Predictive Validity of a Thigh-Worn Accelerometer METs Algorithm in 5- to 12-Year-old Children.
Journal of physical activity & health 2016 ; 13: S78-83.
van Loo CM, Okely AD, Batterham M, Hinkley T, Ekelund U, Brage S, Reilly JJ, Peoples GE, Jones R, Janssen X, Cliff DP
DOI : 10.1123/jpah.2015-0721
PubMed ID : 27392386
PMCID : PMC6207339
URL : https://journals.humankinetics.com/view/journals/jpah/13/s1/article-pS78.xml
To validate the activPAL3 algorithm for predicting metabolic equivalents (TAMETs) and classifying MVPA in 5- to 12-year-old children.
Fifty-seven children (9.2 ± 2.3y, 49.1% boys) completed 14 activities including sedentary behaviors (SB), light (LPA) and moderate-to-vigorous physical activities (MVPA). Indirect calorimetry (IC) was used as the criterion measure. Analyses included equivalence testing, Bland-Altman procedures and area under the receiver operating curve (ROC-AUC).
At the group level, TAMETs were significantly equivalent to IC for handheld e-game, writing/coloring, and standing class activity (P < .05). Overall, TAMETs were overestimated for SB (7.9 ± 6.7%) and LPA (1.9 ± 20.2%) and underestimated for MVPA (27.7 ± 26.6%); however, classification accuracy of MVPA was good (ROC-AUC = 0.86). Limits of agreement were wide for all activities, indicating large individual error (SB: -27.6% to 44.7%; LPA: -47.1% to 51.0%; MVPA: -88.8% to 33.9%).
TAMETs were accurate for some SB and standing, but were overestimated for overall SB and LPA, and underestimated for MVPA. Accuracy for classifying MVPA was, however, acceptable.