Capsaicin for Osteoarthritis

Capsaicin cream or gel gives modest, short-term pain relief in pooled osteoarthritis trials, with low certainty and frequent burning. A 2024 meta-analysis of eight double-blind trials (498 people) found that topical capsaicin (0.0125% to 5%) lowered pain more than placebo (standardised mean difference -0.84, 95% CI -1.48 to -0.19), and raised the chance of burning at the application site about 5.6-fold (risk ratio 5.56, number needed to harm 3). The authors rated the certainty of the evidence low to very low because the trials were short, small and varied widely. That effect size looks large, but its confidence interval is wide and an earlier review found a smaller effect (0.44), so the true benefit is probably smaller.

A network meta-analysis of 28 trials (7,372 people) found that capsaicin beat placebo only at licensed doses (effect size 0.41, 95% CI 0.17 to 0.64) and showed no significant difference from topical NSAID gels, although this is an indirect comparison and no trial compared the two directly. We rate it likely but with limited evidence, and 3/5 effectiveness: the direction is consistent across reviews and the ACR guideline conditionally supports it for the knee, but certainty is low. It is a reasonable short-term add-on for knee pain, not a replacement for exercise, weight management or a topical NSAID. See the osteoarthritis page for the treatments with the strongest support, and the capsaicin page for the oral and nerve-pain evidence.

Capsaicin activates TRPV1, a receptor on pain-sensing nerve fibres. A 2021 review of injectable capsaicin explains that, in enough dose, it disrupts the endings of those fibres so that they cannot be triggered by any stimulus, and that the effect after a single administration lasts weeks to months and is fully reversible. The early arthritis trials framed it differently, as depleting substance P, a nerve chemical implicated in joint inflammation and pain.

The same review describes topical capsaicin as acting on abnormal nerve activity in the superficial layers of the skin. That is a plausible fit for the knee, which sits close to the surface, and a poor one for the hip: the ACR panel did not examine capsaicin for hip osteoarthritis because the depth of the joint beneath the skin makes a meaningful effect unlikely.

The abstracts we read measured pain, stiffness and function, not cartilage or joint structure, so nothing here suggests capsaicin changes the disease itself. It is a pain treatment.

Blinding is the other limit. Capsaicin burns, so people can often tell which cream they have. A 2018 literature review gave its recommendation only grade B and named difficulty of blinding as the most important limitation of the trials, so some of the measured benefit may reflect expectation.

Trials used 0.025% cream four times daily for 4 weeks, 0.0125% gel three times daily for 4 weeks with a crossover to placebo (100 people), and 0.075% cream four times daily on painful hand joints. A 28-day single-blind comparison found 0.25% cream twice daily acted faster than 0.025% four times daily: pain fell 48.0% versus 17.7% after 2 days, although burning on day 2 was also commoner (69% versus 31%).

The over-the-counter 0.025% cream label we reviewed tells adults to apply a thin film and rub it in until absorbed, repeating as needed but no more than 3 to 4 times daily (the 4-week course comes from the trials, not the label), and to wash hands afterwards unless the hands are being treated. Children under 18 should ask a doctor. The one 0.25% trial we read in detail was small and single-blind, although the 2024 review pooled strengths up to 5%. Ask a clinician before using a stronger preparation.

The 2024 meta-analysis (eight double-blind trials, 498 people; five at low risk of bias and three at high risk) found lower pain than placebo with capsaicin of 0.0125% to 5% (standardised mean difference -0.84, 95% CI -1.48 to -0.19, p = 0.01), with benefit at follow-ups of up to 3 months. The authors listed short durations, small samples, high heterogeneity and overall low-to-very-low certainty as limits, and suggested capsaicin for short-term pain in people who cannot take NSAIDs.

An earlier meta-analysis of five double-blind trials and one case-crossover trial (0.025% to 0.075%, 4 to 12 weeks) found a smaller effect, a standardised mean difference of 0.44 (95% CI 0.25 to 0.62) at 4 weeks, with no difference by joint or strength. A 2004 BMJ review that pooled three trials in chronic musculoskeletal pain (368 people) reported a relative benefit of 1.5 (95% CI 1.1 to 2.0) and a number needed to treat of 8.1 (95% CI 4.6 to 34) for 0.025% cream or plaster, and called efficacy 'moderate to poor'. The estimates disagree on size, which is part of why the certainty is low.

