Energy Expenditures during Wheelchair Propulsion Derived from a Body-worn Sensor versus Indirect Calorimetry

Abstract

The Bodymedia SenseWear armband has been previously shown to be a valid and practical tool to assess energy expenditure (EE) in upright, ambulatory adults. However, wheelchair propulsion is a dissimilar movement pattern to walking, employing both bi-lateral symmetry and a greater vigour of arm motion. PURPOSE: The purpose of this study was to identify relationships between energy expenditure (EE) derived from the SenseWear armband versus indirect calorimetry in wheelchair users. METHODS: Ten individuals with spinal cord injury (SCI) performed two different wheelchair treadmill protocols, one with a constant grade (1%) but six different velocities, and the other at a constant velocity (2 km·h-1), but with 6 variable grades. EE was measured concurrently from breath-by-breath indirect calorimetry and the SenseWear armband (AB). To assess internal validity of the AB and its equations employed to transform skin temperature, GSR, heat flux and accelerometry to EE, estimated submaximal EE and VO2 were contrasted using each system. RESULTS: Across all treadmill velocity and gradient combinations (n=140), AB overestimated mean VO2 (681±285 ml·min-1) and mean EE (at 14.3±6. kJ·min-1) compared to indirect calorimetry (VO2=530±176 ml·min-1; EE=11.4±4.0 kJ·min-1 [R2=0.66 and R2=0.62, respectively]). For treadmill tests that employed variable speeds over a flat gradient, R2 between indirect calorimetry and AB VO2 or EE ranged from 0.65-0.82. However, when a constant treadmill velocity was employed with variable gradients, R2 between indirect calorimetry and AB VO2 or EE was much lower, in the range 0.34-0.58. CONCLUSION: This study revealed that an arm-mounted multi-sensor device developed for upright movements overestimated both VO2 and EE compared with indirect calorimetry during wheelchair propulsion. Algorithms utilised by the device apparently overestimated metabolic demand, a finding previously observed utilising other types of accelerometry comparing able-bodied versus SCI individuals. Interestingly, the overestimation and variability was greater when gradient was elevated, than when speed increased. It appears it is necessary to apply wheelchair-specific algorithms to the SenseWear armband to enhance the accuracy of estimating energy expenditure during wheelchair exercise.