Across 79 randomized trials and 6,462 participants, cannabinoids showed moderate evidence for chronic pain, MS spasticity, and chemotherapy nausea — but insufficient evidence for most other conditions patients use cannabis to treat.
Read this if you've heard “cannabinoids work for everything” and want to see which symptoms beat placebo in 79 randomized trials, and by how much.
47% vs 20%complete chemo nausea and vomiting response with cannabinoids vs placebo (3 trials)
The Backstory
By 2015, the gap between public confidence in medical cannabis and the actual clinical evidence had become embarrassing. Millions of patients were using cannabinoids for pain, nausea, anxiety, and dozens of other conditions. Dispensaries were booming. Advocacy organizations cited "thousands of studies." But nobody had systematically asked the obvious question: across all the randomized controlled trials — the gold standard of medical evidence — what do cannabinoids actually work for?
The Swiss Federal Office of Public Health commissioned Penny Whiting's team to find out. The answer, published in JAMA, was the most comprehensive assessment of cannabinoid medicine ever assembled — and it pleased almost nobody.
The Review
Whiting's team searched 28 databases — not just PubMed and MEDLINE, but dozens of specialized repositories — for every randomized controlled trial of cannabinoids for any medical indication, from the earliest databases through April 2015. They included trials of all cannabinoid formulations: smoked cannabis, pharmaceutical THC (dronabinol), synthetic THC (nabilone), THC:CBD spray (nabiximols/Sativex), and oral CBD.
They found 79 trials involving 6,462 participants. Only four were rated at low risk of bias.
Evidence Quality
The State of Cannabinoid Clinical Evidence in 2015
79
RCTs found
across all indications worldwide
4
Low risk of bias
only 5% met the gold standard
6,462
Total participants
across all 79 trials
28
Databases searched
the most comprehensive search to date
Whiting et al. (2015), JAMA 313(24):2456-2473
That last number is worth pausing on. By 2015, tens of millions of people were using cannabinoids therapeutically. The total number of patients ever enrolled in a randomized trial of any cannabinoid for any condition: 6,462. For context, a single large pharmaceutical trial for a new blood pressure drug might enroll more participants than the entire global history of cannabinoid RCTs combined.
What The Evidence Showed
The pattern was clear: moderate evidence for three conditions (pain, spasticity, chemotherapy nausea), low-quality evidence for a few more, and insufficient evidence for most of the conditions that patients were actually using cannabis to treat. The effects that did reach significance were often modest — a pain reduction of 0.46 points on a 10-point scale is real but not dramatic.
The Pain Numbers, Honestly
The pillar article describes roughly 30% of patients achieving meaningful pain relief with cannabinoids versus 20% on placebo. That gives a number needed to treat (NNT) of about 10 — meaning you need to treat 10 patients before one gets meaningful relief beyond placebo.
Cannabinoids for chronic pain
- NNT ~10-24 (varies by analysis)
- Modest effect (0.46 points on 10-point scale)
- Favorable safety vs opioids
- Most studied: neuropathic pain
Modest
Standard comparisons
- NSAIDs for acute pain: NNT ~3
- Gabapentin for neuropathic pain: NNT ~7
- Opioids for chronic pain: NNT ~4 (but serious safety concerns)
- Antidepressants for neuropathic pain: NNT ~5-7
Reference
Whiting et al. (2015); Finnerup et al. (2015); Moore et al. (2015)
These numbers don't make cannabinoids look like miracle drugs. But they don't make them look useless either. For the subset of patients who respond — and particularly for those who have failed other treatments — the modest average effect conceals genuine individual benefit. And cannabinoids have a safety advantage over long-term opioid use that the NNT doesn't capture.
The honest assessment: cannabinoids for chronic pain are a legitimate option, not a first-line treatment. They work modestly for some people, particularly for neuropathic pain. The evidence base supporting their use is thinner than it should be, but it's real.
The Side Effects
Safety
Cannabinoid Side Effects Across 79 Trials
Most common
Dizziness, dry mouth, nausea, fatigue, somnolence, euphoria, vomiting, disorientation, drowsiness, confusion, loss of balance, hallucination
Serious AEs
Increased risk of serious adverse events with cannabinoids vs placebo — though absolute rates were low
Context
Side effect profile is very different from opioids. No respiratory depression, no overdose deaths, no physical dependence in short-term trials. But cognitive and psychomotor impairment is real.
