Validation of Pattern-Recognition Monitors in Children Using Doubly Labeled Water.

• Autores: Miguel A. Calabró, Gregory J. Welk, Jeanne M. Stewart
• Localización: Medicine & Science in Sports & exercise: Official Journal of the American College of Sports Medicine, ISSN 0195-9131, Vol. 45, Nº. 7, 2013, págs. 1313-1322
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • AB Purpose: Accurate assessments of physical activity and energy expenditure (EE) are needed to advance research on childhood obesity prevention. The objective of this study is to evaluate the validity of two SenseWear Armband monitors (the Pro3 (SWA) and the recently released Mini) (BodyMedia Inc., Pittsburgh, PA) under free-living conditions in a youth population. Methods: Twenty-eight healthy children age 10-16 yr wore both monitors for 14 consecutive days, including sleeping time. Estimates of total EE from the monitors were computed using two different algorithms (version 2.2, available in the SenseWear software 6.1 and 7.0, and the newly developed 5.0 algorithms). The EE estimates were compared with estimates derived from doubly labeled water (DLW) methodology using a three-way mixed model ANOVA (sex x monitor x algorithm), correlation analyses, and Bland-Altman plots. Results: The mixed-model ANOVA revealed nonsignificant gender and monitor main effects but a significant main effect for algorithm (P < 0.001). The mean absolute percentage error values were considerably lower with the 5.0 (SWA: 10.9%; Mini: 11.7%) than for the 2.2 algorithm (SWA: 20.7%; Mini: 18.3%). Correlations were high for all comparisons (>0.90), but the Bland-Altman plots revealed consistent bias (greater overestimation at higher EE values). The variance in the differences between methods that was attributable to the mean level of EE ranged from R2 = 0.17 to R2 = 0.44. The magnitude of random error (estimated as the SD of the residuals) ranged from 227 to 299 kcal, but values tended to be lower with the 2.2 algorithm and with the Mini monitor. Conclusions: The newly developed SenseWear Armband 5.0 algorithms outperformed the version 2.2 algorithms for group comparisons, but additional work is needed to understand factors contributing to large individual variability