The Actiheart in adolescents: a doubly labelled water validation.

Background/objective
This study investigated the validity of the Actiheart device for estimating free-living physical activity energy expenditure (PAEE) in adolescents.
Subjects/methods
Total energy expenditure (TEE) was measured in eighteen Canadian adolescents, aged 15-18 years, by DLW. Physical activity energy expenditure was calculated as 0.9 X TEE minus resting energy expenditure, assuming 10% for the thermic effect of feeding. Participants wore the chest mounted Actiheart device which records simultaneously minute-by-minute acceleration (ACC) and heart rate (HR). Using both children and adult branched equation modeling, derived from laboratory-based activity, PAEE was estimated from the ACC and HR data. Linear regression analyses examined the association between PAEE derived from the Actiheart and DLW method where DLW PAEE served as the dependent variable. Measurement of agreement between the two methods was analyzed using the Bland-Altman procedure.
Results
A nonsignificant association was found between the children derived Actiheart and DLW PAEE values (R = .23, R(2) = .05, p = .36); whereas a significant association was found between the adult derived Actiheart and DLW PAEE values (R = .53, R(2) = .29, p < .05). Both the children and adult equation models lead to overestimations of PAEE by the Actiheart compared with the DLW method, by a mean difference of 31.42 kcal.kg(-1).d(-1) (95% limits of agreement: -45.70 to -17.15 kcal.kg(-1).d(-1) and 9.80 kcal.kg(-1).d(-1) (95% limits of agreement: -21.22-1.72 kcal.kg(-1).d(-1)), respectively.
Conclusion
There is relatively poor measurement of agreement between the Actiheart and DLW for assessing free-living PAEE in adolescents. Future work should develop group based branched equation models specifically for adolescents to improve the utility of the device in this population.