Habitual Physical Activity and Walking Endurance Following Inpatient Rehabilitation in Children With Cerebral Palsy

Background: Cerebral palsy (CP) at GMFCS level II is the most functionally heterogeneous ambulatory group, with high intra-group variability in walking endurance and daily physical activity. The relationship between six-minute walk test (6MWT) outcomes and accelerometer-based monitoring following inpatient rehabilitation remains unclear. Aim: The aim of this study was to assess correlations between wear-time-normalised habitual physical activity (PA) parameters and rehabilitation-induced change in walking endurance, and to determine whether objectively measured spatiotemporal gait parameters explain the inverse relationship between metabolic equivalent of task (MET)-based PA classification and baseline 6MWT performance in this population. Methods: Twenty-five children with CP at GMFCS II (female: 10, male: 15; age: 13.40 ± 1.12 years) completed a two-week inpatient rehabilitation programme. The 6MWT was performed on the first and final day. PA was monitored over 72 hours using a wearable device, which was removed during sleep, bathing, and activities that might cause damage to the device; data were not recorded during structured therapy sessions. Participants were classified as active (n = 13) or inactive (n = 12). Analysis included paired t-test, Pearson and Spearman correlations, Mann-Whitney U test, and Cohen's dz. Spatiotemporal gait parameters were obtained from instrumented gait analysis. Results: Significant improvement in 6MWT distance was observed in the entire group (moderate effect), and the active subgroup (medium-to-large effect); no improvement was found in the inactive subgroup. No significant correlations between normalised PA measures and 6MWT improvement were identified. At baseline, the active subgroup showed lower 6MWT distance despite higher step accumulation. Between-group PA differences were significant. Spatiotemporal analysis revealed inefficient gait biomechanics in the active subgroup (lower walking speed, shorter step and stride length, longer stance phase). An exploratory analysis found a negative correlation between baseline 6MWT and rehabilitation-induced change in the inactive subgroup. Conclusions: Two weeks of inpatient rehabilitation produced clinically meaningful improvements in walking endurance, particularly in the active subgroup. Habitual PA intensity and endurance gains were not correlated. Inefficient gait biomechanics in the active subgroup may help explain the paradoxical inverse relationship between MET classification and baseline 6MWT performance.