The Effect of Protein and Glycemic Index on Children’s
Body Composition: The DiOGenes Randomized Study
WHAT’S KNOWN ON THIS SUBJECT: The potential of HP and LGI
diets to promote weight loss in adults has resulted in research to
investigate the effect of these diets on weight control in children;
results have been promising but still conflicting.
WHAT THIS STUDY ADDS: In the DiOGenes study, a European
family-based, randomized, controlled, multicenter, 6-month
energy ad libitum dietary intervention study, the LP/HGI diet
increased body fat, whereas overweight/obesity decreased in
those on the HP/LGI diet.
abstract
OBJECTIVE: To investigate the effect of protein and glycemic index (GI)
on body composition among European children in the randomized,
6-month dietary intervention DiOGenes (diet, obesity, and genes)
family-based study.
PATIENTS AND METHODS: In the study, 827 children (381 boys and 446
girls), aged 5 to 18 years, completed baseline examinations. Families
with parents who lost ⱖ8% of their weight during an 8-week run-in
low-calorie diet period were randomly assigned to 1 of 5 ad libitum
diets: low protein (LP)/low glycemic index (LGI); LP/high GI (HGI); high
protein (HP)/LGI; HP/HGI; and control diet. The target difference was 15
GI U between the LGI/HGI groups and 13 protein percentage points
between the LP/HP groups. There were 658 children examined after 4
weeks. Advice on food-choice modification was provided at 6 visits
during this period. No advice on weight loss was provided because the
focus of the study was the ability of the diets to affect outcomes
through appetite regulation. Anthropometric measurements and body
composition were assessed at baseline, week 4, and week 26.
RESULTS: In the study, 465 children (58.1%) completed all assess-
ments. The achieved differences between the GI and protein groups
were 2.3 GI U and 4.9 protein percentage points, respectively. The LP/
HGI group increased body fat percentage significantly more than the
other groups (P ⫽ .040; partial
2
⫽ 0.039), and the percentage of
overweight/obese children in the HP/LGI group decreased significantly
during the intervention (P ⫽ .031).
CONCLUSIONS: Neither GI nor protein had an isolated effect on body
composition. However, the LP/HGI combination increased body fat,
whereas the HP/LGI combination was protective against obesity in this
sample of children. Pediatrics 2010;126:e1143–e1152
AUTHORS: Angeliki Papadaki, PhD,
a
Manolis Linardakis,
MSc,
a
Thomas M. Larsen, PhD,
b
Marleen A. van Baak,
PhD,
c
Anna Karin Lindroos, PhD,
d
Andreas F. H. Pfeiffer,
MD,
e,f
J. Alfredo Martinez, PhD,
g
Teodora Handjieva-
Darlenska, MD, PhD,
h
Marie Kunesová, MD, PhD,
i
Claus
Holst, MSci, PhD,
j
Arne Astrup, MD, DSc,
b
Wim H. M. Saris,
MD, PhD,
c
and Anthony Kafatos, MD,
a
on behalf of the
DiOGenes Study Group
a
Department of Social Medicine, Preventive Medicine and
Nutrition Clinic, University of Crete, Heraklion, Crete, Greece;
b
Department of Human Nutrition, Faculty of Life Sciences,
University of Copenhagen, Copenhagen, Denmark;
c
Department
of Human Biology, Nutrition and Toxicology Research Institute
Maastricht, University of Maastricht, Maastricht, Netherlands;
d
MRC Human Nutrition Research, Elsie Widdowson Laboratory,
Cambridge, United Kingdom;
e
Department of Clinical Nutrition,
German Institute of Human Nutrition Potsdam-Rehbruecke,
Nuthetal, Germany;
f
Department of Endocrinology, Diabetes and
Nutrition, Charité Universitätsmedizin Berlin, Berlin, Germany;
g
Department of Physiology and Nutrition, University of Navarra,
Pamplona, Spain;
h
Department of Human Nutrition, Dietetics and
Metabolic Diseases, National Transport Hospital, Sofia, Bulgaria;
i
Obesity Management Centre, Institute of Endocrinology, Prague,
Czech Republic; and
j
Institute of Preventive Medicine, Center for
Health and Society, Copenhagen, Denmark
KEY WORDS
childhood adiposity, dietary intervention, DiOGenes, glycemic
index, protein, randomized controlled trial
ABBREVIATIONS
HP—high protein
GI—glycemic index
LGI—low glycemic index
HGI—high glycemic index
DiOGenes—Diet, Obesity, and Genes
LCD—low-calorie diet
LP—low protein
This trial has been registered at www.clinicaltrials.gov
(identifier NCT00390637).
