Background/Purpose
Walking is the most common form of physical activity for adults with intellectual disabilities (ID; Draheim, McCubbin & Williams, 2002). Pedometers are a useful tool for measuring walking behavior. There is preliminary evidence for the accuracy of pedometers for adults with ID (Stanish, 2004); however, there is little validity evidence of using pedometers for adults with Down syndrome (DS). Considering unique gait and body characteristics of individuals with DS, it is important to evaluate pedometer accuracy. The primary purpose of the study was to examine if pedometer accuracy differed between pedometer models, walking speeds, body characteristics, and adults with and without DS to make recommendations for future use of pedometers.
Method
Twenty adults with DS (12 female, 8 male) and 24 adults (14 female, 10 male) without a disability wore two piezoelectric (Omron HJ-112) and two spring-levered (Yamax Digiwalker SW-200) pedometers at the waist on one side of the body. Testing conditions included walking a controlled course for three trials of 2 minutes at self-selected, slow and fast speeds. Steps taken were observed and counted to compare with pedometer measurements. Waist and hip circumferences were also measured for waist-to-hip ratios (WHR).
Analysis/Results
A 2x2x3 (group-by-model-by-speed) repeated measures ANOVA and a follow-up 2x2x3 repeated measured ANCOVA with WHR as a covariate were employed to examine differences on the absolute rate of errors by pedometers. Absolute error was calculated as (|Observed steps – Pedometer| / Observed steps). Error rates ranged from 8% to 22% for DS group and 1% to 16% for control group. There was a significant difference in absolute error between individuals with and without DS, (F(1,42)=9.06, p<0.01, h2=.18). There was also a significant model-by-speed interaction, (F(2,84)=13.14, p<0.001, h2=.24). When the covariate of WHR was added, there remained a significant group difference, (F(1,41)=7.352, p<.05, h2=.15), but no significant within-subjects effects (p>0.1).
Conclusions
Overall, the piezoelectric pedometer was more accurate than the spring-levered pedometer, particularly at slower walking speeds. The WHR covariate indicates that differences between models and speeds can be explained by increasing waist-to-hip-ratios. However, significant group differences demonstrate that pedometers have greater measurement error when used by adults with DS. Given the moderately high magnitude of absolute error observed in pedometry for adults with DS, use of this measurement tool requires further investigation. If pedometers are used to monitor walking behaviors in adults with DS, the use of piezoelectric models is recommended.