In this large cross-sectional study, linking survey and administrative data on a representative sample of Canadian veterans released from the Canadian Regular Force (1998-2018), several sociodemographic, health, and service-related factors were associated with authorization for CMP reimbursement.

We found that about one-fifth of veterans were authorized for CMP reimbursement, with particularly strong associations observed for variables related to PTSD, anxiety and mood disorders (unadjusted ORs>5) and severe pain (OR>7.5). From a physical health standpoint, individuals reporting traumatic brain injury, migraines and bowel ulcers were more likely to be authorized for CMP, whereas those with diabetes or heart diseases were less likely. Considering that these last two conditions have been associated with certain risks for complications following regular cannabis use [22,23], results from the current analyses could possibly point to contraindications preventing healthcare providers from authorizing cannabis for people with this medical profile. Unadjusted analyses also unveiled associations with poorer relations and social supports, as well as provincial differences, with significantly lower CMP authorizations in Quebec, which has a global regulatory approach known to be especially conservative, compared to Ontario. How different levels of the social environment, from individual relationships to political views about cannabis, may influence medical use should be further investigated under psychosocial and health equity lenses. Although retired individuals were twice as likely to be authorized for CMP than those actively working or on reserve, being on disability leave was the work status that was the most strongly associated with CMP (OR>6).

In multivariate analysis, younger age, inability to work, disabling pain that prevents most activities, PTSD and anxiety emerged as key correlates of CMP authorization in the overall sample.

Similar to what was seen in unadjusted analyses, inability to work and severe disabling pain were consistent correlates across PTSD strata, but additional associations differed by PTSD status. Among veterans without PTSD, younger age, multiple physical conditions, and frequent mental health service use were notable correlates, whereas among those with PTSD, fewer years of service, witnessing destruction during service, and suicidal ideation were associated with CMP authorization. These findings suggest that CMP authorization patterns may reflect not only mental and physical health conditions, but also broader functional limitations and service-related exposures.

Our results are consistent with prior studies documenting that younger veterans are more likely to use cannabis compared to older veterans [24]. However, while younger veterans (e.g., ages 18–25) have previously been shown to have higher overall rates of cannabis use, older veterans (especially those aged 60 and above) were more likely to report using cannabis specifically for medical reasons, such as pain management and mental health conditions [25]. The strong associations with inability to work and pain are also in line with previous reports linking cannabis use to self-management of chronic pain [26] and functional impairments [27]. The observed associations with PTSD and anxiety also mirror findings from other studies [15–18]. In veteran populations, cannabis use has been described as a strategy for coping with intrusive symptoms and hyperarousal [28]. However, cannabis use in the context of mental health disorders remains controversial, with some evidence suggesting potential short-term symptom relief but limited evidence of long-term efficacy and growing concerns regarding dependence, psychiatric comorbidity, and cognitive effects [1,29,30]. Longitudinal studies are urgently needed to assess longer-term effects and determine whether evidence supports these concerns.

The stratified analyses by PTSD status highlight the complex interplay of mental health, service experiences, and functional status. In Veterans without PTSD, CMP authorization was more strongly associated with markers of poor physical health and healthcare engagement, whereas in those with PTSD, service-related trauma and suicidality were key correlates. This distinction may reflect differing pathways to CMP authorization: for some Veterans, cannabis may be pursued primarily for physical health conditions, while for others, it may be linked to trauma and its sequelae. However, pain and functional disability seem to be key factors, regardless of the physical or psychological nature of health issues.

A key strength of this study is the use of large, population-level administrative data linked to the survey, which minimizes recall bias and captures real-world patterns of CMP authorization across a representative veteran population in Canada. The high response (72%) and share (92%) rates minimize selection bias. The inclusion of a broad range of variables, including physical and mental health conditions, functional limitations, and service-related exposures, provides a more comprehensive view than many prior studies, which have typically focused on narrower domains. The stratification by PTSD status represents a novel contribution, uncovering differences in correlates that were obscured in pooled analyses. Furthermore, this study examined CMP authorization rather than self-reported cannabis use, thereby offering unique insights into the regulated, physician-mediated process of access to medical cannabis. Together, these strengths enhance the policy and clinical relevance of our findings, particularly in the Canadian context, where authorization is tied to reimbursement through Veterans Affairs Canada.

Several limitations should be acknowledged. First, the cross-sectional design precludes causal inference; associations cannot establish whether these factors drive CMP authorization or instead reflect the characteristics of those already authorized. Second, reliance on administrative data and survey may underrepresent symptom burden, functional impairment, or informal cannabis use. Third, although we included a broad set of potential correlates, residual confounding remains a possibility. For example, reliance on self-reported data, including PTSD diagnosis and symptoms, pain severity, and employment status, introduces potential for recall and social desirability bias and lacks clinical validation. Importantly, information on actual cannabis use and substance use problem/disorder was not available. Nevertheless, the large sample size and use of data sources based on reimbursement strengthen the robustness of our findings.

Building on these study findings, future studies should prioritize longitudinal designs to establish relationships between potential predictors and CMP authorization/cannabis use, as well as to evaluate CMP’s real-world effectiveness on outcomes highly relevant for veteran (e.g., pain interference, PTSD symptoms, functional recovery) and identify individuals who may benefit the most. For policy makers, distinct correlates by PTSD status highlight the importance of integrated pain-mental health services and careful benefit-risk discussions regarding cannabinoids, particularly where poly-morbidity and functional limitations are present.