ABSTRACT
Objective: To examine the validity of accelerometers for characterizing habitual physical activity patterns in Indian children. Design: Cohort study. Setting: Holdsworth Memorial Hospital, Mysore. Subjects: Children (N=103, mean age 6.6 years) selected from an ongoing birth cohort study. Methods: Physical activity was measured over 7 days using accelerometers (MTI Actigraph) and concurrent parent-maintained activity diaries. Actigraph counts per minute representing sedentary (<10), light (<400), moderate (<3000) and vigorous (≥3000) activity were determined using a structured activity session in a separate group of 10 children. In 46 children chosen for validating accelerometers, time spent in different activity levels according to diaries was determined. Energy Expenditure (EE) was calculated from diaries using a factorial method. Results: Ninety-eight children wore the monitor for ≥4 days. Total counts and time spent in different activity levels were similar in boys and girls (P>0.2). Among 46 children chosen for comparisons, time spent in sedentary (r =0.48, P=0.001), light (r=0.70, P<0.001) and moderate activities (r=0.29, P=0.054) according to diaries correlated with those derived from counts, and total Actigraph counts correlated with EE (r=0.42, P=0.004). Bland-Altman analysis showed systematic bias, and wide limits of agreement between these methods for time spent in different activity levels. Conclusions: Accelerometers are a well tolerated and objective way of measuring activity behavior in free-living children. Though accelerometer counts correlate with time spent in activity of varying intensity and energy expenditure derived from parent-maintained diaries, wide limits of agreement show that the limitations of accelerometers need to be recognized in interpreting the data that they generate.