Cannabis produces a small but real improvement in chronic pain — about 10% more patients get meaningful relief vs. placebo, with an NNT of ~24 and manageable side effects.
Chronic pain patients considering cannabis; clinicians making treatment decisions; anyone wanting evidence-based expectations rather than hype.
NNT ~24 — you need to treat 24 patients with cannabis for one to achieve meaningful pain relief beyond placebo
The Backstory
If you want to know whether cannabis works for chronic pain, there are two ways to find out. You can ask the millions of people who use it. They'll tell you yes. Or you can look at the randomized controlled trials — the only study design that can separate a drug's real effects from placebo, expectation, and wishful thinking.
Li Wang's team did the latter. They found 32 trials, enrolled 5,174 patients, used the most rigorous methods available, and came back with an answer that satisfied neither side: cannabis produces a real but small improvement in pain. It also improves sleep and physical function — a little. And it comes with side effects — mostly manageable ones, but at rates higher than placebo.
This is the most rigorous meta-analysis of cannabis for chronic pain ever published. What it shows isn't dramatic. But it's honest. And for the millions of chronic pain patients navigating real decisions about real treatment options, honesty is more useful than hype.
The Numbers
Wang's team searched for every randomized controlled trial comparing medical cannabis or cannabinoids to placebo for chronic pain — both cancer and non-cancer. They found 32 trials spanning one to five and a half months of follow-up. Then they did something unusual for cannabis research: they applied GRADE methodology, the gold standard for rating the certainty of medical evidence.
Pain Evidence
Cannabis for Chronic Pain: The Rigorous Assessment
32
Randomized trials
the full evidence base as of 2021
5,174
Patients enrolled
across all trials
10%
More patients got relief
vs. placebo (moderate certainty)
~24
Number needed to treat
patients treated per one meaningful responder beyond placebo
Wang et al. (2021), BMJ 374:n1034
That number needed to treat (NNT) deserves context. An NNT of 24 means you need to give cannabis to 24 chronic pain patients before one achieves meaningful relief beyond what a sugar pill provides. For comparison:
Strong Performers
- NSAIDs for acute pain: NNT ~3
- Triptans for migraine: NNT ~4-5
- Opioids for acute pain: NNT ~4
- These drugs have clear, measurable effects that dramatically exceed placebo
Clear benefit
Cannabis for Chronic Pain
- NNT ~24 for meaningful pain relief
- Similar to many chronic pain medications
- Comparable to duloxetine for neuropathic pain (NNT ~7-12)
- Better safety profile than long-term opioids
- Most patients do not achieve clinically meaningful benefit beyond placebo
Real but modest
Wang et al. (2021); comparative NNTs from Cochrane reviews
What Improved — and What Didn't
Wang's analysis broke outcomes into specific domains. The pattern was consistent: small improvements in pain, physical function, and sleep. No improvement in emotional, role, or social functioning.
The Side Effect Profile
The trade-off picture was equally specific. Cannabis isn't dangerous for most people in these trials, but it isn't side-effect-free either.
Adverse Effects
Side Effects: What You're Trading For Pain Relief
9-28%
More dizziness
9% at <3 months, 28% at ≥3 months
5%
More drowsiness
vs. placebo
5%
More nausea
vs. placebo
3%
More impaired attention
vs. placebo
2%
More cognitive impairment
vs. placebo
Wang et al. (2021), BMJ 374:n1034
The dizziness finding at longer durations (28% increase over placebo at three months or more) is notable. Most cannabis-for-pain discussions focus on the short-term safety profile. This review showed that some side effects worsen, not improve, with continued use — the opposite of what tolerance-based arguments would predict.
Why The Results Seem Worse Than What Patients Report
There's a disconnect between these modest trial results and the enthusiastic reports from millions of cannabis users who say it helps their pain. Several factors explain the gap:
Process
Why Trials Show Less Benefit Than Patient Reports
Placebo response in chronic pain is enormous
Chronic pain trials routinely show 20-30% of placebo patients reporting improvement. This isn't faking — placebo genuinely activates endogenous pain modulation. Cannabis has to beat this substantial baseline to show a drug effect.
