American Journal of Epidemiology
ª The Author 2012. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of
Public Health. All rights reserved. For permissions, please e-mail: [email protected].
Vol. 175, No. 8
DOI: 10.1093/aje/kwr378
Advance Access publication:
January 27, 2012
Original Contribution
Weight Cycling and Mortality in a Large Prospective US Study
Victoria L. Stevens*, Eric J. Jacobs, Juzhong Sun, Alpa V. Patel, Marjorie L. McCullough,
Lauren R. Teras, and Susan M. Gapstur
* Correspondence to Dr. Victoria L. Stevens, Epidemiology Research Program, American Cancer Society, 250 Williams Street NW,
Atlanta, GA 30303 (e-mail: [email protected]).
Initially submitted June 28, 2011; accepted for publication September 27, 2011.
Weight cycling has been associated with an increased risk of death in some studies, but few studies differentiated weight cycling initiated by intentional weight loss from that initiated by illness. The association of weight
cycling with death was examined among 55,983 men and 66,655 women in the Cancer Prevention Study II Nutrition
Cohort from 1992 to 2008. A weight cycle was defined as an intentional loss of 10 or more pounds (4.5 kg)
followed by regain of that weight, and the lifetime number of weight cycles was reported on a questionnaire
administered at enrollment in 1992. A total of 15,138 men and 10,087 women died during follow-up, which ended
in 2008. Hazard ratios and 95% confidence intervals were estimated using Cox proportional hazards regression
models. When the models were adjusted for age only, weight cycling was positively associated with mortality (P for
trend < 0.0001). However, after adjustment for body mass index and other risk factors, low numbers of weight
cycles (1–4 cycles) were associated with slightly lower mortality rates (hazard ratio (HR) ¼ 0.93, 95% confidence
interval (CI): 0.89, 0.97 in men and HR ¼ 0.93, 95% CI: 0.89, 0.98 in women), whereas high numbers of weight
cycles (20 cycles) were not associated with mortality (HR ¼ 1.03, 95% CI: 0.89, 1.19 in men and HR ¼ 0.99, 95%
CI: 0.88, 1.12 in women). These results do not support an increased risk of mortality associated with weight cycling.
body weight changes; mortality; obesity
Abbreviations: BMI, body mass index; CPS-II, Cancer Prevention Study II.
Obesity is a serious public health problem in the United
States and elsewhere. Approximately two-thirds of American
adults have high body mass indexes (weight (kg)/height (m)2;
BMI) and are classified as overweight (BMI of 25–29.9) or
obese (BMI 30) (1). A high percentage of US adults (55%
of women and 39% of men) are currently trying to lose
weight (2). However, most people who lose weight later
regain it (3). Repeated cycles of weight loss and regain are
referred to as weight cycling. The prevalence of weight
cycling, for which there is no standardized definition, has
been reported to be approximately 18%–34% in men (4, 5)
and 20%–55% in women (4, 6).
Because of the high prevalence of weight cycling, it is
important to understand whether it results in adverse health
consequences. Some of the potentially adverse effects of
weight cycling that might influence mortality rates include
possible redistribution of body fat from peripheral to central
locations and slowing of the basal metabolic rate, which
would make future weight loss harder; however, the scientific evidence supporting these effects is minimal (7). In
several studies, investigators reported that weight cycling
is associated with an increased risk of all-cause mortality
(8–13), whereas others found no association (14–17). All
except one study (16) defined weight cycling based on patterns of body weight either measured by study investigators
or self-reported at different points of time without taking into
account whether the weight loss was intentional. In contrast,
in the study by Field et al. (16) in which investigators found
no association, history of weight cycling was based on selfreported intentional weight loss of specified amounts. It is
critical to differentiate between intentional and unintentional
weight loss because unintentional weight loss is often due to
comorbid conditions that in turn may be strongly associated
with later death. Others examined the associations between
785
Am J Epidemiol. 2012;175(8):785–792
786 Stevens et al.
weight cycling and the risk of conditions that might influence
mortality, including myocardial infarction (18), hypertension
(19, 20), stroke (18), diabetes (18, 21), bone fractures (18, 22),
gallstones (5), and various cancers (23–25). In these studies,
different criteria to define weight cycling were used, and the
results often conflicted.
