Energy Expenditure

EE: Evidence Analysis: Estimating RMR with Prediction Equations (2006)

Estimating RMR with Prediction Equations: What Does the Evidence Tell Us?

The ADA evaluated four common RMR prediction equations using its rigorous Evidence Analysis Process (which includes a systematic review of the literature), to assess their ability to accurately predict RMR within +/- 10% of measured RMR in various healthy[1] populations:

  1. non-obese
  2. obese
  3. various ethnic groups
  4. older age groups (1)

Detail on Equations and Grades

Table 1 details the equations, including those developed by Harris-Benedict, Mifflin-St. Jeor, Owen, and the World Health Organization/Food & Agricultural Organization/United Nations University (WHO/FAO/UNU)[2]. Each article that met the sorting criteria was reviewed and summarized. Based upon this analysis, expert panel members developed conclusion statements, and grades were assigned to communicate the strength of the evidence.

Table 2 defines the grades the ADA uses to determine the strength of the evidence.

What Does the Research Tell Us?

Despite widespread use of the Harris-Benedict equation, the Mifflin-St. Jeor equation performed the best when predicting RMR in non-obese and obese populations, 20-82 years of age (1).

In other populations:

Older adults: In older adults (60-82 yr) across all weight classifications and with all of the equations, error ranges were large (maximum underestimations up to 31% to maximum overestimations of 12%) and none of the equations sufficiently evaluated individuals >80 years of age.

Non-white racial populations: The evidence analysis revealed that none of the equations have been adequately studied for their applicability to U.S.-residing racial and ethnic populations.

Underweight adults: The analysis did not attempt to assess the accuracy of the equations in underweight populations (BMI< 18.5).

It may be advised to measure RMR using indirect calorimetry in older adults, individuals of non-white race, and underweight adults to obtain more accurate RMR information.

Table 3 provides more complete findings from the evidence analysis (1).



[1] Healthy individuals represent adults who do not have illnesses directly affecting RMR (e.g., diabetes mellitus, hypothyroidism, heart failure) or take medications known to affect RMR, with studies using individual telephone screenings, medical examinations, health surveys or self-report to establish. While health is relative, physiological changes with multiple interactions (such as age, multiple chronic and acute occurrences), the current predictive equation errors were developed from volunteer subjects who met the above mentioned criteria and/or who perceived their physical condition as healthy.

[2] The Carol Ireton-Jones equation was not included in the evidence analysis for estimating RMR since it is used for patients in the critical care setting rather than in healthy individuals as defined above. It will be examined as part of a future evidence analysis.