www.pediatrics.org/cgi/doi/10.1542/peds.2009-3633
doi:10.1542/peds.2009-3633
Accepted for publication Jul 16, 2010
Address correspondence to Angeliki Papadaki, PhD, Preventive
Medicine and Nutrition Clinic, Department of Social Medicine,
Faculty of Medicine, University of Crete, Greece, PO Box 2208,
Heraklion 710 03, Crete, Greece. E-mail: agpapadak@edu.med.
uoc.gr
PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
Copyright © 2010 by the American Academy of Pediatrics
FINANCIAL DISCLOSURE: The authors have indicated they have
no financial relationships relevant to this article to disclose.
ARTICLES
PEDIATRICS Volume 126, Number 5, November 2010 e1143
by guest on March 30, 2013pediatrics.aappublications.orgDownloaded from
Adiposity in childhood is associated
with increased risk of mortality.
1–5
Re-
cent increases in energy intake, pro-
duced by untoward changes in macro-
nutrient composition
4,6
and factors
such as portion size, availability and
cost of food and soft drinks, and reduc-
tion in physical activity, have report-
edly played a major role in the rise of
obesity.
4,7
With high-protein (HP) diets,
weight loss is promoted
8
and weight
regain in adults is prevented,
9
most
probably by increasing satiety.
10
How-
ever, this has not been indicated in
obese children.
11,12
Low-glycemic index
(LGI) diets are also suggested to pro-
mote weight loss in adults, mainly by
promoting satiety and reducing hun-
ger.
13,14
However, the clinical useful-
ness of GI in weight regulation has
been questioned.
15
In children, LGI
meals compared with HGI meals signif-
icantly reduced subsequent energy in-
take and hunger in some,
13,16
but not
all,
17
studies.
The effect of HP, LGI diets in children has
not been assessed in large-scale, ran-
domized controlled studies. We aimed to
investigate the effect of protein and GI on
body weight, anthropometric measure-
ments, and body composition among
free-living children after a 6-month,
family-based dietary intervention. It
should be noted that no advice on weight
loss was provided to children because
the intervention gave less emphasis on
energy intake (which was ad libitum),
but focused on the ability of the diets to
regulate appetite and thus body weight
and composition. This study is part of the
European integrated project DiOGenes
(Diet, Obesity, and Genes), which focused
on dietary means of preventing weight
gain
9,18,19
and was performed in 8 Euro-
pean centers.
PATIENTS AND METHODS
Participants
Volunteer families from 8 countries
(Netherlands, Denmark, United King-
dom, Greece, Germany, Spain, Bulgaria,
and Czech Republic) were invited to par-
ticipate during the period from Novem-
ber 2005 to April 2007. Families attended
a screening examination to determine el-
igibility. Eligible families were generally
healthy, with at least 1 parent over-
weight (BMI ⱖ 27 kg/m
2
) and younger
than 65 years, and at least 1 child be-
tween the age of 5 and 18 years.
9
Recruit-
ment strategies and exclusion criteria
for parents are provided in detail else-
where.
