Validity and Reliability of a Novel Method for Physical Activity Surveillance in Toddlers

Abstract
The purpose of this study was to test the validity and reliability of a brief, wearable, sensor-based screening tool for risk of insufficient daily physical activity (PA) in toddlers. Families (N = 119) with 24- to 35-month-old children attending an Early Head Start (EHS) in a major metropolitan area were recruited. Children wore accelerometers for 7 days, including the usual ∼3.5 hours/week during which children attended the EHS. After applying wear time criteria, accelerometer data from 50 children were included in further analyses. For each child, 15-minute samples of raw triaxial accelerometer data were randomly extracted from the period of time during which children were in the EHS. Using a custom scoring tool, PHIT (Physical Inactivity Test), accelerometer signals were scored for the presence of total PA (TPA) and moderate-to-vigorous PA (MVPA). TPA and MVPA PHIT scores derived from 15-minute samples were respectively used to classify children meeting TPA and MVPA recommendations as determined during the 7-day period. Cross-validated misclassification errors were used to evaluate PHIT score performance. The Spearman-Brown Prophecy formula was used to determine the number of 15-minute samples needed to achieve sufficient reliability (r ≥ 0.70). Using one 15-minute sample, misclassification errors for TPA and MVPA PHIT scores were 19% and 13%, respectively. Spearman-Brown analyses showed that three samples yielded TPA and MVPA PHIT score reliabilities of r = 0.79 and r = 0.75, respectively. Using three samples, PHIT score misclassification errors for identifying insufficient TPA and MVPA in toddlers were 28% and 20%, respectively. PHIT was a valid and reliable tool for PA surveillance in toddlers.