Trial participants differ from self-selected users
People who seek out cannabis for pain are already motivated to believe it works. Trial participants are randomly assigned and may have no prior interest in cannabis. The populations aren't comparable.
Most trials used pharmaceutical cannabinoids, not whole plant
30 of 32 trials used oral preparations (pills, capsules, oils). Only 2 used topical. None used inhaled cannabis — which is how most pain patients actually consume it. The entourage effect, if real, would be absent in most trial formulations.
Pain is subjective and multidimensional
Trial endpoints measure pain intensity on a scale. But patients report cannabis helps with pain-related anxiety, sleep disruption, and quality of life — outcomes that may not be fully captured by a VAS score.
Real-world use allows dose titration
Patients adjust their dose in real time based on pain levels. Trials assign fixed doses. The flexibility of real-world use may produce better individual outcomes than rigid trial protocols.
Wang et al. (2021); interpretive analysis
What This Means for Patients
The Bigger Picture
Wang's review appeared alongside a companion clinical practice guideline (Busse et al., 2021) in the same issue of the BMJ. The guideline recommended — with weak certainty — offering a trial of non-inhaled medical cannabis for chronic pain when standard care is insufficient. "Weak recommendation" isn't an insult. It means the evidence exists, the benefit-risk balance is plausible, but patient values and individual circumstances should drive the decision.
This is a meaningful shift from Whiting's 2015 JAMA review, which found moderate-quality evidence for chronic pain but fewer trials and less rigorous grading. Six years later, the evidence base had grown from 8 to 32 trials for pain — but the effect sizes hadn't grown with it. More data confirmed a small benefit, not a hidden large one.
For the millions living with chronic pain conditions who are considering cannabis, this study provides the most honest assessment available: real effects, modest magnitude, manageable side effects, and an evidence base that — while growing — still lags behind what patients deserve.
Related Research
The Cannabis and Pain Evidence Base
Tracing the evidence from the first comprehensive review to the most rigorous meta-analysis.
Cannabinoids for Medical Use: A Systematic Review and Meta-analysis
Whiting et al. (2015)
The first comprehensive assessment — 79 trials across all indications. Wang's review is the pain-specific successor with stricter methodology.
Medical cannabis or cannabinoids for chronic pain: clinical practice guideline
Busse et al. (2021)
The companion guideline published alongside Wang — translating the evidence into a weak recommendation for a trial of cannabis in treatment-resistant pain.
Clinical endocannabinoid deficiency (CECD)
Russo (2004)
Why some pain patients may respond better than others — if endocannabinoid deficiency is real, those patients might derive more benefit than the average trial participant.
Taming THC: potential cannabis synergy
Russo (2011)
The entourage effect hypothesis — possibly explaining why whole-plant cannabis outperforms isolated cannabinoids in patient reports, a gap the trials couldn't test.
Cannabis opioid substitution
Boehnke et al. (2016)
The patient perspective: 64% reduced opioid use after starting cannabis. Real-world outcomes may exceed trial results for reasons the trials couldn't capture.
Does cannabis work for chronic pain?
Yes, but modestly. The most rigorous evidence (32 randomized trials, 5,174 patients) shows cannabis produces a small but real improvement in pain intensity — about 10% more patients achieve meaningful relief compared to placebo. It also provides small improvements in sleep and physical function. However, it does not improve emotional or social functioning, and most individual patients in trials don't achieve clinically meaningful pain reduction beyond placebo. It's a genuine therapeutic option, not a miracle cure.
How does cannabis compare to other pain medications?
The number needed to treat (~24) is worse than acute pain treatments like NSAIDs (NNT ~3) or opioids for short-term pain (NNT ~4). But it's comparable to many medications used for chronic pain, and cannabis has a better long-term safety profile than opioids. For patients who've failed multiple conventional treatments, a modest benefit with manageable side effects can be clinically meaningful.
