Introduction: There is a lack of validation studies of formulas for estimating resting metabolic rate (RMR) in healthy subjects over 70 years of age. Indirect calorimetry allows measuring RMR (RMRm), but is time consuming and costly and therefore formula are generally used to estimate RMR (RMRe). We assessed the degree of agreement between RMRm and RMRe predicted by five popular equations: Harris-Benedict (HB), Mifflin-St Jeor (MJ), Owen (OW), World Health Organization (WHO/FAO/UNU) and Lührmann (LM) in a cohort of elderly subjects.

Methods: In 119 healthy subjects, aged 70-98 yr, RMRm was obtained by indirect calorimetry and RMRe by the HB, MJ, OW, WHO/FAO/UNU and LM equations. Means were compared by paired t-test. The Bland and Altman method was used to assess agreement between RMRm and RMRe. Accuracy was defined as the % of individuals whose RMRe was within +/-10% of RMRm.

Results: The HB showed the lowest mean RMRe-RMRm difference (-40.9 kcal/day), followed by LM (+44.8 kcal/day) and WHO/FAO/UNU (+53 kcal/day). The HB performed the best of the five equations, having 72.4% of the cases within+/-10% of RMRm. In 18.7% of male subjects and 20% of female subjects HB underestimated the measured values.

Conclusions: Large discrepancies exist between RMRm and RMRe in subjects above 70 years of age. HB performs best, but still tends to underestimate in both sexes. In order to develop more accurate equations to estimate RMR in elderly subjects it would be worthwhile to examine whether additionally specific markers of body composition should be taken into consideration.