How valid are
predictive equations
to estimate resting
energy expenditure
in obese elderly?
Bachelor Nutrition and Dietetics
Hogeschool van Amsterdam,
Amsterdam University of Applied
Sciences
Thesis number: 2012200D, June 2012
Authors: Jolanda M Koopman
Esther Ruigrok
1
.
2
How valid are predictive
equations to estimate
resting energy expenditure
in obese elderly?
Authors:
Jolanda M. Koopman
jolandamkoopman@hotmail.com
Esther Ruigrok
Estherruigrok01@hotmail.com
Thesis no:
2012200D
In order of:
Dr. Ir. PJM Weijs
Hogeschool van
Amsterdam,
Amsterdam University of
Applied Sciences
Dr. Meurerlaan 8
1067 SM Amsterdam
Supervisors:
Ir. AM Verreijen,
SE van der Plas
Copyright © 2012,
J.M. Koopman and E. Ruigrok
All rights served. No part of this publication
may be reproduced, stored in a database or
published in any form or by any meanselectronic, mechanical, photocopying,
recording or otherwise – without prior
permission of the publisher.
3
.
4
Table of contents
Table of contents ..................................................................................................................................................... 4
Preface .................................................................................................................................................................... 5
Abstract ................................................................................................................................................................... 9
Introduction ............................................................................................................................................................ 11
Methods ................................................................................................................................................................. 13
Subjects ............................................................................................................................................................. 13
Measurements ................................................................................................................................................... 13
Selection of predictive equations ....................................................................................................................... 14
Statistics ............................................................................................................................................................ 15
Results .................................................................................................................................................................. 19
Subject characteristics ....................................................................................................................................... 19
Predictive equations compared to indirect calorimetry ...................................................................................... 24
Discussion ............................................................................................................................................................. 26
Conclusion ............................................................................................................................................................. 30
References ............................................................................................................................................................ 32
Appendix I: Inclusion and exclusion criteria Muscle Preservation Study ............................................................... 36
Appendix II: Bland-Altman plots ............................................................................................................................ 38
5
6
Preface
This thesis was written to complete our study ‘Nutrition and Dietetics’ at the Amsterdam University of Applied
Sciences. We wrote this thesis during the period February 2012 until June 2012.
We have collected data for the Muscle Preservation Study in the food science laboratory of the Amsterdam
University of Applied Sciences. This data was partly used to write our thesis.
First and foremost we would like to give our special thanks to our supervisor Ir. AM Verreijen for all her support,
assistance and useful advises. She always made time for us and was very interested in our thesis. We would also
like to thank the principal investigator of the Muscle Preservation Study Dr. Ir. PJM Weijs for his vision and
expertise.
We are also very grateful for the support and assistance of SE van der Plas, T Haarsma and our colleagues at
the food science laboratory.
Amsterdam, June 8th 2012
E Ruigrok & JM Koopman
7
8
Abstract
Introduction:
Equations are often used to predict resting energy expenditure (REE), because measuring REE using indirect
calorimetry (IC) is hardly feasible in practice. It is unknown whether predictive equations are valid to estimate
REE in obese elderly. Therefore the aim of this thesis is to investigate the validity of predictive equations to
estimate REE in obese elderly compared to IC.
Methods:
Thirty predictive equations for estimating REE were tested and included, of which twenty equations based on
weight and ten equations based on fat free mass (FFM). REE was measured in 44 Dutch obese elderly (age ≥ 55
y) using IC. The validity of the predictive equations was determined by comparing predicted REE to measured
REE by using the following statistical tests: the percentage of accurate predictions (within 10% of REE
measured), root mean squared error (RMSE), concordance correlation coefficient (CCC) and the mean bias.
Results:
The three equations giving best results for predicting REE in obese elderly when comparing accurate predictions,
RMSE, CCC and bias are the equations of Harris & Benedict 1984 (68% accurate predictions; RMSE 189 kcal/d;
CCC 0.731; bias -72 kcal/d), Huang et al. (68% accurate predictions; RMSE 189 kcal/d; CCC 0.754; bias -50
kcal/d) and Korth et al. (66% accurate predictions; RMSE 188 kcal/d; CCC 0.769; bias 7 kcal/d). Of all equations
in this study most formulas underestimate REE more often than they overestimate REE. Equations using FFM
provided no advantage for accurately estimating REE compared to equations using weight only.
Conclusion:
The equations of Harris & Benedict 1984, Huang et al. and Korth et al. are accurate for predicting REE in obese
elderly. The equations of Huang et al. and Korth et al. are probably not very well known in dietary practice. The
Harris & Benedict 1984 equation however is one of the most commonly used formulas in dietary practice. It is
therefore advisable to use the Harris & Benedict 1984 equation for predicting REE in obese elderly.
Keywords: resting energy expenditure; indirect calorimetry; predictive equations; formulas; obese elderly
9
10
Introduction
The prevalence of obesity has more than doubled worldwide since 1980. According to an estimation of the World
Health Organization, over 200 million adult men and 300 million adult women suffered from obesity in 2008.1 At
the moment the prevalence of obesity is a major health concern and the prevalence is still rising.1,2 The main
causes are a decreased physical activity and an increased consumption of energy-dense foods.1 Energy-dense
foods are available everywhere, are convenient, relatively cheap, and quite tasty.3,4 In The Netherlands almost
19% of the population currently suffers from obesity.5
Not only the prevalence of obesity is increasing, the number of elderly is also rising rapidly.6 In 2002 were
approximately 605 million elderly worldwide. This is expected to grow to 1.2 billion in 2025.7The age group older
than sixty years is the fastest growing age group in most countries in the world.7 The causes are increasing life
expectancy, lower birth rates and in The Netherlands the aging of the baby boomers who were born after World
War II.8,9 Expected is that in Europe the proportion of people aged 65 and older will increase from 14% in 2010 to
25% in 2050.10,11 At the moment, the median age in Europe is already the highest in the world.10
Currently, more than fifteen percent of the Dutch population above 65 years old suffer from obesity and this
percentage is still rising.12 Obesity is clearly related to metabolic risk factors for cardiovascular diseases, diabetes
mellitus and some cancers.1,13-15 Obesity also plays a significant role in non-fatal physical disability in the
elderly.1,16 Voluntary weight loss in obese elderly leads to better metabolic control with better glucose regulation,
lower blood pressure, better pulmonary function and improved functional ability.13,15.
In The Netherlands, a CBO guideline is developed for the treatment of obese adults. This guideline advises an
energy intake of 600 kilocalories less than the daily energy expenditure.17 The daily energy expenditure can be
calculated by multiplying resting energy expenditure (REE) by physical activity level. REE is the amount of energy
the body needs for vital functions such as breathing, regulating body temperature, blood circulation, digestion and
absorption of food.18 The golden standard for measuring REE is the indirect calorimetry (IC).19-21 It measures
oxygen consumption and carbon dioxide production.22 By using the Weir formula, REE can be calculated.23
However, there are some disadvantages to measuring REE with the IC. At first, the dietetic institution needs an
indirect calorimeter which is very expensive. Secondly, the client is not allowed to eat for at least four hours and
exercise for at least 24 hours before the measurement. Thirdly, the measurement is time consuming and the
indirect calorimeter can only be operated by trained personnel.24,25 Since January 2012 dietary counseling is no
longer being reimbursed by basic health insurance in The Netherlands.26 This makes measurements with an
indirect calorimeter extremely expensive for clients. Therefore, the IC is not feasible in practice for the treatment
of obese elderly. To estimate the resting energy expenditure in a cheap and easy way, energy formulas are often
used.22,24,25
REE is different for obese elderly as compared to normal weight adults: obese adults have a higher total REE, but
the REE per kg of body weight is lower than in normal weight adults.25 This is due to a higher percentage of body
fat which has a low metabolic rate.27-33 In elderly REE is lower than in adults.6,25,34,35 REE decreases with aging
because fat free mass (mainly muscle) declines.6 After the age of 20 REE decreases approximately 1 or 2% per
decade.36
Most equations are developed and tested in healthy, normal weight, adult subjects.22,25 In obese people and
elderly, equations are probably less accurate.20,22,25
In obese adults, most predictive equations tend to underestimate REE.22 The Mifflin equation37 is the most
accurate equation to estimate REE in overweight or obese adults in the USA.20,22,25,38 For the Dutch overweight or
obese adults however, no accurate equation has been published. Weijs et al.39 have developed an equation that
is accurate in 81% of Dutch adult obese subjects, but this equation has yet to be published.39 Although the Mifflin
equation37 is accurate in overweight or obese Americans, the Mifflin equations tends to underestimate the REE in
American elderly.25
11
In American adults the Harris-Benedict 191940 equation is accurate for people with a normal bodyweight, but also
for the overweight and obese.25 The Harris-Benedict 191940 equation is also seems accurate in elderly people.25,41
Currently it is unknown whether predictive equations are valid to estimate REE in obese elderly.20 Therefore the
aim of this thesis is to investigate the validity of predictive equations to estimate REE in obese elderly compared
to indirect calorimetry.
