Cross-Sectional Study

D - Click here for explanation of classification scheme.

NEUTRAL: See Quality Criteria Checklist below.

• To examine those predictive equations for energy expenditure that would permit more accurate resting energy expenditure estimates within a young female population
• To revise the chosen equations to adjust individual body mass index (BMI) to the subject's nutritional state (normal weight, overweight, obese)
• To examine how weight loss might influence the REE prediction of anthropometrical equations after a period of dietary treatment.

• Low level of physical activity
• No signs of physical or mental disorders
• No pharmacological treatment
• Not on particular diet
• Maintained a steady weight for at least two months (±3kg)
• Low level of physical activity
• No pharmacological treatment.

• Unhealthy subjects
• Refusal to consent.

Design

Cross-sectional study with a before-after weight loss subset.

Statistical Analysis

• ANOVA was performed to examine the statistical significant differences among three BMI groups
• Bland's and Altman's indications were taken into account for the statistical analysis for comparison of the REE measured and the prediction equations
• Paired T-test was used to determine the statistically significant differences between the measured and predicted REE derived from the applied equations
• The ±95% CI and Bland's and Altman's analysis were used for comparison between measured and predicted REE, due to the limit of the mean
• Pearson coefficient correlation R between measured and predicted REE was calculated
• Stepwise multiple regression was performed using anthropometrical measurements as independent variables and REE as the dependent variable to develop a REE prediction equation.

Study Protocol

• REE was measured with IC and was compared with the principal prediction equations [Harris and Benedict, Owen, Mifflin St. Jeor, WHO, Bernsteine and Robertson and Reid, and one derived from current study (OUR)]
• OUR equation: 11.5wt(kg) + 542.2
• An additional REE measurement was performed on patients that had lost 5% or more of the initial body weight due to a sound low-calorie diet
• Subjects were separated into three groups according to their BMI:
  • BMI 25 or less: Normal weight
  • 25 to 30 BMI: Overweight
  • BMI 30 or more: Obese
• Steady state: RQ, oxygen consumption, minute ventilations were stable for at least five minutes.

Anthropometrics

• Height measured: Yes
• Weight measured: Yes
• Fat-free mass measured: Yes
• Body surface area (BSA), waist and hip ratio (WHR), BMI: Yes.

 Clinical

• Monitored heart rate: No
• Body temperature: No
• Medications administered: No.

Resting Energy Expenditure

• IC type: Canopy system
• Equipment of calibration: Yes
• Coefficient of variation using STD gases: Yes; ±3%
• Rest before measure (state length of time rested if available): 20 minutes
• Measurement length: 25 to 45 minutes
• Steady state: Evaluate if mention of initial acclimatization and then continuous monitoring throughout measurement, if appropriate
• Fasting length: 12 hours
• Exercise restrictions XX hours prior to test: 24 hours
• Room temperature: 25.0°C±0.5°C
• Number of measures within the measurement period: Not specified
• Were some measures eliminated: No
• Set of measurements averaged: Not specified.
• Coefficient of variation in subjects measures: No
• Training of measurer: No
• Subject training of measuring process: No. 

Dietary

Two-hour fast; food intake was not standardized prior to the IC study.

Intervening Factors

• Subjects slept for at least eight hours
• Were kept from smoking, sweet beverages, coffee, tea and drugs for at least 12 hours
• Refrained from doing physical activity in the 24 hours preceding the morning of the measurements.

Statistical Tests

• ANOVA
• Paired T-test
• Pearson coefficient correlation
• Stepwise multiple regression.

Timing of Measurements

• One measurement
• Additional measurement in sub-sample (N=31) who had lost greater than or equal to 5% of initial body weight on a low-calorie diet:
  • Patients were re-evaluated every 15 days per month
  • Given no less than four weeks for losing 5% of initial weight.

Dependent Variables/Outcomes

• Accuracy of REE estimates
• Prediction equation (OUR equation)
• Effect of weight loss on REE
  • IC type: Canopy system
  • Equipment of calibration: Yes
  • Coefficient of variation using STD gases: Yes; ±3%
  • Rest before measure (state length of time rested if available): 20 minutes
  • Measurement length: 25 to 45 minutes
  • Steady state: Measurements were made only when the subjects had reached a steady state condition (RQ, oxygen consumption, minute ventilation were stable for at least five minutes)
  • Fasting length: 12 hours
  • Exercise restrictions XX hour prior to test? 24 hours
  • Room temperature: 25.0°C±0.5°C
  • Number of measures within the measurement period: Not specified
  • Were some measures eliminated: No
  • Were a set of measurements averaged: Not specified
  • Coefficient of variation in subjects measures: No
  • Training of measurer: No
  • Subject training of measuring process: No
  • Monitored heart rate: No
  • Body temperature: No
  • Medications administered: No
  • Dietary: 12-hour fast, food intake was not standardized prior to the IC study
  • Intervening factors:
    • Subjects slept for at least eight hours
    • Were kept from smoking, sweet beverages, coffee, tea and drugs for at least 12 hours
    • Refrained from doing physical activity in the 24 hours preceding the morning of the measurement.