9
Exclusion criteria for children
were special diets, food intolerances,
systemic infections/chronic diseases,
use of medication that might influence
study outcomes, and drug or alcohol
abuse. Informed consent was obtained
from parents and assent was obtained
from the children, according to local leg-
islation. The study was approved by the
local medical ethical committees in the
respective countries.
Study Design and Procedures
A description of the design and study
procedures are provided in an earlier
report.
9
In short, eligible adults under-
went an 8-week low-calorie diet (LCD)
period after their screening, during
which the enrolled children received
no intervention. During the seventh
week of the LCD, children attended a
clinical examination (representing
baseline),
9
when anthropometric pa-
rameters and body composition were
assessed and a 3-day weighed dietary
record was provided. Families with at
least 1 parent who lost ⱖ8% of weight
during the LCD were randomly as-
signed to 1 of 5 ad libitum diets: low
protein (LP)/LGI; LP/HGI; HP/LGI; HP/HGI;
and control diet (national dietary
guidelines, with medium protein con-
tent and no specific instructions on
GI).
19
Eligible families were allocated to
a dietary intervention group by using a
simple block randomization proce-
dure with stratification.
9
Randomization was followed by (1)
in Maastricht and Copenhagen, a
6-month supermarket period (free
food provided to families by laboratory
shops, in addition to dietary instruc-
tions),
20,21
and (2) in remaining cen-
ters, a 6-month dietary instruction-
only period. Families followed their
randomized diet for 6 months. Chil-
dren attended a second clinical exam-
ination 4 weeks into their randomized
diet (same measurements as baseline,
apart from body composition) and a
final examination at the end of the
6-month intervention (week 26, same
measurements as baseline) (Fig 1).
Dietary Instruction Period
On the day families were randomly as-
signed, trained dietitians gave detailed
FIGURE 1
Design of the dietary intervention for children at the supermarket and the instruction centers.
e1144 PAPADAKI et al
by guest on March 30, 2013pediatrics.aappublications.orgDownloaded from
instructions on the ad libitum diets. All
diets were low in fat (25%–30% of en-
ergy). The target was for protein con-
tent to comprise 10% to 15% of energy
intake in the LP and 23% to 28% in the
HP groups, complying with the accept-
able range (10%–30%) for children
aged 4 to 18 years.
22
Children in the
LGI groups were advised to consume
the LGI foods, and those in the HGI
groups to consume the HGI foods
within a food group.
19
The aim was to
achieve a 15-point GI difference be-
tween the HGI and LGI groups. The
assignment of GI values to foods is
reported separately.
23
Children were encouraged to be
present during the randomization vis-
it; otherwise, parents were instructed
to aid their children with their random-
ized diets. During the intervention, chil-
dren were requested to attend 6 coun-
seling sessions, accompanied by their
parents, during which intensive guid-
ance was provided. Dietitians advised
on weight control and reinforced the
diet composition messages
19
through
food-choice and behavior-modification
advice. A detailed description of the di-
ets is provided elsewhere.
19
Anthropometric Measurements
Standard operating procedures were
produced for all investigations under-
taken to ensure standardization
across the centers. Moreover, the
same measurement devices were
used in each center on every occasion
a measurement was provided.
9
Sub-
jects had been fasting for 4 hours be-
fore the clinical examinations, and
they were measured in their under-
wear and with an empty bladder.
Weight was measured on calibrated
digital scales (Seca 861, Hamburg, Ger-
many) to the nearest 0.1 kg, and height
was measured to the nearest millime-
ter with a wall-mounted stadiometer
(Seca 225, Hamburg, Germany), with-
out shoes, on all examinations and di-
etary counseling sessions. BMI was
calculated as weight divided by height,
squared (kg/m
2
), and weight status
was determined according to the age-
and gender-specific International
Obesity Task Force iso-BMI ⱖ 25 lim-
its.
