Disease-specific predictive formulas for energy expenditure in the dialysis population.

OBJECTIVE Metabolic rate is poorly understood in advanced kidney disease because direct measurement is expensive and time-consuming. Predictive equations for resting energy expenditure (REE) are needed based on simple bedside parameters. Algorithms derived for normal individuals may not be valid in the renal population. We aimed to develop predictive equations for REE specifically for the dialysis population. DESIGN Two-hundred subjects on maintenance dialysis underwent a comprehensive metabolic assessment including REE from indirect calorimetry. Parameters predicting REE were identified, and regression equations developed and validated in 20 separate subjects. RESULTS Mean REE was 1,658 ± 317 kCal/day (males) and 1,380 ± 287 kCal/day (females). Weight and height correlated positively with REE (r(2) = 0.54 and 0.31) and negatively with age older than 65 years (r(2) = 0.18). The energy cost of a unitary kilogram of body weight increased nonlinearly for lower body mass index (BMI). Existing equations derived in normal individuals underestimated REE (bias 50-114 kCal/day for 3 equations). The novel derived equation was REE(kCal/day) = -2.497·Age·Factorage+0.011·height(2.023) + 83.573·Weight(0.6291) + 68.171·Factorsex, where Factorage = 1 if 65 years or older and 0 if younger than 65, and Factorsex = 1 if male and 0 if female. This algorithm performed at least as well as those developed for normals in terms of limits of agreement and reduced bias. In validation with the Bland-Altman technique, bias was not significant for our algorithm (-22 ± 96 kCal/day). The 95% limits of agreement were +380 to -424 kCal/day. CONCLUSION Existing equations for REE derived from normal individuals are not valid in the dialysis population. The relatively increased REE in those with low BMI implies the need for higher dialysis doses in this subgroup. This disease-specific algorithm may be useful clinically and as a research tool to predict REE.