Weight cycling and women3 Robert
W Jeffery,
Rena
ABSTRACt weight and
R
Wing,
and
This study examined
cycling
and
101 obese 4.5
cardiovascular
women.
by retrospective of
and cardiovascular
kg,
2) total
cycling lost
between
in three
ways
weight
losses
in prior
weight-loss
and
included
and high-density-lipoprotein cholesterol, 2-h glucose values, waist-to-hip ratio,
attempts,
lowest
blood
weight
as an
pressure,
total and
triglycerides, fasting and percent ofdietary
cal-
ories from fat. Results did not support the hypothesis that weight cycling increased cardiovascular risk factors in either men or
women. with P
0f88 associations examined, only 0.05 and of these, 6 were opposite
1 3.6 kg but < 3 1.8 kg overweight, old,
were
nonsmokers,
beverages
per
consumed
day,
foods, were able were not taking
were
not
traceptives, and agreed who expressed interest the
eligibility
in the
ofthe
Table
forms
1 shows
written
TABLE 1 Characteristics
Subjects and met by
the and
characteristics
values
participants Men = 101)
(n
diseases, oral con-
approved
selected
of study
Women (n
101)
=
to any
ofPittsburgh
as mean
AL
alcoholic
or allergic
University
as well
study
three
in the study. in the study
consent
Boards
University of Minnesota. of the study participants included
diet
ET
if they were 25 and 45 y
had no existing serious medications including
signed
Review
than
to participate in participating
criteria
Institutional
fewer
on a special
to exercise, prescription
and
in the study were between
JEFFERY
for
variables
analysis.
37.26
Married (%)
79 97 63 96.25 ± 7.19 30.95 ± 2.24 0.97 ± 0.04
White
(%)
degree (%) (kg)
College Weight
BMIt WHRt Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg)
Total Measures
5.65
Age (y)
±
I 14.94
37.65
5.22
±
62 87 53 6.49 ± 2.09
83.42
±
31.03
0.86±0.06
± 9.71
108.10
±
10.01
±
8.23
69.74
±
7.76
5.06
±
0.93
4.91
±
0.69
1.07
±
0.18
1.32
±
0.25
1.36
±
0.83
0.95
±
0.53
5.7 ± 0.8 5.6 ± 2.0 40.39 ± 7.23 66.63 ± 4.41
5.3
72.84
cholesterol
(mmol/L) cholesterol (mmol/L) Triglyceride (mmol/L) HDL
Three Two
measures
were
ofweight
based
on
cycling
were
five questions
amount of weight losses in previous were asked to indicate the number 4.5-9.0, 9.1-22, were 1-2 times, number the
median
dieting of times
cycles
of
frequency
attempts. that they
frequency
4.5 kg was calculated
of weight
losses
and Subjects had lost
in each
category
times
the
median
weight-loss-
amount
and summing across categories. An index of maximum fluctuation was generated by comparing two questions,
“What
is the most
nancies?”
and
you have
“What
is the
ever least
weighed you
have
not counting
pregas an
(BMI, the two
BMIs
was
fluctuation. Dependent Waist-to-hip
taken
as an index
of maximum
adult
variables in this study consisted of the ratio (WHR) was assessed by measuring
cumference
of the
circumference
waist
of the
at the
hips
umbilicus
at the
and
maximum
weight
following. the cir-
dividing
point
it by the
of protrusion,
University
of Minnesota
number
of national
lesterol,
high-density-lipoprotein
glycerides ysis
was
were
assessed
performed
of Pittsburgh. Disease
Control.
Serum
calories
fat was
from
from
a 12-h
School
laboratory
blood diet (13).
Prineas
studies
(12).
(HDL)
at the The
fasting load. Habitual Questionnaire 12-h
by Ronald
hypertension
glucose
used
fasting
blood
of Public was
weight (Table Because
ratio of total examination
number
sample.
Anal-
Centers from
ofdietany
fat intake.
the
energy intake to lean body mass. showed that the distribution
of weight
cycles
total
and
and women by gender.
weight
±
0.48
8.33
±
4.38
cycle weight 1 Quartile 2 3
Quartile
4
Maximum BMI
0±0 15.18±7.31 42.97 ± 8.26 117.32 ±69.64
1 2
4.83
±
7.08
±
Quartile
3
8.99
±
Quartile
4
12.54
15.15±8.71 33.41 ±0
62.84
±
138.38
±
14.18
[20] [17] [35]
45.53 [29]
±
1.15 0.65 0.60 3.41
[24] [26] [25] [26]
7.17 ± 1.25
[25]
0.45
[25]
1 1.08 ± 0.50 13.51 ± 1.76
[24] [271
9.60
±
SD.
S
mass
f Waist-to-hip § Ratio of II n
index
in kg/rn2.
ratio.
energy
intake
in Id to LBM
in kg.
in brackets.
cycling
were performed
highly
skewed
by
both
variables,
dividing
measures
weight
analyzed
fluctuation
the using
was
normally
as a categorical
included
age and
BMI
analysis
sample
and
of weight
cycling
quartiles
by
into
analysis
variable
tinuous variable. Analysis were linear models statistical program yses
[18] [31] [31] [21]
[13] [25] [37] [26]
(kg)
Quartile
was
3.34±0.51 7.29 ± 1.93 14.03 ± 2.77
minimum
-
Quartile
1.38±0.41
[26] [23] [34]
of variance.
distributed
by quartile
conducted using (SAS, mc, Cary,
and
the
Maxithus
was
and as a conthe SAS general NC).