12
Methods
In this study, baseline data from the Muscle Preservation Study (MPS) were used. The MPS investigates the
superiority of a specialized oral nutritional supplement on preservation of muscle mass in Dutch obese elderly
during a weight loss program. The participants were measured three times during the study: at baseline, after
seven weeks and after thirteen weeks. We only used data from the baseline measurements. Measurements were
conducted between March 1st 2011 and February 24th 2012 at the laboratory of the Amsterdam University of
Applied Sciences. The study protocol of the MPS was approved by the Medical Ethical Committee of the VU
medical centre.
Subjects
Obese elderly were recruited for participation by the distribution of flyers to citizens in district Amsterdam Nieuw
West. Flyers and posters were placed at general practices, community centres and supermarkets. Additionally an
advertisement was placed in the local newspapers Westerpost and Gazet.
Main inclusion criteria were: age 55-85 years, if women: postmenopausal, healthy, obese: BMI ≥30 kg/m2 or BMI
≥28 kg/m2 and waist circumference ≥88 cm (women) or ≥102 cm (men). Main exclusion criteria were:
Participation in a weight loss diet and/or participation in a resistance exercise program three months before the
start of the study and during the study. The complete inclusion and exclusion criteria of the Muscle Preservation
Study can be found in Appendix I.
All subjects were screened within three weeks before the baseline measurement. All subjects gave informed
consent.
Measurements
The following data were collected: gender, ethnicity, age in years, height in cm measured with a wall mounted
stadiometer (Seca 222, Seca, Hamburg, Germany) to the nearest cm, weight measured in kg by using the
BodPod precision scale (Life Measurement Inc, Concord, CA, USA) to the nearest 0.001 kg, fat mass (FM) and
fat free mass (FFM) in kg measured by using the BodPod system (Life Measurement Inc, Concord, CA, USA) and
REE in kcal/d measured by using IC with ventilated hood system (Vmax Encore n29; Viasys Healthcare, Houten,
The Netherlands).
Measurements with BodPod and IC were performed on the same day. To minimize the thermic effect of food, the
subjects were measured after a 5 h fast. Low caloric beverages were allowed until one hour before the
measurement. The subjects had not been physically active for at least 12 hours before the measurement.
BodPod
Air displacement plethysmography was used to measure body composition by using the BodPod sytem. The
BodPod system was calibrated every morning and additionally before each measurement. During the
measurement subjects had to wear tightly fitting underwear or bathing clothes and a swim cap. They were not
allowed to wear jewellery.
Each subject was measured two times. If the difference in body volume between the two measurements was less
than 0,2 L, the measurement was adopted. The average data of two measurements were used.
13
Indirect calorimetry
REE was measured using the golden standard: the IC. The measurements were performed with a ventilated hood
system. Before using, the Vmax was calibrated each day for volume with two standard gasses. The Vmax also
automatically recalibrated every five minutes.
During the measurement, the subjects were in a supine position and in a quiet environment at room temperature
(20ºC -25ºC). Falling asleep and coughing were not allowed. The ventilated hood was placed over the head and
sealed from outside air. The IC measures oxygen consumption and carbon dioxide production.22 The
measurement took twenty minutes. The steady state period of the measurement was selected according to the
procedures of the Vmax system. A steady state period is a ≥ five minute period in which oxygen consumption and
carbon dioxide production change by <10%.42 The Vmax system calculated REE for a 24h period (kcal/d) by
using the Weir formula.23
Selection of predictive equations
Predictive equations were selected using the US National Library of Medicine’s PubMed database. Used
keywords were: ‘resting energy expenditure’, ‘resting metabolic rate’, ‘predictive equation(s)’, and ‘indirect
calorimetry’. The additional terms were ‘formula’, ‘estimate’, ‘predict’, ‘obesity’, and ‘obese’. All terms were used in
every possible combination. Limitations placed on initial search strategies were: English language, adults (≥18 y)
and human subjects.
In 2008 Weijs22 studied which predictive equations are most valid for predicting REE in US and Dutch overweight
and obese adults aged 18-65 years old. The same formulas as in Weijs 200822 were also used in this study.
Additionally three more recent formulas were included (Lazzer et al. 201036, Lazzer et al. 2010 FFM36 and an
unpublished equation of Weijs et al.39 developed in 2010). All formulas were selected by the following inclusion
criteria: suitable for every age group, n ≥ 50. Exclusion criteria were as follows: formulas developed by using a
study population aged <18y, (critically ill) patients, specific ethnic groups, small sample size (n≤50).
Thirty predictive equations were included in this study of which 20 equations based on weight (Table 1A) and 10
equations based on FFM (Table 1B).
14
Statistics
The following statistical tests were used to evaluate the validity of the different predictive equations using the data
of the IC as reference: bias (mean difference), root mean squared error (RMSE), Bland-Altman plots,
concordance correlation coefficient (CCC) and the percentage of accurate predictions.
Data were analysed by using Predictive Analytics Software, former SPSS (PASW statistics, version 18.0.0, SPSS
Inc., Chicago, IL). A P-value of <0,05 was considered statistically significant. The RMSE and percentage of
accurate predictions were calculated by using MS Excel 2007. To calculate Bland-Altman and CCC MedCalc
software was used (MedCalc, version 12.1.1, Mariakerke, Belgium).
Bias and RMSE
The bias is the mean difference between REE predicted by formula and REE measured by IC. The bias is a
measure of accuracy in groups, but large individual errors can be averaged out.20,22 To investigate the accuracy
on an individual level, the RMSE was used. The RMSE is calculated as follows: individual differences between
predicted REE and measured REE were squared. These squared errors were summated. The root of the sum of
squared errors was taken and divided by n. The lower the RMSE, the more accurate the formula.43
Bland-Altman
A Bland-Altman plot visualises the difference between predicted REE and measured REE. The x-axis represents
REE measured by IC. The y-axis represents the difference between REE predicted by formula and REE
measured by IC. In a Bland-Altman plot every dot represents a subject. If the dots are located at y=0 this means
there is no difference between the formula and the IC. If the dots are located at y>0, this means that the formula
overestimates REE. If the dots are located at y<0, this means that the formula underestimates REE. The mean of
the differences is the bias. The bias is drawn as a line in the Bland-Altman plot. An accurate formula had a bias
line close to y=0. The limits of agreement are two lines located at +1,96 and -1,96 times the standard deviation of
the mean difference. Approximately 95% of the data are located between the limits of agreement. An accurate
formula has narrow limits of agreements.44-46
Concordance correlation coefficient
The CCC was used to measure precision and accuracy. The CCC differs from the Pearsons correlation
coefficient. The Pearsons correlation coefficient measures only the correlation between REE predicted and REE
measured, while the CCC also takes the deviation from the x=y line (Cb) into account. CCC is the product of the
Pearsons correlation coefficient and the Cb, in which the Pearsons correlation coefficient represents the precision
and the Cb represents the accuracy. CCC = 1 means that the formula predicts REE perfectly. CCC = 0 means
that the formula is not accurate at all.47,48
Percentage of accurate predictions
An accurate prediction in a subject is defined as a difference of +/- ≤10% between measured REE and predicted
REE. To examine the accuracy of a formula, the percentage of accurate predictions was calculated.20 The
accuracy of all included formulas were compared to determine the most accurate equation. A difference of <10%
was classified as an underprediction. A difference of >10% was considered an overprediction.22 Similar studies
have also used this method to examine accuracy.20,22
15
Table 1A: Predictive equations for resting energy expenditure using gender, age, weight, height and/or BMI.
Author(s)
and year of
publication
Bernstein
49
et al.
