Normalised plantar pressure measures are reliable, whereas absolute force and gait-curve temporal measures are not: a test–retest reliability study of a plantar pressure platform in patients with knee disorders and healthy controls

Abstract Background Plantar pressure platforms are widely used to characterise foot loading in knee disorders, but the measurement error of individual output parameters is rarely reported. Whether a parameter can detect real change depends on its test–retest reliability and on how large that error is relative to the values measured. We assessed the relative and absolute reliability of static, regional and dynamic plantar pressure and gait parameters in patients with knee disorders and healthy controls. Methods Thirty-seven patients with knee disorders and 15 healthy controls completed three consecutive trials of quiet standing and barefoot walking on an Emat plantar pressure plate (3doeCom, China; 50 × 50 cm, 4 sensors/cm², 300 Hz). We derived 22 dynamic gait-curve, 60 regional (force, contact area and pressure across ten zones) and 12 static load-distribution parameters, and computed ICC(2,1) with 95% confidence intervals, standard error of measurement, minimal detectable change at 95% (MDC95), within-subject coefficient of variation, and the trials needed to reach ICC 0.75. Reporting followed the GRRAS statement. Physiologically impossible values (negative durations or forces) were treated as missing. Results Reliability varied sharply with parameter type. All 12 static percentage load-distribution parameters were reliable (ICC 0.77–0.97), as were regional peak pressures (16/17 ≥ 0.75). Regional force in kilograms reached ICC ≥ 0.75 in only 4 of 20 parameters in patients (13 of 20 in controls), and dynamic gait-curve parameters in only 2 of 22. For unreliable parameters, MDC95 exceeded the group mean, sometimes severalfold (right metatarsal peak force: MDC95 97.3 kg versus a mean of 78.5 kg). Absolute force was measured less reliably in patients than in controls. Most dynamic parameters required three or more trials, and several could not reach acceptable reliability at any number. Conclusions Normalised, ratio-derived plantar pressure measures are reliable in knee disorders; absolute force output and the temporal landmarks of the force curve are not. Repeated-measures studies should be built on normalised pressure measures, record several trials and walking speed, and report ICC, SEM and MDC95 for every parameter used.