Prediction of Energy Expenditure from Accelerometers during Physical Activity in Adults with Down Syndrome: The Effect of Accelerometer Placement

Abstract

Listen

Translate article

INTRODUCTION: For accurate physical activity assessment with accelerometers in adults with Down syndrome (DS), there is a need to examine if the relationship between the rate of oxygen uptake (VO2) and output from hip-and wrist-worn accelerometers across different activities and sedentary behaviors is different between adults with and without DS. In this study, we examined this question and we also evaluated the accuracy of hip- and wrist-worn accelerometers in estimating the VO2. METHODS: The sample included 16 adults with DS (10 men; age 31 ± 15 years) and 19 adults without DS (10 men; age 24 ± 6 years). We measured VO2 with a portable spirometer (K4b2, Cosmed) and accelerometer output (Vector Magnitude [VM]) with a hip- and a wrist-worn accelerometer (wGTX-BT, Actigraph). We used multi-level regression to predict VO2 from VM and group. Additional predictors included body mass index (BMI), age, height, weight, and sex. We evaluated the accuracy of the prediction with the absolute percent error and Bland-Altman plots. RESULTS: For both the hip and the wrist accelerometer, VM and group were significant predictors of VO2 (p ≤0.021). However, when BMI was added to the models, BMI was a significant predictor and DS was no longer significant for both accelerometer models. The final models included Vector Magnitude and BMI (p ≤0.001; R2 = 0.78 and 0.57, for hip and wrist accelerometer model, respectively). For the hip accelerometer, absolute percent error across all tasks and for both groups combined was 22.5 ± 27.4%, whereas, for the wrist accelerometer, error was 37.8 ± 38.0%. Absolute percent error across tasks combined and for each task separately did not differ between groups. The Bland-Altman plots indicated nearly zero mean error for both groups. However, error had somewhat greater 95% confidence intervals for the wrist- than the hip-accelerometer models. CONCLUSION: Adults with DS have different VO2 to VM responses, but this is due to their higher BMI. Predictability of VO2 from accelerometer output is better for a hip- than a wrist-worn accelerometer. These results may help advance physical activity assessment for adults with DS.