Comparative Validity of Accelerometer-Based Measures of Physical Activity for People With Multiple Sclerosis.
ABSTRACT Coote S, O'Dwyer C. Comparative validity of accelerometer-based measures of physical activity for people with multiple sclerosis. OBJECTIVE: To estimate the criterion validity of accelerometer-based devices as measures of steps and energy expenditure in healthy controls and people with multiple sclerosis (MS) with varying disability levels during everyday activities. DESIGN: Cross-sectional study. SETTING: University research room. PARTICIPANTS: People with MS who used at most a stick to walk outdoors (MS-A; n=19), people with MS who used bilateral support for gait (MS-B; n=11), and healthy controls (n=15). INTERVENTIONS: Participants completed 85 minutes of 9 scripted everyday activities. MAIN OUTCOME MEASURES: Estimates of metabolic equivalent thresholds (METs) and kilocalories from a portable metabolic system, and steps counted from video of the activities. Step and MET estimates from an integrative accelerometer and from a uniaxial accelerometer, and kilocalorie estimates from the integrative accelerometer. RESULTS: The uniaxial accelerometer had >30% error for steps for all groups. MET estimates had an intraclass correlation coefficient (ICC) <0.2 for all groups. For the integrative accelerometer, step estimates for controls had an ICC of .69 and <1% error. The step estimates for MS-A and MS-B groups had >20% error. The MET estimates had an ICC of .50 to .65 and 6% to 15% error. Kilocalorie estimates had 2.9% error for controls, 8.16% for MS-A, and 2.56% for MS-B groups. ICCs were all >.67, and mean differences from criterion were <20kcal. CONCLUSIONS: The agreement between steps and MET estimates from both devices and the criterion was poor, particularly for people with MS. Only the step and MET estimates for the control group for the integrative accelerometer were not significantly different from the criterion. Kilocalorie estimates from the integrative accelerometer using the proprietary algorithms of the device provide the most valid estimate of physical activity during activities of daily living for people with a range of walking disabilities resulting from MS.