Cannabinoids for the treatment of mental disorders and symptoms of mental disorders: A systematic review and meta-analysis

Abstract

Background:

Medicinal cannabinoids, including medicinal cannabis, pharmaceutical cannabinoids and their synthetic derivatives, including tetrahydrocannabinol (THC) and or cannabidiol (CBD), have been suggested to have a therapeutic role for certain mental health conditions. The primary objective was to review the evidence for cannabinoids in treating symptoms of depression, anxiety, post-traumatic stress disorder, attention-deficit hyperactivity disorder, Tic/Tourette syndrome, and psychosis, either as the primary condition or secondary to other conditions. Secondary outcomes included quality of life and global functioning.

Methods:

We undertook a systematic review and meta-analysis of published and unpublished studies (1980-2018) using MEDLINE, Embase, PsycINFO, and Cochrane Central Register of Controlled Clinical Trials, clinicaltrials.gov, the EU Clinical Trials Register, and the Australian and New Zealand Clinical Trials Registry. We included randomised controlled trials (RCTs) and non-RCT treatment studies. Two independent reviewers screened all studies and performed data extraction. RCT evidence was synthesised, as odds ratios (ORs) for disorder remission and standardised mean differences (SMDs) for change in symptoms, via random-effects meta-analyses. Evidence quality was evaluated using the Cochrane Risk of Bias and GRADE approaches.

Findings:

A total of k=83 studies (k=40 RCTs, n=3067) were included: k=40 for depression (k=22 RCTs, n=2524), k=31 for anxiety (k=17 RCTs, n=605), k=8 for Tic/Tourette syndrome (k=2 RCTs, n=36), k=4 for attention-deficit hyperactivity disorder (k=1 RCT, n=30), k=12 for post-traumatic stress disorder (k=1 RCT, n=10) and k=11 for psychosis (k=6 RCTs, n=281). Pharmaceutical THC (with or without CBD) improved anxiety symptoms amongst those with other medical conditions (primarily chronic non-cancer pain and multiple sclerosis; SMD=−0.25 [95% confidence interval: −0.49:−0.01]; k=7, n=252). Pharmaceutical THC (with or without CBD) worsened negative symptoms of psychosis in a single study (SMD=0.36 [0.10:0.62]; n=24). Pharmaceutical THC (with or without CBD) did not improve any other primary outcomes but did increase adverse events (OR=1.99 [1.20:3.29]; k=10, n=1495) and withdrawals due to adverse events (OR=2.78 [1.59:4.86]; k=11, n=1621). Very few RCTs examined pharmaceutical CBD or medicinal cannabis.

Interpretation:

There is a lack of evidence that cannabinoids improve depressive disorders and symptoms, anxiety disorders, attention-deficit hyperactivity disorder, Tic/Tourette syndrome, post-traumatic stress disorder, or psychosis. There is very-low-quality evidence that pharmaceutical THC (with or without CBD) leads to a small improvement in symptoms of anxiety amongst those with other medical conditions. There remains insufficient evidence to provide guidance on the use of cannabinoids for mental health conditions within a regulatory framework. More high-quality studies examining the effect of cannabinoids on mental disorders are needed.

Review registration:

PROSPERO CRD42017059372, CRD42017059373, CRD42017059376, CRD42017064996, CRD42018102977

Funding:

Therapeutic Goods Administration, Commonwealth Department of Health; Australian National Health and Medical Research Council, NIH

Introduction

Countries are increasingly allowing cannabinoids to be made available for medicinal purposes, including for the treatment of mental disorders. In this review, based on previous agreed terminology¹, we use the term ‘medicinal cannabinoids’ as an umbrella term to encompass all plant-derived and synthetic derivatives. We use ‘medicinal cannabis’ to refer to any part of the cannabis plant and plant material, such as buds, leaves or full plant extracts (e.g., cannabis sativa). We use ‘pharmaceutical cannabinoids’ to refer to pharmaceutical-grade medicinal extracts with defined and standardised THC and THC/CBD content (e.g., tetrahydrocannabinol [THC], cannabidiol [CBD] extract, or THC:CBD combinations (nabiximols)) and synthetic cannabinoid derivatives¹. Given increasing interest in CBD products for a range of conditions, we also separately grouped studies using pharmaceutical CBD only.

After chronic non-cancer pain (CNCP), mental health is one of the most common reasons for accessing medicinal cannabinoids². In terms of biological plausibility, there is a potential role of the endocannabinoid system (CB1 receptors) in reducing depressive and stress symptoms³ and the emotional and cognitive features of post-traumatic stress disorder (PTSD)⁴. CBD has been proposed as an effective short-term treatment for individuals experiencing social anxiety disorder⁵. Medicinal cannabinoids have been reported to reduce tics in Tourette Syndrome⁶. Many surveys report elevated rates of cannabis use among people living with depression, anxiety, PTSD, and psychosis, and self-medication of symptoms is suggested to be a driver of some of this use^7,8.

Given the interest in using medicinal cannabinoids for these purposes, it is important to thoroughly review the evidence to inform policy and clinical decisions. Previous systematic reviews have been limited in their coverage of mental disorders, study designs, and use of quantitative synthesis (i.e., meta-analysis). A 2015 review by Whiting and colleagues⁹, which included 5 RCTs of mental disorders, found no effect on psychosis or depression, but noted low-quality evidence for some improvement in Tourette syndrome and anxiety. A 2016 review by Wilkinson and colleagues¹⁰ included 40 trials (RCTs and observational studies) of medicinal cannabinoids for PTSD, Tourette syndrome, and Alzheimer’s disease. No RCTs were identified for any condition and no meta-analysis was conducted, thus authors could not draw conclusions regarding efficacy. Crucially, highly prevalent disorders for which medicinal cannabinoids are often sought – such as depression, anxiety, and psychosis – were not included. The 2017 National Academy of Sciences (NAS) review¹¹ reported beneficial effects for Tourette syndrome, anxiety, and PTSD, and no impact on psychosis or depression; however, this review was based largely on findings reported by Whiting and colleagues⁹. There remains no single review that has considered: all types of evidence; the potential differential effects of different types of medicinal cannabinoids; and the safety of using cannabinoids for mental disorders. Disentangling the evidence for different types of cannabinoids for specific mental disorders is needed to direct research efforts and provide clinical guidance.¹

The aim of this systematic review and meta-analysis was to examine evidence for all types of medicinal cannabinoids and all study designs (controlled and observational) to determine:

1. The impact of medicinal cannabinoids on:

   1. Primary outcomes including remission from and symptoms of depression, anxiety, PTSD, and psychosis; and symptoms of attention-deficit hyperactivity disorder (ADHD) and Tic/Tourette syndrome; either as the primary disorder or secondary to other disorders;
   2. Secondary outcomes including global functioning, quality of life, patient or caregiver impression of change; and

2. The safety of medicinal cannabinoids for mental health, including all-cause, serious and treatment-related adverse events and study withdrawals.