All anal-
as covariates.
Lean
bioelectrical by Segal et al was computed
in men separately
3.34
4
mum for
of
different conducted
3
Quartile
weight-cycling
University
the
assessed
also
tn-
2 h after a 75-g oral glucose by the Block Food Frequency ofthis analysis, percent of total
as an index
cycling was substantially 1). Thus, analyses were total
in a cho-
and
Health, by
body mass was estimated from a measure impedance according to the method described ( 14). Finally, a measure of energy requirements by taking the Preliminary
used total
cholesterol,
is certified
sample and was assessed For purposes
and Serum
[18111
Quartile
t Body
following the protocol of recommendations of Lohman et al (1 1). Systolic and diastolic blood pressure were measured with standardized methodology according to protocols developed at the
0±0 1.50±0
Quartile
± 0.6 5.5 ± 1.3 38.34 ± 6.60 49.16 ± 3.44 120.00 ± 43.10
± 54.10
2
Quartile
Each ofthese weights was converted to body mass index in kg/rn2) to adjust for height, and the difference between
adult?”
144.01
Total
ever
weighed
(mmol/L) 2-h glucose (mmol/L)
Quartile
in each
category weight
glucose
LBM (kg) Energy:LBM Total cycle number Quartile 1
weight lost in previous the median frequency
the
Fasting
Energy from fat (%)
by summing
across
categories. A measure of total cycles was generated by multiplying
of losses
in the study.
the
23-35, 36-45, and > 45 kg. Response options 3-5 times, 6-9 times, and 10 times. Total
ofweight
amount weight
employed
regarding
of weight
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/3/641/4715165 by guest on 02 May 2018
of
Results The three weight-cycling measures strongly related to each other (chi-square < 0.0001). Moreover, all three measures
used in this 42. 1-374.6,
were significantly
to BMI (P < 0.01). Thus, subsequent analyses justing for BMI. Age was also used as a covariate because
older
individuals
would
have
study df
had
more
were
in these
were
9; P related =
done adanalyses
opportunity
to
WEIGHT TABLE
CYCLING
2
Cardiovascular
risk
factors
by weight-cycling
quartile
(WCQ)5
WCQ1 Males WHR
0.96
Systolic
blood
Diastolic Total
pressure
blood
(mm
pressure
cholesterol
Hg)
(mm
Hg)
glucose
Energy from LBM (kg) Energy:LBM
(mmol/L)
0.98
5.02
± 0.22
1.03 1.28 6. 1
±
72.86
0.04 0.19 0.2
± ±
WCQ3
0.01
0.96
± 1.70
1 16.31
1.37
72.57
±
1 13.49
1.83
±
6.6
fat (%)
0.01
± 2.27
75.81
(mmol/L)
WCQ2
±
1 17.02
Total HDL (mmol/L) Triglyceride (mmol/L)t Fasting glucose (mmol/L) 2-h
643
±
5.26
± 0.17
1.04 1.57 5.6
±
0.03
±
0.15
±
0. 1
± 0.5
WCQ4
±
0.01
0.96
±
1.70
1 13.26
± 1.37
70.66
5.04
± 0.16
4.81
1.09 1.39 5.6
±
5.6 ± 0.4
± ±
5.5
0.03 0.15 0. 1
± 0.4
0.01
0.1679
± 2.09 ±
0.4361
1.68
0.2481
± 0.20
0.3565
1 . 1 3 ± 0.04 1.09 ± 0.18 5.6 ± 0.2
0.0848
4.9
42.72 63.89 142.55
±
1.74
39.72
± 1.30
39.89
± 1.30
40.1
±
1.01
68.30
±
0.75
65.92
± 0.75
67.59
±
12.80
± 9.58
152.42
± 9.58
146.44
0.86 109.63 71.20 5.06
±
0.01
±
2.26
106.27
±
1.70
67.17
±
0.15
4.90
±
1.23 1.04 5.5 5.5 37.60
±
±
0.06 0.1 1 0.14 0.3 1.50
±
0.78
50.57
± 0.85
134.81
±
P
0.2125 0.0824
± 0.4
0.0843
1 ± 1.60
0.5337
± 0.92
0.0047
±
1 1.76
0.6308
Females
WHR Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Total cholesterol (mmol/L) Total
HDL
(mmol/L)
1.26
Triglyceride (mmol/L)t Fasting glucose (mmol/L) 2-h
glucose
Energy from LBM (kg)
0.90 5.2 5.7 37.43 48.82
(mmol/L)
fat (%)
Energy:LBM
SE adjusted for age and BMI, t Analyzed with a log transformation.
P values
cycle
[empirically,
age
was
measures,
dent measures cycling indices Ofthe significant.
total
related
gorically.
pounds
run
to 2-h
Analyzed
are from
(P
glucose
regression
and
results based and are not