Weight and body composition changes affect resting energy expenditure predictive equations during a 12-month weight-loss intervention.

Objective: Mathematical equations that predict resting energy expenditure (REE) are widely used to derive calorie prescriptions during weight-loss interventions. Although such equations are known to introduce group- and individual-level error into REE prediction, their validity has largely been assessed in weight-stable populations. Therefore, this study sought to characterize how weight change affects the validity of commonly used REE predictive models throughout a 12-month weight-loss intervention.

Methods: Changes in predictive error of four models (Mifflin-St-Jeor, Harris-Benedict, Owen, and World Health Organization/Food and Agriculture) were assessed at 1-, 6-, and 12-month time points in adults (n = 66, 76% female, aged 18-55 years, BMI = 27-45 kg/m ) enrolled in a randomized clinical weight-loss trial.

Results: All equations experienced significant negative shifts in bias (measured - predicted REE) toward overprediction from baseline to 1 month (p < 0.05). Three equations showed reversal of bias in the positive direction (toward underprediction) from baseline to 12 months (p < 0.05). Early changes in bias were correlated with decreased fat-free mass (p ≤ 0.01).

Conclusions: Changes in body composition and mass during a 12-month weight-loss intervention significantly affected REE predictive error in adults with overweight and obesity. Weight history should be considered when using mathematical models to predict REE during periods of weight fluctuation.