Estimation of energy expenditure using CSA accelerometers at hip and wristsites

Purpose: This study was designed to establish prediction models that relate hip and wrist accelerometer data to energy expenditure (EE) in field and l aboratory settings. We also sought to determine whether the addition of a w rist accelerometer would significantly improve the prediction of EE (METs), compared with a model that used a hip accelerometer alone. Methods: Sevent y participants completed one to six activities within the categories of yar dwork, housework, family care, occupation, recreation, and conditioning, fo r a total of 5 to 12 participants rested per activity. EE was measured usin g the Cosmed K4b(2) portable metabolic system. Simultaneously, two Computer Science and Applications, Inc. (CSA) accelerometers (model 7164), one worn on the wrist and one worn on the hip, recorded body movement. Correlations between EE measured by the Cosmed and the counts recorded by the CSA accel erometers were calculated, and regression equations were developed to predi ct EE from the CSA data. Results: The wrist, hip, and combined hip and wris t regression equations accounted for 3.3%, 31.7%, and 34.3% of the variatio n in EE, respectively. The addition of the wrist accelerometer data to the hip accelerometer data to form a bivariate regression equation, although st atistically significant (P = 0.002), resulted in only a minor improvement i n prediction of EE. Cut points for 3 METs (574 hip counts), 6 METs (4945 hi p counts), and 9 METs (9317 hip counts) were also established. Conclusion: The small amount of additional accuracy gained from the wrist accelerometer is offset by the extra time required to analyze the data and the cost of t he accelerometer.