Subjects
N
n = 202
48 M
154 F
1983
FAO/WHO/
50
UNU
n ≈11 000
Age
Predictive equations
Weight or BMI
M: Mean age
40 ±13 y
Weight
60-204 kg
M: 11,02 × WT + 10,23 × HTCM – 5,8 × AGE – 1032 =
kcal/d
F: Mean age
39 ± 12 y
Mean weight
103 kg ± 26
F: 7,48 × WT – 0,42 × HTCM – 3 × AGE + 844= kcal/d
30-82 y
Equations using weight only
M 30-60 y: 11,6 × WT + 879 = kcal/d
M ≥60 y: 13,5 × WT + 487 = kcal/d
F 30-60 y: 8,7 × WT + 829 = kcal/d
F ≥60 y: 10,5 × WT + 596 = kcal/d
1985
Equations using weight and height
M 30-60 y: 11,3 × WT – 16 × HTM + 901 = kcal/d
M ≥60 y: 8,8 × WT + 1128 × HTM – 1071 = kcal/d
F 30-60 y: 8,7 × WT – 25 × HTM + 865 = kcal/d
F ≥60 y: 9,2 × WT + 637 × HTM - 302 = kcal/d
Harris and
40
Benedict
1919
n = 239
136 M
103 F
M: WT × 13,7516 + HTCM × 5,0033 – AGE × 6,755 +
66,473 = kcal/d
M: 16-63 y
Mean age
27 ± 9 y
F: WT × 9,5634 + HTCM × 1,8496 – AGE × 4,6756 +
655,0955 = kcal/d
F: 15-74 y
Mean age
31 ± 14 y
Harris and
Benedict
equation reevaluated by
51
Roza et al.
n = 337
168 M
169 F
M: Mean age
30 ± 14 y
M: 13,397 × WT + 4,799 × HTCM – 5,677 × AGE +
88,362 = kcal/d
F: Mean age
44 ± 22 y
F: 9,247 × WT + 3,098 × HTCM – 4,33 × AGE + 477,593
= kcal/d
1984
52
n = 2901
1544 M
1357 F
Henry
2005
M: Mean age
51 y
M: Mean BMI
23 kg/m2
F: Mean age
49 y
F: Mean BMI
24 kg/m2
Equations using weight only
M 30-60 y: 0,0592 × WT + 2,48 = MJ/d
M ≥60 y: 0,0563 × WT + 2,15 = MJ/d
F 30-60 y: 0,0407 × WT + 2,9 = MJ/d
F ≥60 y: 0,0424 × WT + 2,38 = MJ/d
Equations using weight and height
M 30-60 y: 0,0476 × WT + 2,26 × HTM – 0,574 = MJ/d
M ≥60 y: 0,0478 × WT + 2,26 × HTM – 1,07 = MJ/d
F 30-60 y: 0,0342 × WT + 2,1 × HTM – 0,0486 = MJ/d
F ≥60 y: 0,0356 × WT + 1,76 × HTM + 0,0448 = MJ/d
Huang et
53
al.
2007
Lazzer et
55,56
al.
2007
Lazzer et
36
al.
2010
BMI > 35 kg/m2
n = 104
20-70y
BMI 18-41 kg/m2
50 M
54 F
M: Mean age
37 ± 15 y
M: Mean BMI
26 ± 4 kg/m2
F: Mean age
35 ± 15 y
F: Mean BMI
26 ± 4 kg/m2
n = 346
M: 20-65 y
164 M
182 F
F:19-60 y
M: Mean BMI
45 kg/m2
M: 0,048 × WT + 4,655 × HTM –0,020 × AGE – 3,605 =
MJ/d
F: Mean BMI
46 kg/m2
F: 0,042 × WT + 3,619 × HTM – 2,678 = MJ/d
n = 7368
18-74 y
M: Mean age
46 ± 14 y
M: Mean BMI
42 ± 7 kg/m2
46 × WT – 14 × AGE + 1,14 × GENDER + 3,252 = kJ/d
2000 M
5368 F
Country of
origin:
Italy
F: Mean age
48 ± 14 y
279 M
759 F
2004
Korth et al.
Mean age
45 ± 13 y
n = 1088
54
Mean BMI
46 ± 8 kg/m2
10,158 × WT + 3,933 × HTCM – 1,44 × AGE + 273,821
× GENDER + 60,655 = kcal/d
41,5 × WT + 35,0 × HTCM + 1107,4 × GENDER – 19,1
× AGE – 1731,2 = kJ/d
F:Mean BMI
42 ± 7 kg/m2
16
Livingston
and
57
Kohlstadt
n = 655
229 M
356 F
M: 18-95 y
Mean age
36 ± 15 y
F: 18-77 y
Mean age
39 ± 13 y
2005
De Lorenzo
58
et al.
2001
n = 320
18-59 y
127 M
193 F
M: Mean age
29 ± 11
F: Mean age
41 ± 12 y
Mifflin et al.
37
n = 498
251 M
247 F
1990
M: 19-78 y
Mean age
44 ± 14 y
M weight:
33-278 kg
Mean weight
96 ± 45 kg
M: 293 × WT0,4330 – 5,92 × AGE = kcal/d
F: 248 × WT0,4356 – 5,09 × AGE = kcal/d
F weight:
36-261 kg
Mean weight
90 ± 37 kg
BMI 17-40 kg/m2
M: 53,284 × WT + 20,957 × HTCM – 23,859 × AGE +
487 = kJ/d
M: Mean BMI
27 ± 4 kg/m2
F: 46,322 × WT + 15,744 × HTCM – 16,66 × AGE + 944
= kJ/d
F: Mean BMI
28 ± 5 kg/m2
BMI 17-42 kg/m2
Mean BMI
27 ± 5 kg/m2
9,99 × WT + 6,25 × HTCM – 4,92 × AGE + 166 ×
GENDER - 161 = kcal/d
F: 20-76 y
Mean age
45 ± 14 y
59
Müller et al.
n = 2528
2004
1027 M
1501 F
5-91 y
Mean BMI
27 kg/m2
Equation suitable for everyone
0,047 × WT – 0,01452 × AGE + 1,009 × GENDER +
3,21 = MJ/d
Equation suitable for BMI 25-30
0,04507 × WT – 0,01553 × AGE + 1,006 × GENDER +
3,407 = MJ/d
Equation suitable for BMI ≥ 30
0,05 × WT – 0,01586 × AGE + 1,103 × GENDER +
2,924 = MJ/d
Owen et
60,61
al.
n = 104
1986-1987
62
Schofield
1985
60 M
44 F
n = 1106
> 30 y
M: 18-82 y
Mean age
38 ± 16 y
M: 20-59 kg/m2
Mean BMI
28 ± 8 kg/m2
F:18-65 y
Mean age
35 ± 12 y
F: 18-50 kg/m2
Mean BMI
28 ± 9 kg/m2
M: Mean age
42 y
M: Mean BMI
23 kg/m2
F: Mean age
43 y
F: Mean age
25 kg/m2
M: WT × 10,2 + 879 = kcal/d
F: WT × 7,18 + 795 = kcal/d
Equations using weight only
M 30-60 y: 0,048 × WT + 3,653 = MJ/d
M ≥60 y: 0,049 × WT + 2,459 = MJ/d
F 30-60 y: 0,034 × WT + 3,538 = MJ/d
F ≥60 y: 0,038 × WT + 2,755 = MJ/d
Equations using weight and height
M 30-60 y: 0,048 × WT – 0,011 × HTM + 3,67 = MJ/d
M ≥60 y: 0,038 × WT + 4,068 × HTM – 3,491 = MJ/d
F 30-60 y: 0,034 × WT + 0,006 × HTM + 3,53 = MJ/d
F ≥60 y: 0,033 × WT + 1,917 × HTM + 0,074= MJ/d
Weijs et al.
39
2010
(abstract)
n = 136
41 M
95 F
Mean age
42 ± 12 y
BMI 25-40 kg/m2
M: 16,790 × WT + 4,531 × HTCM + 0,169 × AGE –
420,492
F: 12,753 × WT + 3,869 × HTCM – 1,489 × AGE +
13,150
Data are displayed as mean ± standard deviation; M = male; F = female; D = diabetic; ND = non diabetic; WT = weight in kg;
HTCM = height in cm; HTM =height in m; GENDER: M = 1 F = 0; AGE =age in years
17
Table 1B: Predictive equations for resting energy expenditure using gender, age, FFM and/or FM.
Author(s)
and year of
publication
Bernstein
49
et al.
N
n = 202
Huang et
53
al.
2006
54
2007
Lazzer et
55,56
al.
2007
Lazzer et
36
al.
2010
Mean weight
103 kg ± 26
Mean age
45 ± 13 y
BMI > 35 kg/m2
n = 150
21-64 y
BMI 17-49 kg/m2
43 M
107 F
M: Mean age
47 ± 10 y
M: mean weight
85 ± 18 kg
F: Mean age
42 ± 11 y
F: Mean weight
69 ± 17 kg
n = 104
20-70y
BMI 18-41 kg/m2
50 M
54 F
M: Mean age
37 ±15 y
M: Mean BMI
26 ± 4 kg/m2
F: Mean age
35 ± 15 y
F: Mean BMI
26 ± 4 kg/m2
n = 346
M: 20-65 y
164 M
182 F
F: 19-60 y
M: Mean BMI
45 kg/m2
M: 0,081 × FFM + 0,049 × FM – 0,019 × AGE – 2,194 =
MJ/d
F: Mean BMI
46 kg/m2
F: 0,067 × FFM + 0,046 × FM + 1,568 = MJ/d
n = 7368
18-74 y
M: Mean age
46 ± 14 y
M: Mean BMI
42 ± 7 kg/m2
82 × FFM – 10 × AGE – 44 × GENDER + 3,517 = kJ/d
2000 M
5368 F
F: Mean age
48 ± 14 y
Mifflin et al.
37
n = 498
251 M
247 F
1990
19,02 × FFM + 3,72 × FM – 1,55 × AGE + 236,7 = kcal/d
F: Mean age
39 ± 12 y
279 M
759 F
Johnstone et
63
al.
