Abstract
BACKGROUND: Stiff-Knee gait affects 25-75 % of individuals with post-stroke gait impairment and is typically defined as reduced swing phase knee flexion. Different studies use various measures to identify Stiff-Knee gait, such as peak swing knee flexion angle, timing of peak knee flexion, knee range of motion, and ankle push-off acceleration, leading to inconsistent results. METHODS: This study used univariate cluster analysis to examine the independence, consistency, validity, and accuracy of different definitions in 50 post-stroke individuals (24 with and 26 without Stiff-Knee gait), as determined by a physiatrist. Spearman's rank correlation was used for correlation analysis, and five clustering techniques along with clinician evaluations were used for validity analysis. FINDINGS: Correlation analysis showed that peak knee flexion timing and knee hyperextension are poorly correlated with reduced swing-phase knee flexion angle (ρ = -0.09 and ρ = -0.26 respectively). Validity analysis indicated that the between-limb difference in peak swing knee flexion angle and peak swing knee flexion angle at self-selected gait speeds were the most valid differentiators. At the fastest comfortable gait speed, the between-limb difference of peak knee flexion angle had the highest sensitivity, lowest specificity, and highest F1 scores. INTERPRETATION: We determined thresholds of less than 44.3° for peak swing knee flexion angle and greater than 17.0° for the between-limb difference of peak knee flexion angle identify Stiff-Knee gait during self-selected walking. We recommend using the difference in peak swing knee flexion angle between limbs to diagnose post-stroke Stiff-Knee gait due to its robustness to changes in gait speed.