- Click here for explanation of classification scheme.

• To develop a practical (i.e., based on easily measured variables) and accurate age-specific equation for predicting resting metabolic rate (RMR) in older women
• Compare (cross-validate) the measured RMR values of the study participants with existing equations for predicting RMR in older females.

• Healthy
• Female
• Age 50-81
• Able to give consent.

• Clinical evidence of CHD (i.e. ST-segment depression greater than 1mm at rest or exercise) or cardiomyopathy)
• Hypertension (blood pressure >140/90mm Hg)
• Medications that could affect cardiovascular function or metabolic rate
• Medical history of diabetes or obesity
• Instability of body weight during previous year
• Exercise-limiting non-cardiac disease (arthritis, peripheral vascular disease, cerebral vascular disease, etc.)
• History of oophorectomy.

Recruitment

Methods not specified

Design

Cross-sectional study

Blinding used

Not used

Intervention

Not applicable

Statistical Analysis

• Regression equations used to examine linear and non-linear relationships between RMR and other independent variables
  • Semi-partial F-tests performed to determine whether the quadratic models explained a significantly greater amount of variance in RMR than explained by linear regression models
  • Pearson correlation used to assess the degree of association between variables
• Stepwise multiple regression used to predict RMR from the independent variables
  • Potential predictors of RMR: Standing height, body weight, body surface area, body mass index, fat mass, fat-free mass, skinfolds, VO₂max, leisure time physical activity, daily energy and macronutrient intake, menstrual status, and age
  • Practical equations derived in which only easily measured variables were considered: Body weight standing height, age, skinfolds, menstrual status and leisure time physical activity
• Dependent T-test used to compare the measured and predicted means of RMR
• Independent z test used to compare correlation coefficient (r) between predicted RMR and measured RMR with the multiple R obtained from the regression equation of the original study
• Statistical significance was set at P<0.05.

 

Timing of Measurements

All participants admitted to research center evening before metabolic testing and identical protocol for all participants

Dependent Variable

Resting Metabolic Rate (RMR): 

• Measured: Indirect calorimetry (IC)
  
  • IC type: Used ventilated hood technique for 45 minutes
  • Rest before measure: IC done under inpatient conditions (in research laboratory and measurements done in same room where participant slept); recent work showed inpatient RMR 8% lower under inpatient conditions compared with outpatient condition
  • Measurement length: 45 minutes
  • Fasting length: Overnight
  • Exercise conditioning 24 hours prior to test? IC 36-48 hours after last exercise bout to minimize the effect of physical activity on RMR
  • Room temp: Not reported
  • No of measures repeated? Not reported
  • Coefficient of variation? Not reported
  • Equipment calibration: No machine calibration was mentioned
  • Training of measurer? Not reported
  • Subject training of measuring process: Participants given practice with ventilated hood the evening before to reduce apprehension with testing conditions
• Estimated RMR: Validity of five commonly used prediction equations tested in this data set: 
  • WHO/FAO (>60 years)
  • Harris-Benedict
  • Owen
  • Fredrix
  • Mifflin.

Independent variables

• Height and weight: Methods not specified
• Body composition:
  • Body fat: Underwater weighing to determine body density with simultaneous measurement of residual lung volume by the helium dilution method using the formula of Siri
  • Fat free mass (FFM): Total body weight minus fat weight
  • Skinfolds: Triceps, chest, abdomen and thigh (Lange skinfold caliper); all measurements taken on right side of body; each value was the mean of three consecutive measurements by same investigator
• Physical activity level: Minnesota Leisure Time Physical Activity Questionnaire
• Maximum aerobic power (VO₂ max): progressive and continuos treadmill test to volitional fatigue
• Energy intake: Three-day dietary intake using weighed food records
• Menopausal status: Classified as 1=pre-menopausal (6%); 2=peri-menopausal (6%); or 3=postmenopausal (88%).

