ABSTRACT
BACKGROUND: Walking speed discriminates levels of functionality in stroke survivors, but its relationship with neuromuscular parameters remains unclear. We aimed to (1) verify relationships between walking speed, peak torque, and rate of torque development in individuals with chronic stroke and (2) investigate whether adjusting the predictive model for assistive device use and lower limb motor function improves its accuracy. METHODS: Twenty-nine stroke survivors (time post-stroke ≥6 months) were assessed for walking speed, motor function, torque of knee extensors, and rate of torque development. Hierarchical multiple regression was conducted to infer the contribution of assistive device use and lower limb motor function in the proportion of variance in walking speed explained by neuromuscular parameters. Adequacy of regression models was verified using Shapiro-Wilk test and visual inspection of histograms. FINDINGS: Rate of torque development measures presented higher coefficients of determination (R2 = 0.399-0.457) than peak torque (R2 = 0.333). However, no differences were observed between correlation coefficients (P > 0.05). When adjusted for assistive device use and lower limb motor function, coefficients of determination of all neuromuscular parameters were improved (P < 0.05). Regression models showed good adequacy. INTERPRETATION: Peak torque and rate of torque development from knee extensors are adequate predictors of walking speed in stroke survivors. Adjusting predictive models for assistive device use and lower limb motor function increases predictive capacity. These findings may have repercussions for assessing chronic stroke individuals and developing treatment strategies.