Weight or BMI
Weight
60-204 kg
n = 1088
2004
Age
Predictive equation
M: Mean age
40 ±13 y
48 M
154 F
1983
Korth et al.
Subjects
M: 19-78 y
Mean age
44 ± 14 y
Mean BMI
46 ± 8 kg/m2
14,188 × FFM + 9,367 × FM – 1,515 × AGE + 220,863 ×
GENDER + 521,995 = kcal/d
90,2 × FFM + 31,6 × FM – 12,2 × AGE + 1613 = kJ/d
108,1 × FFM + 1231 =kJ/d
F: Mean BMI
42 ± 7 kg/m2
BMI 17-42 kg/m2
19,7 × FFM + 413 = kcal/d
Mean BMI
27 ± 5 kg/m2
F: 20-76 y
Mean age
45 ± 14 y
59
Müller et al.
n = 2528
2004
1027 M
1501 F
5-91 y
Mean BMI
27 kg/m2
Equation suitable for everyone
0,05192 × FFM + 0,04036 × FM + 0,869 × GENDER –
0,01181 × AGE + 2,992 = MJ/d
Equation suitable for BMI 25-30
0,03776 × FFM + 0,03013 × FM + 0,93 × GENDER –
0,01196 × AGE + 3,928 = MJ/d
Equation suitable for BMI ≥30
0,05685 × FFM + 0,04022 × FM + 0,808 × GENDER –
0,01402 × AGE + 2,818 = MJ/d
Owen et
60,61
al.
1986-1987
n = 104
60 M
44 F
M: 18-82 y
Mean age
38 ± 16 y
M: 20-59 kg/m2
Mean BMI
28 ± 8 kg/m2
F: 18-65 y
Mean age
35 ± 12 y
F: 18-50 kg/m2
Mean BMI
28 ± 9 kg/m2
M: 22,3 × FFM + 290 = kcal/d
F: 19,7 × FFM + 334 = kcal/d
Data are displayed as mean ± standard deviation; M = male; F = female; D = diabetic; ND = non diabetic; WT = weight in kg;
HTCM = height in cm; HTM =height in m; GENDER: M = 1 F = 0; AGE =age in years; FFM = fat free mass; FM = fat mass
18
Results
Subject characteristics
44 subjects were included in this study (Figure 1), 18 were male and 26 were female. Table 2 shows body
composition, anthropometric and REE data of all subjects and for men and women separately.
Males differ significantly from women on weight, height, fat weight, lean weight and REE IC. Women were shorter
(p<0.001), lighter (p=0.002), had more FM (p=0.012), less FFM (p<0.001) and a lower REE (P<0.001) than
males.
82% of the subjects were of Dutch origin. The country of birth of other subjects was Suriname (7%), Netherlands
Antilles (2%), Iran (2%), Egypt (2%), Croatia (2%) and South Africa (2%).
* Excluded from data analysis n = 35:






n=5: were claustrophobic. Because of this, BodPod or
IC could not be conducted.
n=9: did not fast
n=2: did not wear tightly fitting underwear
or bathing clothes and a swim cap
n=1: fell asleep during IC.
n=11: the difference between the two BodPod
measurements was too large.
n=7: the measuring equipment caused trouble.
Figure 1: Number of screened subjects, dropouts and included subjects in data analysis
Table 2: Subject characteristics: body composition, anthropometric data and resting energy expenditure
Total
(n= 45)
Men
(n=19 )
Mean
64 ± 5
102.0 ± 8.7
177 ± 7
32.7 ± 2.4
35.6 ± 6.4
66.4 ± 6.4
2014 ± 239
Women
(n=26)
Mean
61 ± 5
90.9 ± 12.3
165 ± 6
33.5 ± 5.7
43.2 ± 11.1
47.6 ± 5.3
1601 ± 200
P- value
Mean
Range
Age (years)
62 ± 5
55-77
0.126
Weight (kg)
95.4 ± 12.2
74.3-124.3
0.002*
Height (cm)
170 ± 9
154-188
<0.001*
BMI (kg/m2)
33.2 ± 4.6
28.1-51.3
0.508
FM (kg)
40.1 ± 10.1
21.9-77.8
0.012*
FFM (kg)
55.3 ± 10.9
39.0-78.7
<0.001*
REE IC (kcal)
1767 ± 296
1299-2462
<0.001*
Data are displayed as means ± standard deviation
REE = resting energy expenditure; FM = fat mass; FFM = fat free mass; REE IC = resting energy expenditure measured by
indirect calorimetry
* Significant difference between males and females (P <0.05)
Results on validity of the predictive equations are shown in Table 3A and 3B. Data are displayed as: REE in
kcal/d, bias in kcal/d, maximum negative error (underprediction) and maximum positive error (overprediction) in
kcal/d, RMSE in kcal/d and percentage of accurate -, under - and overpredictions.
19
Table 3A: Analysis of predictive equations based on bodyweight
Predictive equation
REE IC
Bernstein et al. 1983 49
FAO/WHO/UNU 1985 weight 50
FAO/WHO/UNU 1985 weight and height 50
Harris and Benedict 1919 40
Harris and Benedict 1984 51
Henry 2005 weight 52
Henry 2005 weight and height 52
Huang et al. 2004 53
Korth et al. 2007 54
Lazzer et al. 2007 55,56
Lazzer et al. 2010 36
Livingston and Kohlstadt 2005 57
De Lorenzo et al. 2001 58
Mifflin et al. 1990 37
Müller et al. 2004 59
Müller et al. 2004 BMI 59
Owen et al. 1986-1987 60,61
Schofield 1985 weight 62
Schofield 1985 weight and height 62
Weijs et al. 2010 39
Mean REE
SD
Bias
Maximum negative
error
Maximum positive
error
RMSE
Kcal/d
Kcal/d
Kcal/d
Kcal/d
Kcal/d
Kcal/d
1770
1377
1708
1705
1698
1718
1684
1665
1720
1777
1811
841
1591
1743
1615
1721
1715
1641
1648
1664
1877
296
176
203
187
228
225
231
219
243
262
194
132
193
211
207
216
227
251
195
210
248
-393
-62
-65
-72
-52
-86
-105
-50
7
41
-929
-179
-27
-155
-49
-55
-129
-122
-106
107
-224
-331
-310
-318
-320
-362
-353
-365
-397
-330
-224
-276
-335
-335
-327
-324
-312
-314
-332
-390
371
399
365
416
409
406
372
399
460
390
332
320
375
366
361
384
415
439
423
457
441
194
202
195
189
201
212
189
188
204
957
256
188
243
190
193
226
226
215
218
CCC
0.282
0.709
0.673
0.731
0.740
0.721
0.687
0.754
0.769
0.664
0.050
0.576
0.728
0.611
0.731
0.733
0.688
0.629
0.670
0.697
Accurate
predictions
Under
predictions
Over
predictions
%
%
%
9
59
61
59
68
59
59
68
66
55
0
52
64
52
57
59
57
61
57
52
91
30
30
30
20
30
34
20
14
9
100
48
16
48
27
25
36
34
36
7
0
11
9
11
11
11
7
11
20
36
0
0
20
0
16
16
7
5
7
41
REE = resting energy expenditure; SD = standard deviation; RMSE= root mean squared error; CCC = concordance correlation coefficient; REE IC = resting energy expenditure measured by indirect calorimetry; BMI= body mass index.
Table 3B: Analysis of predictive equations based on fat free mass
Predictive equation
REE IC
Bernstein et al. 1983 FFM 49
Huang et al. 2004 FFM 53
Johnstone et al. 2006 FFM 63
Korth et al. 2007 FFM 54
Lazzer et al. 2007 FFM 55,56
Lazzer et al. 2010 FFM 36
Mifflin et al. 1990 FFM 37
Müller et al. 2004 FFM 59
Müller et al. 2004 FFM BMI 59
Owen et al. 1986-1987 FFM 60,61
Mean REE
SD
Bias
Maximum negative
error
Maximum
positive error
RMSE
Kcal/d
Kcal/d
Kcal/d
Kcal/d
Kcal/d
Kcal/d
1770
1341
1678
1699
1723
1316
931
1502
1697
1678
1476
296
197
236
220
283
388
208
215
212
216
275
-429
-92
-71
-47
-454
-839
-268
-73
-92
-294
-310
-350
-353
-420
-697
-318
-320
-319
-300
-373
406
394
436
605
714
466
462
350
363
570
465
198
190
198
741
859
326
193
202
349
CCC
0.297
0.736
0.738
0.764
-0.250
0.112
0.477
0.723
0.707
0.505
Accurate
predictions
Under
predictions
Over
predictions
%
%
%
0
59
57
55
34
0
34
59
61
25
100
32
34
25
52
100
66
27
30
75
0
9
9
20
14
0
0
14
9
0
REE = resting energy expenditure; SD = standard deviation; RMSE= root mean squared error; CCC = concordance correlation coefficient; REE IC = resting energy expenditure measured by indirect calorimetry;
FFM= fat free mass; BMI= body mass index.
