Anisha Suri1, Andrea L Rosso2, Jessie VanSwearingen3, Leslie M Coffman3, Mark S Redfern4, Jennifer S Brach3, Ervin Sejdić1,4. 1. Department of Electrical and Computer Engineering, Swanson School of Engineering, University of Pittsburgh, Pennsylvania, USA. 2. Department of Epidemiology, School of Public Health, University of Pittsburgh, Pennsylvania, USA. 3. Department of Physical Therapy, School of Rehabilitation Sciences, University of Pittsburgh, Pennsylvania, USA. 4. Department of Bioengineering, Swanson School of Engineering, University of Pittsburgh, Pennsylvania, USA.
Abstract
BACKGROUND: The relation of gait quality to real-life mobility among older adults is poorly understood. This study examined the association between gait quality, consisting of step variability, smoothness, regularity, symmetry, and gait speed, and the Life-Space Assessment (LSA). METHOD: In community-dwelling older adults (N = 232, age 77.5 ± 6.6, 65% females), gait quality was derived from (i) an instrumented walkway: gait speed, variability, and walk ratio and (ii) accelerometer: signal variability, smoothness, regularity, symmetry, and time-frequency spatiotemporal variables during 6-minute walk. In addition to collecting LSA scores, cognitive functioning, walking confidence, and falls were recorded. Spearman correlations (speed as covariate) and random forest regression were used to assess associations between gait quality and LSA, and Gaussian mixture modeling (GMM) was used to cluster participants. RESULTS: Spearman correlations of ρ p = .11 (signal amplitude variability mediolateral [ML] axis), ρ p = .15 and ρ p = -.13 (symmetry anterior-posterior-vertical [AP-V] and ML-AP axes, respectively), ρ p = .16 (power V), and ρ = .26 (speed), all p <.05 and marginally related, ρ p = -.12 (regularity V), ρ p = .11 (smoothness AP), and ρ p = -.11 (step-time variability), all p <.1, were obtained. The cross-validated random forest model indicated good-fit LSA prediction error of 17.77; gait and cognition were greater contributors than age and gender. GMM indicated 2 clusters. Group 1 (n = 189) had better gait quality than group 2 (n = 43): greater smoothness AP (2.94 ± 0.75 vs 2.30 ± 0.71); greater similarity AP-V (.58 ± .13 vs .40 ± .19); lower regularity V (0.83 ± 0.08 vs 0.87 ± 0.10); greater power V (1.86 ± 0.18 vs 0.97 ± 1.84); greater speed (1.09 ± 0.16 vs 1.00 ± 0.16 m/s); lower step-time coefficient of variation (3.70 ± 1.09 vs 5.09 ± 2.37), and better LSA (76 ± 18 vs 67 ± 18), padjusted < .004. CONCLUSIONS: Gait quality measures taken in the clinic are associated with real-life mobility in the community.
BACKGROUND: The relation of gait quality to real-life mobility among older adults is poorly understood. This study examined the association between gait quality, consisting of step variability, smoothness, regularity, symmetry, and gait speed, and the Life-Space Assessment (LSA). METHOD: In community-dwelling older adults (N = 232, age 77.5 ± 6.6, 65% females), gait quality was derived from (i) an instrumented walkway: gait speed, variability, and walk ratio and (ii) accelerometer: signal variability, smoothness, regularity, symmetry, and time-frequency spatiotemporal variables during 6-minute walk. In addition to collecting LSA scores, cognitive functioning, walking confidence, and falls were recorded. Spearman correlations (speed as covariate) and random forest regression were used to assess associations between gait quality and LSA, and Gaussian mixture modeling (GMM) was used to cluster participants. RESULTS: Spearman correlations of ρ p = .11 (signal amplitude variability mediolateral [ML] axis), ρ p = .15 and ρ p = -.13 (symmetry anterior-posterior-vertical [AP-V] and ML-AP axes, respectively), ρ p = .16 (power V), and ρ = .26 (speed), all p <.05 and marginally related, ρ p = -.12 (regularity V), ρ p = .11 (smoothness AP), and ρ p = -.11 (step-time variability), all p <.1, were obtained. The cross-validated random forest model indicated good-fit LSA prediction error of 17.77; gait and cognition were greater contributors than age and gender. GMM indicated 2 clusters. Group 1 (n = 189) had better gait quality than group 2 (n = 43): greater smoothness AP (2.94 ± 0.75 vs 2.30 ± 0.71); greater similarity AP-V (.58 ± .13 vs .40 ± .19); lower regularity V (0.83 ± 0.08 vs 0.87 ± 0.10); greater power V (1.86 ± 0.18 vs 0.97 ± 1.84); greater speed (1.09 ± 0.16 vs 1.00 ± 0.16 m/s); lower step-time coefficient of variation (3.70 ± 1.09 vs 5.09 ± 2.37), and better LSA (76 ± 18 vs 67 ± 18), padjusted < .004. CONCLUSIONS: Gait quality measures taken in the clinic are associated with real-life mobility in the community.
Authors: Dylan Kobsar; Chad Olson; Raman Paranjape; Thomas Hadjistavropoulos; John M Barden Journal: Gait Posture Date: 2013-09-19 Impact factor: 2.840
Authors: Stephanie Studenski; Subashan Perera; Kushang Patel; Caterina Rosano; Kimberly Faulkner; Marco Inzitari; Jennifer Brach; Julie Chandler; Peggy Cawthon; Elizabeth Barrett Connor; Michael Nevitt; Marjolein Visser; Stephen Kritchevsky; Stefania Badinelli; Tamara Harris; Anne B Newman; Jane Cauley; Luigi Ferrucci; Jack Guralnik Journal: JAMA Date: 2011-01-05 Impact factor: 56.272