The 1,583-Person NIH Trial That Tested Glucosamine Against Celecoxib
In 2006 the US government spent public money to settle whether glucosamine and chondroitin work for knee pain, in a trial big enough to answer and run by no manufacturer. The answer was mostly no, with one subgroup that the supplement industry has quoted ever since.
Evidence strength
Level 1b
Individual RCT
Peer-reviewed refs
7
Reading time
8 min
Key Takeaways
- GAIT randomised 1,583 people with painful knee osteoarthritis to glucosamine hydrochloride 1,500 mg, chondroitin 1,200 mg, both, celecoxib 200 mg or placebo for 24 weeks. It was funded by the NIH, not by a manufacturer.
- 60.1% of people on placebo met the response threshold. Against that, glucosamine added 3.9 percentage points, chondroitin 5.3 and the combination 6.5, none statistically significant. Celecoxib added 10.0, which was.
- In people with moderate-to-severe pain, the combination looked better: 79.2% against 54.3%. That subgroup was small, and the analysis was exploratory. It has not been confirmed by a later trial.
- At two years, no supplement arm slowed joint-space loss significantly, and none improved pain more than placebo.
- GAIT tested glucosamine hydrochloride, not the crystalline sulfate product behind most positive trials. It settles the question for the common US supplement, not for that product.
Glucosamine and chondroitin had been sold for years on the strength of European trials of prescription products when the US National Institutes of Health decided to test them independently. GAIT, the Glucosamine/chondroitin Arthritis Intervention Trial, is still the largest publicly funded trial of either supplement. It is quoted by both sides of the argument: by sceptics for its headline result, and by sellers for one subgroup. Both readings are in the same paper, and the difference between them is a lesson in how to read a trial.
Who Was Studied and What They Took
GAIT enrolled 1,583 people with symptomatic knee osteoarthritis and randomised them to five arms for 24 weeks [1]:
- glucosamine hydrochloride, 1,500 mg a day;
- chondroitin sulfate, 1,200 mg a day;
- both together;
- celecoxib 200 mg a day, a prescription anti-inflammatory, as the active control;
- placebo.
The primary endpoint was the proportion of people whose WOMAC pain score fell by at least 20% by week 24. The trial was double-blind and funded by the NIH. The celecoxib arm mattered: it showed whether the trial was sensitive enough to detect a drug known to work.
The Primary Result
| Arm | Response rate vs placebo | P value |
|---|---|---|
| Placebo | 60.1% responded | — |
| Glucosamine HCl 1,500 mg | +3.9 points | 0.30 |
| Chondroitin 1,200 mg | +5.3 points | 0.17 |
| Glucosamine + chondroitin | +6.5 points | 0.09 |
| Celecoxib 200 mg | +10.0 points | 0.008 |
Only celecoxib beat placebo. [1] The number that stands out is the first one: six in ten people on placebo improved by at least 20%. That ceiling left little room for any treatment to show itself, and even a proven drug added only ten percentage points. It also means that a person who takes glucosamine and feels better has roughly the same odds of feeling better without it.
The Subgroup Everyone Quotes
GAIT's authors also looked separately at people with moderate-to-severe pain at baseline. In that group, 79.2% responded to the combination against 54.3% to placebo (P=0.002). [1] That is the result printed on supplement websites. Three things limit it. The subgroup was a minority of the trial. The authors themselves described the analysis as exploratory, a signal to test rather than a finding to act on. And when the combination was tested again, the results went both ways: a 606-person trial found it matched celecoxib but had no placebo arm, [4] and a 164-person placebo-controlled trial found pain fell less on the combination than on placebo, 19% against 33%. [5]
What the Two-Year Follow-Ups Showed
Two ancillary studies kept part of the GAIT population on treatment for two years. In 572 people with X-rays, joint-space width on placebo fell by 0.166 mm, and no treatment arm differed significantly from placebo; there was a trend in knees with milder disease, and the authors noted the study was underpowered. [2] In 662 people followed for pain, the odds of a 20% improvement against placebo were 1.21 for celecoxib, 1.16 for glucosamine, 0.83 for the combination and 0.69 for chondroitin, none significant. [3] By two years, even celecoxib had stopped separating clearly.