24
Waist circumference (mea-
sured midway between the lower rib
and iliac crest) and hip circumfer-
ence (measured at the widest point
between the hips and buttocks) cir-
cumference were measured twice, to
the nearest 0.5 cm, with a tape in a
vertical plane and with the subject
standing and gently breathing out.
The mean of the 2 measurements
was recorded. Waist-to-hip ratio was
then calculated.
9
Body Composition
Body composition was determined by
dual-energy radiograph absorption
(Lunar Radiation, Madison, WI) or bio-
electric impedance analysis (Quad-
Scan 4000; Bodystat, Douglas, Isle of
Man, United Kingdom). Children who
participated in the determination of
body composition were fasting (no in-
take of foods or liquids for at least 4
hours before the investigations except
for a water intake of 350 –500 mL).
9
Dietary Records
All families were provided with weigh-
ing scales (Soehnle 1208 Actuell Back-
nang, Germany) and were instructed
to record their food and liquid intake
for 3 consecutive days (2 weekdays
and 1 weekend day). Participants were
required to weigh all foods and bever-
ages consumed, as well as leftovers,
and to provide cooking methods and
recipes for composite meals. When
weighing was not possible, partici-
pants were instructed to record their
intake in household measures. Par-
ents were instructed to aid their chil-
dren with the recording process when
appropriate. Dietary records were
checked for completeness on return to
clarify any uncertainties. The method
of analysis of dietary records is re-
ported elsewhere.
9,23
Statistical Analysis
Primary outcome measures were
changes in anthropometric measure-
ments, BMI z score, and body composi-
tion during the intervention. Second-
ary outcomes were changes in the
proportion of overweight and obese
children and changes in waist-to-hip
circumference ratio. Baseline charac-
teristics of participants were com-
pared between the different countries
by using the Kruskal Wallis rank test
and Pearson’s
2
test. Changes in di-
etary intakes during the intervention
were calculated for those participants
who provided food diaries at baseline,
week 4, and week 26, and comparisons
between dietary groups were per-
formed by using the Kruskal Wallis
test. Comparisons of overall changes
in anthropometric measurements dur-
ing the dietary intervention between
dietary groups were performed by us-
ing analysis of covariance, with age,
gender, family structure (single par-
ent, couple with 1 randomized, and
couple with both randomized), and
country as covariates. These analyses
were performed for “completers only”
(all randomized subjects with data
from all clinical examinations), as well
as on an “intention-to-treat” basis, by
using the multiple imputation tech-
nique for all randomly assigned sub-
jects.
25,26
For the completers’ analysis,
the observed power effect size was
72%. In addition, the differences in ef-
fect between the protein groups and
the GI groups and potential interaction
between GI and protein content were
analyzed by analysis of covariance
(same covariates as above). A McNe-
mar test was used to examine changes
in weight status (overweight and
obese children) within the dietary
groups. All analyses were performed
by using SPSS 17.0 (SPSS Inc, Chicago,
ARTICLES
PEDIATRICS Volume 126, Number 5, November 2010 e1145
by guest on March 30, 2013pediatrics.aappublications.orgDownloaded from
IL), and significance was defined as
P ⬍ .05.
RESULTS
A total of 1140 children were regis-
tered to attend screening, and 827 at-
tended the baseline examination. Eight
hundred children were randomly as-
signed, and 465 (201 boys, 264 girls)
attended all clinical examinations
(baseline, week 4, and week 26) and
were included in present analyses. In
total, 335 children (41.9% of those ran-
domly assigned) dropped out or were
excluded during the dietary interven-
tion because of their parents’ with-
drawal, noncompliance with study re-
quirements, or because they did not
attend at least 1 of the above men-
tioned clinical examinations (compl-
eters only) (Fig 2). Baseline charac-
teristics did not differ between
completers and children who did not
provide information at all visits or
dropped out (data not shown).
Baseline characteristics of children
who completed the intervention at the
8 different centers are shown in Table
1. Children differed in various vari-
ables between the different centers.
