Ask people with rheumatoid arthritis what they want most from treatment, and up to 70% give the same answer: less pain. Not fewer swollen joints, not better blood tests. Less pain. Modern drugs have become very good at calming the immune attack on the joints, yet many patients keep hurting even when their inflammation looks controlled on paper. That gap is where cannabis enters the conversation. Rheumatologists are now asked about it regularly, by patients who already use cannabis products on their own or are thinking about adding them to prescribed treatment.
So what does the clinical evidence actually say? A team from the University of Ioannina and the University of Thessaly in Greece went looking for every clinical study of cannabinoids for pain in adults with rheumatoid arthritis published since 1990. Their scoping review, published open access in Rheumatology International, came out in October 2026, and its main finding is a count. Out of 593 records, three studies qualified. One of them was a randomized trial. It enrolled 58 people, ran for five weeks and was published two decades ago.
Key Takeaways
The review by Nikolaos Zintziovas and colleagues searched four databases for clinical studies published between January 1990 and September 2026 and found only three that tested or tracked cannabinoids for pain specifically in adults with rheumatoid arthritis. The single randomized, placebo-controlled trial, run in the UK and published in 2006, gave 58 patients a mouth spray containing THC and CBD in equal parts for five weeks. It was associated with a reduction of 0.95 points in pain on movement and 1.04 points in pain at rest on a 0-to-10 scale, a 1.17-point improvement in sleep quality and a 0.76-point drop in a standard disease activity score, with no serious side effects. The reviewers call these pain reductions modest and stress that statistical significance does not by itself establish a clinically meaningful benefit. The two other studies, a US online survey of CBD users and an analysis of US hospital records, were rated at serious and critical risk of bias and cannot show cause and effect. Five more trials are registered but have no published results. The authors conclude that cannabinoids remain an investigational approach in rheumatoid arthritis and cannot currently be recommended for routine use.
How the Cannabinoids and Rheumatoid Arthritis Pain Review Was Done
A scoping review is a mapping exercise. Instead of pooling numbers from many trials into a single estimate, it asks a simpler set of questions: how much evidence exists, what kind it is, and where the holes are. That format fits a field where nobody is sure there is enough to pool.
The authors followed the PRISMA-ScR reporting standard and searched PubMed/MEDLINE, Scopus, the Directory of Open Access Journals and the Cochrane Central Register of Controlled Trials for studies published between January 1, 1990 and September 10, 2026. Search terms covered rheumatoid arthritis, pain, and a broad list of cannabinoid names, from cannabis, THC and CBD to the pharmaceutical products nabiximols (sold as Sativex), dronabinol and nabilone. To be included, a study had to be a full, peer-reviewed randomized trial or observational study in adults with confirmed rheumatoid arthritis, and it had to report a pain-related outcome.
The search returned 593 records: 399 from Scopus, 114 from PubMed, 46 from the open-access directory and 34 from Cochrane. After 66 duplicates were removed, 527 were screened by title and abstract, and 511 were dropped. The breakdown is revealing. No fewer than 392 were reviews, editorials, letters or conference abstracts, which means papers discussing the evidence outnumbered the studies that produced it by more than a hundred to one. Another 43 measured something other than pain, 38 were animal studies, 27 dealt with a different disease, six were not in English and five were clinical trials with no published results. Of the 16 papers read in full, nine mixed rheumatoid arthritis patients with other conditions without reporting them separately, and four did not report pain. Three remained.
Two authors screened and extracted data independently, with a third settling disagreements. The quality of the randomized trial was rated with the Cochrane RoB 2 tool and the observational studies with ROBINS-I, both standard checklists for spotting ways a study could mislead. With only three studies of very different designs, the authors did not attempt a meta-analysis and described the results one by one.
Why Rheumatoid Arthritis Pain Outlasts Inflammation
Rheumatoid arthritis is an autoimmune disease: the immune system attacks the lining of the joints. The backbone of treatment is a group of medicines called disease-modifying antirheumatic drugs, or DMARDs, which slow that attack and protect joints from permanent damage. For a long time, pain was treated as a simple by-product of inflammation. Bring the inflammation down, the thinking went, and the pain follows.
