Resumen de Indices of Resting Metabolic Rate Accurately Reflect Energy Deficiency in Exercising Women
Nicole C.A. Strock, Kristen J. Koltun, Emily A. Southmayd, Nancy I. Williams, Mary Jane De Souza
Energy deficiency in exercising women can lead to physiological consequences. No gold standard exists to accurately estimate energy deficiency, but measured-to-predicted resting metabolic rate (RMR) ratio has been used to categorize women as energy deficient. The purpose of the study was to (a) evaluate the accuracy of RMR prediction methods, (b) determine the relationships with physiological consequences of energy deficiency, and (c) evaluate ratio thresholds in a cross-sectional comparison of ovulatory, amenorrheic, or subclinical menstrual disturbances in exercising women (n = 217). Dual-energy X-ray absorptiometry (DXA) and indirect calorimetry provided data on anthropometrics and energy expenditure. Harris–Benedict, DXA, and Cunningham (1980 and 1991) equations were used to estimate RMR and RMR ratio. Group differences were assessed (analysis of variance and Kruskal–Wallis tests); logistic regression and Spearman correlations related ratios with consequences of energy deficiency (i.e., low total triiodothyronine; TT3). Sensitivity and specificity calculations evaluated ratio thresholds. Amenorrheic women had lower RMR (p < .05), DXA ratio (p < .01), Cunningham1980 (p < .05) and Cunningham1991 (p < .05) ratio, and TT3 (p < .01) compared with the ovulatory group. Each prediction equation overestimated measured RMR (p < .001), but predicted (p < .001) and positively correlated with TT3 (r = .329–.453). A 0.90 ratio threshold yielded highest sensitivity for Cunningham1980 (0.90) and Harris–Benedict (0.87) methods, but a higher ratio threshold was best for DXA (0.94) and Cunningham1991 (0.92) methods to yield a sensitivity of 0.80. In conclusion, each ratio predicted and correlated with TT3, supporting the use of RMR ratio as an alternative assessment of energetic status in exercising women. However, a 0.90 ratio cutoff is not universal across RMR estimation methods.
