The study
A randomised trial published in Aging Clinical and Experimental Research on 28 October 2026, run at Henan Provincial Hospital of Traditional Chinese Medicine in China.
Fifty-six adults aged 50 to 70 with knee osteoarthritis (Kellgren–Lawrence grade 2 or 3, pain at least 3 out of 10) were split evenly into two groups for 12 weeks. The massage group had 20-minute sessions twice a week, all given by one traditional Chinese medicine physician with more than five years' experience, using pressure to the patient's tolerance. The control group had a 20-minute health education phone call twice a week, plus a booklet. The people measuring outcomes didn't know who was in which group. Participants and the therapist did.
The trial was registered before it started (ITMCTR2025001524). The published analysis changed from the registered plan: it switched statistical method and dropped the planned subgroup analyses.
The treatment
The main text doesn't describe the massage techniques in any detail, and nobody checked whether sessions were delivered the same way each time. With one therapist giving every session, the result may say as much about that therapist as about massage in general.
Reading the numbers
Pain was measured on the WOMAC pain scale, which runs from 0 to 20. The trial used 1.2 points as the smallest change patients tend to notice (the minimal clinically important difference, or MCID). The 95% confidence interval (CI) is the range the true effect plausibly sits in.
What they found
Pain fell 1.88 points more with massage than with education (95% CI 0.62 to 3.14; p = 0.004). The average beat the 1.2-point MCID. The low end of the range, 0.62, sits below it, so the true benefit could be smaller than patients would notice.
Stiffness and function improved more with massage. Stiffness fell 1.13 points more (0.47 to 1.80) and function 9.60 points more (5.54 to 13.66).
Knee loading didn't change. There was no difference in the first or second peak knee adduction moment (p = 0.93 and 0.25) or in knee adduction angular impulse (p = 0.60). Walking speed rose by 0.03 m/s, which wasn't significant (p = 0.49).
Medial muscle co-contraction dropped. Co-contraction of vastus medialis and semitendinosus in early stance fell by 10.34% of maximum voluntary contraction. Lateral co-contraction didn't change.
Three people dropped out (one massage, two education), attendance was above 95% in both groups, and no adverse events were reported.
Funding and conflicts
Funded by Henan provincial research grants and the National Natural Science Foundation of China. The authors declared no conflicts of interest. No industry funding was reported.
What it means for your practice
The comparison group got phone calls. There was no sham or other hands-on control, and participants knew which group they were in. Some of the 1.88-point difference could come from the attention and expectation that come with in-person treatment, and pain was self-reported.
Pain changed while knee adduction moment stayed the same. Whatever massage did for pain here, it doesn't appear to have worked through reduced medial knee load. The drop in medial co-contraction is one possible route, and the trial can't show it caused the pain change.
Results stop at 12 weeks. There was no follow-up after sessions ended, so we don't know how long the benefit lasts.
The sample was 50 to 70 years old with moderate OA and fairly low BMI. People with severe OA or obesity weren't represented.
The paper has a reporting inconsistency. The education group's pain change differs between the text and the table.
Bottom line
In 56 adults with moderate knee OA, 12 weeks of twice-weekly massage reduced pain more than phone-based education, with no change in knee loading. It seems like massage may help short-term pain, though the true effect could be smaller than the average and we don't know how long it lasts.