For many patients it does not. The review describes two mechanisms behind this. The first is at the level of the nerves in and around the joint. Signaling proteins that drive the disease, including TNF-α, IL-1β, IL-6 and IL-17, appear to act directly on pain-sensing nerve cells and make them fire more easily. The second is in the spinal cord and brain, and is called central sensitization. A useful picture is a microphone with the gain turned up too far: a normal voice comes out as a roar, and even background noise becomes loud. The nervous system amplifies pain signals, and can keep doing so when the joint itself is no longer visibly inflamed. The same mechanism featured in a Charité pilot study of a cannabinoid extract for endometriosis pain, another condition where pain and visible disease often fail to line up.
Psychological factors matter too. Anxiety, low mood and a tendency to expect the worst from pain have all been linked to poorer long-term outcomes and lower pain tolerance in rheumatoid arthritis. The practical consequence, the authors write, is that simply stepping up immune-targeting drugs may not be enough when part of the pain is not inflammatory in the first place.
Why Cannabinoids Look Plausible on Paper
The interest in cannabinoids rests on the endocannabinoid system, a signaling network the body uses to regulate, among other things, pain and immune activity. It has three parts: receptors, the body’s own cannabis-like molecules that switch those receptors on, and the enzymes that make and break those molecules down. The two best-known molecules, anandamide and 2-arachidonoylglycerol, are produced on demand rather than stored. They travel backwards across the junction between two nerve cells and tell the sending cell to quiet down, a little like a listener raising a hand to ask a speaker to slow down.
There are two main receptors. CB1 sits mostly in the brain and spinal cord, where it influences how pain is perceived. CB2 sits mostly on immune cells and in tissues outside the brain, where it is more closely tied to dampening inflammation. Both receptors, and the body’s own cannabinoids, have been found in joint tissue. Plant cannabinoids also act on other targets, including TRPV1, the same channel that responds to the heat of chili peppers, and a family of receptors inside the cell nucleus called PPARs.
Animal work supports the idea. In an often-cited example, a study in mice with collagen-induced arthritis, published in 2000, reported that oral CBD had an anti-arthritic effect. The review’s central point is the distance between this kind of result and proof in people. A mechanism that makes sense and an experiment that works in mice explain why a treatment is worth testing. They do not show that it works.
The One Randomized Trial: 58 Patients, a THC and CBD Spray, Five Weeks
The only trial that directly tested a cannabinoid medicine in rheumatoid arthritis is the study by Blake and colleagues, published in the journal Rheumatology in 2006 and conducted in the UK. It was double-blind and placebo-controlled, meaning neither patients nor doctors knew who was getting the real product. It enrolled 58 people with active rheumatoid arthritis who were already taking DMARDs. The spray was therefore tested as an add-on to standard treatment, not as a replacement for it.
The product was nabiximols, a spray applied inside the mouth that delivers THC and CBD in a fixed one-to-one ratio. Patients used it in the evening only, for five weeks.
According to the review, nabiximols was associated with a mean reduction of 0.95 points in pain on movement on a 0-to-10 numerical rating scale (p=0.044), a reduction of 1.04 points in pain at rest (p=0.018) and an improvement of 1.17 points in sleep quality (p=0.027). The DAS28, a score doctors use to track disease activity across 28 joints, fell by 0.76 points (p=0.002). Morning stiffness did not change significantly. Side effects were mostly mild to moderate, and nobody in the active-treatment group had a serious adverse event or left the trial because of side effects.
On the risk-of-bias checklist, the trial scored well: low risk across all domains. Its weakness is not how it was run but how small and short it was. And the reviewers are blunt about the size of the effect. A drop of about one point on a ten-point scale is, in their words, modest, and its clinical importance remains uncertain. For illustration, that is the difference between rating your pain a 6 and rating it a 5. They write that the statistically significant findings should not be read as establishing a clinically meaningful painkilling benefit.
They add a second caution about the DAS28 result, which might look like evidence that the spray reduced inflammation. The score is a composite. It combines counts of swollen and tender joints and a blood marker with the patient’s own rating of how they feel, so it can improve because a person feels better, without the underlying inflammation changing. The review says the DAS28 drop does not provide direct evidence of reduced inflammation.
Effects of this size are a recurring pattern in controlled cannabinoid pain research. The phase 3 trial discussed in our coverage of a 10-year study of medical cannabis for chronic low back pain found a difference of 0.6 points over placebo on the same scale.
