THE VALIDITY OF THREE DIFFERENT ACCELEROMETERS FOR THE ASSESSMENT OF LIFESTYLE PHYSICAL ACTIVITY

Abstract

583 This study compared the validity of 3 different accelerometers for the assessment of "lifestyle" physical activity. Fourty-one adults (mean age = 30) completed 9, 6 min. bouts of activity while wearing a CSA (CSA), Tritrac (TRI) and Biotrainer (BIO) activity monitor on their hip. The 9 activities included 3 treadmill paces (2,4, and 6 mph), 3 gardening tasks (mowing, raking and shoveling), and 3 home tasks (vacuuming, sweeping and stacking). Indirect calorimetry data obtained during the 3 treadmill (Sensormedics 2900) and 6 lifestyle (Aerosport KB1-C) activities served as the criterion measure. Mean correlations among the 3 monitors were (r = .82 and r = .78) for the treadmill and lifestyle activities, respectively. Mean correlations of the monitors with VO2 were r = .85 and r = .33 for the respective activities. For the treadmill paces, the estimated METS from the CSA (prediction equation) were not significantly different from measured METS (βp = .374); however, estimates of METS from the TRI and BIO were both significantly higher than measured METS (βp < .001). For the lifestyle activities, all 3 monitors significantly underpredicted METS (βp < .001). The magnitude of underprediction, when averaged across the 6 lifestyle activities, was 42%, 42% and 46% for the CSA, TRI and BIO, respectively. Highest correspondence occurred for the lawn mowing task for all 3 monitors. This study reveals that the validity of activity monitors varies by setting (lab vs. field), and by type of activity but is comparable among monitors. Supported by the International Life Sciences Institute