, but it cannot determine whether cannabis use contributes to, results from, or simply accompanies the health and socioeconomic differences observed.

The profile of self-defined medical cannabis users found here, including older age, greater medical burden, higher frequency of use, and pain as the predominant indication, is broadly consistent with findings from U.S. population-based surveys of medical cannabis users conducted around the same period. Ontario-specific studies had previously suggested similar patterns in Canadian sub-populations, but this is one of the first analyses to confirm those patterns at the national level with population weights. The finding that SDMNM users report co-occurring non-therapeutic pharmaceutical use extends earlier observations from regional convenience samples by establishing this association in a representative national dataset. Taken together, the study confirms rather than challenges prior literature, while offering a methodological upgrade through its nationally representative scope and complex-survey analytic approach.

The most consequential analytic choice was the survey’s skip-logic structure, which excluded one-time cannabis users from being classified as SDMNM. This is not a statistical analysis decision the authors made but a structural feature of the questionnaire itself, and it inflated the apparent frequency gap between SDMNM and NMO groups. Had the survey permitted one-time users to indicate medical purposes, the daily-use disparity between groups might have been smaller and the SDMNM group might have appeared less extreme in its use patterns. Additionally, replacing the area-based income proxy with individual-level household income data could have meaningfully altered the magnitude and significance of income-related associations, potentially attenuating or strengthening the observed relationship between low income and SDMNM use.

The most likely overinterpretation is treating self-defined medical and non-medical (SDMNM) cannabis users as representative of Health Canada’s formally authorized medical cannabis patients. These are fundamentally different populations. The CTADS did not verify medical authorization, meaning SDMNM users include an unknown mixture of authorized patients, unauthorized self-medicators, and individuals who may frame recreational use as medical. Any headline claiming this study shows that “medical cannabis patients are sicker and poorer” fundamentally exceeds what the evidence supports.

A related misreading is interpreting the association between SDMNM use and worse health as evidence that cannabis harms health. Cross-sectional data cannot distinguish cause from consequence. People in worse health may turn to cannabis for relief, cannabis may contribute to health decline in some users, or both patterns may be driven by a third set of factors entirely. The study’s design simply cannot answer that question.

This study provides a methodologically sound, nationally representative pre-legalization baseline of Canadian cannabis use. It identifies a self-defined medical user group that is older, sicker, poorer, and uses cannabis far more frequently than recreational-only users. It does not establish whether cannabis use causes, treats, or simply accompanies these health characteristics. For clinicians, it is a prompt to ask about cannabis use in vulnerable patients. For policymakers, it is a reference point, not a verdict.

Does this study prove that medical cannabis makes people sicker?

No. The study found that people who reported using cannabis for medical reasons also reported worse health, but it cannot determine whether cannabis caused that worse health, was being used to treat it, or was simply present alongside it. This is a limitation inherent to all cross-sectional surveys: they capture a single moment in time and cannot establish cause and effect.

Were the people in this study actual medical cannabis patients?

Not necessarily. The survey asked people whether they used cannabis for medical reasons, but it did not verify whether a doctor had authorized their use through Health Canada’s official program. The “self-defined medical” group likely includes a mix of formally authorized patients, people self-medicating without medical oversight, and individuals who may describe recreational use as medical.

Why does this 2015 study still matter after cannabis was legalized in 2018?

This study provides a crucial baseline. To understand how legalization has changed cannabis use patterns, public health officials need to know what those patterns looked like before the law changed. The 2015 CTADS data serve as that reference point for tracking shifts in prevalence, user demographics, health correlates, and use frequency over time.

Should I be concerned if I use cannabis daily for pain?

Daily use was common among self-defined medical users in this study, and pain was the most frequently cited reason for medical use. Higher-frequency use is associated with greater dependence risk in population data, though it may also reflect genuine therapeutic need. This is exactly the kind of pattern that benefits from open conversation with a healthcare provider who can help evaluate whether cannabis is serving you well and whether other approaches might complement or improve your care.

Have thoughts on this? Share it: