Convergent Validity of Four Accelerometer Cutpoints With Direct Observation of Preschool Children's Outdoor Physical Activity

Purpose: More than one fifth of American preschool-aged children are classified as overweight/obese. Increasing physical activity is one means of slowing/reversing progression to overweight or obesity. Measurement of physical activity in this age group relies heavily on motion sensors such as accelerometers. Output is typically interpreted through application of one of several cutpoints to quantify the amount of time spent at various physical activity levels. Cutpoint nonequivalence has resulted in widely varying estimates of physical activity. Thus, we sought to ascertain the convergent validity of four ActiGraph accelerometer cutpoints with the Observation System for Recording Activity in Children-Preschool (OSRAC-P). Method: Two studies were conducted with sixty-nine 4- and 5-year-olds during unstructured outdoor free play at a single university campus preschool in Southern California. In Study 1 (N = 57), percentages of time spent in sedentary behavior (SED) and moderate-to-vigorous physical activity (MVPA) were compared at the aggregate level using the two-proportion z-test and Bland Altman plots. In Study 2 (N = 12), time-matched 5-s intervals between accelerometery and the OSRAC-P were compared using percent agreement, kappa, and classification functions. Results: In Study 1, cutpoints by Sirard, Trost, Pfeiffer, Dowda, and Pate (2005) best converged with the OSRAC-P estimates of SED and MVPA. In Study 2, Shard et al. cutpoints and those by Pate, Almeida, McIver, Pfeiffer, and Dowda (2006) were more sensitive in detecting SED and MVPA, respectively. Conclusions: Choosing a cutpoint in this age group should be based on the exposure of interest, epoch/observation interval duration, and required degree of stringency for classifying physical activity level.