Does Collagen Help Joints in Human Clinical Trials?

Clinical trials of collagen for joints measure pain scores, not repaired cartilage. The published evidence shows small pain reductions on visual rating scales. It does not show cartilage rebuilding.

Pooled analyses show that trial results depend on the rating scale. A 2019 meta-analysis and a 2025 network meta-analysis both found that oral collagen improved scores on a Visual Analogue Scale (VAS). However, neither analysis found an improvement on the pain subscore of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). Furthermore, the largest positive trials carry direct manufacturer funding. A 2025 placebo-controlled trial found no pain difference at all.

Collagen Lowers VAS Pain but Not WOMAC Pain

The clinical evidence for oral collagen shows modest reductions in subjective joint pain. The magnitude of that benefit depends on which rating scale researchers choose. A 2019 meta-analysis by Garcia-Coronado and colleagues pooled randomised trials in osteoarthritis. Total WOMAC scores improved, as did Visual Analogue Scale (VAS) pain scores. The weighted mean difference for VAS pain was -16.57. That difference was statistically significant. However, the WOMAC pain subscore showed no significant difference from placebo. Physical function subscores also failed to change.

Recent pooled analyses confirm this exact divergence between rating instruments. A 2025 network meta-analysis in Nutrients of 39 dietary supplement trials encompassing 4,599 osteoarthritis patients included five collagen studies. Versus placebo, collagen was reported significant on VAS pain (mean difference 16.65). In contrast, the WOMAC pain difference was not significant (mean difference 0.22). Stiffness showed no improvement either. A 2023 meta-analysis by Lin and colleagues pooled four trials of hydrolysed collagen peptides in 507 knee osteoarthritis patients (three trials, 375 patients, contributed to the pain analysis). They reported a significant reduction in VAS pain. However, Cochrane Risk of Bias tools rated all four included trials at high risk of bias.

Trial Evidence on Whether Collagen Helps Joints in Osteoarthritis

Randomised controlled trials of hydrolysed collagen in knee osteoarthritis present conflicting findings that range from large pain reductions to no significant difference from placebo. The largest positive study was published by Carrillo-Norte and colleagues in 2024. They enrolled 120 adults with Kellgren-Lawrence grade II to III knee osteoarthritis. Participants received 10 grams daily of bovine collagen peptides with calcium ascorbate or placebo for six months. VAS pain dropped by 22.6 millimetres more in the collagen group. Analgesic use fell to 3.3 percent in the collagen arm versus 41.7 percent in placebo. However, the study carries important methodological limitations. The manufacturer of the peptides and the intellectual property owner funded the trial. Baseline analgesic use differed substantially between arms, sitting at 71.7 percent in placebo and 28.3 percent in collagen.

In contrast, a 2025 randomised trial found no significant pain difference from oral collagen. Researchers randomised 68 knee osteoarthritis patients to tablets combining hydrolysed collagen with a small amount of undenatured type II collagen or placebo for 12 weeks. All participants received exercise therapy and rescue analgesia. Pain scores fell by 2.21 points with collagen and 2.29 points with placebo. That between-group difference was 0.08 points. It did not reach statistical significance. Knee Injury and Osteoarthritis Outcome Score (KOOS) subscales and patient satisfaction showed no separation between trial arms. Funding came from a medical council foundation; the tablets were supplied by a pharmaceutical company whose role the paper does not describe. The trial authors noted that their study carried roughly 5 percent post hoc statistical power.

Dose and timing are covered in collagen dosage and timing.

Undenatured Type II Collagen Evidence and Trial Discrepancies

Undenatured type II collagen (UC-II) is a different product from hydrolysed collagen, with a different proposed mechanism, but its clinical trial record shows sharp contradictions between positive sponsor-run studies and lower network meta-analysis rankings. The primary positive trial was conducted by Lugo and colleagues in 2016. They tested 40 milligrams daily of UC-II against placebo and a glucosamine-chondroitin combination across 191 knee osteoarthritis patients. Over 180 days, UC-II reduced total WOMAC scores by 551 points versus 414 points for placebo (p = 0.002). Rescue medication use was lower in the UC-II arm. Glucosamine and chondroitin did not separate from placebo. However, knee flexion range of motion did not differ between groups. Blood biomarkers of inflammation and cartilage breakdown did not move either. The company manufacturing UC-II funded the trial, two authors were company employees, and the third consulted for it.