Methods

This review was registered on PROSPERO (depression: CRD42017059376; anxiety: CRD42017059373; PTSD: CRD42017064996; ADHD and Tic/Tourette syndrome: CRD42017059372; psychosis: CRD42018102977).

Search strategy

We searched MEDLINE, Embase, PsycINFO, the Cochrane Central Register of Controlled Clinical Trials (CENTRAL), and the Cochrane Database of Systematic Reviews via Ovid from 1980 to May 2018. Five separate searches were conducted to identify studies that evaluated the efficacy of plant-based and pharmaceutical cannabinoids in reducing or treating symptoms of depression, anxiety, PTSD, ADHD and Tic/Tourette syndrome, and psychotic disorders. The detailed search strategies for each condition are shown in Appendix A. To identify ongoing or unpublished studies, we additionally searched clinicaltrials.gov, the EU Clinical Trials Register and the Australian and New Zealand Clinical Trials Registry using keywords ‘cannabis’, ‘cannabinoids’, ‘marijuana’ and each of the five mental disorders. We also hand searched reference lists of included studies and topical reviews for potentially relevant articles. No restrictions were placed on language, publication status, or type.

Inclusion and exclusion criteria

Types of populations:

We included studies examining medicinal cannabinoids for adults aged ≥ 18 years for the purpose of treating depression, anxiety, ADHD and Tic/Tourette syndrome, PTSD and psychosis either as the primary condition or as secondary to other medical conditions (such as CNCP). We chose to review these specific conditions because they are widely cited as reasons for accessing medicinal cannabinoids² and have onset in young adulthood and thus have impact across the lifespan¹². We did not include neurocognitive disorders such as dementia as they have a markedly different pathophysiology and have onset later in life and thus warrant a separate, specific review.

Types of cannabinoids:

We considered studies examining any type and formulation of medicinal cannabinoid: tetrahydrocannabinol; cannabidiol; combination tetrahydrocannabinol + cannabidiol; cannabis sativa; and other cannabinoids e.g. tetrahydrocannabinolic acid, cannabidiolic acid, cannabidivarin, and the synthetic delta-9-tetrahydrocannabinol formulations nabilone and dronabinol. We categorised these into pharmaceutical grade THC (with or without CBD; labelled here as THC:CBD), pharmaceutical grade CBD, and medicinal cannabis.

Types of study designs:

As per existing reviews examining the efficacy of medicinal cannabinoids for CNCP¹³ and epilepsy¹⁴, we included both experimental and observational study designs, that is, randomised controlled trials (RCTs), non-RCTs, quasi-experimental, before and after studies, prospective and retrospective cohort studies, case control studies, analytical cross-sectional studies, observational studies, self-report, and N-of-1 studies. This approach allows researchers, clinicians, and policymakers to map current research activity and to identify knowledge gaps. For studies with a comparison group, we considered any type of comparator, including placebo, waitlist controls, and other interventions. We excluded reviews of mechanisms of cannabinoid systems, commentary articles, and clinical overviews that did not assess and synthesise individual studies.

Types of outcomes:

To be eligible for inclusion an article had to report on at least one primary outcome, that is, either mental disorder remission or change in mental disorder symptomology (see Table 1 for the full list of outcomes).

Table 1:

Primary and secondary outcomes considered for each of the disorders

	Primary Outcomes	Secondary outcomes
Depression	• Remission – absence of a depressive disorder diagnosis using validated scales • Change in depressive symptoms using self-report scales or items	• Measures of global functioning – including quality of life, patient or caregiver global impression of change, and satisfaction with treatment
Anxiety	• Remission – the absence of an anxiety disorder diagnosis using validated scales • Change in anxiety symptoms using self-report scales or items	• Measures of global functioning – including quality of life, patient or caregiver global impression of change, and satisfaction with treatment
ADHD	• Change in ADHD symptom-related behaviour using standardised measures – any context • Change in ADHD symptom-related behaviour in the home using standardised measures • Change in ADHD symptom-related behaviour in school using standardised measures	• Measures of global functioning – including quality of life, patient or caregiver global impression of change, and satisfaction with treatment • Change in cardiovascular effects • Weight changes
Tic/Tourette syndrome	• Change in Tic severity measured using standardised measures	• Measures of global functioning – including quality of life, patient or caregiver global impression of change, and satisfaction with treatment • Change in cardiovascular effects • Weight changes
PTSD	• Remission – the absence of PTSD diagnosis using valid and reliable clinician-rated scales • Change in severity of self-reported traumatic stress symptoms using self-report scales or items	• Measures of global functioning – including quality of life, patient or caregiver global impression of change, and satisfaction with treatment • Change in severity of depressive symptoms using a standardised measure • Change in severity of anxiety symptoms using a standardised measure • Change in sleep quality • Change in frequency of nightmares
Psychosis	• Whether patients still meet criteria for a diagnosis post-treatment • Change in positive and negative symptoms of psychosis	• Measures of global functioning – including quality of life, patient or caregiver global impression of change, and satisfaction with treatment • Change in cognitive functioning • Measures of emotional functioning – including depression, anxiety, mood, and social skills
All 6 Disorders		• Adverse events (AEs) - all-cause • Serious adverse events (SAEs; as defined by authors) - all-cause • Treatment-related adverse events (TAEs) - all-cause • Study withdrawals - all-cause • Study withdrawals - due to AEs

Study screening and selection

Two reviewers independently examined titles and abstracts using the web-based systematic review program Covidence¹⁵. Relevant articles were obtained in full and assessed for inclusion in the review independently by two authors. Inter-reviewer disagreement was resolved via discussion to reach consensus, with a third reviewer consulted where consensus could not be reached by the two initial reviewers.

