Fig. 2: Primary endpoint and responder analyses.

Between-group differences in A treatment-related change in walking speed; B responder rates—with responder defined as (1) a change in walking speed larger than the 0.16 m/s MCID or (2) a change in walking speed >0.16 m/s MCID and a post-training walking speed above 0.80 m/s; and C the time course of walking speed improvement. The trial’s intent-to-treat analyses included 72 study participants (n = 40 in the InTandem group [shown in green] and n = 32 in the Active Control group [shown in blue]). A reports individual subject walking speed changes and the mean values ± standard error for each treatment group. Shown are the results of a 2 × 2 General Linear Mixed Model (GLMM) used to evaluate between-group differences in the treatment effect (i.e., the treatment × time interaction: F(1,49) = 6.6, p = 0.013). B reports individual subject data points relative to different clinically meaningful speed change thresholds. Shown are the results of two Chi-Square (χ²) tests used to evaluate between-group differences in the number of responders (responder analysis 1: χ²(1) = 6.70, p = 0.01; responder analysis 2: χ²(1) = 6.47, p = 0.01). Note: the horizontal axes for the InTandem (left) and Control (right) groups are oriented in opposite directions. C reports the results of a 2 × 17 × 2 GLMM used to evaluate between-group differences in the time course of change in walking speed (i.e., the Treatment × Session interaction: F(16,206) = 1.79, p = 0.034). Each data point is the inter-subject session average within each group. Source data are provided as a Source Data file.