Two Observational Studies: A CBD Survey and 3.3 Million Hospital Records
The other two studies did not assign treatment to anyone. They observed people who had already used cannabis products, which makes them far weaker for answering whether those products work.
The first is a cross-sectional survey by Frane and colleagues, run in the United States and published in 2022 in the Journal of Cannabis Research. It collected answers from 428 adults with arthritis who had used CBD, 142 of whom had rheumatoid arthritis. Across the whole group, 83% reported an improvement in pain, 66% in physical function and 66% in sleep. Respondents reported an average 44% reduction in pain and a drop of 2.58 points on the 0-to-10 scale (p<0.001). Most said they had cut back or stopped other medication, including anti-inflammatories, acetaminophen and opioids.
Those figures need three qualifications, all spelled out in the review. They describe everyone in the survey, most of whom had other forms of arthritis, and should not be read as results for rheumatoid arthritis. Within the rheumatoid arthritis subgroup, responses were generally less pronounced than in osteoarthritis. And the design invites bias at every step: participants were recruited online through advocacy networks and social media, they chose to take part, they reported their own outcomes from memory, and there was no comparison group. People who feel a product helped them are more likely to answer a survey about it. The reviewers rated the study at serious risk of bias.
The products themselves were a further unknown. Respondents used whatever CBD they had chosen, in different forms, doses and durations. Self-directed use of this kind is widespread, as a six-country survey of 79,644 people on why they use CBD shows, and what is in the bottle is not guaranteed either: when scientists tested 186 CBD oils sold in Europe, half failed.
The second study, by Shrestha and colleagues, published in 2025, used the National Inpatient Sample, a large US database of hospital stays. It covered 3,347,284 hospitalizations related to rheumatoid arthritis, of which 42,415, or 1.27%, also carried a diagnostic code for cannabis use. Patients with that code had lower adjusted odds of dying in hospital and lower odds of chronic pain (odds ratio 0.45), depression (0.55) and anxiety (0.55). They also had higher odds of opioid use (1.1), nicotine dependence (1.35) and alcohol use (1.35). Length of hospital stay did not differ significantly.
An odds ratio of 0.45 sounds striking, but the reviewers rated this study at critical risk of bias, the worst category. A diagnostic code for cannabis use says nothing about what the person used, how much, for how long, by what route, or whether it had anything to do with pain. Lower odds of chronic pain alongside higher odds of opioid use is not a pattern with one obvious explanation, and the dataset cannot supply one. The review describes these associations as non-causal and difficult to translate into clinical terms.
What Is Known About Safety
Not much, and only for the short term. In the Blake trial, no serious adverse events or side-effect-related withdrawals occurred in the nabiximols group, but that is 58 patients over five weeks. In the Frane survey, 41% of CBD users reported at least one side effect, most often dry mouth and sleepiness, followed by appetite changes, dry eyes, trouble concentrating, dizziness, headache and stomach complaints. Most were described as mild. The hospital database was not designed to capture cannabinoid side effects at all.
Rheumatoid arthritis is a lifelong condition, and people who have it usually take several medicines for years. The reviewers conclude that the available data are insufficient to establish the long-term safety or tolerability of cannabinoid-based treatment in this group.
Five Registered Trials, No Published Results
One of the more telling tables in the paper lists studies that were registered but never produced an eligible publication. A randomized, placebo-controlled trial of CBD in rheumatoid arthritis (NCT04911127) is marked as completed. A phase 2 trial of a cannabinoid formulation called IHL-675A (NCT05942911) was terminated. A study of cannabichromene, a lesser-known cannabinoid (NCT07087938), is suspended. A crossover trial in rheumatoid and psoriatic arthritis (NCT04269993) was terminated after its funding ended. And a pilot trial of a topical cannabinoid preparation (ISRCTN29199098), registered in 2014, is listed as completed.
None had peer-reviewed results the reviewers could use. As they note, registering or even completing a trial does not mean it will be published. For readers, the list is a reminder that the thin evidence base is not for lack of trying: two of these trials are listed as completed, yet the reviewers found no peer-reviewed publication for either.