Understanding the effects of weight cycling has important
public health implications. Over half of the US population is
overweight or obese, and most weight loss is not maintained. Therefore, the prevalence of weight cycling is likely
to increase. If weight cycling does increase risk of death,
greater emphasis should be placed on prevention of weight
gain and maintenance of weight loss.
To specifically address the question of whether weight
cycling resulting from repeated periods of intentional weight
loss followed by regain is associated with increased mortality
rates, we conducted analyses using the large American Cancer Society Cohort, the Cancer Prevention Study II (CPS-II)
Nutrition Cohort from 1992 to 2008. Analyses were adjusted
for measures of weight history to determine the association
between weight cycling and mortality independent of these
factors.
MATERIALS AND METHODS
Study population
Subjects for this study (86,402 men and 97,786 women)
were participants in the Nutrition Cohort of the CPS-II. The
Nutrition Cohort, which has been described in detail elsewhere (26), was initiated in 1992 as a subgroup of the CPS-II,
a prospective study of cancer mortality involving approximately 1.2 million Americans that was begun by the American Cancer Society in 1982. Participants in the Nutrition
Cohort were recruited from a pool of CPS-II members who
resided in 21 states and were between 50 and 74 years of age.
At enrollment in 1992/1993, participants completed a selfadministered questionnaire that included questions about
demographic, anthropometric, medical, and lifestyle information. Usual dietary intake over the past year was assessed
using a 68-item food frequency questionnaire developed by
Block (27, 28). All aspects of the CPS-II Nutrition Cohort
study were approved by the Emory University Institutional
Review Board (Atlanta, Georgia).
Deaths were ascertained by biennial automated linkage
to the National Death Index (29). Multiple cause-of-death
codes were obtained for 99.3% of known deaths in CPS-II.
Follow-up ended at the date of death or December 31, 2008,
whichever came first.
Exclusion criteria for this analysis were as follows: report
of prevalent cancer in 1992 (10,129 men and 13,501 women);
a history of heart attack, stroke, emphysema, or other lung
disease in 1992 (14,777 men and 8,338 women); missing
information on BMI in 1982 or 1992 or a BMI <18.5 or >50
(1,726 men and 3,917 women); missing information on the
number of times a participant purposely lost 10 pounds or
more or the number of times a participant regained weight
that was purposefully lost (926 men and 1,774 women);
a report of never having purposefully lost 10 pounds or of
10 or more pounds being the most weight ever purposefully
lost (908 men and 871 women); and report of a number of
times that weight was purposely lost that differed from
the number of times weight was regained by more than
2 (1,953 men and 2,730 women). After these exclusions,
a final cohort of 55,983 men and 66,655 women were
available for this analysis.
Identification of participants who experienced weight
cycling
We determined whether individual CPS-II Nutrition
Cohort participants experienced weight cycling based on
their responses to 2 questions about intentional weight loss
and regain on the 1992 baseline questionnaire. The first
question asked, ‘‘How many times in your life have you
purposefully lost 10 pounds or more?’’ The second question
asked, ‘‘How many time in your life have you regained as
much as 10 pounds that you previously had lost?’’ A write-in
answer was required, with spaces for 2 digits provided.
Thus, answers could range from 0 to 99. A weight cycle
included both a purposeful loss and a regain, and cohort
participants were classified by the number of weight cycles
they reported. These criteria resulted in the identification of
32,451 men and 28,823 women who had not experienced
weight cycling (hereafter referred to as noncyclers) and
23,532 men and 37,832 women who had (hereafter referred to as weight cyclers). The total number of weight
cycles reported was used to further classify the weight
cyclers into 4 groups of 1–4, 5–9, 10–19, and 20 or more
weight cycles.
