The reliability, validity, and stability of a measure of physical activity in the elderly.

OBJECTIVE: To assess the reliability, validity, and stability of an accelerometer-based monitor of physical activity in an elderly population. DESIGN: Six studies were conducted to assess the Tritrac, a newly available accelerometer capable of storing three-dimensional activity data collected in minute epochs. After initial bench testing, the waist-worn devices were assessed for test-retest reliability during sitting and treadmill walking at 1mph and 2mph. Validity testing examined the Tritrac's ability to discriminate between different levels of physical activity and also compared the Tritrac with an established wrist-worn acclerometer, the Actigraph. Stability of the measure was examined over a period of 3 to 7 days, and in a subset of subjects this measurement was performed for a second week. SETTING AND SUBJECTS: Residents of a VA Nursing Home (n = 40; mean age = 76), participants in an in-home nonaerobic exercise program (n = 36; mean age = 77), and community participants in an aerobic exercise program (n = 10; mean age = 71). RESULTS: Intraclass correlation coefficients (ICC) for both the bench testing and test-retest reliability were .97. The device discriminated among subjects independently categorized as sedentary, moderately active, or active (F = 49.4, p = .0001) and between specific activities of varying intensity (F = 114.5, p = .0001). Tritrac and Actigraph measurement comparison showed a correlation of r = .77, p = .0001. Stability of the measure was demonstrated by ICCs = .81 and .78 for mean activity values and proportion of time spent in sedentary activity, respectively. No significant differences were observed when comparing activity measured for two separate weeks. CONCLUSION: The Tritrac is a reliable and valid instrument producing activity measurement that was stable over time in the elderly population we studied. During our longitudinal testing, 20% of the subjects did not comply with wearing the device; this noncompliance issue must be considered in any use of the Tritrac.

Duke Scholars

Author Carl F. Pieper Biostatistics & Bioinformatics, Division of Biostatistics

Author Eleanor Schildwachter McConnell School of Nursing

Author Maragatha Kuchibhatla Biostatistics & Bioinformatics, Division of Biostatistics