How much of a knee’s future depends on which way the toes point? A year ago I’d have laughed at that question. In my head, knee arthritis only ever went one way: ibuprofen, ice, a little less cartilage every season, and eventually a surgeon with a titanium joint. Then I read about a year-long trial where people with arthritic knees changed their foot angle by a handful of degrees. They beat a convincing fake version of the same training on pain, and an MRI marker of their cartilage health held up better than the sham group’s.
The trial, led by Scott Uhlrich of the University of Utah with collaborators at Stanford and NYU, was published in The Lancet Rheumatology in 2025 and is getting a second wave of attention this fall. It enrolled 68 people with mild-to-moderate arthritis in the inner, or medial, compartment of the knee (mean age 64.4) and split them evenly: 34 got personalized gait retraining and 34 got a sham.
The prescription came from a lab. Each person walked on an instrumented treadmill with force sensors underfoot and motion-capture cameras tracking every stride, while the team tested toes turned inward or outward by 5 or 10 degrees and measured which tweak took the most load off the worn inner knee. Whichever angle won became that person’s target. Some did best toeing in and others toeing out, and anyone for whom no tested angle reduced the load was screened out before enrollment. That tells you right away this isn’t a one-angle-fits-all trick. The sham group got everything else. They had the same six weekly sessions and the same little shin device that buzzed when the foot drifted off target, and they were coached to hold a “new” angle that was simply the one they already walked with. I’ll admit the sham is what won me over. I came in expecting another posture fix that falls apart the moment someone runs a control group, and these researchers made their control nearly as convincing as the treatment.
Wait, though: how does rotating a foot a few degrees reach all the way up into the knee? I had to sit with that one. On every step the ground pushes back up through your leg, and in a knee worn on the inside, that force acts like a lever tipping your body weight onto the strip of cartilage that’s already thinning. Biomechanists measure that lever as the knee adduction moment, a standard stand-in for how hard the inner knee is being loaded. Swing the foot in or out and the force runs along a slightly different line through the leg, the lever shortens, and the inner compartment carries less. Over the year, peak knee adduction moment fell by 0.17 in the retrained group and crept up by 0.08 in the sham group. It’s mechanical, and it’s personal, because the angle that unloads one person’s knee can load another’s.
On a 0-to-10 scale, pain fell 2.5 points in the retrained group and 1.3 points in the sham group. That leaves a 1.2-point advantage over placebo, which Uhlrich has placed somewhere between what you’d expect from an over-the-counter drug like ibuprofen and a narcotic like OxyContin. That comparison is his yardstick from the drug literature, since the trial had no medication arm. And look at how much the sham group improved on attention and structure alone. That’s exactly why a convincing control matters.
| Group | Value (points) |
|---|---|
| Retrained | 2.5 |
| Sham | 1.3 |
In an exploratory analysis, about 91 percent of the retrained group reported a clinically meaningful drop in pain, against 66 percent of the sham group. People also kept it up. After the supervised sessions they were asked to practice at least 20 minutes a day, and check-ins found them holding their prescribed angle to within about one degree on average.
The MRI is the result I keep coming back to. The ibuprofen-to-opioid range Uhlrich used as his benchmark treats pain, and none of those drugs changes how much force lands on the cartilage. In this trial, a T1rho scan, which reads cartilage composition, suggested slower deterioration of the inner-knee cartilage in the retrained group. The between-group difference in medial cartilage relaxation time was 3.74 milliseconds in the retrained group’s favor (95% CI 1.05 to 6.42). A second MRI measure showed no significant difference between the groups, so this is one encouraging sign of slower wear. It isn’t a rebuilt knee, and it doesn’t prove anyone will put off a replacement.
The money behind this is worth a look. The work ran on federal research grants from the Department of Veterans Affairs, the National Institutes of Health and the National Science Foundation, and the authors reported no relevant financial disclosures. My read is that a treatment with no molecule to patent and no monthly refill has no obvious corporate sponsor, and this one took taxpayer money to get a proper placebo-controlled test. Uhlrich’s explanation for why earlier gait studies gave muddy results fits too: “Previous trials prescribed the same intervention to all individuals,” and some of those people ended up not reducing, or even increasing, the load on their knees.
There are limits, and they’re specific. This is one center and 68 people, and the group was about 60 percent women and 79 percent white, which nurse practitioner J. Nicholas Manwaring flagged in his commentary for Healio even as he called the approach an “excellent nonpharmacologic, nonsurgical option.” There were no severe adverse events, but two retrained participants and one sham participant withdrew because their knee pain got worse. The authors also say plainly that the method isn’t ready for routine clinics. It currently needs motion capture, an instrumented treadmill and repeated visits, so Uhlrich’s group is working on smartphone video and motion-sensing “smart shoes” to bring it to physical therapy offices.
I’m not going to start walking pigeon-toed on a hunch, because the tweak that unloads one knee can overload another. But if I had early medial knee arthritis, I wouldn’t settle for a decade of rotating anti-inflammatories while the cartilage thins. I’d ask a physical therapist whether they can measure my knee loading while I walk, and I’d make that call before anyone mentioned the word replacement.
Sources
- Uhlrich et al., “Personalised gait retraining for medial compartment knee osteoarthritis: a randomised controlled trial,” The Lancet Rheumatology (2025)
- Healio: Gait foot-angle retraining slows knee osteoarthritis progression vs sham
- ScienceDaily: A surprisingly simple way to relieve knee arthritis pain without pills or surgery
- News-Medical: Personalized gait retraining reduces pain and slows knee damage in osteoarthritis
- Newswise: A potential way to relieve arthritic knees without drugs or surgery