The reliability of gait variability measures for individuals with Parkinson's disease and healthy older adults - The effect of gait speed

Background: Step-to-step variability is a marker of reduced motor control and a frequently studied outcome measure in neurodegenerative disorders such as Parkinson's disease (PD) as compared to healthy older adults (HOA). To challenge motor control of gait, walking should be tested at different gait speeds. Good reliability is essential, and gait variability estimates show good reproducibility when sampled at normal gait speed. The aim was therefore to investigate if gait variability could be reliably sampled at slow and fast speeds for individuals with PD and HOA by evaluating test-retest reliability. Methods: 29 (14 males) subjects with idiopathic PD, Hoehn & Yahr 2 (n = 18) and 3,= 60 years, and 25 age matched HOAwere included. Spatiotemporal gait data was collected (GAITRite) during slow, normal, and fast walking on two occasions. Results: Measurement error was lowest for gait variability estimates based on 40 steps in both groups. This was true across all speeds in HOA, but only for normal and fast gait speeds in the PD cohort. Due to increased homogeneity in the variability estimates intraclass correlation coefficients (ICC) were low for HOA, except for step width variability. In the PD cohort ICCs were good to excellent for temporal-and step width gait variability across speeds. Conclusion: HOA demonstrated reliable gait variability estimates across all speeds, whereas Individuals with PD were reliable at normal and fast gait speeds only Estimates should be based on at least 40 steps. Step width variability was overall the most reliable variable across groups and speed conditions.