Data extraction

Data were extracted by two reviewers using a pre-piloted, standardised data extraction tool in Microsoft Excel. We extracted data on details of the populations; interventions; comparisons; outcomes of significance to the mental disorder (PICO); study methods; cannabinoid dose and route of administration; placement in the therapeutic hierarchy; adverse events and study withdrawals. When data were not reported in full, we contacted authors for additional information. When authors reported multiple analyses (e.g., intention to treat [ITT], available case, or per protocol), we extracted the more conservative with a preference for ITT analyses. We reported AEs according to high-level Medical Dictionary for Regulatory Activities (MedDRA; https://www.meddra.org/) categories. We used Review Manager (RevMan) version 5.3¹⁶ to perform calculations or transformation on available data to impute missing data (e.g., confidence intervals, number of cases) in order to calculate required outcome data (ORs, SMDs).

Primary and secondary outcomes

Table 1 outlines the primary and secondary outcomes for each condition. We planned to examine remission from the target mental disorder (where appropriate) and changes in symptoms of the target mental disorder as the primary outcomes. Secondary outcomes included changes in distal factors related to the mental disorder, including global functioning, cardiovascular effects, weight, and sleep (see Table 1). All-cause, serious, and treatment-related adverse events, as well as all-cause study withdrawals and study withdrawals due to adverse events were examined as secondary outcomes for all disorders.

Assessment of risk of bias and grading of evidence

For RCTs, risk of bias was assessed using the Cochrane risk of bias tool (see Appendix D for further details of the tool used as well as for risk of bias plots)¹⁷, which includes assessment of indicators of selection bias, performance bias, detection bias, attrition bias, and reporting bias. Risk of bias assessments were completed independently by two reviewers. Inter-reviewer disagreement was resolved via discussion to reach consensus, with a third reviewer consulted where consensus could not be reached by the two initial reviewers.

We used the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach to rate the quality of the evidence for each outcome¹⁸. This was conducted by one reviewer and checked by a second reviewer, with disagreements resolved via discussion with two further reviewers. In this approach, RCT evidence is allocated ‘high quality’ initially, but can be downgraded up to three levels to ‘moderate quality’, ‘low quality’, or ‘very low quality’ due to five categories of limitations. High quality indicates we are confident that the true effect is similar to the estimated effect; very low quality indicates that the true effect is likely to be substantially different to the estimated effect. Limitations considered are (1) risk of bias (i.e., whether limitations in the study design and execution would bias the effect estimate), (2) indirectness of evidence (e.g., if effects of cannabinoids on mental health disorders had to be inferred from indirect evidence amongst those without the disorder), (3) inconsistency of results (i.e., high, unexplained heterogeneity) (4) imprecision (i.e., wide confidence intervals, including potentially covering appreciable benefit and harm), and (5) publication bias (i.e., selective publication of studies leading to a systematic bias in the effect estimate).

Data analysis

All analyses were conducted using Review Manager (RevMan) version 5.3¹⁶. Meta-analyses included parallel and cross-over RCTs. Continuous and dichotomous outcomes were pooled as standardised mean differences (SMD) and odds ratios (ORs), respectively, using random-effects, generic inverse variance meta-analyses. A common rule of thumb for interpreting SMDs is: 0.2, 0.5, and 0.8 represent small, medium, and large effects, respectively¹⁹. Heterogeneity was assessed using the I² statistic. I² values of 0-39%, 40-74%, and 75-100% can be considered unimportant, moderate/substantial, and high levels of inconsistency across studies, respectively²⁰.

Analyses were stratified by mental health condition, cannabinoid used (pharmaceutical THC:CBD, pharmaceutical CBD, medicinal cannabis), and comparator used (active, placebo). For each of these, we first pooled the evidence from all eligible RCTs, regardless of population studied. Where applicable (depression and anxiety studies only), we then conducted sensitivity analyses restricted to only those RCTs enrolling participants with the mental health disorder. Where heterogeneity was substantial and sample sizes were sufficient, we conducted exploratory analyses to examine potential reasons for the heterogeneity. Finally, we pooled the evidence across RCTs (regardless of mental health condition) on the incidence of adverse events and withdrawals. Narrative synthesis of results from observational studies was conducted by summarising key results from each study, using the same stratification as for RCTs where possible. For the interested reader, further details on the meta-analytic approach–including methods employed to manage variations in study design and avoid unit-of-analysis errors–are provided in Appendix G.

Role of the funding source

The funders had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report, or in the decision to submit the paper for publication.

Results

Results of searches

The PRISMA flowchart is shown in Figure 1, and the list of studies excluded at the full-text screening stage is listed in Appendix B. Appendix E shows the number of studies according to study designs of eligible studies for each mental health outcome and the characteristics of each individual included study. After screening, there were 83 eligible studies (40 RCTs): 42 for depression^21–62 (22 RCTs), 31 for anxiety^{21–24,26,27,29,32–34,39,40,42–44,46,48,50,53,58,63–73} (17 RCTs), 8 for Tic/Tourette syndrome^{6,43,66,70,74–77} (2 RCTs), 3 for ADHD^6,75,78 (1 RCT), 12 for PTSD^{37,71,72,79–87} (1 RCT) and 11 for psychosis^88–98 (6 RCTs). Appendix C details ongoing and incomplete trials identified in the clinical trials registry.

Figure 1:

PRISMA flowchart

Description of included RCTs

Table 2 summarises the characteristics of included RCTs. By and large, medicinal cannabinoids were investigated as adjuvant medicines. The RCTs were typically very small (with median sample size between 10-39 across mental health outcomes), with short follow-up periods (median trial length was 4-5 weeks). Across disorders, the majority of RCTs examined pharmaceutical THC (with or without CBD; labelled here as THC:CBD); most commonly, these were nabiximols and nabilone. The exception was RCTs of psychosis, which primarily examined pharmaceutical CBD. Very few RCTs examined medicinal cannabis as the treatment.

