The results of the CAMS-22 survey further underline the dramatic recent increase in the proportion of people using legally-prescribed rather than illicitly-sourced medical cannabis, consistent with recent TGA data indicating dramatic growth in the number of medical cannabis prescriptions in Australia since 2019 [8, 9, 16]. Almost three quarters of our sample had accessed mainly prescribed cannabis—compared with just over a third in our previous CAMS-20 survey [30], and less than 3% in our 2016 and 2018 surveys [4, 29]—allowing us for the first time to report the experiences of a large number of people who had been prescribed medical cannabis.

In CAMS-22, prescribed medical cannabis users appeared to have a different ‘profile’ than illicit users: they were more likely to be male, employed, and have tertiary qualifications. In addition, prescribed users tended to have started using cannabis for medical reasons later in life, used less cannabis for non-medical reasons, and were less likely to have used cannabis non-medically before commencing medical use. These differences suggest that for some illicit users the demarcation between non-medical and medical use is less clear than for prescribed users.

Medical cannabis—both illicit and prescribed—was primarily being used by respondents to treat pain (37%), mental health (36%) and sleep conditions (15%), consistent both with previous Australian surveys [4, 29] and with findings from studies of medical cannabis users around the world [18–20]. However, there were differences between prescribed and illicit users concerning which conditions they were more likely to be treating, with prescribed users more likely to use cannabis to treat a pain condition and less likely to treat a mental health condition than illicit users [29]. This is encouraging as there is better evidence for the effectiveness of cannabis for treating pain than for treating mental health conditions [1, 2, 38, 39]. Nonetheless, mental health conditions remain the second most commonly identified group of conditions treated with medicinal cannabis. The differences in main condition treated between prescribed and illicit users may stem from prescribed users’ receiving guidance from trained health professionals whose job is to be informed about the existing evidence for clinical indications for medications. Results from a recent survey of n = 505 Australian general practitioners support this interpretation, with respondents indicating they are much less comfortable with the idea of prescribing cannabis for anxiety, depression and sleep conditions than they are for pain, epilepsy, and chemotherapy-induced nausea [21]. Alternatively, it may be that the type of person who is already well-informed about the clinical evidence for effectiveness of cannabis for certain conditions is also more likely to seek their medical cannabis through legal channels. Unfortunately our survey is unable to give definitive answers on the reasons for the differences in conditions treated; future research among both clinicians and consumers may be able to explore these reasons in greater depth.

In CAMS-22 prescribed users were significantly more likely than illicit users to be mainly treating a sleep condition, reflecting the large increase in the proportion of SAS-B approvals granted for sleep disturbances in the last two years [16]. As with mental health disorders, there is scarce empirical evidence that medical cannabis is efficacious for sleep disorders [40–42], once again raising concerns over the apparent willingness of many Australian clinicians to prescribe medical cannabis for disorders for which there is limited evidence of efficacy.

Both prescribed and illicit users overwhelmingly endorsed the effectiveness of their medical cannabis in treating their main health condition. Over 95% of respondents who reported their main indication was pain, mental health, or sleep indicated that their condition was a little, much, or very much better since starting medical cannabis. In contrast to these overwhelmingly positive consumer reports, empirical evidence concerning efficacy of cannabis for mental health or sleep conditions is weak, and even for pain conditions, the available evidence suggests medicinal cannabis has modest efficacy [1, 2, 38, 43]. Recent observational and longitudinal studies of patients attest to a more general improvement in health-related quality of life in patients prescribed medical cannabis, seemingly irrespective of whether their primary condition is being specifically treated [44–47]. It may be that cannabis leads to a shift in hedonics, and broader health and social wellbeing, rather than specific clinical outcomes such as pain intensity or sleep duration.