The network meta-analysis found topical NSAIDs beat placebo overall (effect size 0.30, 95% CI 0.19 to 0.41), while capsaicin beat placebo only at licensed doses (0.41, 95% CI 0.17 to 0.64). The difference between the two was not significant (as licensed: -0.09, 95% CI -0.34 to 0.16). No trial compared them directly, and the placebo comparison is the only link between them.

A series of n-of-1 trials gives a more personal angle. In 22 people with painful knee osteoarthritis who tried 5% ibuprofen gel and 0.025% capsaicin cream in random order, mean pain reduction on a 0-10 scale was 1.2 on ibuprofen and 1.6 on capsaicin (P = 0.221). Even so, 59% responded better to one than the other: nine people (41%) to capsaicin and four (18%) to ibuprofen. Anyone who gets nothing from one topical treatment may fare better on the other.

In a 1991 double-blind trial, 70 people with osteoarthritis and 31 with rheumatoid arthritis applied 0.025% cream or placebo four times daily to painful knees for four weeks. Pain fell 33% in the osteoarthritis group (P = 0.033 versus placebo), and 23 of 52 capsaicin-treated patients felt transient burning. A crossover trial of 100 people with mild to moderate knee osteoarthritis found 0.0125% gel better than vehicle gel on pain and WOMAC scores (p < 0.05), with burning in about 67% and no withdrawals; all participants were female farmers, so it may not generalise.

Not every result was large. In a 6-week trial of 200 pain-clinic patients, 0.025% capsaicin lowered pain by a mean 0.5 points on a 0-10 scale from a baseline of 6.40; adding glyceryl trinitrate cream, a nitrate drug, lowered it by 1.1. In hand osteoarthritis, a 21-person trial of 0.075% cream reduced pain and tenderness in the 14 people with osteoarthritis (p less than 0.02), but the guideline panel judged direct evidence too thin to recommend it for hands.

The 2019 ACR/Arthritis Foundation guideline conditionally recommends topical capsaicin for knee osteoarthritis, citing small effect sizes and wide confidence intervals, and conditionally recommends against it in hand osteoarthritis. It strongly recommends exercise, weight loss if overweight, and topical NSAIDs for knee osteoarthritis. A comparison of three guidelines found that the ACR and VA/DoD guidelines recommend topical capsaicin while the OARSI guideline does not.

The ACR recommendation is strong for topical NSAIDs and conditional for capsaicin. The 2024 review suggested capsaicin for short-term pain in people who cannot take NSAIDs.

A single injection of purified trans-capsaicin into the knee (CNTX-4975) was tested in a phase II trial of 172 people: the 1.0 mg dose lowered walking pain more than placebo at week 12 (least squares mean difference -1.6, P < 0.0001) and week 24 (-1.4, P = 0.0002), and the 0.5 mg dose did not reach significance. It is investigational, and a related injected TRPV1 agonist (RTX-GRT7039) failed a 2026 phase 3 trial of 469 people: the week-12 difference from placebo was -0.11 points (95% CI -0.50 to 0.27, p = 0.569). Injected TRPV1 drugs are not proven.

Still missing: long-term trials beyond a few months, a head-to-head trial against topical NSAIDs, trials in hip and hand osteoarthritis that would settle the guideline caution, and any trial of oral capsaicin. Many trials are decades old, small, and hard to blind.

Application-site burning is the expected side effect. It affected 35% to 100% of capsaicin-treated people across trials, peaking in week 1 and declining after, with a number needed to harm of 3 in the 2024 meta-analysis. In individual trials, burning occurred in about 67% of people on 0.0125% gel (none withdrew) and in 23 of 52 on 0.025% cream (two withdrew). Reviews describe systemic toxicity as absent, and serious problems were not a feature of the trials we read.

The cream label warns of a burning sensation that generally disappears after several days of regular use. It says not to use on wounds or damaged skin or if allergic to capsicum or chili peppers, to avoid the eyes, lips, nose and mucous membranes, not to wrap or bandage the area tightly, and not to expose it to heat or direct sunlight immediately before or after use. Stop and ask a doctor if the condition worsens, redness or skin irritation develops, or burning is severe. The label's 7-day rule is written for minor aches; with a long-term condition like osteoarthritis, it is a reason to involve your clinician if you plan regular use. If pregnant or breastfeeding, ask a health professional first.

None of the trials or the label we read reported a drug interaction, and reviews describe no systemic toxicity. The label does advise against heat and tight wrapping around the time of application. We found no trial of capsaicin used together with a topical NSAID, so combining them rests on your clinician's judgement.