• Initial N: Not given
• Attrition (final N): N=75 females
• Age: 61±8 years  (50-81)
• Ethnicity: Not reported
• Other relevant demographics: None reported
• Anthropometrics:

Women	Mean±SD	Range
Weight, kg	63.3±7	51.8-92.3
Height, cm	163±7	149-178
Body fat, percent	30.4±5	20.1-43.2
Fat-free body mass, kg	43.8±4	36.4-54
Fat mass, kg	19.4±5	11.1-39.8
Measured RMR range, kcal per day	1,302±100	1,094-1,513

• Location: University of Vermont.

Major Results

• Potential predictors of RMR were: Standing height, body weight, body surface area, body mass index, fat mass, fat-free mass, four skinfolds, VO₂ max, leisure time physical activity, daily energy and macronutrient intake, menstrual status and age
• Correlations (r) with RMR:
  • Age: -0.26
  • Height: 0.48
  • Weight 0.68
  • Fat free mass: 0.89
  • Leisure-time activity: -0.09
  • VO₂ max: 0.42
• Fat-free mass was the strongest and only significant predictor of RMR in the total group of 75 older women (R=0.89; P<0.01); explaining 79% of the total variance. In a laboratory setting, the measurement of FFM predicted RMR in older women within ±46kcal per day.
  • Regression equation: RMR (kcal per day)=21 (FFM, kg) + 369 (r²=0.79) (SEE= ±46kcal per day)
• Focus of the study was to develop a simple equation for predicting RMR after removal of variables deemed impractical or difficult to measure (i.e., FFM) in a clinical (non-research) setting. A practical equation was derived in which only easily measured variables were considered: Body weight, standing height, age, skinfolds, menstrual status and leisure time physical activity.
  • Of these variables, body weight (kg) accounted for 47% of total variance in RMR (R=0.68; P<0.01)
    • Standing height (cm) explained an additional 8% of variance in RMR
    • Menopausal status explained 4% of total variance
    • Together, the three variables—weight, standing height and menopausal status accounted for 59% (R²) of the variance in RMR and the new practical equation predicted RMR within ±66kcal per day, P<0.01
    • Regression equation for this sample: RMR (kcal per day)=7.8 (weight, kg) + 4.7 (standing height, cm) - 39.5 (menopausal status, 1,2,3) + 144 (R=0.77, R²=0.59, SEE=66kcal per day, P<0.01)

Cross-validation results

• Harris-Benedict underestimated (7%) RMR in older population; differences not significant
• Owen et al significantly (P<0.05) under-predicted RMR by an average of 5% (group); with a range of -27% to 15% on an individual basis
• Mifflin et al significantly (P<0.05) under-predicted measured RMR by 11% (group); ranging from -31% to 7%
• WHO/FAO and Fredrix et al (P<0.05) overestimated measured RMR by 3%; ranging from -8% to 12% and -24% to 20%, respectively
• WHO/FAO predicted RMR in older women with SEE of ±94kcal per day compared to ±66kcal per day in this study
• Outpatient conditions in Fredrix study may have contributed to overestimated RMR; in previous study done by author, outpatient overestimated by approximately 8% compared with inpatient conditions.

Comparison of the original prediction equation of the five studies used for cross-validation analyses and the practical equation from the present study

Study	Original equation
WHO (females, N=NA>60 years)	RMR (kcal per day)=[9.2 (W) + 6.4(H)] -302 (R2=0.67)
Harris Benedict (females, N=16>50 years)	RMR (kcal per day)=[1.8 (H) + 9.6 (W) - 4.7 (A)] +655 (R2=0.59)
Owen et al (females, N=8>50 years)	RMR (kcal per day)=[7.2 (W)] +795 (R2=0.55)
Fredrix et al (females, (N=22>50 years)	RMR (kcal per day)=[10.7 (W) - 9 (A) - 203 (2)] + 1641 (R2=0/84)
Mifflin et al (females, N=50>50 years)	RMR (kcal per day)=[10 (W) + 6.25 (H)=5 (A)] - 161 (R2=0.71)
Arciero et al (females, N=75>50 years)	RMR (kcal per day)=[7.8 (W) + 4.7 (H) - 40 (M)] +144 (R2=0.59)