The disparity between self-reported effectiveness of medical cannabis from patient surveys and cohort studies compared to evidence from ‘gold standard’ randomised trials underscores many of the challenges facing consumers, clinicians, regulators and researchers. For many clinical conditions the efficacy of medical cannabis has not been studied thoroughly, with the majority of studies being cross-sectional surveys or longitudinal cohort studies [1, 2, 38–42]. Such studies: (1) lack comparison groups and therefore provide lower quality evidence than randomised trials, and (2) are subject to limitations in interpretation including the potential for patient and/or clinician expectancy/placebo effects and often a ‘survivor bias’ caused by failure to include data from patients who discontinued treatment due to limited effectiveness or unpleasant side effects [48, 49]. For mental health or sleep conditions, the second and third most common conditions respondents treated with medical cannabis, randomised controlled trials are either underpowered, suffer from methodological limitations that hamper meta-analyses, or simply have never been conducted [38–42]. In addition, studies typically exclude patients with multiple comorbidities limiting generalisability of experimental findings to real world patients. There is a need for consensus approaches to undertaking clinical trials with medicinal cannabis products to better allow pooling of data across trials. In addition to their clinical endpoints, future clinical trials should incorporate patient-reported outcome measures that capture ‘what is important to patients’, so as to disentangle the seeming contradiction between what patients report and researchers find.

Whilst we advocate for more and better-conducted research, patients, clinicians, and regulators should address this evidence gap in the ‘here and now’. There is greater need for independent consumer education that is not led by the medical cannabis industry, with more explicit information about safety and effectiveness of using different medical cannabis products for different conditions. Similarly, in most settings, medical cannabis is prescribed ‘off label’. Healthcare professionals need to consider how they undertake ‘off label’ prescribing of unregistered medicines while abiding by local professional standards [50]. There are important issues surrounding: how prescribers of medical cannabis obtain and document informed consent; the role of published clinical guidance; seeking second opinions with relevant specialists; and understanding medico-legal issues such as insurance. Regulators must also consider the level of autonomy they wish to allow clinicians. For example, in some jurisdictions regulators limit the range of cannabis products or conditions that a doctor can prescribe for, while in others, such as Australia, the regulatory framework leaves this to the discretion of the clinician and patient [3].

Leaving aside the issue of efficacy, our findings suggest that people who are using prescribed medical cannabis may experience health benefits over people who self-medicate with illicit cannabis. Prescribed users reported experiencing significantly fewer side-effects than illicit users, were more likely to consume their medical cannabis via safer routes (oral or vaporised rather than smoking) and consumed significantly less tobacco. Prescribed users were also significantly more certain of the composition of their medical cannabis, exposing them to less risk of harm via under- or overdosing or unknowingly ingesting harmful contaminants.

This iteration of the CAMS survey was the first with a sufficient number of respondents who had used both prescribed and illicit cannabis to allow for an analysis of their preferences. Among these ‘dual-users’, respondents were far more likely to prefer prescribed medical cannabis for its consistency of dose, ease of getting supplies, effectiveness in treating their condition, better side-effects profile, and reduced risk of legal issues. This highlights the benefits to patients of being able to access medical cannabis of known potency through legal channels rather than having to obtain it from illicit sources. The only feature in which dual-users preferred illicit cannabis was with respect to cost, and many users of illicit medical cannabis have cited this as a barrier to seeking a prescription [4, 29, 30]. Interestingly however, these user preferences may not reflect reality, for when we excluded respondents who did not pay for their illicit medical cannabis (e.g. who grew their own or were given cannabis as gifts) prescribed users actually paid less on average per week than illicit users. This departure from earlier CAMS surveys may reflecting the diminishing cost of THC-based medical cannabis products as supply has increased and access has improved, though no studies have yet published reliable figures of how costs have changed over time.

There were several limitations to the CAMS-22 survey that may have biased results. First, self-report inevitably leads to inaccuracies in data, whether due to mistaken recall, lack of insight into effectiveness, fatigue in later sections of the survey, or incorrect reporting of diagnostic conditions and products used. Second, any research method that uses convenience sampling, such as an anonymous online survey, is likely to suffer from selection bias, with people who have had positive experiences with medical cannabis, or who have personal or political agenda promoting medical cannabis, more inclined to respond than those who have had negative experiences or who disapprove of medical cannabis. Such bias can significantly limit the generalisability of any findings [51], hence to what extent our results are representative of the general population of Australian medical cannabis users remains unclear. Third, naturally a higher proportion of the 14% of respondents who were recruited through medical cannabis providers were prescribed users (18% vs 4% illicit users). This imbalance may have biased sampling and affected the generalisability of our results. Finally, the main sources of recruitment have changed with each iteration of CAMS (e.g. CAMS-20 recruited mainly through Twitter whereas CAMS-22 recruited mainly through Facebook), meaning that changes across surveys may reflect changes in the sample rather than trends over time in the broader community.