- Click here for explanation of classification scheme.

1. Examine the original Harris Benedict data to investigate the relationship between REE and both sex and age
2. Examine the relationship between BCM and REE
3. Examine the accuracy of the Harris Benedict equation in predicting REE in normally nourished and malnourished patients

Healthy adults

None reported in reanalysis

Definitions:

“Body cell mass” was calculated used Moore equation

Study Protocol

ANTHROPOMETRIC:

Ht measured? Yes

Wt measured ? Yes

CLINICAL:

Monitored heart rate? Not reported

Body temperature? Not reported

Resting energy expenditure:

(Taken from Frankenfield, 1998)

Shizgal’s measurements taken on hospitalized patients not abstracted onto worksheet)

C type: Universal respiration apparatus

Equipment of Calibration:

Coefficient of variation using std gases:  Yes or No

Rest before measure (state length of time rested if available): 30 mins

(1935 subjects were studied in own homes- true basal)

Measurement length: 15 min

Steady state:

Motion detector

Fasting length: 12 hr

Exercise restrictions XX hr prior to test?:  Traveled to site by automobile

Room temp:  None reported?

No. of measures within the measurement period: One

Were some measures eliminated? Not reported

Were a set of measurements averaged? No

IF avg, identify length of each measure & no. of measurements?

Coefficient of variation in subjects measures?

Not reported

Training of measurer? Yes

Subject training of measuring process? Yes, with successive measures

Statistical tests:

Pearson correlation data; multiple linear regression; 95% confidence limits were calculated as 1.96 times the SE of the estimate; Each regression contained a multiple correlation coefficient and the statistical significance.

1. [(VO2, l/min), VCO2 (l/min; ml/kg/min), heat production (l/min)].
2. Independent variables of age, weight, height, body surface area, pulse rate, free mass, fat mass

Blinding used: None

N= 337 subjects aged

Men: n=136 age range 16-91;

Mean 30±14 yr

Women: n=103 age range 15-88 yr mean 40 ± 22 yr

Harris Benedict regression equations:

Men:

Equation 1. Harris Benedict equation based on 1919 data (n=136)

REE= 66.473 + 5.003 (ht) + 13.752 (wt) – 6.755 (age)

Equation 2: Original data recalculated (n=136):

77.607 + 4.923 (ht) + 13.702 (wt) – 6.673 (age)

r=0.86 ; F=122.4 p<0.001 95% CL= +210.5 kcal

Equation #3: Total data (n=168)

REE = 88.362 + 4.799 (ht) + 13.397 (wt) – 5.677 (age)

r=0.88 F=192.3  p<0.001 95% CL= + 312.0 kcal

Women:

Equation 1.  Harris Benedict equation based on 1919 data (n=103)

REE= 655.096 + 1.850 (ht) + 9.563 (wt) – 4.676 (age)

Equation 2: Original data recalculated (n=103):

REE: 667.051 + 1.729 (ht) + 9.740 (wt) – 4.737 (age)

r=0.77 ; F=37.8 p<0.001 95% CL= +211.9 kcal

Equation #3: Total data (n=169)

REE = 447.593 + 3.098 (ht) + 9.247 (wt) – 4.330 (age)

r=0.83 F=119.2  p<0.001 95% CL= + 201.0 kcal

In a sample 70-kg man, the 95% confidence limits represent 13.7% of the estimated REE.

There was a highly significant correlation between REE and BCM for both men and women, 0.86 and 0.80 respectively.

The REE as dependent variable was correlated with age and BCM as the independent variables and were highly significant (P<0.001) in multiple linear regression equations.  The regression coefficient associated with the BCM was statistically significant in both equations (p <0.001) while the coefficient associated with age was statistically significant (p<0.05) for women only.

IN both men and women body weight did not change with age . . however, the BCM as a fraction of body weight decreased significantly with age for both men and women.  With aging there is a significant (p<0.001) decline in the proportion of body weight due to the BCM, correlation coefficients= -0.81 for both sexes.

The REE of the women is lower that that of the men primarily because the BCM of the women tends to be smaller.

As stated by the author in body of report:

“In our study, we derived the Harris-Benedict equation by using the original 1919 Harris Benedict data to correlate REE, as the dependent variable, with height, weight, age, and sex as the independent variables. The regression equations derived using the original 1919 data are virtually identical to the equations obtained with the data from the larger number of subjects. ”

“The correlation coefficients, SE of the estimate, and the 95% confidence limits are similar for all the equations.  The Harris Benedict equation is therefore equally valid for both younger and older individual.  The 95% confidence limits of both the original equations and that derived from the larger group of subjects represent + 14% of the estimated REE.  Hence, the Harris Benedict regression thus estimates  REE of a normal individual with a precision of 14%.

“The assumption of multiple linear regression is that there is a linear relationship between REE and the independent parameters . . .however, the relationship may not be truly linear and therefore at the extremes of age, the resultant estimates may be in error. 

“It is valid to use the BCM equations to calculate the BCM of the Harris Benedict subjects, who were a representative sample of the normal population. 

When examining the independent contribution of both age and BCM on REE, the small regression coefficients associated with age for both men and women indicates age plays an insignificant role in both sexes as a determinant of REE

“The relationship between the BCM and body weight is not a constant one throughout life; in the Harris-Benedict subjects, body weight did not change significantly with advancing age, however, the BCM as a fraction of the body weight did decrease significantly with increasing age for both men and women.  Age is a factor to account for the changing relationship between the BCM and body weight with advancing age.”

“Equations for predicting REE were derived from a normal, healthy population and thus are best applied to similar populations. . . to assess REE accurately in the ill and malnourished patient, direct measurement is preferable to the use of the Harris Benedict equations.”

Government:	Medical Research Council (Canada)
University/Hospital:	McGill University, Royal Victoria Hospital (Canada)

Strengths:

• Used statistical analyses to report 95% confidence intervals

Generalizability/Weaknesses: 

• ”Search strategy to locate studies not discussed ; Frankenfield re-analysis included additional “supplemental” subjects to 1918 data;  Data on 1935 series is different than Frankenfield re-analysis (i.e., missing one male subject)
•  Does not describe the indirect calorimetry analysis in article [retrieved from Frankenfield article]
• Does not discuss the effects of obesity in Harris-Benedict subjects;
• “Describes the possibility for nonlinearity with aging and wide variances and then summarizes that HB equation useful for healthy populations; Inconsistent with data