Eight Functional outcomes, two dimensions: a cross-sectional analysis of scale dependence in spatiotemporal gait...

Spatiotemporal gait parameters have been linked to clinical outcomes after stroke, but whether determinants differ between outcomes is untested in one cohort.

To determine how many dimensions eight routinely used outcomes represent, and whether between-outcome differences in determinants reflect construct or scale.

Cross-sectional study, 244 patients with subacute stroke (Functional Ambulation Category [FAC] ≥2). Fifteen walkway parameters were modeled, with one predictor pool, against Berg Balance Scale (BBS), gait velocity, Timed Up and Go (TUG), 10-meter (10MWT) and 2-minute (2MWT) walk tests, FAC, and Functional Independence Measure (FIM). Principal component analysis assessed dimensionality; partial-F tests and reciprocal transformation tested determinant exchangeability.

The outcomes reduced to two components (90.6% of variance): walking capacity (78.0%), loaded by all walking measures and FAC, and one defined by FIM (12.6%); velocity and 10MWT shared 99.4%. Bilateral step length explained BBS, velocity and 2MWT; step-time asymmetry and cadence explained TUG and 10MWT (adjusted R² 0.83-0.92). The temporal block added variance only for time-denominated outcomes (ΔadjR² = 0.073, 0.077), and reciprocal transformation reversed every profile. Cadence predicted FAC (OR = 1.027, 1.002-1.054, p = 0.037); FIM was unexplained (0.05).

These outcomes represent two dimensions, and which parameters appear dominant depends on whether an outcome is expressed as a time or a rate, so outcome-specific models are not evidence for outcome-specific training targets. The implication is one of measurement: one primary walking outcome suffices, and independence must be assessed separately.