There were no baseline differences in
anthropometric measurements be-
tween the 5 dietary groups (compl-
eters’ analysis).
Only 113 children (24.3% of those who
attended all clinical examinations) re-
corded their dietary intake at all 3 time
points. At all time points, self-reported
dietary protein content had increased
significantly in the HP groups, result-
ing in a 4.9-percentage point differ-
ence between the HP and LP groups at
week 26 (P ⬍ .001) (Table 2). Between
baseline and 26 weeks, the proportion
of energy from carbohydrates was re-
duced more in the HP group than in the
LP group (⫺2.6 vs 4.8%; P ⬍ .001). GI
was reduced in the 2 LGI groups at
week 26, resulting in a 2-point differ-
ence between the LGI and HGI groups
(P ⫽ .013). Glycemic load was, as ex-
pected, reduced more in the HP group
than in the LP group during the inter-
vention (⫺37.9 vs ⫺12.8 g/day; P ⫽
.006). The proportion of energy from
fat did not differ among groups and
was within the recommended levels of
25% to 30% at 4 and 26 weeks.
Between baseline and 26 weeks, there
was a significant increase among the
overall sample (completers) in body
weight (53.2 vs 54.4 kg; P ⬍ .001),
height (1.53 vs 1.55 m; P ⬍ .001), hip
circumference (85.5 vs 85.9 cm; P ⫽
.041), and fat-free mass (37.7 vs 39.0
kg; P ⬍ .001), whereas waist-to-hip ra-
tio (0.855 vs 0.846; P ⬍ .001) and waist-
to-height ratio (0.477 vs 0.467; P ⬍
.001) were significantly reduced. No
statistically significant differences in
changes in outcome measures among
the dietary groups were observed dur-
ing the intervention, apart from a sta-
tistically significant increase in body
fat percentage in the LP/HGI group (P
⫽ .040; partial
2
⫽ 0.039) and a sta-
tistically significant reduction in the
percentage of overweight/obese chil-
FIGURE 2
Flow diagram in which the progress of the participants from screening to the end of intervention is
shown.
e1146 PAPADAKI et al
by guest on March 30, 2013pediatrics.aappublications.orgDownloaded from
dren in the HP/LGI group (Table 3; Figs
3 and 4). When the main effect of pro-
tein and GI and the interaction be-
tween protein and GI were assessed,
no significant changes in outcome
measures were observed between the
groups. Multiple imputation confirmed
that the LP/HGI group had the highest
increase in body fat percentage: impu-
tation number 4 (P ⫽ .034) and impu-
tation number 5 (P ⫽ .013).
DISCUSSION
Excess adiposity is said to be the most
serious health problem that affects
children in developed countries.
4
The
potential of HP
8
and LGI
14
diets to pro-
mote weight loss in adults has re-
sulted in research that investigates
the effect of these diets on weight con-
trol in children, with promising but still
conflicting results. The DiOGenes study
has several strengths compared with
earlier studies because it involved a
prospective large-scale, randomized
controlled design. With the interven-
tion we examined the effectiveness of
the diet (application in a real-life set-
ting), and the family-based plan en-
couraged support between family
members that could positively affect
treatment in children.
4,9
Our most significant finding, that the
combination of LP and HGI increased
body fat percentage, may suggest that
this dietary combination does not af-
fect body composition favorably. LP di-
ets hinder satiety and energy expendi-
ture and inhibit fat oxidation by
increasing insulin response.
10
In addi-
tion, carbohydrates with HGI are rap-
idly digested, thereby increasing insu-
lin levels, stimulating hunger, and
promoting food intake and the accu-
mulation of body fat.
14,27–29
These mech-
anisms may explain the increase in
body fat observed in the LP/HGI group.
