Cannabis-based medicinal products (CBMPs) are a collective term to describe a preparation or other product that contains cannabis or its derivatives for medicinal use in humans [1]. There are significant barriers to the integration of CBMPs within treatment pathways including ongoing stigma, cost, education, complex pharmacology and a paucity of evidence to inform international and national guidelines [2, 3]. Limited evidence, does, however, support the role of CBMPs in conditions such as chronic pain, neurological disorders, and psychiatric disease [4]. There is also growing evidence of side effects and how the severity and incidence of side effects may differ between patients [4]. The quality of evidence, however, is often insufficient in the opinion of insurers, regulators, and guideline bodies [5].

The National Institute for Health and Care Excellence in the UK has only recommended licensed CBMPs for a limited range of indications [6]. Changes to scheduling as recommended by the World Health Organisation, and within individual countries, recognises the potential medicinal value of cannabis and removes barriers for clinical and research use [1, 7]. However, widespread stigma, complex pharmacology, funding, and challenges in sustaining adequate supply of consistent products continue to act as barriers for clinical research.

Randomised controlled trials (RCTs) are necessary and should continue to be the standard against which medical evidence is upheld. However, they are expensive, time consuming and subject to their own limitations [8]. Whilst these are awaited, there is a requirement to generate evidence of potential benefits and harms to inform policy and clinical practice.

RCTs are not infallible—they are expensive and time consuming. Globally $100 billion USD is spent on biomedical research [9]. In the UK, the National Institute for Health Research (NIHR) provides £80 million GBP in funding for clinical trials [10]. Yet, their narrow scope can lack ecological validity to real-world circumstances and therefore lack generalisability in more diverse populations. There are also specific barriers to conducting RCTs using CBMPs.

Complex Pharmacology

In addition to cannabidiol (CBD) and (−)-trans-Δ⁹-tetrahydrocannabinol (THC) there are over 140 cannabinoids, as well as flavonoids, terpenes, and other compounds within the flower of different cannabis plants [8]. These can each potentially affect the clinical outcomes observed between CBMPs due to their individual and collective effects [11]. The concentrations of each compound are influenced by the genetics and environment each plant is grown in producing a distinct chemical profile. The result of a clinical trial for one finished pharmaceutical product, therefore, cannot be extrapolated to all CBMPs, due to their heterogeneity. However, current evidence reviews often fail to account for this [12, 13].

The route of administration further affects the pharmacokinetics of CBMPs and the associated outcome of any trial. CBMPs can be administered sublingually, trans-dermally, via inhalation, or orally [14]. This subsequently affects the distribution, biotransformation and elimination of active compounds. Heat exposure and vaporisation of dried flower or extracted oils changes the underlying phytocannabinoid composition compared to the original unprocessed dry flowers, increasing the proportion of decarboxylated cannabinoids [15, 16]. Assessment of efficacy using RCTs in isolation will therefore ultimately fail to identify the most appropriate CBMP for each clinical scenario [17].

Placebo-control

An appropriately blinded assessment against placebo or active therapy is the optimal design for RCTs. It has been difficult to identify a placebo that cannot be distinguished against an active CBMP according to absence of both vasoactive and psychoactive effects, as well as the typical aroma associated with cannabis [15]. This presents a challenge to adequate blinding.

Cost

Production methods and import costs mean that CBMPs are typically expensive, adding further to high research costs [18]. Research has therefore focused on compounds under patent as opposed to generic CBMPs where research outcomes fail to provide a similar return on investment for licensed producers and pharmaceutical companies. Historically, clinical trials on CBMPs were funded privately, which may be associated with potential reporting biases [19].

RCTs are possible with CBMPs; however, the above issues present legitimate challenges. In many chronic diseases there is a need for novel therapeutics and CBMPs are therefore being utilised based on best available evidence. Due to the challenges in developing CBMPs through a traditional drug development pipeline, the exploration of its utility should not be limited to traditional methods. It is important that we capture a suite of real-world evidence (RWE) to inform prescribing guidelines, regulations, and clinical trials. By leaning on RWE there is an opportunity to improve the quality and design of RCTs and clinical evidence in general, via a top-down approach [20].

RWE is defined as evidence derived from health data sourced from non-interventional studies, registries, electronic health records and insurance data as opposed to the highly controlled setting of RCTs [21]. There is an abundance of this unstructured data, however, the necessary frameworks and governance are needed for the application of this data [22]. It is currently used extensively to monitor post-approval pharmacovigilence [23]. There is clear evidence of benefit in using population-based data to detect safety events associated with specific medications to implement restrictions to reduce harm [21].

Consistent use of RWE to aid regulatory decision making is yet to be normalised, but the promise is apparent [21]. Recently, regulator-supported initiatives have highlighted the desire to incorporate RWE into licensing and guidelines, developing a framework which can incorporate its insights into decisions regarding safety and effectiveness [21, 22]. It is important that studies standardise their methodology according to those set out by regulatory authorities to ensure research has the greatest impact [21, 22]. Moreover, they should seek to directly address questions set out by governing bodies as areas where there is insufficient research [24].

NHS England and NHS improvement published a review on the barriers to accessing CBMPs in the UK [3]. Their recommendations included the need for the collection of structured data, and the development of methods to further support the generation of new evidence, for patients who cannot enrol onto relevant RCTs.

RWE is already being incorporated into the scientific literature on cannabis (Table 1). Early examples utilised state-level records to examine the effects of cannabis laws on opioid misuse. Subsequently there have been examples of online and self-administered survey tools analysing national outcomes. More recently there has been a focus on collecting evidence from clinical registries and databases with evidence generated from patient-reported outcome measures and long-term pharmacovigilance.