Table 2:

Summary of randomised controlled trials (RCTs) of medicinal cannabinoids studies for the treatment of mental health

	Depression	Anxiety	ADHD	Tic/Tourette syndrome	PTSD	Psychosis
	N = 23	N = 17	N = 1	N = 2	N = 1	N = 6
Region
North America	8	6	0	0	1	3
Western Europe	12	10	1	2	0	1
Other and multiple regions	3	1	0	0	0	2
Year of study
1980-1990	0	1	0	0	0	0
1991-2000	0	0	0	0	0	0
2001-2010	13	9	0	2	0	2
2011-onwards	10	7	1	0	1	4
Conflict of interest declared?
Yes – none	9	6	0	0	1	2
Yes – potential conflict	9	5	0	1	0	3
Not declared	5	6	1	1	0	1
Participant characteristics
Total number of participants in RCTs	2551	605	30	36	10	281
Median no. participants	34	30	30	18	10	39
Median % women	52.8%	50.0%	36.7%	14.6%	0%	34.6%
Median age	49.8	47.6	NR	33.5	44	34.7
Primary health condition of study participants
Depression	0	0	0	0	0	0
Anxiety disorder	0	3	0	0	0	0
Tourette syndrome	1	2	0	2	0	0
ADHD	0	0	1	0	0	0
PTSD	0	0	0	0	1	0
Psychotic disorder	0	0	0	0	0	6
Multiple sclerosis	7	2	0	0	0	0
Chronic non-cancer pain	10	7	0	0	0	0
Parkinson’s disease	0	0	0	0	0	0
Other	5	3	0	0	0	0
Primary indication
Depression	2	1	0	0	0	0
Anxiety	1	4	0	0	0	2
Analgesia	14	9	0	0	0	0
Tic severity	1	2	0	2	0	0
Sleep	2	2	0	0	0	1
ADHD symptoms	0	0	1	0	0	0
PTSD symptoms	0	0	0	0	1	0
Spasticity	5	1	0	0	0	0
Antipsychotic	0	0	0	0	0	4
% cannabinoid naïve (n studies reporting)	38.5%/10	71%/7	33.3%/1	56.3%/2	NR/1	17.17%/2
Cannabinoid used
Cannabis sativa	5	1	0	0	0	0
THC extract	2	3	0	2	0	1
Nabiximols	7	3	1	0	0	0
THC:CBD extract	1	1	0	0	0	0
Cannabidiol (CBD)	0	2	0	0	0	5
Dronabinol	5	2	0	0	0	0
Nabilone	3	5	0	0	1	0
THC-HS	0	0	0	0	0	0
Unknown	0	0	0	0	0	0
Pharmaceutical grade
Yes	18	15	1	2	2	5
No	4	1	0	0	0	0
Unsure/unknown	1	1	0	0	0	1
Route of administration
Vapourised	2	0	0	0	0	0
Smoked	3	1	0	0	0	0
Oral	10	12	0	2	1	3
Oral mucosal spray	8	4	1	0	0	0
Mixed routes	0	0	0	0	0	0
Not recorded/unclear	0	0	0	0	0	2
Intravenous	0	0	0	0	0	1
Rectal	0	0	0	0	0	0
Median treatment (weeks)	5	4	6	3.1	7	3.5
Place in therapeutic hierarchy
Primary	0	3	1	0	0	1
Adjuvant	20	12	0	2	1	5
Not reported, unclear	3	2	0	0	0	0

Note: THC – Δ-9 tetrahydrocannabinol. CBD – cannabidiol.

In most of the RCTs for depression and anxiety, the primary indication for the cannabinoid was some other medical condition, with chronic non-cancer pain (CNCP), followed by multiple sclerosis, being the most common primary conditions. In studies of other mental health conditions, the mental health outcome was the primary indication for the cannabinoid.

Risk of bias of included studies

A summary of the risk of bias of included studies is provided in Appendix D. Briefly, most RCTs reported adequate randomisation sequence generation and concealment; however, the majority were of unclear or high risk of bias for blinding of participants, personnel and outcome assessors. Most studies had other potential, albeit unclear, sources of bias, such as use of post-hoc analyses and unclear adjustment for cross-over trials.

RCT evidence on the effects of medicinal cannabinoids on symptoms of mental disorders, adverse events, and withdrawals

Results of all meta-analyses of RCTs of cannabinoids for the treatment of mental health are described below and reported in full in Tables 3 (pharmaceutical THC:CBD), Table 4 (pharmaceutical CBD), and Table 5 (medicinal cannabis). Adverse events and withdrawals for each of THC:CBD, CBD, and medicinal cannabis are described below and reported in full in Table 6. Forest plots for primary outcomes are displayed in Appendix F.

Table 3:

Summary of RCT evidence on the use of pharmaceutical THC (THC alone, or THC:CBD preparations)¹ for the treatment of mental health

Disorder	Outcome	Comparator	Studies (participants)	Risk of Bias	Indirectness	Inconsistency	Imprecision	Publication Bias	Pooled SMD [95% CI]a	I2	Favours	GRADE
Depression	Remission from disorder	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in depressive symptoms†	Active	1 (52)	Not serious	Very serious	Serious	Serious	Undetected	0.00 [−0.17, 0.17]	NA	Neither	Very low
	Change in depressive symptoms†	Placebo	12 (1656)	Not serious	Very serious	Serious	Not serious	Likely	−0.05 [−0.20, 0.11]	67%	Neither	Very low
	Change in global functioning	--	0 (0)	--	--	--	--	--	--	--	--	--
Anxiety	Remission from disorder	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in anxiety symptoms†	Active	1 (52)	Not serious	Very serious	Serious	Serious	Undetected	−0.12 [−0.30, 0.05]	NA	Neither	Very low
	Change in anxiety symptoms†	Placebo	7 (252)	Serious	Serious	Serious	Serious	Likely	−0.25 [−0.49, −0.01]	65%	THC:CBD	Very low
	Change in global functioning	--	0 (0)	--	--	--	--	--	--	--	--	--
ADHD	Change in ADHD symptoms - any location†	Placebo	1 (30)	Not serious	Not serious	Serious	Serious	Undetected	−0.67 [−1.41, 0.07]	NA	Neither	Low
	Change in ADHD symptoms - home	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in ADHD symptoms - school	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in global functioning	Placebo	1 (30)	Not serious	Not serious	Serious	Serious	Undetected	0.00 [−0.72, 0.72]	NA	Neither	Low
	Cardiovascular effects	--	0 (0)	--	--	--	--	--	--	--	--	--
	Weight change	Placebo	1 (30)	Not serious	Not serious	Serious	Serious	Undetected	0.14 [−0.58, 0.85]	NA	Neither	Low
Tic/Tourette syndrome	Change in Tic/Tourette symptoms†	Placebo	2 (41)	Not serious	Not serious	Serious	Serious	Undetected	−0.46 [−1.32, 0.40]	68%	Neither	Low
	Change in global functioning	Placebo	2 (41)	Not serious	Not serious	Serious	Very serious	Undetected	−0.84 [−2.10, 0.42]	68%	Neither	Very low
	Cardiovascular effects	--	0 (0)	--	--	--	--	--	--	--	--	--
	Weight change	--	0 (0)	--	--	--	--	--	--	--	--	--
PTSD	Remission from disorder	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in PTSD symptoms	--	0 (0)	--	--	--		--	--	--	--
	Change in global functioning	Placebo	1 (19)	Not serious	Not serious	Serious	Serious	Undetected	−1.13 [−1.48, −0.77]	NA	THC:CBD	Low
	Change in depressive symptoms	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in anxiety symptoms	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in sleep quality	Placebo	1 (19)	Not serious	Not serious	Serious	Serious	Undetected	−0.10 [−0.38, 0.18]	NA	Neither	Low
	Change in nightmare frequency	Placebo	1 (19)	Not serious	Not serious	Serious	Serious	Undetected	-1.11 [−1.46, −0.76]	NA	THC:CBD	Low
Psychosis	Remission from disorder	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in total symptoms	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in positive symptoms†	Placebo	1 (24)	Not serious	Not serious	Serious	Serious	Undetected	−0.20 [−0.45, 0.06]	NA	Neither	Low
	Change in negative symptoms†	Placebo	1 (24)	Not serious	Not serious	Serious	Serious	Undetected	0.36 [0.10, 0.62]	NA	Placebo	Low
	Change in global functioning	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in cognitive function	Placebo	1 (24)	Not serious	Not serious	Serious	Serious	Undetected	1.08 [0.71, 1.45]	NA	Placebo	Low
	Change in emotional functioning	--	0 (0)	--	--	--	--	--	--	--	--	--

