Abstract:
INTRODUCTION: Accelerometer based prediction equations are used to calculate physical activity energy expenditure (PAEE) among children. Currently, accelerometer-derived PAEE prediction equations validated against a criterion method do not exist for Sri Lankan children. Objective: To assess the validity of published prediction equations to estimate PAEE in Sri Lankan children against the doubly labelled water (DLW) technique. Method: Ninety-six children aged 11-13 years from an urban area of Sri Lanka were included in the study. Energy expenditure was assessed using the DLW technique over 10 days and participants wore ActiGraph accelerometers during the same period. Correlation between the measured and predicted PAEE was assessed by the Pearson correlation coefficient. Validity of equations was assessed by the paired t-test and the level of agreement using the Bland Altman analysis. Results: Predicted PAEE values were significantly (p<0.05) correlated with the measured PAEE except for the equations of Treuth and Schmitz. Prediction equations of Ekelund, Freedson, Mattock and Zhu significantly overestimated measured PAEE (p<0.05) whereas, Trost and Puyau equations significantly underestimated PAEE. A wide limit of agreement with a large mean bias was observed in all estimated PAEE, except for the equation of Zhu. Conclusions: Existing accelerometer-based PAEE equations have low accuracy in predicting PAEE in Sri Lankan children.
