With rising obesity and rapidly decreasing levels of physical activity (PA) in young children, accurate PA measurement is needed for early screening and intervention evaluation. Although the combination of accelerometry (ACC) with HR recordings is accurate in assessing PA intensity in older children, its utility in correctly classifying PA in preschoolers is unknown.
Thirty-three children (64% were boys) were recruited from four German preschools. Data included direct observation and Actiheart (CamNtech, Cambridge, UK) monitoring during 2.5 T 0.7 h (derivation data) and 1.5 T 0.3 h (validation data). Observers assessed PA using the Children’s Activity Rating Scale (CARS). Moderate-to-vigorous PA (MVPA) was defined by a CARS score of 4 or 5, and sedentary behavior (SB) was defined by a CARS score of 1 or 2. Actiheart recordings were linked to CARS level, with means and 95% confidence intervals calculated for ACC and HR at each observed CARS level and for MVPA/SB. Using receiver operating characteristic analysis, gender-specific ACC and HR cutoffs for correctly classifying MVPA and SB were first determined in a derivation data set and then tested in a separate validation data set of observations.
By combining HR and ACC cutoffs, 91% and 87% of the 15-s intervals observed as MVPA were correctly classified in girls and boys, respectively. Although generally lower, correct classification rates for SB were highest when only ACC cutoffs were applied (69% for girls and 67% for boys) rather than when combined cutoffs were used.
Devices that combine HR and ACC data yield an accurate classification of MVPA in preschoolers but perform less well for classifying SB. These differences underscore the need to match evaluation methods with the objectives of future PA interventions.