Cannabis products cut chronic pain by about half a point on a ten-point scale

Cannabis-based products produce small reductions in chronic pain — on the order of half a point on a 0-to-10 scale for the best-studied preparation — and the effect is concentrated in neuropathic pain, over trials lasting one to six months, with substantially more dizziness, sedation and nausea than placebo [s1]. Beyond six months, there is essentially nothing to report, because essentially nothing has been studied.

The updated review

An evidence synthesis funded by the US Agency for Healthcare Research and Quality, published in Annals of Internal Medicine, searched Ovid MEDLINE, PsycINFO, Embase, the Cochrane Library and Scopus to 28 July 2025 and included 25 randomised placebo-controlled trials lasting one to six months, covering 2,303 participants, of whom 64% had neuropathic pain [s1].

Products were classified by their THC-to-CBD ratio, by source (synthetic, purified or extracted from whole plant) and by route of administration — a level of granularity earlier reviews lacked, and one that turns out to matter.

Oral synthetic or purified high THC-to-CBD products — effectively THC alone — may slightly reduce pain severity, at a pooled difference of −0.78 points [s1]. Oromucosal extracted products with comparable THC-to-CBD ratios probably slightly reduce it, at −0.54 points [s1]. Within the THC-only category the two agents diverged sharply: nabilone moderately reduced pain severity at −1.59 points, while dronabinol did not, at −0.23 [s1]. Low THC-to-CBD interventions may not improve outcomes at all [s1].

The parallel living systematic review from the same programme reports the confidence intervals for the best-evidenced comparison: extracted, comparable-ratio THC-to-CBD oral spray against placebo gave a mean difference of −0.54 (95% CI −0.95 to −0.19) across 7 randomised trials in 878 participants, with I² of 39% and moderate strength of evidence [s2]. Function improved by −0.42 (−0.73 to −0.16, I² 32%, moderate strength of evidence) — a figure the reviewers note sits slightly below their own threshold for even a small effect [s2].

Harms are not proportionally small

The comparable-ratio and high-ratio products come with moderate or large increases in dizziness, sedation and nausea [s1]. CBD alone may not increase harms; low THC-to-CBD mixed products may increase dizziness, sedation and nausea [s1]. Withdrawals due to adverse events did not differ from placebo in the living review's pooled analysis [s2], which suggests the side effects, while common, were mostly tolerable within short trials.

That last qualifier matters. A trial running one to six months in participants who consented to take a cannabinoid is not a good instrument for detecting cognitive, psychiatric or dependence outcomes over years.

The opioid-sparing claim

One of the strongest public arguments for medical cannabis is that it lets people take fewer opioids. A systematic review examined that specifically in people with chronic pain already receiving prescription opioids [s3].

It found five randomised trials, all in patients with chronic cancer pain, and twelve observational studies [s3]. Every randomised trial instructed participants to maintain their opioid dose, which makes them structurally incapable of demonstrating opioid reduction; the resulting estimate was very low certainty for little or no impact on opioid use, at a weighted mean difference of −3.4 morphine milligram equivalents (95% CI −12.7 to 5.8) [s3].

Those same trials provided high-certainty evidence that adding cannabis had little or no effect on pain relief in chronic cancer pain (WMD −0.18 cm, 95% CI −0.38 to 0.02, on a 10 cm visual analogue scale) or on sleep disturbance (WMD −0.22 cm, −0.4 to −0.06, against a minimally important difference of 1 cm) [s3]. Adding cannabis likely increased nausea (RR 1.43, 1.04–1.96; risk difference 4%, 0% to 7%) and vomiting (RR 1.5, 1.01–2.24; RD 3%, 0% to 6%), both moderate certainty [s3].

Eight observational studies did report reduced opioid use, at a weighted mean difference of −22.5 morphine milligram equivalents (−43.06 to −1.97), but the reviewers rate that very low certainty [s3]. The gap between the observational and randomised findings is the whole story: people who choose cannabis and reduce opioids differ from people who do not, in ways an observational study cannot fix.

Fibromyalgia and the "nociplastic" claim

A scoping review of cannabis-based medicines specifically for nociplastic pain — the category that includes fibromyalgia — found ten studies, predominantly reviews, opinion pieces and consensus documents rather than primary data [s4]. The signals it identified suggest modest benefits in fibromyalgia and chronic musculoskeletal pain, with improvements more evident for sleep, affective domains and quality of life than for direct analgesia [s4].

Adverse effects were common, including sedation, dizziness and psychiatric symptoms, particularly with THC-dominant products [s4]. The review notes low and variable oral bioavailability of around 6%, food and hepatic effects notably for cannabidiol, a prolonged terminal half-life, and psychomotor and cognitive impairment among the pharmacodynamic effects [s4]. No included study applied standardised nociplastic instruments, and no phenotype-specific opioid-sparing evidence was identified [s4].

The limitations the reviewers name themselves

The Annals review lists them: variability within its own product categories, a lack of product detail in the source trials, uncertainty about whether the products studied are even available in the United States, and restriction to English-language studies [s1]. Its conclusion calls for studies on long-term outcomes and on other cannabis product types [s1].

That last point is the one a reader is most likely to be affected by. Almost none of this evidence comes from products people actually buy. The trials tested pharmaceutical-grade oromucosal sprays, synthetic capsules and purified compounds at known concentrations; dispensary flower, edibles and vape products at unknown or variable potency are largely unstudied.

What to watch

Whether any adequately powered trial runs past six months, and whether the categories the reviewers built — ratio, source, route — hold up as more trials accumulate. The divergence between nabilone and dronabinol, two THC-only products with a difference of more than a point between them [s1], suggests that "cannabis for pain" is not one question.

This article is informational and is not medical advice. It does not describe doses or recommend any product.

Sources

• Cannabis-Based Products for Chronic Pain: An Updated Systematic Review — Annals of Internal Medicine, 2025-12-22
• Living Systematic Review on Cannabis and Other Plant-Based Treatments for Chronic Pain: 2025 Update — Agency for Healthcare Research and Quality, 2025-07-01
• Opioid-sparing effects of medical cannabis or cannabinoids for chronic pain: a systematic review and meta-analysis of randomised and observational studies — BMJ Open, 2021-07-28
• Cannabis-Based Medicines for Nociplastic Pain: A Scoping Review of Efficacy, Safety, and Patient Outcomes — Journal of Pain & Palliative Care Pharmacotherapy, 2026-07-22