A validation of two motion sensors in the prediction of child and adult physical activity levels

Abstract

The authors sought to assess the reliability and validity of two motion activity sensors in the prediction of observed physical activity levels in adults and children by means of two studies at North Dakota State University in Fargo, North Dakota. In study 1, 50 adults were observed in fall 1984 for one hour of recreational activity while they simultaneously wore two motion sensors: an activity counter (i.e., LSI Moving Activity Monitor (GMM Electronics Inc., Verona, PA)) and an electronic accelerometer (i.e., Caltrac Personal Activity Computer (Hemokinetics, Inc., Vienna, VA)). The results indicated that both the activity counter (r = 0.65) and the accelerometer (r = 0.69) strongly correlated to observed physical activity levels. The two motion sensors also were highly correlated to each other (r = 0.83). A correlation of self-reports of time spent in various types of physical activity to observed physical activity ranged from 0.21 to 0.88 (mean = 0.60). Study 2 evaluated these same two sensors in winter 1984 in 30 preschool children observed in a freeplay situation. Results of this study indicated very modest relationships between observed physical activity levels and accelerometer readings (r = 0.35) and activity counter readings (r = 0.40). The two motion sensors intercorrelated at very modest levels (r = 0.42). The results indicate that while both motion sensors closely corresponded to observed levels of physical activity in adults, neither sensor correlated substantively to observed levels of physical activity in children. Implications for validating and utilizing these sensors in epidemiologic research are discussed.