In contrast, the percentage of over-
weight/obese children in the HP/LGI
group decreased significantly during
TABLE 1 Baseline (First Examination) Characteristics of Study Participants Who Completed the Intervention (N ⫽ 465) at the 8 Centers
Netherlands Denmark United Kingdom Greece Germany Spain Bulgaria Czech Total
Boys
Age, mean ⫾ SD (n), y 10.8 ⫾ 3.1 (44) 11.5 ⫾ 3.5 (53) 12.3 ⫾ 2.8 (26) 11.9 ⫾ 3.7 (12) 12.6 ⫾ 3.4 (17) 12.4 ⫾ 3.6 (32) 13.9 ⫾ 3.4 (13) 11.7 ⫾ 2.4 (4) 11.9 ⫾ 3.4 (201)
Weight, mean ⫾ SD (n), kg 46.8 ⫾ 20.9 (44) 52.8 ⫾ 25.3 (53) 53.2 ⫾ 20.3 (26) 69.6 ⫾ 29.4 (12) 54.7 ⫾ 20.2 (17) 51.5 ⫾ 18.6 (32) 71.0 ⫾ 25.8 (13) 58.6 ⫾ 18.7 (4) 53.8 ⫾ 23.2 (201)
a
Height, mean ⫾ SD (n), m 1.48 ⫾ .19 (44) 1.52 ⫾ .22 (53) 1.56 ⫾ .16 (26) 1.56 ⫾ .22 (12) 1.59 ⫾ .19 (17) 1.53 ⫾ .17 (32) 1.64 ⫾ .14 (13) 1.57 ⫾ .19 (4) 1.53 ⫾ 0.19 (201)
BMI, mean ⫾ SD (n) 20.1 ⫾ 4.5 (44) 21.5 ⫾ 5.2 (53) 21.3 ⫾ 4.8 (25) 27.0 ⫾ 5.1 (12) 20.9 ⫾ 5.0 (17) 21.3 ⫾ 4.4 (32) 25.7 ⫾ 5.8 (13) 23.2 ⫾ 2.6 (4) 21.8 ⫾ 5.1 (200)
b
Weight status, OW/OB, % 13.6/18.2 32.1/17.0 33.3/8.3 25.0/66.7 23.5/11.8 34.5/10.3 38.5/38.5 50.0/25.0 28.1/19.4
b
WC, mean ⫾ SD (n), cm 70.8 ⫾ 15.6 (43) 72.9 ⫾ 15.1 (53) 74.9 ⫾ 14.3 (26) 86.8 ⫾ 14.9 (12) 72.3 ⫾ 13.7 (17) 72.2 ⫾ 12.4 (32) 76.3 ⫾ 11.4 (4) 77.6 ⫾ 5.1 (4) 73.6 ⫾ 14.6 (191)
HC, mean ⫾ SD (n), cm 73.8 ⫾ 15.4 (43) 84.3 ⫾ 17.0 (53) 88.1 ⫾ 12.8 (24) 99.9 ⫾ 16.9 (12) 85.2 ⫾ 13.9 (17) 85.7 ⫾ 12.7 (32) 87.3 ⫾ 11.9 (4) 90.8 ⫾ 9.3 (4) 83.9 ⫾ 16.1 (189)
b
Waist/hip ratio, mean ⫾ SD (n) 0.960 ⫾ 0.053 (43) 0.866 ⫾ 0.047 (53) 0.846 ⫾ 0.063 (24) 0.870 ⫾ 0.064 (12) 0.848 ⫾ 0.053 (17) 0.842 ⫾ .053 (32) 0.874 ⫾ 0.064 (4) 0.857 ⫾ 0.036 (4) 0.879 ⫾ 0.069 (189)
b