Note:

^1.THC:CBD refers to pharmaceutical THC + CBD combinations such as nabiximols. In all comparisons the control group (placebo/active) is the reference group.

^†indicates outcomes for which forest plots are available in Appendix F.

White cells are primary outcomes and shaded cells are secondary outcomes. NA = not applicable.

Table 4:

Summary of RCT evidence on the use of pharmaceutical cannabidiol (CBD) for the treatment of mental health

Disorder	Outcome	Comparator	Studies (participants)	Risk of Bias	Indirectness	Inconsistency	Imprecision	Publication Bias	Pooled SMD [95% CI]a	I2	Favours	GRADE
Depression	--	--	0 (0)	--	--	--	--	--	--	--	--	--
Anxiety	Remission from disorder	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in anxiety symptoms†	Placebo	2 (44)	Not serious	Not serious	Serious	Very serious	Undetected	−0.87 [−2.01, 0.27]	85%	Neither	Very low
	Change in global functioning	--	0 (0)	--	--	--	--	--	--	--	--	--
ADHD	--	--	0 (0)	--	--	--	--	--	--	--	--	--
Tic/Tourette syndrome	--	--	0 (0)	--	--	--	--	--	--	--	--	--
PTSD	--	--	0 (0)	--	--	--	--	--	--	--	--	--
Psychosis	Remission from disorder	--	0 (0)	--	--	--	--	--	--	--	--	--
	Change in total symptoms†	Active	1 (39)	Not serious	Not serious	Serious	Serious	Undetected	−0.02 [−0.65, 0.60]	NA	Neither	Low
	Change in total symptoms†	Placebo	2 (122)	Not serious	Not serious	Serious	Serious	Undetected	0.05 [−0.50, 0.61]	52%	Neither	Low
	Change in positive symptoms†	Active	1 (39)	Not serious	Not serious	Serious	Serious	Undetected	−0.10 [−0.73, 0.53]	NA	Neither	Low
	Change in positive symptoms†	Placebo	2 (122)	Not serious	Not serious	Serious	Serious	Undetected	−0.17 [−0.69, 0.35]	47%	Neither	Low
	Change in negative symptoms†	Active	1 (39)	Not serious	Not serious	Serious	Serious	Undetected	−0.48 [−1.12, 0.16]	NA	Neither	Low
	Change in negative symptoms†	Placebo	2 (122)	Not serious	Not serious	Not serious	Serious	Undetected	0.08 [−0.27, 0.44]	0%	Neither	Moderate
	Change in global functioning	Placebo	1 (86)	Not serious	Not serious	Serious	Serious	Undetected	−0.62 [−1.14, −0.09]	NA	CBD	Low
	Change in cognitive function	Placebo	3 (150)	Not serious	Not serious	Not serious	Serious	Undetected	−0.01 [−0.33, 0.32]	0%	Neither	Moderate
	Change in emotional functioning	Active	1 (39)	Not serious	Not serious	Serious	Serious	Undetected	0.27 [−0.36, 0.90]	NA	Neither	Low
	Change in emotional functioning	Placebo	2 (122)	Not serious	Not serious	Serious	Serious	Likely	0.10 [−0.49, 0.69]	57%	Neither	Very low

Note:

^†indicates outcomes for which forest plots are available in Appendix F.

White cells are primary outcomes and shaded cells are secondary outcomes. NA = not applicable; CBD = cannabidiol. In all comparisons the control group (placebo/active) is the reference group.

Table 5:

Summary of RCT evidence on the use of medicinal cannabis for the treatment of mental health

Disorder	Outcome	Comparator	Studies (participants)	Risk of Bias	Indirectness	Inconsistency	Imprecision	Publication Bias	Pooled SMD [95% CI]a	I2	Favours	GRADE
Depression	Remission from disorder	--	0 (0)	--	--	--		--	--	--	--	--
	Change in depressive symptoms†	Placebo	1 (42)	Not serious	Very serious	Serious	Serious	Likely	−0.14 [−0.33, 0.05]	NA	Neither	Very low
	Change in global functioning	--	0 (0)	--	--	--		--	--	--	--	--
Anxiety	--	--	0 (0)	--	--	--		--	--	--	--	--
ADHD	--	--	0 (0)	--	--	--		--	--	--	--	--
Tic/Tourette syndrome	--	--	0 (0)	--	--	--		--	--	--	--	--
PTSD	--	--	0 (0)	--	--	--		--	--	--	--	--
Psychosis	--	--	0 (0)	--	--	--		--	--	--	--	--

Note:

^†indicates outcomes for which forest plots are available in Appendix F.

White cells are primary outcomes and shaded cells are secondary outcomes. NA = not applicable. In all comparisons the control group (placebo/active) is the reference group.

