Using real-time biofeedback to alter running biomechanics: a randomized controlled trial

Key points

1. 01Eighty percent of runners switched to a non-rearfoot strike (NRFS) right after a 2-hour group training session (91 of 114, P < .001).
2. 02Most who switched kept the new pattern at 1 year (control 69%, biofeedback 75%), with no difference between groups.
3. 03The biofeedback sensor group did no better than control, mainly because sensor compliance was very low (14% of running mileage).
4. 04Overall 1-year injury rates were similar for heel strikers (37%) and non-heel strikers (30%), a difference that was not significant.
5. 05Heel strikers had nearly 6 times the risk of knee injury versus non-heel strikers (relative risk 5.64, 95% CI 1.90 to 16.8).

How it was conducted

Design

Prospective randomized controlled trial with 1-year follow-up (NCT02495792)

Participants

128 rearfoot-strike runners (Department of Defense beneficiaries, West Point NY, ages 18 to 50); 114 completed the 1-year follow-up

Groups

Control group received the 2-hour gait training only; biofeedback group also got an IMU tibial sensor and mobile app giving real-time feedback

Intervention

4-week home strengthening program, then a 2-hour group session teaching a non-rearfoot strike with a target cadence near 180; sensor alerted when tibial shock exceeded 6 g

Outcomes

Foot strike pattern (treadmill video at 240 Hz) at baseline, post-training, 6 months, 1 year; injuries via weekly email surveys and medical record review

What they found

• Eighty percent of runners showed a non-rearfoot strike after the training session (91/114, P < .001).
• Retention at 1 year did not differ between groups (control 69%, P = .29; biofeedback 75%, P = .36); no difference at 6 months (chi-square = 0.67, df = 1, P = .41) or 1 year (chi-square = 0.31, P = .55).
• Self-reported sensor compliance was only 14%.
• Cadence rose significantly in both groups: control +10 steps/min post-training and +7 at 1 year; biofeedback +11 post-training and +8 at 1 year (all P < .001).
• At 1 year, 12 of 32 heel strikers (37%) and 25 of 82 non-heel strikers (30%) were injured, no significant difference (chi-square = 0.51, P = .47, df = 1; effect size 0.05, post-hoc power 0.08).
• Relative risk of knee injury in heel strikers was 5.64 (95% CI 1.90 to 16.8; P = .02); foot pain RR for NRFS 2.60 (95% CI 0.81 to 8.41, P = .11) and lower leg pain RR 1.33 (95% CI 0.46 to 3.83, P = .60).
• Mean cadence did not differ between injured (173 +/- 12.3 steps/min) and non-injured runners (173 +/- 10.9), t = 0.12, P = .89 (95% CI -4.77 to 4.19).
• Inter-rater agreement was substantial for foot strike (kappa 0.78) and excellent for cadence (ICC2,1 0.94).

Limitations

• Very low compliance with the biofeedback sensor (14% of mileage) means the device's true potential could not be tested.
• The study was underpowered for the injury comparison; the authors estimate about 2000 runners would have been needed to detect the observed 5% difference.
• Foot strike was assessed only on a treadmill in the lab where runners knew they were observed, so it is unknown whether they used the same pattern running outdoors.
• Participants were a specific military beneficiary population, which may limit how far results generalize to other runners.

Why it matters

For patients

If you are a heel striker with knee problems, learning a non-heel-strike form through a structured gradual program may help your knees, but switching is unlikely to lower your overall injury risk.

For clinicians

A brief group-based gait retraining session can reliably shift and sustain foot strike pattern without a wearable sensor, and may be worth considering specifically to reduce knee injuries in heel strikers.

For readers

Real-time biofeedback wearables only help if people actually use them, and changing one biomechanical habit shifts where injuries happen more than whether they happen.