Background: Gait and balance impairments, including increased gait variability, are prevalent in people with bilateral vestibulopathy (BVP). Wearable accelerometers may provide a clinically feasible method to objectively assess gait variability but have not yet been explored in BVP.

Research question: Is accelerometer-based assessment of step time variability during treadmill walking valid in people with BVP?

Methods: Adults with BVP and age-sex-matched healthy controls walked at 0.6 m/s, 0.8 m/s, and 1.0 m/s on the treadmill of the Computer Assisted Rehabilitation Environment. We examined differences in step time means and coefficients of variation (CoV) between accelerometery (single lower back sensor; MOX1) and 3D motion capture and force plates (Vicon). Validity was assessed using intraclass correlation coefficients (ICC_3,1), Pearson and Spearman correlation coefficients, and Bland-Altman analyses to determine agreement, association and consistency between the methods. Validity was additionally assessed by comparing statistical significant differences and the effect sizes between the groups using each method.

Results: Mean step time showed moderate to excellent agreement between methods, while step time CoV showed poor agreement and proportional bias. Accelerometery showed consistent between-group significance and effect size values, particularly at 0.6 m/s, although effect sizes were larger in motion capture data than in accelerometer data.

Significance: An accelerometer-based assessment is valid for assessing mean step time in people with BVP. For assessing step time variability, it can distinguish between known groups (particularly at slower speeds) but does not demonstrate criterion validity. Before clinical application, test-retest reliability and sensitivity to change should be assessed in BVP.