Units for equations: W: Weight, kg; H: Height, cm; A: Age; M: Menopausal, 1=pre, 2=peri, 3=post; NA=not available

Comparisons of measured RMR and RMR predicted by five published equations and the correlation coefficient of predicted and measured RMR vs. the multiple R from the original prediction equation

	Criterion #1			Criterion #2
Study	Measured (kcal per day)	Predicted (kcal per day)	P-value	r vs. R
WHO (>60 years)	1,286.6	1,320.0	0.12	0.79 vs. 0.82
Harris-Benedict	1,301.6	1,274.6	0.07	0.75 vs. 0.78
Owen	1,301.6	1,249.2	<0.01	0.68 vs. 0.74*
Fredrix	1,301.6	1,359.3	<0.01	0.69 vs. 0.92*
Mifflin	1,301.6	1,185.3	<0.01	0.75 vs. 0.84*

* correlation coefficient is significantly different from the multiple R, P<0.05

• Currently available equations to predict RMR in older individuals have not been age- or sex-specific, have generally been based on data extrapolated from younger individuals, and have relied generally on small sample sizes (N=50)
• In a laboratory setting, the measurement of FFM predicted RMR in older women within ±46kcal per day
• However, in a clinical or non-research setting; body weight alone best predicted RMR, accounting for 47% of the explained variance in RMR in older females. The use of body weight alone would permit an estimate of RMR with a standard error of estimate of ±74kcal per day
• Standing height significantly (R²=8%) strengthened the prediction of RMR and addition of menopausal status contributed an additional 4% of explained variance. The independent two (variables (plus body weight) increased the R² to 59%, which would permit an estimate of RMR within ±66kcal per day
• There was a significant linear decline in RMR in females >50 years of age with onset of menopause
• Because of finding that menopausal status is a significant factor contributing to the variation in RMR in older women (hormone status), authors suggested may need to develop separate equations for older men and women
• When five previously published equations were applied to predict RMR in cross validation-independent predicted values deviated by -31% to 20% from the measured values
• With the exception of the WHO/FAO equation and the Harris-Benedict equation, none of the published equations cross-validated successfully with the measured RMR values from the 75 older women. The WHO/FAO equation predicted RMR in older women with a standard error of estimate of ±94kcal per day compared with ±66kcal per day in the present age- and sex-specific equation.

Major Conclusion

“In summary, we offer an accurate equation to predict RMR in older women that is both age-specific and practical for use in the clinical or non-research setting using body weight, standing height and menopausal status.”

Government:

NIH

Strengths

• Large sample size in comparison to other studies
• Inpatient conditions
• Measurement standardized for optimum basal conditions
• IC measurements performed 36-48 hours after last exercise bout to avoid effects of exercise on RMR
• Subjects familiarized with equipment prior to study to minimize anxiety.

Limitations

• Caucasians
• Healthy older women
• So limited generalizability; can’t apply findings to females in other age or ethnic groups
• Cross-sectional; no cause and effects
• Small sample of “oldest old” >80 years
• Direct measurements recommended for precise measurements of RMR
• Self-selection bias with convenience sample; may be different than general population
• With diet diaries and weighed diet; may alter diet; not usual diet; may pick foods easier to weigh and measure
• Large range of calorie intake among participants; range 1,000-3,223kcal per day; mean 1,758±429kcal per day which could affect results; not mentioned what effect it had on RMR (?)
• Large range of leisure time activity; range 77-839kcal per day; mean 263±166kcal per day); not mentioned what effect it had on RMR (?)
• Not sure why collected data on variables that were considered by authors as impractical or hard to measure.

If height and weight were significant predictors, what about BMI?