Statistical analyses
BMIs of participants in 1982 and 1992 were calculated
using weights reported at each of those times, whereas BMIs
at 18 years of age were calculated using recalled weights
reported in 1992. Weight change between age 18 years of
age and 1982 was calculated as the difference between the
recalled weight at age 18 years and the reported weight in
1982. Weight change between 1982 and 1992 was calculated
using the weights reported at each of these times. The validity
of recalled weights at either 18 years of age by middle-aged
adults (30) or at 20 years of age by elderly subjects (31) has
been shown to be very good.
We used Cox proportional hazards regression analysis
(32) to calculate hazard ratios and corresponding 95% confidence intervals for the association between weight cycling
and mortality. P values for linear trend were estimated by
modeling the number of weight cycles as a categorical
variable, with the median value determined for each category. The Cox proportional hazards assumption was tested
by modeling multiplicative interaction terms between weight
cycling categories and time. The statistical significance of
the interaction terms was assessed using the likelihood ratio
test (33). No violation of the proportional hazards assumption was found.
All Cox models were stratified on the exact year of age of
the CPS-II Nutrition Cohort participants in 1992. Additional
covariates included in the multivariate-adjusted models were
race (white, black, or other/missing); alcohol consumption (0,
Am J Epidemiol. 2012;175(8):785–792
Weight Cycling and Mortality
<1 drink/week, 1–6 drinks/week, 1 drink/day, 2 drinks/day,
or missing); current smoking status (never, former, current,
ever, or missing); for current smokers, number of cigarettes
smoked per day (10, 11–20, 21–30, or >30); for former
smokers, combinations of years since quitting (<9, 10–19,
or 20) and number of cigarettes formerly smoked per day
(20 or >20); educational level (some high school, high
school graduate, some college, college graduate, or missing);
BMI in 1982 (18.5–22.4, 22.5–24.9, 25–27.4, 27.5–29.9,
30–34.9, 35–50, or missing); weight change from 18 years
of age to 1982 (lost >5 pounds, lost 5 pounds to gained
5 pounds, gained 5.1–20 pounds, gained 20.1–40 pounds,
gained 40.1–60 pounds, gained >60 pounds, or missing);
history of high blood pressure (yes or no); history of diabetes (yes or no); total energy intake in gender-specific
quintiles; and physical activity in metabolic equivalents
(0, 0.1–6.9, 7–17.4, 17.5–24.4, 24.5–31.4, 31.5 hour/week,
or missing). Physical activity was assessed on the 1992
enrollment questionnaire by asking about the average time
per week during the past year spent performing any of several activities. Summary metabolic equivalent hours/week
were calculated by multiplying the lowest number of hours
in the response by the metabolic equivalent score for each
activity according to the Compendium of Physical Activities
(34) to provide conservatively estimated summary measures.
Usual energy intake was estimated at baseline from individual responses the 68-item Block food frequency questionnaire. Energy intake from each food was estimated
using the Diet Analysis System, version 3.8a (28); usual
energy contributions from all individual foods were summed
to provide overall average intakes.
BMI in 1982 rather than BMI in 1992 was included in the
model because the association between BMI and mortality
differs substantially by age (35), and BMI in middle age is
more strongly associated with mortality than is BMI later in
life (36). (In 1982, the average age of the CPS-II Nutrition
Cohort participants was 54 years for men and 52 years for
women.) In addition, adjustment for BMI in 1982 had
a larger effect on the hazard ratio than did adjustment for
BMI in 1992.
RESULTS
Weight cycling was a prevalent behavior among CPS-II
Nutrition Cohort participants. A total of 42% of men and
56.8% of women reported intentionally losing and regaining
at least 10 pounds 1 or more times in their lifetime. Although
the majority of weight cyclers reported between 1 and 4 weight
cycles (30.6% of men and 36.7% of women), some men
(1.3%) and women (3.4%) reported weight cycling 20 times
or more.