What the Study Cannot Tell You
It tested hydrochloride. GAIT used glucosamine hydrochloride, the common US supplement, not the crystalline glucosamine sulfate behind most positive European trials. A meta-analysis of 25 trials later found that brand explained 41% of the variation in results, with the crystalline sulfate product at an effect size of −1.07 and everything else at −0.11. [6] GAIT settles the question for the product most Americans buy. It does not test that one product.
The placebo response may have hidden a small effect. In osteoarthritis trials the placebo effect on pain averages an effect size of about 0.5, and it grows with baseline pain. [7] A 60% placebo response is high even by that standard. A supplement with a real but small effect could disappear in it. That is a statement about the trial's sensitivity, not evidence that the effect exists.
Six months is short for structure. The primary trial measured pain. The structural question needed years and more participants than the two-year ancillary study had.
What This Changes, and What It Does Not
GAIT moved the US guideline. In 2019 the American College of Rheumatology strengthened its position from a conditional to a strong recommendation against glucosamine, citing the gap between industry-sponsored and publicly funded trials, and GAIT is the largest publicly funded one. For someone standing in a pharmacy, the practical reading is simple: glucosamine hydrochloride with or without chondroitin, the common combination product, did not beat placebo in the best independent test.
What it does not change: whether the crystalline sulfate product or pharmaceutical-grade chondroitin work. That question is open, and its evidence is laid out in Glucosamine and Chondroitin for Knee Osteoarthritis: What 20 Years of Trials Show. If you decide to try one, the profiles for glucosamine and chondroitin give the trial doses, and the Knee Osteoarthritis Supplement Trial sets the 12-week window and the stop rule. What would change our read of GAIT itself: a placebo-controlled trial in moderate-to-severe pain that confirms the subgroup. The one attempt so far went the other way.
Scientific References
- [1]Clegg DO, Reda DJ, Harris CL, et al.. Glucosamine, chondroitin sulfate, and the two in combination for painful knee osteoarthritis — The New England Journal of Medicine (2006)Oxford 1bPMID 16495392
- [2]Sawitzke AD, Shi H, Finco MF, et al.. The effect of glucosamine and/or chondroitin sulfate on the progression of knee osteoarthritis: a report from the glucosamine/chondroitin arthritis intervention trial — Arthritis and Rheumatism (2008)Oxford 1bPMID 18821708
- [3]Sawitzke AD, Shi H, Finco MF, et al.. Clinical efficacy and safety of glucosamine, chondroitin sulphate, their combination, celecoxib or placebo taken to treat osteoarthritis of the knee: 2-year results from GAIT — Annals of the Rheumatic Diseases (2010)Oxford 1bPMID 20525840
- [4]Hochberg MC, Martel-Pelletier J, Monfort J, et al.. Combined chondroitin sulfate and glucosamine for painful knee osteoarthritis: a multicentre, randomised, double-blind, non-inferiority trial versus celecoxib — Annals of the Rheumatic Diseases (2016)Oxford 1bPMID 25589511
- [5]Roman-Blas JA, Castañeda S, Sánchez-Pernaute O, et al.. Combined Treatment With Chondroitin Sulfate and Glucosamine Sulfate Shows No Superiority Over Placebo for Reduction of Joint Pain and Functional Impairment in Patients With Knee Osteoarthritis — Arthritis & Rheumatology (2017)Oxford 1bPMID 27477804
- [6]Eriksen P, Bartels EM, Altman RD, et al.. Risk of bias and brand explain the observed inconsistency in trials on glucosamine for symptomatic relief of osteoarthritis: a meta-analysis of placebo-controlled trials — Arthritis Care & Research (2014)Oxford 1aPMID 24905534
- [7]Zhang W, Robertson J, Jones AC, et al.. The placebo effect and its determinants in osteoarthritis: meta-analysis of randomised controlled trials — Annals of the Rheumatic Diseases (2008)Oxford 1aPMID 18541604
Follow us on Google Search
Add BiohackingHub as a preferred source to see more of our research in Google results.
Related protocols
Partner with BiohackingHub
Editorial placements, sponsorships, and product features for evidence-minded brands.