Figure 2A: Percentage accurate resting energy expenditure (REE) prediction (within 10% of measured REE)
FFM= fat free mass; W= weight; H= height; BMI= body mass index.
Equations are ranked from high accuracy to low percentage accuracy. The three equations with the highest percentage accurate
predictions are coloured pink.
Figure 2B: Root mean squared error (RMSE) in kcal/day
FFM= fat free mass; W= weight; H= height; BMI= body mass index
Equations are ranked from low RMSE to high RMSE. The three equations with the lowest RMSE are coloured pink.
21
Figure 2C: Concordance correlation coefficient (CCC)
FFM= fat free mass; W= weight; H= height; BMI= body mas index.
Equations are ranked from high CCC to low CCC. The three equations with the highest CCC are coloured pink.
Figure 2D: Bias in kcal/day
FFM= fat free mass; W= weight; H= height; BMI= body mass index.
Equations are ranked from high bias to low bias. The three equations a bias closest to zero are coloured pink.
22
23
Predictive equations compared to indirect calorimetry
The comparison between REE predicted by equations and REE measured by IC is shown in Table 3A and Table 3B.
The bias varied from -929 until 107 kcal/d. The largest negative error was -697 kcal/d. The largest positive error was 714 kcal/d.
The RMSE varied from 188 until 957 kcal/d. CCC varied from -0.250 until 0.769. The highest percentage accurate predictions was
68%. The lowest percentage accurate predictions was 0%. Most formulas underestimate REE more often than they overestimate
REE.
The predictive equations with the highest percentage of accurate predictions are Harris & Benedict 198451, Huang et al.53 and Korth
et al.54 with 68%, 68% and 66% respectively. In all other formulas the percentage of accurate predictions were ≤ 64%. The
equations of Lazzer et al. 201036, Lazzer et al. 2010 FFM36 and Bernstein et al. FFM49 did not predict REE accurately in any subject
(Figure 2A).
RMSE’s were the lowest in the formulas of Korth et al.54 and De Lorenzo et al.58, both 188 kcal/d. The equations of Harris &
Benedict 198451 and Huang et al.53 had the second lowest RMSE’s, both 189 kcal/d (Figure 2B). CCC was the highest in the
formula of Korth et al.54 with a CCC of 0.769 (Figure 2C). The bias was remarkably low in the Korth et al.54 equation with a bias of
only 7 kcal/d (Figure 2D).
The FFM equations had a lower percentage of accurate predictions than the weight and height equations. The FFM equations with
the highest percentage of accurate predictions were Müller et al. FFM with BMI59, Müller et al. FFM59 and Huang et al. FFM53, with
61%, 59% and 59% respectively. In all other FFM equations accurate predictions were ≤ 57%. The FFM equations of Bernstein et
al.49 and Lazzer et al. 201036 did not predict REE accurately in any subject, they show 100% underprediction. All FFM equations
underestimate REE more often than they overestimate REE.
Bland-Altman plots of the Harris & Benedict 198451, Huang et al.53 and Korth et al.54 equations are shown in Figure 3A,3B,3C.
Bland-Altman plots of all other formulas are shown in Appendix II.
24
1000
800
600
400
+1.96 SD
308,9
200
Mean
-52,1
0
-200
-1.96 SD
-413,1
-400
-600
-800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
Figure 3A: Bland-Altman plot of the Harris & Benedict 198451 equation in kcal/day
1000
800
600
400
+1.96 SD
312,8
200
Mean
-49,7
0
-200
-1.96 SD
-412,3
-400
-600
-800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
Figure 3B: Bland-Altman plot of the Huang et al.53 equation in kcal/day
1000
800
600
+1.96 SD
379,5
400
200
Mean
7,1
0
-200
-1.96 SD
-365,2
-400
-600
-800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
Figure 3C: Bland-Altman plot of the Korth et al.54 equation in kcal/day
25
Discussion
This study shows that the three equations giving best results for predicting REE in obese elderly based on percentage accurate
predictions, RMSE, CCC and bias are the equations of Harris & Benedict 198451, Huang et al.53 and Korth et al.54 Of all equations in
this study most formulas underestimate REE more often than they overestimate REE. Equations using FFM provided no advantage
for accurately estimating REE compared to equations using weight only.
The Harris and Benedict 198451, Huang et al.53 and Korth et al.54 equations all have a RMSE of ≥188 kcal/d. For dieticians this
means a deviation of approximately one bottle of standard enteral formula a day.64
The equations of Harris & Benedict51 and Huang et al.53 had the highest percentages of accurate predictions, 68%. The Huang et
al.53 equation is developed on a study group of obese adults (age 45 ± 13 y). The mean BMI of the study group was 46 ± 8 kg/m2.
The results of our study show that the equation of Huang et al.53 is suitable for obese elderly. This result could be explained by the
obese study group of Huang et al.53
The Harris & Benedict51 1984 equation was based on a study group of normal weight adults (age men 30 ± 14 y; age women 44 ±
12 y). The Korth et al.54 equation, which had 66% accurate predictions, was based on a study group of normal weight and
overweight adults (age men 37 ± 15 y; age women 35 ± 15 y). The mean BMI of the study group was 26 ± 4 kg/m2. Hence, it is
remarkable that the Harris & Benedict 198451 and the Korth et al.54 equations predict REE well in obese elderly.
Our results are comparable to other studies. In a similar study of Weijs 200822, who studied the validity of predictive equations in
obese Dutch adults, the Korth et al.54 equation had the highest percentage of accurate predictions with 67%. The Harris & Benedict
198451 equation estimated fairly well with 63% accurate predictions. The Huang et al.53 equation however preformed poorer with 49
% accuracy. The bias, CCC and RMSE of the Korth et al.54, Harris & Benedict 198451 and Huang et al.53 equations were ranked in
the same way (bias: -0.9%, -3.8%, -8.1%; CCC: 0.792, 0.753, 0.683; RMSE: 198, 211, 251 in kcal/d, respectively).
Weijs & Vasant65 showed that the Huang et al.53 equation is accurate in normal weight to morbid obese Belgium women with 71 %
accuracy. The Harris & Benedict 198451 equation also gave accurate results with 68 % accuracy. The formula of Korth et al.54 was
accurate in 63%. The bias and RMSE of the Huang et al.53, Harris & Benedict 198451 and Korth et al.54 equations were ranked in
the same way (bias: -1.7%, 3.0%, 4.5%; RMSE: 168, 169,179 in kcal/d, respectively).
Hofsteenge et al.66 studied the validity of predictive equations in obese adolescents. In this study the equation of Korth et al.54 gave
accurate predictions in 72% of all subjects with a bias of 3.4% and a RMSE of 187 kcal/d. Harris & Benedict 198451 equation was
accurate in 70%, had a bias of -0.1% and a RMSE of 186 kcal/d. The Huang et al.53 equation was less accurate with 55%, a bias of
-6.9% and a RMSE of 229 kcal/d. However, comparing Hofsteenge et al.66 to our study is difficult because of the major difference
in age between obese adolescents and obese elderly.
Many studies show that equations using FFM do not provide any advantage over equations using weight. Authors of equations
using weight often also made an equation using FFM. The equations using FFM usually had a lower percentage of accurate
predictions and a slightly higher RMSE.22,65-68 This confirms our findings.
There are also differences between this study and other studies. In other studies, the Mifflin equation turned out to be the most
accurate equation for predicting REE in overweight and obese subjects.20,22,25,38,67,69 Most of these studies are performed on an
American study group. Contrary to other studies, this study used only Dutch subjects. Furthermore, this study used elderly subjects
while other studies used adults. This could explain the different results. The Mifflin equation is known to underestimate REE in
elderly25. In our study, the Mifflin equation underestimates approximately 155 kcal. Compared to the other equations in this study, a
bias of -155 kcal/d is relatively large.
There are only a few studies which examined the validity of Harris & Benedict 198451 equation, the Huang et al.53 equation and/or
the Korth et al.54 equation. This makes it difficult to compare our results to other studies. Most validation studies examined the
validity of the original Harris & Benedict 191940 equation and a few other commonly used equations. The original Harris & Benedict
191940 equation often preformed well in elderly and in obese adults.25,41,70,71 The Harris & Benedict 198451 equation was based on
the same study group as the original Harris & Benedict 191940 equation, however 98 additional subjects were added.51
26
Incorrect equation
The equation of Lazzer et al. 2007 using FFM55,56 gives extremely inaccurate results. The CCC was exceptionally low, -0.250. The
equation did not predict REE accurately in any male subject, in women however the equation had 58% accurate predictions. The
lectureship weight management of the Hogeschool van Amsterdam, contacted sir S. Lazzer to ask for an explanation. The
published article of Lazzer et al.55,56 in 2007 appeared to contain a publication error in the FFM equation for males. The formula
developed by Lazzer et al. using FFM for males which should have been published is REE (MJ/d) = 0.081 * FFM + 0.049 * FM 0.019 * AGE + 2.194. This corrected formula preformed well with 57% accurate predictions, a RMSE of 181 kcal/d, a CCC of 0.753
and a bias of -25 kcal/d. In this study only equation as published (incorrect equation) was included.
