For one person with knee osteoarthritis, the best foot angle might be toes turned in by 5 degrees, and for the next it might be toes turned out by 10. A year-long trial tested what happens when that choice is made individually instead of handed out as a general rule, and the answer was a pain improvement that held up against a sham comparison, with some important limits on how far it can be trusted.
The work was published in The Lancet Rheumatology in August 2025 by a team led by Scott Uhlrich at the University of Utah, with collaborators at New York University and Stanford. It enrolled 68 adults with mild to moderate osteoarthritis on the inner side of the knee, split evenly into two groups of 34, with an average age near 64. Everyone had knee pain of at least 3 on a 10-point scale and could walk on a treadmill for 25 minutes without an aid.
Setting the foot angle took lab work. Each person walked on an instrumented treadmill while force sensors and motion-capture cameras measured how much load their walking put on the inner knee. They then tried toeing in or out by 5 or 10 degrees, and the angle that cut that load the most became their personal target. Volunteers for whom no adjustment lowered the load by at least 5% were left out of the trial. Six weekly treadmill sessions followed, using a shin-mounted device that vibrated whenever the foot drifted more than 2 degrees from the target, plus a goal of 20 minutes of practice walking each day. The comparison group went through the same sessions and the same vibration cues, except that their target was simply their natural foot angle.
At one year, pain fell by 2.5 points in the retraining group and by 1.3 points in the sham group, a between-group difference of 1.2 points (95% confidence interval 0.5 to 1.9). About 91% of the retraining group had at least a 1-point improvement, compared with 66% of the sham group.
The sham result deserves as much attention as the headline one. A 1.3-point drop in people who were told to keep walking naturally shows how much attention, practice and expectation can ease knee pain by themselves, and a 1-point improvement is a fairly low bar for counting someone as a responder. The true gain from the personalized angle is the 1.2-point gap, and the pain scores were collected by investigators who were not blinded during the sessions, so some assessor bias is possible.
The study also measured the knee itself. A mechanical measure of inner-knee loading dropped by 7.5% more in the retraining group than in the sham group at one year, which supports the idea that the new walking pattern really changed what the joint experiences. On MRI, one marker of cartilage health in the inner thigh-bone cartilage, called T1rho, pointed toward slower degeneration in the retraining group, while a second MRI measure showed no meaningful difference. Because the cartilage findings were secondary outcomes without correction for multiple comparisons, they read as a hint worth testing in a larger trial. Nothing in the data shows that cartilage was rebuilt or that the disease stopped progressing.
Several details limit how far the result can travel:
- The trial planned for 80 participants and reached 68, and it ended about four months early when COVID-19 shutdowns began.
- Seven people were missing from the final visits, and the 13 laboratory visits the protocol required contributed to dropouts.
- Eligibility covered mild to moderate disease and a BMI under 35, so the results may not apply to severe arthritis or to larger bodies.
- Only the thigh-bone cartilage was assessed on MRI.
- Funding came from the US Department of Veterans Affairs, which the paper says had no role in the design, conduct, analysis or reporting.
The practical catch is that the method depends on measurements most clinics do not have. Picking the right angle meant motion capture and force-sensing equipment, and a person who simply turns their toes in or out at home is not doing what the trial tested. The researchers describe the next step as streamlining the process with mobile sensing so that retraining could happen in a physical therapy clinic and on everyday walks. A smaller University of Bath trial of 50 people, published in Clinical Biomechanics this April, found that feedback on walking pattern cut knee loading by 7.6%, though it did not study cartilage.
Anyone who wants the protocol can find the trial’s ClinicalTrials.gov record. The full text is available as an open-access manuscript. For readers building a routine around arthritis, this trial fits alongside earlier coverage of a simple daily routine linked to less pain and the exercise switch that keeps aging muscles strong.
A personalized foot angle, learned over six supervised sessions, beat a sham routine on knee pain by a modest margin over a year, but that is a small trial that needs a cheaper way to find the right angle before it can reach most people with knee arthritis.
