Applicability of resting energy expenditure prediction equations in Chinese nursing home residents

This study aims to evaluate the applicability of 11 commonly used resting energy expenditure prediction equations among elderly residents in Chinese nursing homes and to investigate the key factors influencing their predictive performance.

This study employed a cross-sectional design and included 156 elderly residents of nursing homes as study participants. The study conducted a comparative analysis of resting energy expenditure values measured using indirect calorimetry and the values predicted by 11 different prediction equations. In addition, the performance of each equation was comprehensively evaluated through comparisons of bias, precision, and accuracy, as well as Bland-Altman plots. Subgroup analysis and restricted cubic spline models were used to investigate the factors influencing the predictive performance of the equations.

A total of 156 older adults residing in nursing homes were included in the study, with a mean age of 83.4 ± 5.6 years; 62.2% were women. REE measured by indirect calorimetry was 1378.6 ± 268.0 kcal/d. Among the 11 predictive equations, only the Owen equation (1348.9 kcal/d, p = 0.113) and the Cunningham equation (1351.5 kcal/d, p = 0.155) showed no statistically significant difference between their predicted values and the measured REE; the remaining 9 equations all significantly underestimated REE (all p < 0.001). Among all equations, the Owen and Cunningham equations exhibited the lowest bias (0.2% and 0.7%), the best precision (150.7 kcal/d and 153.7 kcal/d), and the highest prediction accuracy (48.1% and 45.5%). However, Bland-Altman analysis showed that the 95% limits of agreement were wide (Owen: -401.2 to 460.6 kcal/d; Cunningham: -419.7 to 473.9 kcal/d), suggesting that prediction errors at the individual level remain substantial for both equations. Subgroup analyses revealed that age, muscle mass, and nutritional status were significantly associated with predictive performance; restricted cubic spline analysis further confirmed that age and skeletal muscle index exhibit nonlinear relationships with prediction errors.

The Owen and Cunningham equations can serve as relatively preferred tools for estimating REE in older adults residing in Chinese long-term care facilities; however, their accuracy in predicting individual REE remains limited, and they are unable to fully account for the significant individual metabolic variations within this population. Overall, the application of existing REE prediction equations among older adults in Chinese nursing homes still has certain limitations. Therefore, when conditions permit, indirect calorimetry should remain the method of choice for assessing individual REE in older adults. Future large-sample, multicenter studies are needed to develop and validate REE prediction models better suited to the Chinese elderly population, thereby improving the accuracy of individualized energy requirement assessments.