Prediction of resting energy expenditure in healthy older adults: A systematic review

Background & aims: Estimates of energy requirements, based on measured or predicted resting energy expenditure (REE), are needed to avoid undernutrition or overnutrition (and their clinical consequences) in elderly subjects. The aims of this systematic review were to evaluate the prediction accuracy of REE in healthy elderly subjects and to ascertain which equation is more reliable at group level and/or individual level. Methods: Studies assessing prediction of REE in general elderly population were systematically searched using PubMed, EMBASE, Web of Science and CINAHL until March 2020. Prediction accuracy of REE was assessed at both group (bias) and individual (precision) level for each equation. Results: Fourteen studies met the inclusion criteria of this systematic review. Bias was reported in 8 papers and calculated in another 5 from absolute values. There was a prevalent tendency towards an overestimation of REE across the studies. The least bias was observed for the Mifflin (−0.3%) and Harris–Benedict (+2.6%) equations, with values above 5% for the FAO/WHO/UNU, Fredrix and Muller equations. Precision widely varied between studies for the same equation. The higher precision was observed using the Harris–Benedict equation (~70%), while the Henry and Mifflin equations provided estimates within 10% of measured values in 65% and 61% of elderly individuals, respectively. Conclusions: None of the prediction equations considered provides accurate and precise REE estimates in healthy older adults. However, the best prediction is given by the Mifflin equation at group level and by the Harris–Benedict equation at individual level. Further studies with strong quality design are needed to evaluate the variability and accuracy of REE in the elderly general population.