Independent Variables

• Principal prediction equations (Harris and Benedict, Owen, Mifflin, WHO, Bernstein and Robertson and Reid, and one derived from current study (OUR equation: 11.5wt (kg) + 542.2
• Anthropometric variables: Height (stadiometer to nearest 5cm); weight (in underclothes, precision balance scale to nearest 1kg), waist-hip ratio (WHR) [waist (inferior margin of last rib and the iliac crest), hip (tranchater level), BMI (kg/m²), body surface area]
• Weight loss of 5% of initial weight in no less than four weeks.

Control Variables

Subjects were separated into three groups according to their BMI:

• BMI=25: Normal weight
• BMI=25 to 30: Overweight
• BMI=30: Obese.

• Final N: N=157 females
• Age: 18 to 35 years
• Normal weight: 23.78±3.79 years
• Overweight: 25.39±5.42 years
• Obese: 23.82±5.49 years
• Ethnicity: Italian 
• Anthropometrics:

	Normal Weight (N=41)	Overweight (N=58)	Obese (N=58)
Height (cm)	160.58±5.7	161.27±6.86	161.45±6.75
Weight (kg)***	58.73±6.04	71.37±7.77	90.88±10.55
BMI (kg/m2)***	22.75±1.71	27.37±1.43	34.87±3.64
Waist circumference (cm)***	71.48±6.78	82.46±7.21	97.89±8.9
Hip circumference (cm)***	97.14±5.78	105.16±5.75	118.53±7.71
WHR***	0.73±0.04	0.78±0.06	0.82±0.07
BSA (m2)***	1.61±0.103	1.75±0.13	1.94±0.13

*** Statistical difference among normal weight, overweight and obese subjects on ANOVA test (P<0.001).

• Location: University "Federico II," Naples, Italy.

REE

• The mean numerical bias and the mean percentage bias estimates were 9.7 (under-estimation = -0.75%) and 8.81 (-0.76%) for Owen’s and OUR, respectively
• In the overweight group, Bernstein’s equation and OUR proved to be the most accurate, reporting a mean numerical bias and a mean percentage bias of –11.15 (over-estimation= ­-0.93%) and -4.75 (­0.4%), respectively
• Finally, in the third group of the obese subjects, the most reliable equation were Robertson and Reid’s and OUR with errors of -10.47 (­0.66%) and -1.28 (­0.1%), respectively. 

BMI Groups: Bias Percent ±95% CI  (Note: Negative number indicates and over-estimation.)

	Normal	Overweight	Obese
HB	-16.06 to -9.44	-13.16 to -8.39	-10.27 to -4.98
OWEN	-3.01 to 4.52	0.89 to 6.59	6.3 to 12.53
MIFFLIN	-9.95 to -2.86	-7.78 to -2.7	-5.94 to -0.32
BERNSTEIN	-7.8 to -0.51	-3.64 to 1.77	1.22 to 7.22
WHO	-12.4 to -5.65	-12.04 to -7.06	-12.02 to -6.38
R/R	-10.39 to -3.59	-7.87 to -2.91	-3.51 to 2.19
OUR	-2.98 to 4.52	-3.03 to 2.21	-2.88 to 2.67

• Normal weight group:
  • The ±95% CI of the MREE was 1,177.35 to 1,275.61
  • Owen’s equation and OUR stayed within such range, reporting REE prediction values of 1,203.04 to 1,230.41 and of 1,195.74 to 1,239.59, respectively
  • The same two equations showed a ±95% of the bias percentage within -3.01% to 4.53% and -2.98% to 4.52%.
• Overweight group:
  • The ±95% CI for MREE was 1,313.69 to 1,402.92
  • Bernstein’s for the first and OUR for the second had REE prediction in this interval: ±95% CI of 1,352.81 to 1,386 and 1,339.54 to 1,387.57, respectively
  • Moreover, they showed a ±95% CI of the bias percentage of -3.64% to 1.77% and -3.03% to 2.21%.
• Obese group:
  • The ±95% CI for MREE was 1,531.69 to 1,640
  • The most accurate equations were Robertson and Reids (1,566.42 to 1,626.42) and OUR (1,555.42 to 1,619.25) with ±95% CI of -3.51% to 2.19% and -2.88% to 2.67%, respectively. 