Earlier Reviews, and Where US and UK Guidance Stands
The conclusion is not new, which is part of the story. A Cochrane review of neuromodulators for rheumatoid arthritis pain in 2012 and a systematic review of cannabinoid trials in rheumatic diseases in 2016 both leaned on the same 2006 trial and reached similarly cautious verdicts. More than a decade later, the count of randomized trials in rheumatoid arthritis is unchanged at one.
Official guidance in the two countries where the reviewed studies were carried out reflects that. In the UK, home of the Blake trial, nabiximols has been licensed since 2010, but for a single use: muscle spasticity in multiple sclerosis. The NICE guideline on cannabis-based medicinal products tells NHS clinicians not to offer a combination of CBD with THC, or THC, nabilone or dronabinol, to manage chronic pain in adults, and not to offer CBD for chronic pain outside a clinical trial.
In the United States, where both observational studies were done, nabiximols has never been approved. Its developer, Jazz Pharmaceuticals, discontinued the US program in 2022 after a late-stage multiple sclerosis trial missed its main goal. Patient access runs instead through state programs, which together count some 3.9 million registered medical marijuana patients. The Arthritis Foundation’s CBD guidance, issued in 2019, says there have been no rigorous clinical studies of CBD in people with arthritis and that CBD should never replace the disease-modifying drugs that prevent joint damage. At the federal level, an April 2026 order moved FDA-approved marijuana products and marijuana covered by a state medical license into Schedule III, while a broader rescheduling decision is still pending.
What Would Change the Picture
The authors list what future research needs: trials large enough to detect a real effect, run across several centers, using standardized cannabinoid products, with longer treatment and follow-up. They want pain, disease activity, sleep, disability, mood and safety measured with standard tools, and careful records of the other medicines patients take. They also raise a question the current data cannot answer: when a cannabinoid helps, is it reducing inflammation, dulling pain in a general way, or easing overlapping problems such as central sensitization, fibromyalgia or secondary osteoarthritis? In rheumatoid arthritis, where pain and inflammation so often part ways, that distinction decides what the treatment would actually be for.
The review has limits of its own, which the authors state. Only three studies qualified, and they differ so much that they cannot be combined. Only English-language papers were included. The protocol was not registered in advance. The authors also disclose that they used ChatGPT (GPT-5.6) for language editing between August 1 and September 11, 2026, and state that all scientific content and conclusions were reviewed and approved by them. They report no conflicts of interest and no specific funding.
Their bottom line is that, until better data exist, cannabinoids should remain an investigational approach in rheumatoid arthritis.
FAQ
The evidence is too thin to say. One randomized trial of 58 patients found that a THC and CBD mouth spray was associated with about a one-point reduction in pain on a 0-to-10 scale over five weeks. The authors of the review call that effect modest and of uncertain clinical importance, and no larger or longer randomized trial has been published since 2006.
The survey included 428 adults with arthritis, of whom 142 had rheumatoid arthritis, and 83% of the whole group reported pain improvement. Those figures were not broken out as treatment effects for rheumatoid arthritis, responses in that subgroup were generally weaker than in osteoarthritis, and the survey had no comparison group. It shows what self-selected users said, not what CBD does.
In the randomized trial, side effects were mostly mild to moderate and none were serious, but the trial lasted only five weeks. In the survey, 41% of CBD users reported at least one side effect, most commonly dry mouth and sleepiness. There are no reliable long-term safety data for cannabinoids in rheumatoid arthritis.
No. The FDA has not approved a cannabinoid medicine for rheumatoid arthritis, and nabiximols, the spray used in the only randomized trial, is not approved in the US for any condition. Access runs through state medical marijuana programs, whose qualifying conditions and product rules differ from state to state. Since April 2026, marijuana covered by a state medical license has been in federal Schedule III, while other marijuana remains in Schedule I pending a broader decision.
Disclaimer
This article is for informational and educational purposes only and does not constitute medical, legal or pharmaceutical advice. It summarizes a published scoping review and the studies it covers and does not recommend the use of cannabis, CBD or any cannabinoid product for rheumatoid arthritis or any other condition. Disease-modifying treatment for rheumatoid arthritis should never be stopped or changed without the guidance of a treating physician, and cannabinoids can interact with other medicines. Laws governing cannabis and cannabis-based medicines vary between countries and, in the United States, between states, and they change over time; readers are responsible for checking the rules that apply where they live. The Cannex does not endorse any product or provider.