Two other equations of sir S. Lazzer (Lazzer et al. 2010 and Lazzer et al. 2010 using FFM36) also gave inaccurate results. It is
unknown whether these equations also contain publication errors.
Limitations and strengths
This study has a couple of strengths and limitations. A limitation of this study is the small number of subjects. The study started
with 82 subjects. Two subjects dropped out before the end of the study. The 35 subjects were removed from the dataset mainly
because of measurement errors like a to large difference between two BodPod measurements. Also some subjects did not fast or
worn clothes that were not tightly fitting which could influence our results. A few subjects were claustrophobic making it impossible
to conduct IC or BodPod measurements. Only 44 subjects were actually included. This means that each included subject accounts
for approximately 2% in this study. A larger study group would give a more reliable result. However, a reliable dataset is also very
important.
Another limitation of this study is the allowance of coffee until before the measurement. According to the MPS protocol, subjects
were allowed to drink low caloric beverages (including coffee) until one hour before the measurement. Coffee however contains
caffeine, which is known to elevate REE. The consumption of 200 mg caffeine elevates REE approximately 7% during three hours
(200 mg caffeine is about two small cups of coffee = 250 ml in total or four small cups of tea, which also contains caffeine = 500 ml
in total).72 This could have influenced our results.
A comment on this study is that the subjects were measured in three different groups. The first group was measured in February
2011, the second group in September 2011 and the third group in February 2012. Data were collected by three different teams of
approximately eight trained persons. One of them participated in collecting data of all three groups and made sure that the
measurements were performed the same way in all three groups. All data were collected in the same dataset. To get the most
reliable dataset, all entered data were double checked.
In order to get results as reliable as possible each subject was measured twice in the BodPod. The average of two measurements
was used as data. The maximum difference in volume between two measurements of the BodPod in the MPS protocol was set at
0,15L. This study adopted a maximum of 0,2L. This resulted in a difference in fat percentage of <0,5% between two BodPod
measurements. A difference in volume >0,2L would give major differences in FM and FFM and would result in unreliable data.
Therefore, subjects with volume differences >0,2L were excluded from the study.
Another comment on this study is that the stadiometer (Seca 222, Seca, Hamburg, Germany) was mounted 7mm too low on the
wall. This meant a measurement error for all subjects of -7mm in height. The measurement error was detected after all
measurements were conducted. To correct this measurement error, 7 mm was added to the height of all subjects afterwards.
Unfortunately all calculations were already done before the measurement error was detected. This meant that all results had to be
recalculated.
A strength of this study is that few subjects (18%) were from foreign origin: 7% is Surinamese, 2% is Netherlands Antilleans, 2% is
Iranian, 2% is Egyptian, 2% is Croatian and 2% is South African. The difference in body composition between people from different
countries has influence the results. However, the largest part 82% of the subjects in this study are from Dutch origin. Therefore the
subjects born in other countries will have limited influence on the results of this study.
27
Recommendations
It is not recommended to use equations using FFM, because these equations are less accurate than equations using weight.
Measuring FFM is also time consuming in dietary practice.
This study showed that the three equations giving best results for predicting REE in obese elderly are the equations of Harris &
Benedict 198451, Huang et al.53 and Korth et al.54 The equations of Huang et al.53 and Korth et al.54 are probably not very well
known in dietary practice. The Harris & Benedict 198451 equation however is one of the most commonly used formulas in dietary
practice38, so this equation is well known to dieticians. Our findings that the Harris & Benedict 198451 equation predicts REE
accurately in the obese are also confirmed by other studies.22,65 Therefore it is advisable to use the Harris & Benedict 198451
equation for predicting REE in obese elderly.
28
29
Conclusion
This study investigated the validity of 30 predictive equations in obese elderly. There was a major difference in accuracy between
the equations. The accuracy of the equations varied between 0 and 68% accurate predictions. Most formulas underestimate REE
more often than they overestimate REE.
Equations using FFM provided no advantage over equations using weight for accurately estimating REE. Therefore, it is not
recommended to use equations using FFM. Measuring FFM is also time consuming in dietary practice.
This study shows that the three equations giving best results for predicting REE in obese elderly based on percentage accurate
predictions, RMSE, CCC and bias are the equations of Harris & Benedict 1984, Huang et al. and Korth et al. The equations of
Huang et al. and Korth et al. are probably not very well known in dietary practice. The Harris & Benedict 1984 equation however is
one of the most commonly used formulas in dietary practice, so this equations is well known to dieticians. Our findings that the
Harris & Benedict 1984 equation predicts REE accurately in the obese are also confirmed by other studies. It is therefore advisable
to use the Harris & Benedict 1984 equation for predicting REE in obese elderly.
30
31
References
1 World Health Organization. Media centre. Factsheets. Obesity and overweight. May 2012. Available from:
http://www.who.int/mediacentre/factsheets/fs311/en/index.html
2 World Health Organization. Programmes and projects. Global health observatory. Obesity. Available from:
http://www.who.int/gho/ncd/risk_factors/obesity_text/en/ Date of visit: May 2012
3 Peters JC. The challenge of managing body weight in the modern world. Asia Pac J Clin Nutr. 2002;11: S714-S717
4 Binkley JK, Eales J, Jekanowski M. The relation between dietary change and rising US obesity. Int J Obes Rel Metab Disord. 2000;24:1032-1039
5 World Health Organization. Programmes and projects. Publications. Noncommunicable diseases country profiles. NCD profiles by country. Netherlands. 2011.
Available from: http://www.who.int/nmh/countries/en/index.html
6 Ritz. P. Factors affecting energy and macronutrient requirements in elderly people. Public Health Nutr. 2001;4:561-8
7 World Health Organization. Programmes and projects. Nutrition topics. Nutrition for the elderly. Available from:
http://www.who.int/nutrition/topics/ageing/en/index.html Date of visit: May 2012
8 Bie van der R, Latten J. Baby Boomers: Indrukken vanuit de statistiek. Haarlem/Heerlen: Centraal Bureau voor de Statistiek; 2012. p. 6-11,42,43
9 World Health Organization. Health topics. Ageing. Available from: http://www.who.int/topics/ageing/en/ Date of visit: May 2012
10 World Health Organization Europe. What we do. Health topics. Healthy aging. Available from: http://www.euro.who.int/en/what-we-do/health-topics/Lifestages/healthy-ageing Date of visit: May 2012
11 Centraal bureau voor de statistiek. Thema’s. Bevolking. Publicaties. Artikelen en persberichten. 2003. Na 2010 slaat de vergrijzing toe. April 2003. Available
from: http://www.cbs.nl/nl-nl/menu/themas/bevolking/publicaties/artikelen/archief/2003/2003-1175-wm.htm
12 Centraal bureau voor de statistiek. Statline. Tabellen per thema. Gezondheid en welzijn. Leefstijl en preventief gedrag. Gezondheid, leefstijl, zorggebruik; t/m
2009. March 2010. Available from: http://statline.cbs.nl/StatWeb/publication/?DM=SLNL&PA=03799&D1=267-271&D2=0-17&D3=0&D4=a&VW=T
13 Salihu HM, Bonnema SM, Alio AP. Obesity: What is an elderly population growing into? Maturitas. 2009;63:7-12
14 Lavie CJ, MiIani RV, Ventura HO. Obesity and cardiovascular disease. Risk factor, paradox, and impact of weight loss. J Am Coll Cardiol. 2008;53:959-70
15 Bales CW, Buhr G. Is obesity bad for older persons? A systematic review of the pros and cons of weight reduction in later life. J Am Med Dir Assoc.
2008;9:302-12
16 Zamboni M, Mazzali G, Zoico E, et al. Health consequences of obesity in the elderly: a review of four unresolved questions. Int J Obes. 2005;29:1011-29
17 Kwaliteitsinstituut voor de Gezondheidszorg CBO. Richtlijn: diagnostiek en behandeling van obesitas bij volwassenen en kinderen. Alphen aan den Rijn: Van
Zuiden Communications; 2008. p. 130 Available from: http://www.cbo.nl/thema/Richtlijnen/Overzicht-richtlijnen/Voedingsgerelateerde-aandoeningen/?p=295
18 Thomas B. Manual of Dietetic Pratice. Third edition. Oxford: Blackwell Publishing; 2007. p. 136,137
19 Schoeller DA. Making indirect calorimetry a gold standard for predicting energy requirements for institutionalized patients. J Am Diet Assoc. 2007;107:390-2
120 Frankenfield D, Roth-Yousey L, Compher C. Comparison of predictive equations for resting metabolic rate in healthy nonobese and obese adults: a
systematic review. J Am Diet Assoc. 2005;105:775-89
21 Cooper JA, Watras AC, O’Brien MJ, et al. Assessing validity and reliability of resting metabolic rate in six gas analysis systems. J Am Diet Assoc.
