Using wearable sensors to characterize gait after spinal cord injury: evaluation of test-retest reliability and construct validity.

Abstract

Quantitative cross-sectional study. Evaluate the test-retest reliability and the construct validity of inertial measurement units (IMU) to characterize spatiotemporal gait parameters in individuals with SCI. Two SCI rehabilitation centers in Canada. Eighteen individuals with SCI participated in two evaluation sessions spaced 2 weeks apart. Fifteen able-bodied individuals were also recruited. Participants walked 20 m overground under five conditions that challenged balance to varying degrees. Five IMU were attached to the lower-extremities and the sacrum to collect the mean and the coefficient of variation of five gait parameters (gait cycle time, double-support percentage, cadence, stride length, stride velocity). Intra-class correlation coefficients (ICC) were used to evaluate the test-retest reliability. Linear mixed-effects models were used to compare the five walking conditions to evaluate known-group validity while Spearman's correlation coefficients were used to characterize the level of association between gait parameters and the Mini BESTest (MBT). Cadence was reliable across all walking conditions. Reliability was higher for the mean (ICC = 0.55-0.98) of the parameters compared to their coefficient of variation (ICC = 0.16-0.97). Cadence collected with IMU had construct validity as their values differed across walking conditions and groups of participants. The coefficient of variation was generally better than the mean to show differences across the five walking conditions. The MBT was moderately to strongly associated with mean cadence (ρ ≥ 0.498) and its coefficient of variation (ρ ≤ -0.49) during most walking conditions. IMU provide reliable and valid measurements of gait parameters in ambulatory individuals with SCI.