Table 6:

Summary of RCT evidence on the impact of medicinal cannabinoids on adverse events and withdrawals

Treatment	Outcome	Comparator	Studies (participants)	Risk of Bias	Indirectness	Inconsistency	Imprecision	Publication Bias	Pooled OR [95% CI]	I2	More AE/Withdrawals occurred in…	GRADE
THC:CBD	Adverse events
	AEs – all cause†	Active	1 (60)	Not serious	Serious	Serious	Very serious	Undetected	1.59 [0.57, 4.45]	NA	Neither	Very low
	AEs – all cause†	Placebo	10 (1495)	Not serious	Serious	Serious	Not serious	Undetected	1.99 [1.20, 3.29]	59%	THC:CBD	Low
	SAEs – all cause	Placebo	4 (954)	Not serious	Serious	Not serious	Serious	Undetected	1.29 [0.94, 1.77]	0%	Neither	Low
	TAEs – all cause	Placebo	2 (385)	Not serious	Serious	Not serious	Serious	Undetected	1.32 [0.79, 2.20]	0%	Neither	Low
	Withdrawals
	Withdrawals – all cause	Placebo	15 (2299)	Not serious	Serious	Not serious	Serious	Likely	1.51 [0.96, 2.36]	42%	Neither	Very low
	Withdrawals – due to AEs†	Active	2 (252)	Not serious	Serious	Not serious	Serious	Undetected	0.54 [0.17, 1.68]	0%	Neither	Low
	Withdrawals – due to AEs†	Placebo	11 (1621)	Not serious	Serious	Not serious	Not serious	Undetected	2.78 [1.59, 4.86]	22%	THC:CBD	Moderate
CBD	Adverse events
	AEs – all cause†	Placebo	1 (88)	Not serious	Not serious	Serious	Serious	Undetected	0.97 [0.40, 2.33]	NA	Neither	Low
	SAEs – all cause	Placebo	1 (88)	Not serious	Not serious	Serious	Very serious	Undetected	0.34 [0.01, 8.60]	NA	Neither	Very low
	TAEs – all cause	Placebo	1 (88)	Not serious	Not serious	Serious	Serious	Undetected	1.06 [0.39, 2.87]	NA	Neither	Low
	Withdrawals
	Withdrawals – all cause	Active	1 (42)	Not serious	Not serious	Serious	Very serious	Undetected	3.33 [0.32, 34.99]	NA	Neither	Very low
	Withdrawals – all cause	Placebo	1 (88)	Not serious	Not serious	Serious	Very serious	Undetected	1.61 [0.26, 10.16]	NA	Neither	Very low
	Withdrawals – due to AEs†	Placebo	1 (88)	Not serious	Not serious	Serious	Very serious	Undetected	1.05 [0.06, 17.30]	NA	Neither	Very low
Cannabis	Adverse events
	AEs – all cause	--	0 (0)	--	--	--	--	--	--	--	--	--
	SAEs – all cause	--	0 (0)	--	--	--	--	--	--	--	--	--
	TAEs – all cause	--	0 (0)	--	--	--	--	--	--	--	--	--
	Withdrawals
	Withdrawals – all cause	Placebo	3 (209)	Serious	Serious	Not serious	Very serious	Undetected	1.41 [0.51, 3.88]	7%	Neither	Very low
	Withdrawals – due to AEs	--	0 (0)	--	--	--	--	--	--	--	--	--

Note:

^†indicates outcomes for which forest plots are available in Appendix F.

NA = not applicable. THC = Δ-9 tetrahydrocannabinol. THC:CBD includes pharmaceutical THC alone and pharmaceutical THC + CBD combinations. CBD = pharmaceutical cannabidiol. In all comparisons the control group (placebo/active) is the reference group.

Depression

Pharmaceutical THC:CBD did not significantly improve symptoms of depression, compared to either active comparators⁴⁸ or placebo comparators^{22,25,31,39,42,43,49,50,53,55,59,61} (Table 3). The evidence GRADE was very low, in part due to indirectness because none of the included RCTs included participants with a primary diagnosis of depression; most included participants with multiple sclerosis. Following reviewer suggestion, we conducted an exploratory analysis to examine whether length of follow up contributed to the substantial heterogeneity seen (67%). One study⁴³ administered pharmaceutical THC:CBD and assessed participants on a single day, whereas the remaining used longer treatment and follow-up periods (2-15 weeks). Removing the single shorter study made minimal difference to the effect size and heterogeneity (SMD=−0.05, 95%CI −0.22:0.13; k=11, n=1632; I²=70%).

No RCTs examining CBD for depression outcomes were identified. A single, small RCT examining medicinal cannabis for depression outcomes amongst participants with CNCP found no change in depressive symptoms compared to placebo (Table 5)⁵⁷.

Anxiety

Pharmaceutical THC:CBD led to significantly greater reductions in anxiety symptoms than did placebos (SMD=−0.25, 95%CI −0.49:−0.01; k=7, n=252; I²=65%)^{22,39,42,43,50,53,73}, with no difference seen in the single, small study that used an active comparator (Table 3)⁴⁸. The evidence GRADE was very low, in part because none of the studies included participants with a primary diagnosis of anxiety; most included participants with CNCP or multiple sclerosis. Reporting bias also contributed to the very low GRADE rating; outcomes of three RCTs could not be included in this synthesis due to incomplete data reporting^44,65,67. One showed a beneficial effect of pharmaceutical THC:CBD over placebo, whereas the other two showed no significant difference. Given the confidence intervals of the effect are close to zero (−0.49:−0.01), had it been possible to include these studies it is likely that the benefit of pharmaceutical THC:CBD over placebo would no longer be significant.

We conducted an exploratory analysis to check whether varying lengths of follow up contributed to the substantial heterogeneity seen in the pharmaceutical THC:CBD versus placebo comparison (65%). One study⁴³ administered pharmaceutical THC:CBD and assessed participants on a single day, whereas the remaining used longer treatment and follow-up periods (3-12 weeks). Removing the single shorter study reduced the heterogeneity to an unimportant level and the beneficial effect of pharmaceutical THC:CBD remained significant (SMD=−0.34, 95%CI −0.53:−0.14; k=6, n=228; I²=36%).

Two studies examined the effect of CBD – both in participants with social anxiety – and did not find a significant improvement in anxiety symptoms compared to placebo (Table 4)^63,64. No RCTs examined the impact of medicinal cannabis on anxiety outcomes.