2009;109:128-32
22 Weijs PJM. Validity of predictive equations for resting energy expenditure in US and Dutch overweight and obese class I and II adults aged 18-65 y. Am J
Clin Nutr. 2008;88:959-70
23 Weir JB. New Methods for calculating metabolic rate with special reference to protein metabolism. J Physiol. 1949;109:1-9
24 Foltz MB, Schiller MR, Ryan AS. Nutrition screening and assessment: current practices and dietitians’ leadership roles. J Am Diet Assoc. 1993;93:1388-95
32
25 Hasson RE, Howe CA, Jones BL, et al. Accuracy of four resting metabolic rate prediction equations: effects of sex, body mass index, age and race/ethnicity.
J Sci Med Sport. 2011;14:344-51
26 Ministerie van Volksgezondheid, Welzijn en Sport (Dutch Ministry of Healthy, Welfare and Sport). Veranderingen in het basispakket. Dieetadvies.
http://www.veranderingenindezorg.nl/zorg/dieetadvies.html Date of visit: May 2012
27 Forbes GB. Human body composition: growing, ageing, nutrition and activity. New York: Springer. 1982.
28 Forbes GB. Lean body mass – fat interrelationship in humans. Nutr Rev. 1987;45:225-31
29 James WPT. In energy metabolism: Tissue determinants and cellular corollaries. 1992.
30 Jequier E. Energy expenditure in obesity. Clin Endocrinol Metab. 1984;13:563-80
31 Miller AT, Blyth CS. Lean body mass as a metabolic reference standard. J Appl Physiol. 1953;5:311-16
32 Prentice AM, Black AE, Murgatroyd PR, et al. Metabolism or appetite: questions of energy balance with particular reference to obesity. J Hum Nutr Diet.
1989;2:95-104
33 Rucker RB. Elevated metabolic rates in obesity. Lancet. 1978;8080:106-7
34 Blanc S, Schoeller DA, Bauer D, et al. Energy requirements in the eighth decade of life. Am J Clin Nutr. 2004;79:303-10
35 Roberts SB, Dallal GE. Energy requirements and aging. Public Health Nutr. 2005;8:1028-36
36 Lazzer S, Bedogni G, Lafornuna CL, et al. Relationship between basal metabolic rate, gender, age, and body composition in 8,780 white obese subjects.
Obesity. 2010;18:71-78
37 Mifflin MD, St Jeor ST, Hill LA, et al. A new predictive equation for resting energy expenditure in healthy individuals. Am J Clin Nutr. 1990;51:241-7
38 Oliveira de EP, Orsatti FL, Teixeira O, et al. Comparison of predictive equations for resting energy expenditure in overweight and obese adults. J Obes.
2011:1-5
39 Weijs PJM, Hesseling MG, Bast D, et al. Development and cross-validation of resting energy expenditure prediction equation for overweight and obese
adults. Clinical Nutrition Supplements 2010;3:141 [abstract]
40 Harris JA, Benedict FG. A biometric study of human basal metabolism in man. Washington, DC: Carnegie Institute of Washington, 1919
41 Gaillard C, Alix E, Sallé A, et al. Energy requirements in frail elderly people: A review of the literature. Clin Nutr. 2006;26:16-24
42 McClave SA, Spain DA, Skolnick JL, et al. Achievement of steady state optimizes results when performing indirect calorimetry. J Parenter Enteral Nutr.
2003;27:16-20
43 Weijs PJM, Kruizenga HM, Dijk van AE, et al. Validation of predictive equations for resting energy expenditure in adult outpatients. Clin Nutr. 2008;27:150-7
44 Bland JM, Altman DG. Measurement in Medicine: the Analysis of Method Comparison Studies. The Statistician 1983;32:307-17
45 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement.. Lancet 1986;1:307-10.
46 Twisk JWR. Inleiding in de toegepaste biostatistiek. Second edition. Amsterdam: Elsevier Gezondheidszorg; 2010. p. 324-326
47 Lin LI. A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989;45:225-68
48 Lin L, Torbeck LD. Coefficient of accuracy and concordance correlation coefficient: new statistics for methods comparison. PDA J Pharm Sci Technol
1998;52:55-9
49 Bernstein RS, Thornton JC, Yang MU, et al. Prediction of the resting metabolic rate in obese patients. Am J Clin Nutr. 1983;37:595-602
50 FAO/WHO/UNU. Energy and protein requirements. Geneva, Switzerland: World Health Organ Tech Rep Ser, 1985
51 Roza AM, Shizgal HM. The Harris Benedict equation reevaluated: resting energy requirements and the body cell mass. Am J Clin Nutr. 1984;40:168-82
52 Henry CJK. Basal metabolic rate studies in humans: measurement and development of new equations. Public Health Nutr. 2005;8:133-52
53 Huang KC, Kormas N, Steinbeck K, et al. Resting metabolic rate in severely obese diabetic and nondiabetic subjects. Obes Res. 2004;12:840-5
54 Korth O, Bosy-Westphal A, Zschoche P, et al. Influence of methods used in body composition analysis on the prediction of resting energy expenditure. Eur J
Clin Nutr 2007;61:582-9
55 Lazzer S, Agosti F, Resnik M, et al. Prediction of resting energy expenditure in severely obese Italian males. J Endocrinol Invest. 2007;30:754-61
56 Lazzer S, Agosti F, Silvestri P, et al. Prediction of resting energy expenditure in severely obese Italian women. J Endocrinol Invest. 2007;30:20-7
33
57 Livingston EH, Kohlstadt I. Simplified resting metabolic rate – predicting formulas for normal-sized and obese individuals. Obes Res. 2005;13:1255-62
58 De Lorenzo A, Tagliabue A, Andreoli A, et al. Measured and predicted resting metabolic rate in Italian males and females, aged 18-59 y. Eur J Clin Nutr.
2001;55:208-41
59 Müller MJ, Bosy-Westphal A, Klaus S, et al. World Health Organization equations have shortcomings for predicting resting energy expenditure in persons
from a modern, affluent population: generation of a new reference standard from a retrospective analysis of a German database of resting energy expenditure.
Am J Clin Nutr. 2004;80:1379-90
60 Owen OE, Holup JL, D’Alessio DA, et al. A reappraisal of the caloric requierements of healthy men. Am J Clin Nutri. 1986;44:1-19
61 Owen OE, Holup JL, D’Alessio DA, et al. A reappraisal of caloric requierements in healthy women. Am J Clin Nutri. 1986;44:1-19
62 Schofield WN. Predicting basal metabolic rate, new standards and review of previous work. Hum Nutr Clin Nutr. 1985;39C:5-41
63 Johnstone AM, Rance KA, Murison SD, et al. Additional anthropometric measures may improve the predictability of basal metabolic rate in adult subjects.
Eur J Clin Nutr. 2006;60:1437-44
64 Genuchten van S, Heijden van der H. Compendium dieetproducten en voedingssupplementen. 35e edition. Houten: Bohn Stafleu van Loghum; 2008. p. 113
65 Weijs PJM, Vansant GAAM. Validity of predictive equations for resting energy expenditure in Belgian normal weight to morbid obese women. Clin Nutr.
