Premium
A class of models to estimate energy requirements
Author(s) -
Plucker Andrew,
Powell Michael,
Broskey Nick,
Martin Corby K,
Heymsfield Steven B,
Schoeller Dale,
Thomas Diana,
Redman Leanne
Publication year - 2017
Publication title -
the faseb journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.709
H-Index - 277
eISSN - 1530-6860
pISSN - 0892-6638
DOI - 10.1096/fasebj.31.1_supplement.311.5
Subject(s) - anthropometry , energy requirement , waist , linear regression , statistics , covariate , resting energy expenditure , mathematics , hydrostatic weighing , regression analysis , doubly labeled water , residence , energy expenditure , regression , demography , physical activity , body weight , body mass index , medicine , physical therapy , sociology
Background Accurately estimating energy requirements represents a standard activity for developing effective diet and exercise interventions. Mathematical models that predict energy requirements as a product of physical activity level (PAL) and a resting energy expenditure (REE) formula is a commonly applied method to provide a first pass estimate. These estimates require knowledge of an individual's PAL and an accurate prediction of REE. Access to different anthropometric data or body composition and even REE measurements can improve and personalize predictions without making assumptions involving PAL. Methods Total energy expenditure measured by DLW and metabolic chamber from 733 subjects obtained from compiled study database of baseline measurements measured at Pennington Biomedical Research Center was applied as two different output variables. The DLW measures were applied to develop free‐living energy requirement models and the chamber data was applied to develop in‐residence energy requirement models. Twenty‐eight different linear regression models were developed that included different combinations of input variables that may be accessible to investigators and clinicians. The input variables were age, height, gender, weight, waist circumference, fat mass, fat free mass, and REE. The simplest model predicting DLW measured energy expenditures was validated on the Institute of Medicine DLW database (N=473) and compared to the product of 1.6 and the Mifflin St. Jeor prediction of REE. Results The adjusted R 2 values for the models predicting free‐living energy requirements in males ranged from 0.65 with minimal covariates of age, height, and weight to 0.73 in models that included body composition or REE. For females adjusted R 2 ranged from 0.68 to 0.74. The adjusted R 2 values for the models predicting in‐residence energy requirements were lower (males 0.43–0.45, females (0.32–0.33). The bias in the newly developed models was −95±461 kcal/d while the bias obtained from using 1.6 times REE predicted by Mifflin St. Jeor yielded a bias of −315±444 kcal/d. Conclusions The newly developed class of models offers an improved alternative to estimating a PAL value and energy requirements using REE formulas. Additionally, when available, the models include additional covariates that improve predictions even further.