Commentary on Baltes‐Flueckiger et al.: Rethinking expectations and evidence in cannabis policy

In a randomized control trial (RCT), legal cannabis access does not appear to significantly improve or harm self‐reported mental health outcomes, underscoring the need to revisit which outcomes are prioritized in RCTs and to complement them with broader methodological approaches that capture the full societal impact of cannabis regulation.

Recreational cannabis regulation (RCR) should be guided by evidence and grounded in public health principles. However, few jurisdictions with liberal RCR have relied on evidence to guide the process [1, 2]. Switzerland’s 2021 legal change allowing pilot RCR trials has made it possible to conduct randomized comparisons between individuals with and without legal cannabis access, offering a unique opportunity to inform on individual‐level effects of RCR [3]. Baltes‐Flueckiger et al. [4] present findings from the first such randomized controlled trial (RCT). While this offers valuable insights into the relationship between cannabis regulation and mental health outcomes, it opens the door for a broader conversation about how we evaluate RCR impacts. In this commentary, we build on these findings to reflect on outcomes that matter, explore the strengths and limitations of RCTs in this context, and advocate for a pluralistic approach to evidence that captures both individual and societal dimensions of legalization.

Baltes et al. [4] tested a public health model with regulated pharmacy sales, 20% tetrahydrocannabinol (THC) caps, no advertising and harm reduction measures. The intervention group had access to regulated cannabis while the control group remained on a 6‐month waiting list, purchasing from illegal markets. Their key finding showed that legal pharmacy access to cannabis did not significantly improve self‐reported mental health outcomes and only modestly reduced cannabis misuse, as measured by the Cannabis Use Disorders Identification Test – Revised (CUDIT‐R). These findings balance polarized expectations: neither confirming fears of widespread harm nor hopes of significant mental health benefits.

The methodological rigor of the study strengthens the authors’ claim that differences in outcomes likely reflect the effects of the intervention, rather than being artifacts of the study design.

While the study provides insights into the effects of one regulatory model, it is essential to recognize that RCR varies in form and impact. Baltes‐Flueckiger et al. [4] examined a public health‐oriented regulatory model, while many other studies focus on commercial models [5]. RCR effects likely depend on the specific regulatory model and implementation tested [6]. Future research should therefore assess how public health‐focused versus commercial RCR models produce different outcomes, rather than viewing access as simply legal or illegal.

The neutral RCT outcomes in the Baltes‐Flueckiger et al. [4] study highlights the need to re‐evaluate RCR assessment measures. Cannabis legalization affects multiple domains beyond mental health, including drug use norms, commercial practices and law enforcement. While RCTs establish causal effects on individual‐level outcomes, they cannot fully capture broader social, legal and cultural impacts of regulatory change, which require complementary methods and observational designs. Studies based on observational population level surveys, ethnographic data, media analyses and administrative data (e.g. arrests, hospitalizations) are needed to understand societal effects. Combined, these approaches can inform evidence‐based policy that consider individual, subpopulation and broader societal outcomes.

Beyond their implications for policy evaluation methods, the findings from Baltes‐Flueckiger et al. [4] raise questions about the traditional indicators used to assess problematic cannabis use in regulated environments [7]. Tools like CUDIT‐R, developed within prohibitionist frameworks, may lose validity where cannabis is legally accessible. In the Baltes‐Flueckiger et al. [4] study, only the item about time spent obtaining cannabis showed significant difference, potentially reflecting improved access rather than reduced misuse. Similarly, other items in cannabis misuse scales, such as decreased concern from friends/family, may indicate social acceptance rather than a decline in misuse. We therefore argue that cannabis use disorder/misuse definitions need revision in the RCR era.

It is also important to note that while the internal validity of the Baltes‐Flueckiger et al. [4] study is uniquely strong, the external validity of the trial is inherently limited. RCTs often exclude vulnerable groups, and this was also the case in this trial. The law in Switzerland allowing the conduct of RCR trials precludes the inclusion of non‐users, minors and people with serious psychiatric disorders [8, 9]. Also, those who participated were willing to go to a clinic for health assessments when participating in the trial. As a result, this and other RCTs may not provide insight into the populations most likely to be impacted by RCR, such as those at risk of initiating cannabis use [10] or experiencing changes in their patterns of use [11].

In conclusion, the study by Baltes‐Flueckiger et al. [4] marks an important step toward evidence‐informed RCR. It urges us to rethink not only the expected outcomes of RCR but also the forms of evidence we prioritize. RCR should be judged on the impact on individual‐level mental health outcomes and also by the influence on wider social processes, including, but not limited to, stigma reduction, criminal justice engagement and safe consumer practices.

AUTHOR CONTRIBUTIONS

Sharon R. Sznitman: Conceptualization (equal); writing—original draft (lead); writing—review and editing (lead). Reto Auer: Conceptualization (supporting); writing—original draft (supporting); writing—review and editing (supporting). Barbara Broers: Conceptualization (supporting); writing—original draft (supporting); writing—review and editing (supporting).

DECLARATION OF INTERESTS

No financial or other relevant links to companies with an interest in the topic of this article.

DATA AVAILABILITY STATEMENT

Not applicable.