2009;29:347-51
66 Hofsteenge GH, Chinapaw MJ, Delemarre-Waal van de HA, et al. Validation of predictive equations for resting energy expenditure in obese adolescents. Am
J Clin Nutr. 2010;91:1244-54
67 Ruiz JR, Ortega FB, Rodríguez G, et al. Validity of resting energy expenditure predictive equations before and after an energy-restricted diet intervention in
obese women. PloS ONE. 2011;6:1-11
68 Wilms B, Schmid SM, Ernst B, et al. Poor prediction of resting energy expenditure in obese women by established equations. Metabolism Clinical and
Experimental 2010;59:1187
69 Frankenfield DC, Rowe WA, Smith JS, et al. Validation of several established equations for resting metabolic rate in obese and nonobese people. J Am Diet
Assoc. 2003;103:1152-59
70 Melzer K, Laurie Karsegard V, Genton L, et al. Comparison of equations for estimating resting metabolic rate in healthy subjects over 70 years of age. Clin
Nutr. 2007:26:498-505
71 Assessment of resting energy expenditure of obese patients: Comparison of indirect calorimetry with formulae. Alves VGA, Da Rocha EEM, Gonzalez MC, et
al. Clin Nutr. 2009;28:299-304
72 Koot P, Deurenberg P. Comparison of changes in energy expenditure and body temperature after caffeine consumption. Ann Nutr Metab. 1995;39:135-42
34
35
Appendix I: Inclusion and exclusion criteria Muscle Preservation Study
Inclusion criteria Muscle Preservation Study
-
-
Age 55-85 years
If women: postmenopausal
BMI ≥30 kg/m2 or BMI ≥28 kg/m2 and waist circumference ≥88 cm (women), ≥102 cm (men)
Ability to sign informed consent
Willingness and ability to comply with the protocol, including:
o Maintaining current dietary habits
o Participation in study visits
o Taking the study product every day
o Ability to comply with the complete study protocol
o Ability to understand and fill out questionnaires
Physiotherapist’s professional view that the subject is physically fit and it is safe to participate in the resistance exercise
program
Exclusion criteria Muscle Preservation Study
-
-
-
-
-
Any malignant disease during the last five years except for adequately treated prostate cancer without evidence of
metastases, localized bladder cancer, cervical carcinoma in situ, breast cancer in situ or non-melanoma skin cancer
Previously known:
o Kidney failure
o Liver failure
o Anaemia
o (Chronic) inflammatory status
Medication
o Corticosteroids for systemic use
o Immunosuppressant’s
o Insulin
Dietary of lifestyle characteristics:
o Participation in a weight loss diet three months before staring and during the study
o Use of protein-containing or amino acid-containing nutritional supplements three months before staring and
during the study
o Participation in a resistance exercise program three months before staring and during the study
o Current alcohol or drug abuse
Indications related to interaction with the study product:
o More than 10 μg (400 IU) of daily Vitamin D intake from medical sources
o More than 500 mg of daily calcium intake from medical sources
o Known allergy to milk and milk products
o Known galactosaemia
Uncertainty about the willingness or ability of the subject to comply with the protocol requirements
Participation in any other study involving investigational or marketed products concomitantly or within four weeks prior to
entry into the study
36
37
Appendix II: Bland-Altman plots
600
400
200
0
-200
-400
-600
-800
-1000
1200
1000
800
600
400
200
0
-200
-400
-600
-800
-1000
1200
Bernstein et al. 1983 FFM - REE measured
800
1000
800
600
400
200
+1.96 SD
1,0
Mean
-393,4
1400 1600
1800 2000 2200
REE measured
400
200
0
-200
-400
-600
-800
-1000
1200
Mean
-428,9
-400
1400 1600
2400 2600
2800
-1.96 SD
-785,8
-800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
1000
800
600
+1.96 SD
304,0
400
Mean
-61,5
0
-1.96 SD
-427,0
+1.96 SD
313,6
200
Mean
-64,8
-200
-1.96 SD
-443,2
-400
-600
-800
1800 2000 2200
REE measured
2400 2600
2800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
1000
Harris & Benedict 1984 - REE measured
600
+1.96 SD
-72,0
-600
-1.96 SD
-787,9
1000
800
0
-200
FAO/WHO/UNU 1985 W&H - REE measured
1000
1400 1600
800
600
+1.96 SD
287,8
Mean
-72,1
-1.96 SD
-432,0
400
+1.96 SD
308,9
200
Mean
-52,1
0
-200
-1.96 SD
-413,1
-400
-600
-800
1800 2000 2200
REE measured
2400 2600
2800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
38
1000
800
800
Henry 2005 W&H - REE measured
1000
600
400
200
0
-200
-400
-600
-800
-1000
1200
1400 1600
1800 2000 2200
REE measured
600
+1.96 SD
274,2
Mean
-86,4
-1.96 SD
-447,0
400
200
0
-200
-400
-600
-800
-1000
1200
0
-400
2400 2600
2800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
1400 1600
800
600
+1.96 SD
312,8
400
Mean
-49,7
0
-1.96 SD
-412,3
-600
-800
1800 2000 2200
REE measured
Korth et al. 2007 - REE measured
200
0
-200
-400
-600
-800
1400 1600
2400 2600
2800
-1000
1200
1800 2000 2200
REE measured
+1.96 SD
278,6
400
Mean
-71,1
0
-1.96 SD
-420,8
Lazzer et al. 2007 - REE measured
-200
-400
-600
-800
1400 1600
1800 2000 2200
REE measured
2800
Mean
7,1
-1.96 SD
-365,2
-400
-600
-800
2400 2600
2800
-1000
1200
800
0
2400 2600
+1.96 SD
379,5
-200
1000
200
1800 2000 2200
REE measured
200
800
400
1400 1600
600
1000
600
-1.96 SD
-440,2
-400
800
400
Mean
-91,7
-200
1000
600
+1.96 SD
256,9
200
800
-1000
1200
-1.96 SD
-470,8
-600
1000
-1000
1200
Mean
-105,4
-200
1000
Huang et al. 2004 FFM - REE measured
600
+1.96 SD
260,1
200
-800
1000
800
400
1400 1600
1800 2000 2200
REE measured
600
+1.96 SD
334,2
400
Mean
-46,9
0
-1.96 SD
-428,0
2400 2600
2800
+1.96 SD
437,7
200
Mean
40,6
-200
-1.96 SD
-356,5
-400
-600
-800
2400 2600
2800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
39
1000
800
600
400
200
0
-200
-400
-600
-800
-1000
-1200
-1400
1200
Lazzer et al. 2010 - REE measured
1400 1600
1800 2000 2200
REE measured
1400 1600
+1.96 SD
707,0
Mean
-453,6
-1.96 SD
-1614,2
2400 2600
2800
1000
800
600
400
200
0
-200
-400
-600
-800
-1000
-1200
-1400
-1600
1200
-400
-600
-800
-1000
1200
1400 1600
Mean
-838,7
-1.96 SD
-1207,4
1800 2000 2200
REE measured
2400 2600
2800
1800 2000 2200
REE measured
-800
-1000
1200
Müller et al. 2004 - REE measured
-200
-400
-600
-800
-1000
1200
1400 1600
1800 2000 2200
REE measured
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
600
+1.96 SD
342,2
Mean
-27,2
-1.96 SD
-396,6
400
+1.96 SD
215,7
200
0
Mean
-155,1
-200
-400
-1.96 SD
-525,9
-600
-800
2400 2600
2800
-1000
1200
800
0
-1.96 SD
-541,3
-600
1000
200
Mean
-179,2
-400
800
400
2800
+1.96 SD
182,9
-200
1000
600
2400 2600
0
+1.96 SD
-470,0
Mifflin et al. 1990 - REE measured
-200
1800 2000 2200
REE measured
200
800
0
1400 1600
400
1000
200
-1.96 SD
-1390,6
600
800
400
Mean
-928,7
800
1000
600
+1.96 SD
-466,8
1000
Livingston & Kohlstadt 2005 - REE measured
1000
800
600
400
200
0
-200
-400
-600
-800
-1000
-1200
-1400
-1600
-1800
-2000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
600
400
+1.96 SD
100,3
200
Mean
-267,5
-200
-1.96 SD
-635,3
-600
2400 2600
2800
+1.96 SD
314,8
Mean
-48,7
0
-1.96 SD
-412,1
-400
-800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
40
800
600
400
200
0
-200
-400
-600
-800
-1000
1200
1000
Müller et al. 2004 FFM - REE measured
1000
1400 1600
800
600
+1.96 SD
310,4
Mean
-55,4
-1.96 SD
-421,2
400
200
0
-200
-400
-600
-800
-1000
1200
0
1800 2000 2200
REE measured
2400 2600
2800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
1000
1400 1600
800
600
+1.96 SD
265,8
Mean
-91,6
-1.96 SD
-449,0
400
0
-400
-800
1800 2000 2200
REE measured
Schofield 1985 W - REE measured
200
0
-200
-400
-600
-800
1400 1600
2400 2600
2800
-1000
1200
1800 2000 2200
REE measured
+1.96 SD
80,1
200
Mean
-293,9
-200
-1.96 SD
-667,8
Weijs et al. 2010 - REE measured
-200
-400
-600
-800
1400 1600
1800 2000 2200
REE measured
2800
-600
2400 2600
2800
Mean
-121,9
-400
-1.96 SD
-499,5
-800
-1000
1200
800
0
2400 2600
+1.96 SD
255,7
0
1000
200
1800 2000 2200
REE measured
400
800
400
1400 1600
600
1000
600
-1.96 SD
-497,0
-600
800
400
Mean
-129,4
-200
1000
600
+1.96 SD
238,1
200
800
-1000
1200
-1.96 SD
-428,0
-400
1000
-1000
1200
Mean
-72,7
-200
-800
Owen et al. 1986/1987 - REE measured
600
+1.96 SD
282,6
200
-600
1000
800
400
1400 1600
1800 2000 2200
REE measured
600
+1.96 SD
264,7
Mean
-106,0
-1.96 SD
-476,8
2400 2600
2800
+1.96 SD
484,7
400
200
Mean
106,7
0
-200
-1.96 SD
-271,4
-400
-600
-800
2400 2600
2800
-1000
1200
1400 1600
1800 2000 2200
REE measured
2400 2600
2800
41
42