The gap
Most trials used pharmaceutical-grade products (dronabinol, nabilone, nabiximols) — not the variable-potency flower and concentrates from dispensaries
Whiting et al. (2015), JAMA
Why This Study Changed the Conversation
The Whiting review was uncomfortable for everyone. Cannabis advocates couldn't cite it as proof that cannabis is medicine — the evidence was modest and the evidence base was thin. Cannabis opponents couldn't cite it as proof that cannabis doesn't work — the effects for pain, spasticity, and nausea were real. Policymakers got the answer they didn't want: "it's complicated, and we don't have enough trials."
Myth vs. Reality
Thousands of studies prove medical cannabis works.
By 2015, only 79 randomized controlled trials of cannabinoids for any medical condition had ever been conducted worldwide, involving a total of 6,462 participants. Only 4 met low-risk-of-bias standards. The evidence supports cannabinoids for a few specific conditions (chronic pain, spasticity, chemo nausea) with moderate quality, while evidence for most other conditions remains insufficient.
The Evidence
Whiting et al. searched 28 databases comprehensively. The 'thousands of studies' cited by advocates include preclinical research (cell and animal studies), observational studies, case reports, and reviews — not the randomized trials that constitute the strongest evidence for therapeutic efficacy.
Whiting et al. (2015), JAMA
The review became the reference document for the 2017 National Academies report on cannabis, which used it as a foundation for its own evidence assessments. It remains the most-cited systematic review of cannabinoid therapeutics — not because its findings were dramatic, but because it was honest about what we know and don't know.
For patients considering medical cannabis, the Whiting review's legacy is a framework for realistic expectations: cannabinoids can help, modestly, for specific conditions. They're not a panacea. And the evidence base, while growing, remains far thinner than what exists for most prescription drugs. For a detailed look at what the evidence supports, see our guide to the evidence-based medical benefits of cannabis and the specific research on cannabis and chronic pain.
Cannabinoids for Medical Use: A Systematic Review and Meta-analysis
Whiting PF, Wolff RF, Deshpande S, Di Nisio M, Duffy S, Hernandez AV, Keurentjes JC, Lang S, Misso K, Ryder S, Schmidlkofer S, Westwood M, Kleijnen J (2015) · JAMA
Does this study prove medical cannabis works?
It provides moderate-quality evidence that cannabinoids help with chronic pain, MS spasticity, and chemotherapy nausea. For other conditions, the evidence was insufficient or low-quality. "Works" depends on the condition and what you mean by "works" — the effects are real but modest for the best-supported indications.
Why were there so few trials?
Cannabis's Schedule I status in the United States made clinical research extremely difficult — researchers needed DEA licenses, could only use government-supplied cannabis (often of poor quality), and faced institutional barriers. Other countries had similar regulatory obstacles. The result: one of the most widely used therapeutic substances in the world had almost no high-quality clinical trial data.
Do these findings apply to dispensary cannabis?
Partially. Most trials used pharmaceutical-grade products (dronabinol, nabilone, nabiximols) with precise dosing. Dispensary products have variable potency, different cannabinoid ratios, and additional compounds (terpenes, minor cannabinoids) not present in pharmaceutical formulations. The assumption that dispensary products produce identical results is common but unproven.
Has the evidence improved since 2015?
Yes, substantially. The Epidiolex trials for epilepsy (2017-2018), the Wang et al. BMJ chronic pain review (2021), and several Sativex trials for MS have added to the evidence base. But the fundamental picture — modest effects for specific conditions, insufficient evidence for most — remains largely accurate.
What the researchers found
Across 79 RCTs, cannabinoids were linked to better symptom outcomes than placebo for several indications, but effects were often small and not consistently statistically significant across trials. The clearest pooled benefit was for complete chemotherapy-related nausea and vomiting response (47% vs 20%, OR 3.82), while chronic pain and spasticity showed smaller average improvements on common rating scales. Cannabinoids were also associated with a higher risk of short-term adverse events, including serious adverse events.
Why it matters
In 2015, medical cannabis policy and prescribing debates were moving faster than the clinical trial evidence for specific conditions and outcomes. This review tried to answer a basic but disputed question: for which indications do cannabinoids outperform placebo in RCTs, and what harms show up in the same trials. It also set a quality benchmark by applying the Cochrane risk of bias tool and showing how much of the evidence base depended on trials that did not meet low-bias standards.