Waist/height ratio, mean ⫾ SD (n) 0.478 ⫾ 0.067 (43) 0.480 ⫾ 0.061 (53) 0.479 ⫾ 0.065 (25) 0.557 ⫾ 0.070 (12) 0.456 ⫾ 0.066 (17) 0.473 ⫾ .065 (32) 0.499 ⫾ 0.088 (4) 0.497 ⫾ 0.031 (4) 0.482 ⫾ 0.067 (190)
FFM, mean ⫾ SD (n), kg 35.5 ⫾ 16.6 (23) 40.2 ⫾ 17.5 (29) 40.7 ⫾ 11.5 (19) 49.5 ⫾ 20.1 (11) 42.7 ⫾ 16.2 (16) 37.9 ⫾ 13.8 (14) — 47.2 ⫾ 15.2 (4) 40.5 ⫾ 16.1 (116)
FM, mean ⫾ SD (n), kg 10.8 ⫾ 8.0 (23) 19.9 ⫾ 10.8 (29) 13.4 ⫾ 7.6 (19) 22.9 ⫾ 11.3 (11) 11.2 ⫾ 7.2 (16) 15.5 ⫾ 7.2 (14) — 11.5 ⫾ 4.0 (4) 15.3 ⫾ 9.6 (116)
b
Fat, mean ⫾ SD (n), % 20.9 ⫾ 10.0 (42) 29.4 ⫾ 9.5 (26) 37.0 ⫾ 14.1 (12) 31.7 ⫾ 8.0 (11) 19.8 ⫾ 9.2 (16) 28.5 ⫾ 6.4 (14) — 19.7 ⫾ 3.5 (4) 25.8 ⫾ 11.8 (125)
b
Total energy intake, mean ⫾ SD
(n), kJ/d
7717 ⫾ 2664 (37) 8541 ⫾ 2392 (51) 8450 ⫾ 3095 (19) 8421 ⫾ 2462 (12) 8982 ⫾ 2580 (12) 7499 ⫾ 2471 (23) — 7223 ⫾ 1317 (4) 8176 ⫾ 2564 (158)
Girls
Age, mean ⫾ SD (n), y 12.3 ⫾ 3.2 (63) 11.8 ⫾ 3.5 (73) 13.2 ⫾ 3.0 (26) 11.9 ⫾ 3.6 (7) 12.5 ⫾ 3.8 (16) 11.7 ⫾ 3.8 (29) 14.1 ⫾ 3.3 (28) 12.9 ⫾ 3.9 (22) 12.4 ⫾ 3.5 (264)
Weight, mean ⫾ SD (n), kg 51.3 ⫾ 20.0 (63) 50.7 ⫾ 19.4 (73) 51.7 ⫾ 16.9 (26) 61.3 ⫾ 24.4 (7) 53.6 ⫾ 22.9 (16) 49.9 ⫾ 18.9 (29) 59.9 ⫾ 19.2 (28) 57.2 ⫾ 21.1 (22) 52.8 ⫾ 19.8 (264)
Height, mean ⫾ SD (n), m 1.52 ⫾ .15 (62) 1.51 ⫾ .18 (73) 1.56 ⫾ .13 (26) 1.51 ⫾ .15 (7) 1.53 ⫾ .18 (16) 1.49 ⫾ .19 (29) 1.57 ⫾ .14 (28) 1.54 ⫾ .16 (22) 1.52 ⫾ 0.16 (263)
BMI, mean ⫾ SD (n) 21.6 ⫾ 5.2 (62) 21.4 ⫾ 4.5 (73) 20.6 ⫾ 5.7 (24) 25.7 ⫾ 6.5 (7) 21.7 ⫾ 5.6 (16) 21.5 ⫾ 4.1 (29) 23.5 ⫾ 4.7 (28) 23.4 ⫾ 5.4 (22) 21.9 ⫾ 5.0 (261)
Weight status, OW/OB, % 25.4/15.9 38.6/14.3 23.1/7.7 14.3/57.1 18.8/18.8 35.0/5.0 42.9/10.7 68.2/9.1 34.5/13.9
WC, mean ⫾ SD (n), cm 72.3 ⫾ 12.8 (61) 70.5 ⫾ 12.2 (73) 71.3 ⫾ 12.1 (25) 85.0 ⫾ 15.3 (7) 69.4 ⫾ 12.5 (15) 71.1 ⫾ 10.4 (28) 78.9 ⫾ 20.1 (9) 73.8 ⫾ 11.5 (22) 72.1 ⫾ 12.7 (240)