ADHD

The single, small identified RCT for ADHD compared pharmaceutical THC:CBD with placebo amongst participants with ADHD⁷⁸. No significant effect was seen on the primary outcome, ADHD symptoms (Table 3). Of the secondary outcomes, the study also demonstrated no significant effect of pharmaceutical THC:CBD versus placebo on global functioning or weight change. No studies examined the impact of CBD or medicinal cannabis on ADHD outcomes.

Tic/Tourette syndrome

The two identified, small RCTs for Tic/Tourette syndrome compared pharmaceutical THC:CBD with placebo amongst participants with Tic/Tourette syndrome^43,70. The pooled effect from these two, small studies demonstrated no significant benefit of pharmaceutical THC:CBD compared to placebo on Tic/Tourette symptoms (Table 3). Similarly, no significant effect was seen on the secondary outcome, global functioning. No studies examined the impact of CBD or medicinal cannabis on Tic/Tourette syndrome outcomes.

PTSD

A single, small RCT with participants with PTSD was identified; this RCT did not report either of our primary outcomes⁸². Of the secondary outcomes, this study found a significant benefit of pharmaceutical THC:CBD compared to placebo in improving global functioning and nightmare frequency, and no significant effect on sleep quality (Table 3). No studies examined the impact of CBD or medicinal cannabis on PTSD outcomes.

Psychosis

A single, small RCT reported on the use of pharmaceutical THC:CBD amongst participants with psychosis⁹⁰. This study found no significant change in positive symptoms (Table 3) but a worsening of negative symptoms (SMD=0.36, 95%CI 0.10:0.62; n=24), compared to placebo. Of the secondary outcomes, this study also found that pharmaceutical THC:CBD worsened cognitive functioning (SMD=1.08, 95%CI 0.71:1.45; n=24).

The remaining included psychosis RCTs examined CBD. Across the 1-2 studies that reported on primary outcomes, CBD did not significantly improve total symptoms, positive symptoms, or negative symptoms, compared to placebo^89,96 or active⁹⁴ comparators (Table 4). Of the secondary outcomes, CBD led to an improvement in global functioning compared to placebo in the single study reporting this outcome (SMD=−0.62, 95%CI −1.14:−0.09; n=86)⁹⁶, but did not significantly improve cognitive or emotional functioning^89,92,94,96.

No studies examined the impact of medicinal cannabis on psychosis outcomes.

Adverse events and withdrawals

We pooled adverse events and study withdrawals from all RCTs (Table 6). Pharmaceutical THC:CBD led to significantly more adverse events (OR=1.99, 95%CI 1.20:3.29; k=10, n=1495; I²=59%) and withdrawals due to adverse events (OR=2.78, 95%CI 1.59:4.86; k=11, n=1621; I²=22%) than did placebos. The evidence GRADE was low to moderate, due to inconsistency and indirectness (i.e., participants in most of the analysed studies did not have a mental disorder). It is estimated that one additional participant would experience an adverse event for every 7 (95%CI 5:25) participants treated with pharmaceutical THC:CBD (number needed to treat to harm). Further, one additional participant would withdraw due to an adverse event for every 14 (95%CI 7:39) participants treated with pharmaceutical THC:CBD. No significant differences between pharmaceutical THC:CBD and comparators were seen on serious adverse events, treatment-related adverse events, or all-cause withdrawals.

Very few RCTs examined adverse events and withdrawals due to CBD or medicinal cannabis, and these found no significant increases compared to active and placebo comparators (Table 6).

Observational evidence on the effects of medicinal cannabinoids on symptoms of mental disorders

The findings of all included observational studies are detailed in Appendix E. Here we summarise the findings of studies in which mental health was the primary indication in open-label or prospective cohorts. There were no open-label or prospective cohort studies in which depression was the primary outcome; there were 10 observational studies where depression was a secondary outcome in CNCP or multiple sclerosis patients in open-label (k=7) and prospective cohort studies (k=3). There were eight open-label and prospective cohort studies that reported on anxiety outcomes. Anxiety was a primary outcome in only one study of n=5⁶⁷, which found that nabilone significantly reduced anxiety. There were no open-label or observational studies for ADHD or Tic/Tourette syndrome. There were two open-label and two prospective cohort studies where PTSD was the primary outcome; three studies involved cannabis and one, THC extract. Three studies found reductions in PTSD symptoms^83,85,86 and one found that PTSD symptoms worsened with cannabis use in people with PTSD and comorbid mental health disorder⁸⁷. There was one open-label study where psychosis was the primary outcome, which found that CBD reduced psychosis symptoms⁹⁷.

Discussion

To our knowledge this is the most comprehensive systematic review examining the available evidence for medicinal cannabinoids in treating mental disorders and symptoms. There is a notable lack of high-quality evidence where mental disorders are the primary target of treatment, and most evidence is derived from studies where mental disorders are secondary to another medical condition, commonly CNCP and multiple sclerosis. Most of the included studies were conducted among persons where depression or anxiety was secondary to another medical condition, and of these we found no impact of pharmaceutical THC (with or without CBD; THC:CBD) on depression symptoms, and a small reduction in anxiety symptoms. Of the few studies in which participants had an anxiety disorder, we did not see a significant benefit of CBD on symptoms of anxiety. Single studies found that pharmaceutical THC:CBD improved global functioning in PTSD and pharmaceutical CBD improved global functioning in psychosis. Across the small numbers of included studies, we did not find evidence that any type of cannabinoid significantly improves primary outcomes of ADHD, Tic/Tourette syndrome, PTSD, or psychosis. In fact, we found evidence that pharmaceutical THC:CBD worsened negative symptoms of psychosis.

Cannabinoids are often advocated for as a treatment of various mental health conditions. It is likely that countries that allow medicinal cannabinoid use will see increased demand for such use. Clinicians and consumers need to be aware of the limited quality and quantity of evidence on the effectiveness and the potential risk for adverse events. Most studies are based on pharmaceutical cannabinoids, rather than medicinal cannabis, but plant products are most often used by those using cannabinoids for medicinal purposes in the USA⁸. Although there are 16 trials underway to examine the effectiveness of pharmaceutical CBD for specific conditions, including seven in psychosis, to date there are very few or no clinical studies examining the effectiveness of CBD for depression, anxiety, Tic/Tourette syndrome or ADHD (see Appendix C).