Why do patients report better results than the trials show?
Several factors: chronic pain trials have large placebo responses (~25% improvement), trial participants differ from self-selected cannabis users, most trials used pharmaceutical preparations rather than inhaled whole-plant cannabis, real-world patients can titrate their dose, and cannabis may help with pain-adjacent symptoms (anxiety, sleep, quality of life) that trial endpoints don't fully capture.
What are the main side effects?
Dizziness (9-28% above placebo), drowsiness (5%), nausea (5%), impaired attention (3%), and cognitive impairment (2%). Notably, dizziness worsened at longer treatment durations. Most side effects were self-limiting and not serious. No increase in serious adverse events was found compared to placebo.
What the researchers found
Cannabis provides small to very small improvements in pain intensity (WMD -0.50 on 10cm VAS), physical functioning (WMD 1.67 on 100-point SF-36), and sleep (WMD -0.35 on 10cm VAS) compared to placebo, with moderate-to-high certainty evidence. NNT approximately 24 for meaningful pain relief.
Why it matters
This is the most rigorous assessment of cannabis for chronic pain published. It provides the evidence base for clinical guidelines and shows that while cannabis is a legitimate treatment option, effects are modest — setting realistic expectations for patients and clinicians.
The numbers in context
Trials: 32. Patients: 5,174. Pain relief WMD: -0.50 cm on 10 cm VAS. Physical functioning WMD: 1.67 points on 100-point scale. Sleep quality WMD: -0.35 cm on 10 cm VAS. Dizziness RD: 9% (<3 months) to 28% (3+ months). Drowsiness RD: 5%. Nausea RD: 5%.
How the study worked
Systematic review and meta-analysis of 32 RCTs enrolling 5,174 adults with chronic non-cancer and cancer pain. GRADE methodology applied. 30 oral, 2 topical preparations. Follow-up 1-5.5 months. Risk of bias assessed. Outcomes include pain intensity, physical functioning, sleep, emotional/social functioning, and adverse events.
What this study cannot tell us
Most trials used non-inhaled pharmaceutical preparations, limiting generalizability to whole-plant cannabis. Follow-up was short (1-5.5 months). Trials enrolled different pain conditions. No trial compared cannabis to active comparators (other pain drugs). Patient populations in trials differ from self-selected cannabis users.
How to read the evidence
Moderate to high certainty using GRADE methodology — unusually strong evidence for cannabis research. The review included 32 RCTs and applied the most rigorous evidence-grading framework in medicine.
When this study was published
Published 2021 — represents the current evidence standard. Includes trials through 2021.
The bigger picture
Cannabis for chronic pain occupies a middle ground: better safety than opioids, comparable efficacy to many approved chronic pain medications, but far from a panacea. The evidence supports considering cannabis for patients who have failed standard treatments — but with realistic expectations.
Questions still open
- Would inhaled whole-plant cannabis show different results? Is the entourage effect responsible for the gap between patient reports and trial results? Which specific pain conditions respond best? What is the optimal THC:CBD ratio for pain?
Common questions
Read the original research
Medical cannabis or cannabinoids for chronic non-cancer and cancer related pain: a systematic review and meta-analysis of randomised clinical trials.
BMJ (Clinical research ed.), 374, n1034
Citation
Wang, Li; Hong, Patrick J; May, Curtis; Rehman, Yasir; Oparin, Yvgeniy; Hong, Chris J; Hong, Brian Y; AminiLari, Mahmood; Gallo, Lucas; Kaushal, Alka; Craigie, Samantha; Couban, Rachel J; Kum, Elena; Shanthanna, Harsha; Price, Ira; Upadhye, Suneel; Ware, Mark A; Campbell, Fiona; Buchbinder, Rachelle; Agoritsas, Thomas; Busse, Jason W. (2021). Medical cannabis or cannabinoids for chronic non-cancer and cancer related pain: a systematic review and meta-analysis of randomised clinical trials.. BMJ (Clinical research ed.), 374, n1034. https://doi.org/10.1136/bmj.n1034