Pooled analyses present a less favourable picture of undenatured type II collagen. A second 2025 network meta-analysis evaluating 22 randomised trials ranked, as reported, undenatured type II collagen lowest for osteoarthritis pain, joint stiffness, and physical function. It did not outperform placebo on any WOMAC subscale. The paper carried internal reporting inconsistencies, and authors urged caution regarding small sample sizes. A 2025 narrative review covering 12 studies noted no ongoing knee osteoarthritis UC-II trials registered on ClinicalTrials.gov, CTRI or ChiCTR as of August 2024. They highlighted publication bias as a prominent concern. Our guide on collagen types details these structural differences across categories.

Cartilage Repair Claims and the Missing Human Tissue Data

None of the sources reviewed for this page reports a human study measuring whether ingested collagen peptides accumulate in joint cartilage. In 1999, Oesser and colleagues administered radiolabelled gelatin hydrolysate to mice. Radioactivity in cartilage doubled compared to proline controls. However, the researchers tracked radioactive tracers rather than intact collagen peptides. In 2017, Dar and colleagues fed hydrolysed type I collagen to male mice with surgically induced joint trauma. A high dose equivalent to 7.4 grams daily in humans reduced cartilage loss. The trial included only male rodents, omitted several analyses, and carried manufacturer co-authorship.

The only human imaging trial among the sources reviewed is a small pilot. In 2011, McAlindon and colleagues randomised 30 patients with mild knee osteoarthritis to collagen hydrolysate or placebo for 24 weeks. They used delayed gadolinium-enhanced magnetic resonance imaging (dGEMRIC) to assess proteoglycan content. Tibial dGEMRIC scores rose in the collagen group while falling in placebo. This difference was statistically significant. However, four other knee regions showed no significant difference. Structural T2 values showed no difference between groups. Clinical pain and function scores also failed to separate from placebo. The trial authors explicitly described their findings as preliminary. Furthermore, a 2020 scoping review by Honvo and colleagues noted that three out of four in vitro studies found hydrolysed collagen ineffective or detrimental to osteoarthritis cartilage cells. The published evidence does not show that collagen rebuilds joint cartilage.

Collagen in Athletes and Active Adults

Clinical trials examining collagen in athletes and active adults evaluated exercise-related discomfort in participants without diagnosed joint disease, meaning their findings cannot be transferred to osteoarthritis. Athletic trials exclude participants with joint degeneration. In 2008, Clark and colleagues tested 10 grams daily of collagen hydrolysate in 147 collegiate athletes at Penn State over 24 weeks. Among 97 evaluable athletes, walking pain, standing pain, and resting pain improved more than placebo. However, 50 enrolled participants were excluded from the analysis without explanation. The PubMed record contains no funding or conflict-of-interest information.

Later studies in active adults evaluated specific branded peptides. A 2021 trial by Zdzieblik and colleagues randomised 218 adults aged 18 to 30 (180 completed) with activity-related knee pain to 5 grams daily of a branded porcine peptide or maltodextrin. Subject-rated pain during activity fell by 21.9 millimetres with collagen versus 15.6 millimetres with placebo. That between-group difference was 6.3 millimetres. Secondary outcomes showed no significant difference. Pain during squats showed p = 0.961, resting pain showed p = 0.628, and joint range of motion did not differ. The product manufacturer supplied the peptides and partially funded the trial. Authors stated their results cannot be transferred to other collagen products. A 2024 trial by Zdzieblik and colleagues randomised 182 adults (154 analysed per protocol) with functional knee or hip pain, osteoarthritis and rheumatoid arthritis excluded. Physician-rated pain at rest and walking favoured collagen, while physician-rated stair pain did not reach significance (p = 0.056) and joint stiffness did not differ. A systematic review by Khatri and colleagues reported that collagen peptide trials in active adults rely predominantly on commercial funding.

Tendon and Ligament Outcomes in Clinical Trials

Evidence evaluating collagen for tendon conditions and ligament instability shows no significant between-group difference on objective physical measurements. In 2018, Dressler and colleagues randomised 50 athletes with chronic ankle instability to 5 grams daily of collagen peptides or placebo for six months. Subjective ankle instability questionnaire scores improved significantly in the collagen group. However, objective mechanical joint stability measured by an ankle arthrometer showed no significant between-group change. Self-reported stability improved more with collagen; one author is at the Collagen Research Institute.

Studies on Achilles tendinopathy show a similar lack of objective divergence. In 2019, Praet and colleagues conducted a crossover pilot trial in 20 patients with mid-portion Achilles tendinopathy. Participants completed calf-strengthening exercises alongside 2.5 grams of collagen peptides twice daily or placebo. Symptom scores improved under both treatments. Measures of tendon microvascularity decreased under both treatments without significant group differences. The pilot had no power calculation and received manufacturer financial support. A 2022 systematic review of six trials evaluating dietary supplements in tendinopathy found no functional benefit over physiotherapy alone. The review could not produce a collagen-specific estimate due to sparse data. In the Praet pilot, symptom scores improved under both collagen and placebo.

