Physical activity across abilities: how assistive devices influence accelerometer measurements

Physical activity is an important factor of health for individuals of all ages and activity abilities. Many studies have developed and validated physical activity measurements for healthy populations. However, the accuracy and ability to apply these measures to populations with gait deviations and/or walking aids is less studied. This study had two main aims: 1) evaluate a step count algorithm by Maylor et al. at different walking speeds and accelerometer placement and 2) evaluate effect of AD use on acceleration profile metrics and step count accuracy.

Thirty healthy adults (15F, age 32 ± 12.5 years) walked 30.5m under several conditions: three walking speeds, four types of assistive devices, and up to three different weight bearing conditions. ActiGraph™ GT3x accelerometers were worn on the wrist, hip and ankle to measure total acceleration and extract steps counts through post processing. Observed steps were assessed with a hand counter.

Using previously reported step count parameters resulted in underestimation of step counts varying from 26% to 89% for the wrist and hip wear locations. Optimized parameters reduced error by 29.5%, 35.0% and 53.6% at wrist, hip, and ankle locations. While total accelerations were least affected by assistive device use at the hip, step counts were least affected at the ankle - showing acceptable results (< 25% mean percent error) while using cane, crutches and wheeled walker without altered weight bearing (i.e., for balance only).