The numbers in context
• Chemotherapy nausea and vomiting: 47% complete response with cannabinoids vs 20% with placebo (OR (odds ratio) 3.82, 95% CI: 1.55-9.42; 3 trials). This is a large relative difference, but it comes from only 3 trials so precision is limited.
• Pain responder outcome: 37% with cannabinoids vs 31% with placebo (OR 1.41, 95% CI: 0.99-2.00; 8 trials). This is a modest effect, and the CI includes values near no difference.
• Pain intensity on a 0-10 scale: WMD (weighted mean difference) -0.46 (95% CI: -0.80 to -0.11; 6 trials). A change under 1 point on a 10-point scale is small and many people would not notice it.
• Spasticity (Ashworth scale): WMD -0.36 (95% CI: -0.69 to -0.05; 7 trials). This is a small average change, even though the CI excludes zero.
How the study worked
This was a systematic review and random-effects meta-analysis, meaning the authors searched broadly for RCTs and then statistically combined results when studies measured similar outcomes. They searched 28 databases from inception to April 2015 and included RCTs of cannabinoids for a defined list of conditions, including chemotherapy nausea and vomiting, HIV/AIDS appetite and weight outcomes, chronic pain, and multiple sclerosis-related spasticity. Two reviewers independently handled each stage, and trial quality was rated with the Cochrane risk of bias tool. The biggest weakness was study quality: only 4 of 79 trials were judged at low risk of bias, which limits how confidently pooled effects can be interpreted.
Who was studied
79 randomized clinical trials (N=6,462 pooled participants), countries not specified in abstract
What this study cannot tell us
Only 4 of the 79 included trials were rated low risk of bias, so many pooled estimates rest on trials with design or reporting problems. Several key pooled findings came from a small number of trials for that outcome (for example, 3 trials for chemotherapy nausea and vomiting), which makes the summary estimates less stable. Countries, ages, and sex distribution were not reported in the abstract, making it hard to know how well the pooled sample matches specific real-world patient groups.
How to read the evidence
Rated moderate: a large meta-analysis of 79 RCTs (6,462 people), but only 4 trials were low risk of bias and several key outcomes pooled just a few studies.
When this study was published
Published in 2015, before today’s much larger medical and adult-use markets expanded access and product types (flower, high-THC concentrates, and many edibles). Most RCT evidence in this review came from standardized cannabinoid medicines, so real-world products and doses may not match the trial exposures.
The bigger picture
This review exposed the gap between public confidence in medical cannabis and the actual clinical trial evidence. It became the reference document for the 2017 National Academies report and shaped medical cannabis policy worldwide.
Replication
Not stated in abstract.
Funding
Not reported in abstract.
Conflicts of interest
Not reported in abstract.
Questions still open
- Which specific cannabinoid products and doses (THC-dominant vs CBD-dominant vs balanced, synthetic vs plant-derived) drove the pooled benefits and adverse events across these 79 trials?
- Why does the pooled nausea and vomiting effect look much larger than the pooled pain and spasticity effects. Is it a difference in biology, outcome measurement, or trial design quality?
- How different would the pooled effect sizes look if analyses were restricted to the 4 trials rated low risk of bias?
Common questions
In this JAMA review, which condition had the strongest cannabinoid result?
How big was the pain improvement in this meta-analysis, in actual points?
How solid were the trials behind these results?
Read the original research
Cannabinoids for medical use: A systematic review and meta-analysis
JAMA, 313(24), 2456-2473
JAMA is a top-tier, highly selective, peer-reviewed general medical journal with strong editorial and statistical review standards.
Citation
Whiting, Penny F.; Wolff, Robert F.; Deshpande, Sohan; Di Nisio, Marcello; Duffy, Steven; Hernandez, Adrian V.; et al.. (2015). Cannabinoids for medical use: A systematic review and meta-analysis. JAMA, 313(24), 2456-2473.
Explore the wider topic
- The Proven Medical Benefits of Cannabis: What Research Supports
- Cannabis for Chronic Pain: What the Research Actually Supports
- Cannabis and Multiple Sclerosis: Spasticity, Pain, and the Sativex Story
- Cannabis for Chemotherapy Nausea: What the Evidence Actually Shows
- CBD Oil Quality Guide: How to Avoid Snake Oil
- Cannabis and Migraines: Prevention, Treatment, and What Studies Show