Validity of Resting Metabolic Rate Prediction Equations in Normal, Overweight and Obese Adults

Abstract

The assessment of resting metabolic rate (RMR) using prediction equations is often used to estimate energy balance in relation to weight management. These prediction measures may not be generalizable and thus may not be valid across body mass index (BMI) groups. PURPOSE: To compare 4 different RMR prediction equations, Harris-Benedict (HB), Owen (O), Mifflin (M), and World Health Organization (W), with measured RMR (criterion) assessed by a validated RMR analyzer in normal (NO), overweight (OW) and obese (OB) adults. METHODS: One hundred subjects (NO, BMI<25.0 kg/m2, n = 48; OW, 25.0≤BMI<30.0 kg/m2, n=28; OB≥30 kg/m2, n=24) rested for 10 min following a 4-h restriction of food, caffeine and exercise prior to RMR measurement. A two-way ANOVA was used to compare RMR estimated by the 4 prediction equations to the criterion RMR. RESULTS: There were no significant differences for the O and M predicted RMR values compared to the criterion across BMI groups (p>0.05). As indicated in Table 1 the HB equation significantly overestimated RMR compared to criterion for NO and OW groups (p=0.048, p=0.006) but not for the OB group (p=0.576) while the W equation overestimated RMR for the OW group but not for NO or OB groups. The correlations between the criterion and the prediction equations ranged from r = 0.70 to r = 0.76.Table 1: Percent Difference Between the Criterion and the Prediction Equations.CONCLUSIONS: The M and O equations are valid for estimating RMR across BMI groups making them useful in the field for group energy balance assessment. The HB and W equations are not valid for all BMI groups. Although several mean differences between the criterion and predicted RMR values were not statistically different, 24-hr individual estimates differed from the 24-hr criterion RMR by as much as 930 kcal/day.