Agreement between equations-estimated resting metabolic rate and indirect calorimetry-estimated resting metabolic rate in low-income obese women

ABSTRACT Objectives Indirect calorimetry is established as a gold standard to determine the resting metabolic rate (RMR), however, its clinical use is limited, especially in low-income settings. Thus, the use of predictive equations appear as an alternative to estimate the RMR, but its precision is debatable, especially in obese individuals and in populations without specifically developed equations. To evaluate the agreement between the RMR estimated by equations and by indirect calorimetry in low-income obese women. Subjects and methods A cross-sectional study with adult and obese women, which estimated the RMR by indirect calorimetry and compared with 13 predictive equations using the concordance correlation coefficient, root mean square error (RMSE) and Bland-Altman methods. The maximum allowed differences were predefined as 10%. Results No equation presented its confidence intervals for the Bland-Altman limits of agreement inside the predefined acceptable range. The Harris-Benedict equation achieved better agreement (bias of 2.9% and RMSE of 274.3kcal) whereas the Henry-Rees equation achieved better precision (42.3% of the sample within the 10% maximum allowed difference). Conclusion None of the studied equations satisfactorily estimated the RMR estimated by indirect calorimetry. In the absence of specific equations for this population, the use of the Harris-Benedict and Henry-Rees equations could be considered.

摘要 研究背景与目的：间接测热法（indirect calorimetry）是测定静息代谢率（resting metabolic rate, RMR）的金标准，但其临床应用存在明显局限，尤其在低收入地区。为此，预测方程法成为估算静息代谢率的替代方案，但其准确性尚存争议，尤其是针对肥胖人群以及未专门开发对应预测方程的群体。本研究旨在评估低收入肥胖女性群体中，预测方程估算的静息代谢率与间接测热法测定结果的一致性。
对象与方法：本研究为横断面研究，纳入成年肥胖女性受试者，通过间接测热法测定其静息代谢率，并与13种预测方程的估算结果进行对比，采用一致性相关系数、均方根误差（root mean square error, RMSE）以及Bland-Altman分析法进行评价。预设最大允许差异为10%。
结果：所有方程的Bland-Altman一致性界限的置信区间均未落入预设的可接受范围内。其中，Harris-Benedict方程表现出更优的一致性（偏差为2.9%，均方根误差为274.3千卡），而Henry-Rees方程则具备更高的精准度（42.3%的样本误差落在10%的最大允许差异范围内）。
结论：本研究所涉及的所有预测方程均无法精准匹配间接测热法测定的静息代谢率。在缺乏针对该人群的专属预测方程的情况下，可考虑选用Harris-Benedict方程与Henry-Rees方程。