Stepwise Regression Analysis

Weight was a significant variable, resulting in the following equation: 542.2KG+11.5kg; R=0.59.

Bland-Altman Analysis 

• Approach accurately evaluating the over-estimation, the under-estimation and the gradual development of the error as REE increased
• The means of predicted and measured REE [measured + predicted)/2] plotted against the differences between the two procedures (i.e., measured – predicted) within two SDs from the mean (i.e., agreement lines) for all subjects (N=157) was reported for Harris-Benedict, Owen, Mifflin-St. Jeor, and Bernstein
• Harris-Benedict (Bland-Altman):
  • HB ranged from an over-prediction of 96kcals to an under-prediction of 48kcals (two SDs); three individuals had over-predictions more than 96kcals and three individuals had under-predictions higher than 48kcal.
• Owen (Bland-Altman):
  • Owen ranged from an over-prediction of 60kcals to an under-prediction of approximately 96kcals (two SDs); three individuals had over-predictions higher than 60kcals and six individuals had under-predictions higher than 96kcal.
• Mifflin-St. Jeor (Bland-Altman):
  • MSJE ranged from an over-prediction of 86kcals to an under-prediction of approximately 57kcal (two SDs); three individuals had over-predictions more than 86kcal and six individuals had under-predictions more than 57kcal.
• WHO/FAO/UNU (Bland-Altman):
  • WHO ranged from an over-prediction of 115kcal to an under-prediction of approximately 38kcal (two SDs); three individuals had over-predictions more than 115kcal and five individuals had under-predictions more than 38kcal.

Effect of Weight Loss

The effect of weight loss on the error of prediction was assessed in a group of 31 subjects. Only Owen’s equation maintained the error of prediction within acceptable limits. 

• Harris-Benedict had a group mean over-prediction of 9.92% prior to weight loss and this increased to a mean over-prediction of 18.25 with ±5% weight loss
• Owen had a group mean under-prediction error of 5.82% prior to weight loss and this changed to a group mean over-prediction error of 4.54%. Mifflin-St. Jeor had a group mean over-prediction error of 4.68% to an over-prediction error of 13.18% after ±5% weight loss.
• Owen had a group mean under-prediction error of 5.82% prior to weight loss and this changed to a group mean over-prediction error of 4.54%
• Mifflin-St. Jeor had a group mean over-prediction error of 4.68% to an over-prediction error of 13.18% after ±5% weight loss
• WHO/FAO/UNU had a group mean over-estimation error of 10.83% prior to ±5% weight loss and this increased to a 17.81% group mean error after weight loss.

• The equation of Owen in normal weight, Bernstein in overweight and of Robertson and Reid in obese subjects should be chosen when we have to predict REE in young women. Due to metabolic adaptations occurring during therapeutic or spontaneous energy restriction, we suggest using Owen’s equation.
• We evaluated the precision of the predictive equations over the entire population. Our results indicate that in the age group ranging 18 to 35 years, the most accurate equations were Bernstein’s and OUR, followed by Owen’s and Mifflin’s, which presented slight under-estimation and over-estimation errors respectively
• From a clinical perspective, the REE prediction after the loss of 5% of the initial weight was inappropriate so that all equations, including OUR, had to be considered inaccurate
• In the context of a clinical evaluation and therapeutic approach, it is important to evaluate the intensity and duration of the nutritional restriction because use of prediction equations could yield considerable RMR over-estimations
• From a physiological point of view, the error analysis indicated that energy restriction influenced RMR with consequent increase of the metabolic efficiency.

University/Hospital:

University Federico II (Naples Italy)

Strengths 

• Objective comparison of several equations was performed in the consideration of different weight status
• Sophisticated, appropriate statistics
• Careful evaluation of each equation being compared
• IC protocol was clearly presented. Intervening factors, such as smoking, physical activity, tea and coffee consumption were monitored.
• Conclusions were supported by the results.

Generalizability/Weaknesses

• Questionable validity of indirect calorimeter
• Menstrual cycle was not controlled in the study design
• The resulting equation (OUR) was able to apply to young female population (age 18 to 35) with different body weight (normal, overweight or obese).

[Evidence Analyst note: Bias% was (REE measured minus REE predicted) x 100 /REE predicted, which is a different proportion than IC study.]