Can Gait Retraining Ease Early Medial Knee OA Symptoms?

Changing the way your foot points during walking may reduce load on the inner side of an arthritic knee and improve symptoms over six weeks. The most informative randomized trial so far enrolled 64 people with Kellgren-Lawrence grade I–II medial knee osteoarthritis and found that sensor-guided gait retraining reduced peak knee adduction moment (KAM) and improved patient-reported symptoms relative to control.

That is encouraging, but it is not a license to tell everyone with knee pain to walk toe-in or toe-out. The effect of foot angle on knee loading differs sharply between people. Gait retraining looks like a plausible, individualized option for early medial knee OA—not a proven replacement for exercise therapy, braces, insoles, or standard care.

Why foot angle can change knee load

The KAM estimates the tendency of ground-reaction forces to load the medial, or inner, compartment of the knee during walking. It is a widely used biomechanical marker in medial knee OA because higher values track with greater medial compartment loading. Researchers often evaluate gait interventions against a roughly 5% to 10% KAM reduction, a range considered clinically meaningful for this mechanical target.

Foot progression angle—the angle of the foot relative to the direction of travel—is one way to alter that load. A person may walk with the toes more inward or outward, usually while a clinician gives cues or real-time feedback. The goal is not to perform a fashionable new walk. It is to find a gait change that lowers load for that specific knee without creating a new problem elsewhere.

A 2022 meta-analysis of foot progression angle modification found that toe-in and toe-out changes can affect the first and second KAM peaks and KAM angular impulse. It also found the problem that makes blanket advice unreliable: the relationship is nonlinear and participant-specific. The same instruction can lower KAM in one person, do little in another, and move it in an unwanted direction in a third.

The best trial tested real-time sensor feedback

In a 2021 randomized controlled trial by Wang and colleagues, 64 adults with early medial knee OA completed six weeks of treadmill-based gait retraining with real-time feedback from wearable sensors. The study standardized walking speed during assessment and training, which helped isolate the effect of foot-angle modification rather than letting changes in speed explain a change in KAM.

The intervention used a machine-learning algorithm to estimate KAM in real time. Against laboratory measurement, the algorithm achieved an R² of 0.956. That result supports the feasibility of sensor-based feedback in a clinical setting; it does not mean every consumer wearable can measure knee loads accurately.

The trial's sample-size calculation targeted an effect size of d=0.80, using a more conservative assumption than a prior laboratory-based trial with d=1.01. The intervention group had statistically significant reductions in peak KAM and better patient-reported symptom outcomes than the control group. The publication details supplied here do not provide the absolute KAM change, symptom effect size, or confidence intervals, so those results should not be inflated beyond the reported direction and statistical significance.

Speed control also narrows the question the study answers. It shows what happened under standardized treadmill conditions. People do not spend daily life walking at one fixed speed on a treadmill, so free-living effects may differ.

Several reviews support the mechanical signal, with limits

A 2022 systematic review and meta-analysis by Silva et al. pooled neuromuscular gait-modification strategies in medial knee OA and found KAM reductions consistent with the 5% to 10% benchmark. The interventions differed across studies, which supports the idea that changing gait can alter knee loading but does not identify one universally best instruction.

A broader 2022 review by Rynne et al. reported improvements in KAM and symptoms across gait-retraining interventions for hip and knee OA, including different delivery approaches. In-clinic training, wearable-guided feedback, and telehealth-hybrid care all have feasibility or pilot-level support. No comparative trial has shown that one format produces better mechanical or clinical results than another.

The delivery method may matter less than measurement and coaching. A vague instruction to "walk differently" gives no assurance that knee loading changed. Real-time feedback offers a way to verify the immediate target, while an in-clinic gait assessment may identify the direction of foot-angle change that helps a particular person. Neither approach has yet produced a validated clinical algorithm that reliably predicts the right cue before treatment.

Pain relief and lower KAM are related hypotheses, not the same finding

The randomized trial improved both a biomechanical outcome and symptoms, but current evidence does not prove that the KAM reduction caused the symptom improvement. Supervised movement, attention from a clinician, confidence in walking, and greater activity could each affect pain reports. Small clinic pilots have also reported symptom improvement without detectable KAM differences, although their limited sample sizes make biomechanical null results hard to interpret.

This distinction matters because KAM is a surrogate measure. It may be a useful target for medial loading, but it is not a direct measure of pain, cartilage health, or a person's ability to climb stairs without planning the operation like a lunar landing.

Other conservative treatments may also change biomechanical loading. A 2025 meta-analysis of walking interventions reported quantifiable changes in knee OA biomechanical outcomes, including KAM-related measures. A 2023 review of exercise therapy likewise found measurable effects on biomechanical load. Cross-study comparisons cannot establish whether gait retraining beats structured walking or strengthening, because no head-to-head randomized trial has tested those options using a shared KAM primary outcome. The same gap applies to lateral wedge insoles and unloader bracing.

A one-year MRI association is intriguing, not proof of cartilage protection

The most interesting longer-term finding comes from a 2026 retrospective observational cohort study by Seagers et al.. Among people with medial knee OA, those whose gait modification immediately reduced compressive and shear knee forces had slower cartilage degeneration on T2 MRI after 12 months.

This extends the case beyond KAM, but it does not establish that retraining protected cartilage. The study was retrospective, so people who responded mechanically could have differed at baseline in cartilage status, disease severity, muscle function, physical activity, body-weight trajectory, or adherence to the modified gait. T2 MRI is an objective structural outcome; the observed association still requires confirmation in a prospective randomized trial.

Wearables are becoming measurement tools, not proven home treatments

Wearable technology may make individualized assessment more practical. In a 2025 biomechanics study by Snyder et al., custom instrumented insoles and machine learning predicted medial joint contact force, KAM, and knee flexion moment during daily activities. The study included walking, running, sit-to-stand, and stand-to-sit tasks.

That is a measurement study, not a gait-retraining trial. It supports the technical possibility of estimating knee load outside a traditional gait lab, but no independent randomized trial has shown that instrumented insoles can deliver effective real-time home retraining or improve long-term OA outcomes.

Who should consider gait retraining?

The evidence applies most directly to early medial knee OA, particularly Kellgren-Lawrence grades I–II. It does not establish benefit for grades III–IV, where structural changes may limit what redistributing load can accomplish.

• Seek an assessment that measures your own KAM response before committing to toe-in or toe-out walking.
• Favor a supervised, approximately six-week program with measurable feedback; that model has the clearest trial support.
• Treat symptom improvement as useful, but do not assume it proves reduced cartilage damage.
• Keep strengthening, general activity, and other conservative treatments in the discussion rather than assuming gait retraining has surpassed them.
• Ask how the clinician will monitor symptoms at the hip, ankle, and lateral knee. Completed trials have not adequately characterized compensatory loading at those sites.

Gait retraining has a credible biomechanical rationale and a promising early clinical signal. Its next test is not whether people can change their foot angle; it is whether individualized, durable load reduction improves outcomes better than established conservative care in a sufficiently large trial.