Laboratory-based turning performance during walking in people with mild cognitive impairment and dementia

BackgroundMobility impairments during straight-line walking show utility in detecting dementia-related cognitive impairment (e.g., Alzheimer's disease). Wearable technologies can now derive mobility outcomes related to turning, considered a more cognitively complex movement. However, it is unclear which turning variables would best detect cognitive impairment.ObjectiveThis study aimed to develop a data-driven model of turning performance relevant to cognitive impairment and compare selected turning variables between people with cognitive impairment and cognitively-intact older adults.Methods77 people with cognitive impairment (31 mild cognitive impairment, 46 dementia) and 28 cognitively-intact older adults. performed six intermittent walking trials with a 180° turn while wearing an inertial measurement unit (APDM Opal) attached to their lower back. 102 spatiotemporal and signal-based turning variables were captured. Following data reduction, 15 variables underwent exploratory factor analysis. One-way ANCOVA, adjusting for age, sex and height, explored between-group differences (Bonferroni correction applied).ResultsThree factors emerged: turn initiation, magnitude and smoothness, accounting for 67% of the total variance in turning performance. The cognitively impaired group demonstrated significant differences in turn initiation (start-phase root mean square (RMS) in vertical direction, start-phase jerk RMS in vertical/mediolateral/anterior-posterior directions) and magnitude (mid-phase RMS in mediolateral direction, mean RMS in vertical direction) domains (p < 0.003) with moderate-large effect sizes (Partial η² = 0.09-0.17). Turn initiation and magnitude factor scores were associated with cognitive performance.ConclusionsKey findings suggest that people with cognitive impairment exhibit smaller and slower turning movements specifically during turn initiation and demonstrate lower signal magnitude (RMS) across the turn. Further work should consider use of these variables in predicting incidence of cognitive impairment in at-risk populations.