Personalized gait retraining for knee osteoarthritis: a one-year sham-controlled trial

On October 10, 2026, ScienceDaily published a University of Utah release revisiting a randomized, placebo-controlled trial of personalized gait retraining for knee osteoarthritis, first published in The Lancet Rheumatology in August 2025 ^[1]. Over one year, participants taught to walk with an individually chosen foot angle reported larger pain reductions than those given a sham version of the same training, and one MRI marker of cartilage health declined more slowly ^[1]. The release also summarizes a smaller April 2026 trial from the University of Bath ^[1].

Why the trial matters

Osteoarthritis, a condition in which the cartilage cushioning the ends of bones gradually breaks down, affects roughly 22% of adults over 40, and the story notes that no established treatment reliably restores lost cartilage ^[1]. Gait retraining, meaning a deliberate change in walking mechanics, targets the load on the knee rather than pain alone ^[1]. The inner (medial) compartment of the knee normally carries more load during walking, which the story connects to how common inner-knee osteoarthritis is ^[1].

Co-lead investigator Scott Uhlrich said biomechanical approaches are not new but have lacked randomized, placebo-controlled evidence ^[1]. He emphasized personalization: "Previous trials prescribed the same intervention to all individuals," and some people in those trials did not reduce, or even increased, knee loading ^[1].

What the trial did

The trial enrolled people with mild to moderate medial knee osteoarthritis ^[1]. At two laboratory visits, participants had MRI scans and walked on a force-sensing treadmill under motion capture while testing toe-in or toe-out changes of 5° or 10°; investigators then selected the change that most reduced each person's inner-knee load ^[1]. Prospective participants for whom no tested adjustment reduced loading were excluded ^[1].

Sixty-eight participants were randomized into two groups of 34 ^[1]. The intervention group was trained to their personalized foot angle; the sham group underwent the same process but kept its natural angle, which allowed mechanical effects to be separated from expectation and attention ^[1]. Both groups attended six weekly sessions using biofeedback, a technique that gives immediate signals so a person can learn to control an automatic movement; here, a shin-worn device vibrated when the foot left the target angle ^[1]. Participants were then encouraged to practise for at least 20 minutes daily, and on average held their prescribed angles to within about one degree ^[1].

Results

On a 0 to 10 pain scale, the intervention group improved by an average of 2.5 points versus 1.3 points with sham ^[1]. In an exploratory analysis, about 91% of the intervention group had a clinically meaningful pain reduction, compared with 66% of the sham group ^[1]. The intervention also reduced a mechanical measure associated with inner-knee loading ^[1]. Uhlrich characterized the pain benefit over placebo as falling between what would be expected from ibuprofen and from an opioid such as OxyContin ^[1].

On MRI, T1rho, a measure that detects changes in cartilage composition before structural damage appears on conventional scans, suggested slower medial cartilage deterioration with the intervention; a second MRI measure showed no significant group difference ^[1]. No severe adverse events were reported, although two intervention participants and one sham participant withdrew because of increased knee pain ^[1].

What to watch next

The Bath trial, published in Clinical Biomechanics, randomized 50 people with knee osteoarthritis to one of three six-week biofeedback strategies ^[1]. All three groups, including the control, improved in activity-related pain and function, but only the group given feedback on its walking pattern achieved a lasting 7.6% reduction in a knee-loading measure, still present one month later ^[1]. No group significantly changed foot angle, suggesting other walking adjustments can also alter knee forces ^[1]. That study did not examine long-term cartilage deterioration, so it cannot establish whether these strategies slow progression ^[1].

Limitations

• The main trial randomized 68 people, and the 91% versus 66% responder comparison was exploratory ^[1].
• Excluding people whose loading did not respond to any tested adjustment means the results apply to a preselected group ^[1].
• Only one of two MRI measures favoured the intervention, and the findings do not show cartilage was rebuilt or progression permanently stopped ^[1].
• Selecting each person's adjustment relied on laboratory testing with a force-sensing treadmill and motion capture ^[1].
• This account is based on a university press release, not the full trial report ^[1].