Abstract
How lower-limb kinematic differences are distributed across prolonged running remains unclear. We examined sagittal-plane lower-limb kinematics in 25 amateur long-distance runners during a constant-speed treadmill running-to-fatigue task. Target speeds were individualized using runners’ latest half-marathon or marathon performances. Markerless three-dimensional motion capture assessed kinematics and spatiotemporal parameters using Baseline recordings and those assigned to nominal 20%, 40%, 60%, 80%, and 100% stages. Analyses adjusted for multiplicity, with separate sensitivity analyses excluding early-terminating participants and those with larger timing deviations from nominal stages. Compared with Baseline, peak knee flexion angle during braking increased at 20% (RPE = 12.64), and this finding was retained in both sensitivity analyses. However, broader findings were sensitive to participant inclusion, with only three of 11 primary-analysis outcomes retained after BH-FDR correction in the endpoint analysis and seven in the timing-deviation analysis. At 40% (RPE = 13.72), more outcomes differed from Baseline, involving the hip, knee, and ankle. Later differences involved multiple gait phases, mainly at the hip and knee. Spatiotemporal parameters showed no significant overall stage effects. Kinematic differences were detectable at early and intermediate stages while mean RPE was not yet high. Repeated sampling can capture kinematic information missed by Baseline-to-final-stage comparisons.