Establishing school day pedometer step count cut-points using ROC curves in low-income children

• 5505 pedometer steps discriminate children who achieve 30 min of school MVPA.
• The accuracy associated with this cut-point was 76%.
• There was greater pedometer diagnostic ability in girls compared to boys.
• Additional cut-points were determined using sensitivity and specificity balancing.

Previous research has not established pedometer step count cut-points that discriminate children that meet school day physical activity recommendations using a tri-axial ActiGraph accelerometer criterion. The purpose of this study was to determine step count cut-points that associate with 30 min of school day moderate-to-vigorous physical activity (MVPA) in school-aged children. Participants included 1053 school-aged children (mean age = 8.4 ± 1.8 years) recruited from three low-income schools from the state of Utah in the U.S. Physical activity was assessed using Yamax DigiWalker CW600 pedometers and ActiGraph wGT3X-BT triaxial accelerometers that were concurrently worn during school hours. Data were collected at each school during the 2014–2015 school year. Receiver operating characteristic (ROC) curves were used to determine pedometer step count cut-points that associated with at least 30 min of MVPA during school hours. Cut-points were determined using the maximum Youden's J statistic (J max). For the total sample, the area-under-the-curve (AUC) was 0.77 (p < 0.001) with a pedometer cut-point of 5505 steps (J max = 0.46, Sensitivity = 63%, Specificity = 84%; Accuracy = 76%). Step counts showed greater diagnostic ability in girls (AUC = 0.81, p < 0.001; Cut-point = 5306 steps; Accuracy = 78.8%) compared to boys (AUC = 0.72, p < 0.01; Cut-point = 5786 steps; Accuracy = 71.4%). Pedometer step counts showed good diagnostic ability in girls and fair diagnostic ability in boys for discriminating children that met at least 30 min of MVPA during school hours.