HC, mean ⫾ SD (n), cm 80.8 ⫾ 15.1 (61) 86.3 ⫾ 15.4 (73) 91.4 ⫾ 14.1 (25) 100.2 ⫾ 19.8 (7) 84.7 ⫾ 14.9 (15) 87.9 ⫾ 14.8 (28) 93.9 ⫾ 18.7 (9) 91.5 ⫾ 14.7 (22) 86.7 ⫾ 15.7 (240)
b
Waist/hip ratio, mean ⫾ SD (n) 0.898 ⫾ 0.055 (61) 0.821 ⫾ 0.070 (73) 0.777 ⫾ 0.050 (24) 0.853 ⫾ 0.069 (7) 0.821 ⫾ 0.056 (15) 0.816 ⫾ 0.076 (28) 0.834 ⫾ 0.071 (9) 0.809 ⫾ 0.048 (22) 0.836 ⫾ 0.073 (239)
b
Waist/height ratio, mean ⫾ SD (n) 0.477 ⫾ 0.062 (61) 0.468 ⫾ 0.058 (73) 0.451 ⫾ 0.070 (23) 0.559 ⫾ 0.072 (7) 0.457 ⫾ 0.055 (15) 0.477 ⫾ 0.051 (28) 0.497 ⫾ 0.088 (9) 0.480 ⫾ 0.049 (22) 0.474 ⫾ 0.062 (238)
a
FFM, mean ⫾ SD (n), kg 37.1 ⫾ 12.0 (46) 34.0 ⫾ 9.9 (47) 35.5 ⫾ 7.0 (22) 34.5 ⫾ 11.5 (6) 36.8 ⫾ 13.7 (16) 32.4 ⫾ 14.3 (21) — 41.0 ⫾ 13.4 (19) 35.8 ⫾ 11.7 (177)
FM, mean ⫾ SD (n), kg 15.2 ⫾ 7.8 (46) 18.6 ⫾ 11.5 (47) 17.0 ⫾ 10.0 (22) 20.5 ⫾ 9.2 (6) 17.1 ⫾ 10.2 (16) 15.7 ⫾ 7.3 (21) — 17.0 ⫾ 10.1 (19) 16.9 ⫾ 9.6 (177)
Fat, mean ⫾ SD (n), % 26.8 ⫾ 8.6 (61) 32.4 ⫾ 10.1 (48) 32.2 ⫾ 6.9 (19) 36.4 ⫾ 7.1 (6) 29.9 ⫾ 9.1 (16) 31.9 ⫾ 7.5 (21) — 28.3 ⫾ 7.2 (19) 30.0 ⫾ 8.9 (190)
b
Total energy intake, mean ⫾ SD
(n), kJ/d
7782 ⫾ 1752 (53) 7827 ⫾ 2536 (62) 7939 ⫾ 2167 (18) 9032 ⫾ 661 (7) 7737 ⫾ 2414 (15) 6535 ⫾ 2303 (22) — 7981 ⫾ 2051 (16) 7727 ⫾ 2208 (193)
OW indicates overweight; OB, obese; WC, waist circumference; HC, hip circumference; FFM, fat free mass; FM, fat mass. Age and gender were recorded at screening; the other variables are from baseline. The distributions of weight status at the different
centers were compared by using the Pearson
2
test. The other variables were compared (between centers) by using the Kruskal Wallis rank test:
a
P ⬍ .05;
b
P ⬍ .01.
ARTICLES
PEDIATRICS Volume 126, Number 5, November 2010 e1147
by guest on March 30, 2013pediatrics.aappublications.orgDownloaded from