The risk of adverse outcomes among those using medicinal cannabis products is indicated by a large body of research on the adverse effects of non-medical cannabis use. This suggests that cannabis use can increase the occurrence of depression, anxiety, and psychotic symptoms^11,99–103. The evidence of cannabis’ risks is not derived solely from observational studies of people using cannabis non-medically. For example, there is experimental evidence, using a double-blind, randomised, placebo-controlled and crossover design, of the acute effects of smoked cannabis (containing 13% THC) on psychosis symptoms, which found that cannabis increased risk of acute psychotic symptoms¹⁰⁴. Additionally, young adults (the age group at greatest risks of depression, anxiety, and psychosis) who use cannabis daily over extended periods are at risk of developing dependence upon cannabis⁹⁹. These risks, and the limitations of existing evidence, need to be weighed when considering using medicinal cannabinoids to treat symptoms of common mental disorders. Those who decide to proceed should be carefully monitored for positive and negative mental health effects of using medicinal cannabinoids.

Limitations and future directions

The strengths of our review included our comprehensive search strategies (including clinical trials registries); consideration of the full range and potential distinct effects of different types of cannabinoids; and range of outcomes considered. Compared to previous reviews we identified more studies (e.g., for psychosis we identified six RCTs vs. two in a previous review⁹). Nonetheless our analyses and conclusions are necessarily limited by the small amount of available data, small study sizes, and heterogeneity of findings across studies. Small study sizes are of particular concern as it has previously been identified that effects are larger in small studies of medicinal cannabinoids for CNCP¹³. It is also important to consider that a number of independent analyses were conducted and hence may not retain significance if adjustment for multiple comparisons is made. However, there is no recommended approach for addressing multiplicity in systematic reviews, and we attempted to minimise this by: choosing few primary outcomes, keeping subgroups to a minimum, and testing effects at a single time-point only^105,106. There have been few RCTs, typically of very small size, conducted to date, so the lack of significant effects for ADHD, Tic/Tourette syndrome could well reflect the limited evidence base. Studies of medicinal cannabinoids primarily for people diagnosed with depression and anxiety are lacking. It is possible that the reductions in anxiety symptoms identified in this review may have been due to improvements in the primary medical condition (CNCP or multiple sclerosis). It is crucial that future research focuses on the effectiveness of cannabinoids in patients diagnosed with primary depression and anxiety.

Conclusions

There is increasing use of pharmaceutical cannabinoids and medicinal cannabis to treat symptoms of mental disorders. This is the most comprehensive review of the evidence to date, including both randomised controlled trials and observational studies of depression, anxiety, attention-deficit hyperactivity disorder, Tic/Tourette syndrome, post-traumatic stress disorder, and psychosis. It found very little evidence on the effectiveness of pharmaceutical CBD or medicinal cannabis for the treatment of any of these mental disorders. There was some very-low-quality evidence on the use of pharmaceutical THC (with or without CBD) in treating anxiety symptoms amongst those with other medical conditions, such as chronic non-cancer pain and multiple sclerosis. We need high-quality, randomised controlled trials to properly assess the effectiveness and safety of medicinal cannabinoids, compared to placebo and standard treatments, for the treatment of mental disorders. This evidence is essential before clinical guidelines can be provided on the medicinal use of cannabinoids for these disorders. In light of the paucity of evidence and the lack of good quality evidence, and the known risk of cannabinoids, the use of cannabinoids as treatments for mental health disorders cannot be justified at this time.

Research in context.

Evidence before this study

We searched PubMed up to 12^th July 2019 for reviews of cannabis use and mental health using the Medical Subject Headings (MeSH) terms (((“medical marijuana”[MeSH Terms] OR (“medical”[All Fields] AND “marijuana”[All Fields]) OR “medical marijuana”[All Fields] OR (“medical”[All Fields] AND “cannabis”[All Fields]) OR “medical cannabis”[All Fields]) AND (“mental health”[MeSH Terms] OR (“mental”[All Fields] AND “health”[All Fields]) OR “mental health”[All Fields])) AND Review[ptyp]); this led to 152 results, of which 9 were relevant reviews (or summaries of reviews, as in the case of the US National Academies of Science (NAS)) of cannabis or cannabinoids for mental health problems.

The different reviews included varied study designs to examine the impacts of cannabinoids on mental disorders; some concentrated on cross-sectional studies, others were limited to randomised controlled trials (RCTs), and some further limited this to only studies where the mental health symptoms were the primary indication for the cannabinoid. Some reviews pooled studies quantitatively on one outcome for a given mental health condition (Whiting), but other features of their eligibility criteria and date of the publication meant that there were very few studies included (e.g., zero for depression, one for anxiety, two for psychosis). All reviews agreed that the evidence was limited but in many instances some concluded that no data yet existed for some mental health outcomes (e.g., depression).

No previous reviews defined a priori both primary and secondary outcomes of cannabinoids used for different mental health symptoms, nor systematically compiled both RCT and observational study designs. Most described potential adverse outcomes of cannabinoid use by relying on evidence from studies of people with recreational cannabis use or generally pooling adverse events from any study of medicinal cannabinoids, rather than specifically extracting and pooling data on adverse events and treatment withdrawals. Reviews varied in the clarity with which the specific cannabinoids were documented and the characteristics of the study populations and the studies themselves were extracted and reported.

Added value of this study

Our study represents the most up to date and detailed review of evidence for cannabinoids for mental health symptoms. It pre-specified primary and secondary outcomes to examine for each mental health condition, we included studies where the condition was primary or secondary, we systematically collated non-RCT evidence, and we pooled all outcomes and adverse event data quantitatively wherever possible. We also made clear which cannabinoids were studied, where the data and gaps were across primary and secondary outcomes.

We concluded that there is very-low-quality evidence for the effectiveness of cannabinoids in improving symptoms of anxiety. There is a lack of evidence to suggest that cannabinoids improve depressive disorders, symptoms of depression, anxiety disorders, attention-deficit hyperactivity disorder, Tic/Tourette syndrome, post-traumatic stress disorder, or psychosis.

Implications of all the available evidence

Our findings have direct policy relevance. In countries where cannabis and cannabinoids are being made available for medicinal use, and in which mental health problems are a common reason for requesting access to cannabinoids for medicinal purposes, this review makes clear where the evidence exists and the quality of such evidence. This review also makes clear a real need for investment of high-quality research efforts to study the impact of different cannabinoids upon a range of outcomes for people with mental health disorders.