What Clinical Guidelines Say and Do Not Say

Neither the American College of Rheumatology nor the Osteoarthritis Research Society International mentions collagen in their clinical practice guidelines for osteoarthritis. The 2019 American College of Rheumatology and Arthritis Foundation guideline text explicitly restricted its review to pharmaceutical-grade formulations. This methodological scope eliminated most nutraceuticals from formal consideration. The word collagen appears zero times in the guideline document. Similarly, the 2019 Osteoarthritis Research Society International guideline evaluated 60 interventions, establishing arthritis education and structured land-based exercise, with or without dietary weight management for knee osteoarthritis, as Core Treatments. Its text also contains zero mentions of collagen. This absence does not indicate a recommendation against the supplement. It is an absence of a recommendation, not a finding that collagen does not work.

The guidelines provide useful context through their decisions on neighbouring supplements. The American College of Rheumatology strongly recommended against glucosamine in knee, hip, and hand osteoarthritis. The authors cited discrepancies between industry-sponsored and publicly funded trials, raising serious concerns regarding publication bias. They noted that trials at low risk of bias failed to demonstrate important benefits over placebo. The guideline also strongly recommended against chondroitin sulfate for knee and hip osteoarthritis, while conditionally discouraging fish oil and vitamin D. On its public resources, the National Center for Complementary and Integrative Health states that evidence on dietary supplements beyond glucosamine and chondroitin is too limited for conclusions.

The Commercial Funding Pattern Across the Literature

In the trials reviewed for this page, positive results were funded or supplied by the commercial manufacturers of the tested products. This commercial sponsorship pattern spans both hydrolysed peptides and undenatured collagen. The primary 2016 trial on undenatured type II collagen was funded by its manufacturer. Two study authors were company employees. The third author consulted for it. The 2024 trial by Carrillo-Norte reporting a 22.6-millimetre VAS pain reduction was funded by the peptide manufacturer and the patent holder. Both trials used a branded peptide provided by its manufacturer; the 2021 trial was partly funded by it and had an author at a collagen research institute, and the 2024 trial had an author who is a co-inventor on collagen peptide patents. The Achilles tendinopathy pilot by Praet was supported financially by the supplement maker.

In the two sources reviewed without manufacturer funding, one trial found no pain difference and one meta-analysis found a pain benefit but rated all four trials high risk of bias. The 2025 trial of 68 knee osteoarthritis patients was funded by a medical council foundation; the tablets were supplied by a pharmaceutical company whose role the paper does not describe. It found no significant pain difference between collagen and placebo. A 2023 meta-analysis by Lin declared no funding and found a pain benefit, but rated all four included trials at high risk of bias. A 2020 scoping review by Honvo and colleagues observed that all included in vivo preclinical and clinical studies concluded beneficial effects regardless of quality. The review did not formally assess risk of bias, noted a grant from a collagen ingredient company, and found three of four in vitro studies ineffective or detrimental to osteoarthritis cartilage. Furthermore, trial results cannot be transferred between different commercial products.

Who the Trials Excluded and What Would Change Our Reading

The clinical evidence on collagen covers a narrow subset of joint conditions, leaving acute injuries, inflammatory arthritis, and non-knee joints largely unexamined. The osteoarthritis literature enrolled participants with Kellgren-Lawrence grade II to III knee osteoarthritis. The athletic literature enrolled individuals with activity-related joint discomfort and no diagnosed joint disease. Someone with a joint injury, a swollen or locked joint, inflammatory arthritis, or worsening pain is not asking a supplement-evidence question; that is a conversation for a clinician. Furthermore, the evidence centers almost entirely on knees, providing negligible data on hip or hand osteoarthritis.

Our editorial reading of the literature would change if specific missing evidence were published. An independently funded, adequately powered, pre-registered trial in knee osteoarthritis measuring WOMAC pain and objective MRI structural changes together would alter our assessment. Demonstrating that ingested peptides enter human joint cartilage in tissue-uptake studies would resolve the current mechanistic gap. Formal evaluation by guideline committees would also represent a meaningful shift. Regarding tolerability, a 2023 meta-analysis of four trials found an adverse-event odds ratio of 1.66 (95% CI 0.99 to 2.78, p = 0.05, GRADE very low). Our overview of collagen side effects reviews adverse event reporting across trial durations.

Related

• Collagen vs Collagen Peptides
• Collagen Types Explained
• Collagen Dosage and Timing
• Collagen Side Effects in Clinical Trials
• Menopause and Joint Pain