Does walking strategy in older people change as a function of walking distance? Bijan Najafi a,b, *, Jorunn L. Helbostad c , Rolf Moe-Nilssen d , Wiebren Zijlstra e , Kamiar Aminian b a Center for Lower Extremity Ambulatory Research (CLEAR) at Dr. William M. Scholl College of Podiatric Medicine, Rosalind Franklin University of Medicine and Science, North Chicago, II 60064, USA b Ecole Polytechnique Federale de Lausanne, Laboratory of Movement Analysis & Measurement, CH-1015 Lausanne, Switzerland c Department of NeuroScience, Norwegian University of Science and Technology (NTNU), Trondheim, Norway d Department of Public Health and Primary Health Care, University of Bergen, N-5009 Bergen, Norway e Center for Human Movement Sciences, University Medical Center Groningen, The Netherlands 1. Introduction Gait is usually assessed using laboratory-based systems. Such measurements are considered as gold standard, despite significant limitations [1,2]. A major disadvantage of such an approach is that laboratory-based measurement of human movement does not replicate the true conditions that the subjects are active in. In addition, unless a treadmill is used, space limitations often limit gait analysis to a few steps. Advances in the technology of body-worn sensors during the last decade has encouraged investigators to use these sensors for measuring various aspects of human performance. These included spatio-temporal parameters of gait [1,3,4], joint and segment angles [5–10], monitoring of daily physical activity [11–17], and evaluation of the risk of falling [18,19] or fear of falling [20]. These studies are based on the use of miniaturized and integrated sensors in combination with lightweight, small measuring devices that can be carried without interfering with normal activity [1,2,11]. One of the main advantages of body-worn sensors compared to labora- tory-based measuring systems is that they are ambulatory and can be used in free conditions continuously over long periods of time. Despite this key advantage, clinical protocols based on gait analysis in outdoor and long walking distance have not yet been established. On the other hand, since such technologies are not restricted to gait lab environment, there is no existing evidence on whether the extracted gait parameters in different walking distance conditions are comparable, or on the number of strides that should be considered sufficient to extract a reliable estimation of both the mean and variability of the parameters. Recently, there has been a growing interest in measuring gait and gait variability for evaluation of risk of falling and for fall prevention [1,2,21–28]. Gait measurements over long walking distances (e.g. 20 m) may enable the design of new paradigms for improved evaluation of the risk of falling. Furthermore, using a longer walking distance can enable researchers to accurately Gait & Posture 29 (2009) 261–266 ARTICLE INFO Article history: Received 23 October 2007 Received in revised form 4 September 2008 Accepted 8 September 2008 Keywords: Gait Elderly people Long walking vs. short walking distance Test–retest reliability Gait variability Gait strategy Ambulatory system Body-worn sensor Physilog GaitRite Fall prevention ABSTRACT This study investigates whether the spatio-temporal parameters of gait in the elderly vary as a function of walking distance. The gait pattern of older subjects (n = 27) over both short (SWD < 10 m) and long (LWD > 20 m) walking was evaluated using an ambulatory device consisting of body-worn sensors (Physilog 1 ). The stride velocity (SV), gait cycle time (GCT), and inter-cycle variability of each parameter (CV) were evaluated for each subject. Analysis was undertaken after evaluating the errors and the test– retest reliability of the Physilog device compared with an electronic walkway system (GaitRite 1 ) over the SWD with different walking speeds. While both systems were highly reliable with respect to the SV and GCT parameters (ICC > 0.82), agreement for the gait variability was poor. Interestingly, our data revealed that the measured gait parameters over SWD and LWD were significantly different. LWD trials had a mean increase of 5.2% (p < 0.05) in SV, and a mean decrease of 3.7% (p < 0.05) in GCT compared with SWD trials. Although variability in both the SV and GCT measured during LWD trials decreased by an average of 1% relative to the SWD case, the drop was not significant. Moreover, reliability for gait variability measures was poor, irrespective of the instrument and despite a moderate improvement for LWD trials. Taken together, our findings indicate that for valid and reliable comparisons, test and retest should be performed under identical distance conditions. Furthermore, our findings suggest that the older subjects may choose different walking strategies for SWD and LWD conditions. ß 2008 Elsevier B.V. All rights reserved. * Corresponding author at: Center for Lower Extremity Ambulatory Research (CLEAR); Rosalind Franklin University of Medicine & Science; 3333 Green Bay Rd.; North Chicago, IL 60064. Tel: +1 847-578-8456. E-mail addresses: najafi.bijan@gmail.com, Bijan.najafi@rosalindfranklin.edu (B. Najafi). Contents lists available at ScienceDirect Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost 0966-6362/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2008.09.002