Age-adjusted frequencies of baseline characteristics of the
CPS-II Nutrition Cohort participants by number of weight
cycles are shown in Table 1. Compared with noncyclers,
both male and female weight cyclers were more likely to
be younger, to be former smokers, and to have a history of
diabetes or high blood pressure. Weight cyclers were also
less likely to be current or never smokers and were more
likely to have a higher BMI at 18 years of age, in 1982, and
Am J Epidemiol. 2012;175(8):785–792
787
in 1992 and to have gained more weight between 18 years of
age and 1992 and between 1982 and 1992. Female weight
cyclers were also more likely to have a higher energy intake
(kcal/day) and to drink less alcohol, whereas male weight
cyclers exercised slightly more. For most of these characteristics, the difference between noncyclers and weight cyclers increased with increasing numbers of weight cycles.
The associations between weight cycling and mortality
are shown in Table 2. In age-adjusted models, there was
a dose-related statistically significant positive association
between the number of weight cycles and the risk of allcause mortality in men and in women. These associations
were attenuated but remained statistically significant after
adjustment for typical mortality risk factors, such as current
smoking. However, after additional adjustment for BMI in
1982 and weight change between 18 years of age and 1982,
low levels of weight cycling (1–4 or 5–9 weight cycles in
men and 1–4 weight cycles in women) were statistically
significantly inversely associated with all-cause mortality.
All other levels of weight cycling were not associated with
all-cause mortality rates in either men or women. In the fully
adjusted models, weight cycling at any level except 1–4 times
was not associated with death from cardiovascular disease
in either men or women. In men, weight cycling 5–9 times
was inversely associated with death from cancer and weight
cycling 1–4 times or 5–9 times was inversely associated with
death from other causes. No other weight cycling categories
in men and no categories in women were associated with
death from cancer or other causes.
To determine whether the association between weight
cycling and all-cause mortality varied across categories of
BMI in middle age, analyses were stratified by participants’
BMIs in 1982. These results are shown in Figure 1. Weight
cycling at any level was not associated with mortality in
normal-weight (BMI 18.5–24.9) or obese (BMI 30) men.
Among men who were overweight (BMI 25–29.9) during
middle age, cycling 1–4 and 5–9 times was associated with
a statistically significant lower risk of death (7% and 12%,
respectively). No other levels of weight cycling in overweight men were associated with all-cause mortality. Although no increased risk of mortality was found for women
in any weight category, weight cycling for 1–4 or 10–19
cycles by overweight women was associated with a statistically significant decreased risk of mortality (12% and 21%,
respectively), but no trend was observed. Results were also
examined by weight change from 18 years of age to 1982
(middle age). No significant associations were found for
either men or women who maintained their weight (lost
5 pounds to gained 5 pounds) or who gained any amount of
weight these analyses (data not shown).
Smoking influences both body weight (37) and mortality
risk (38), and adjusting for it in the statistical model may not
adequately control for its effect on the association between
weight cycling and mortality. Therefore, we performed
a sensitivity analysis that was restricted to never smokers.
The associations were null and unchanged (data not shown).
Limiting the analysis to men and women who were 70 years
of age or younger at baseline in 1992 also did not change
the association between weight cycling and all-cause
mortality.
Number of Weight Cycles
Men
Variable
0
(n 5 32,451)
%
Age in 1992, years
Mean
1–4
(n 5 17,121)
%
64.3
Mean
Women
5–9
(n 5 3,520)
%
62.8
Mean
10–19
(n 5 2,154)
%
61.7
Mean
0
(n 5 28,823)
‡20
(n 5 737)
%
61.5
Mean
%
60.8
Mean
1–4
(n 5 24,445)
%
62.9
Mean
5–9
(n 5 6,414)
%
61.5
Mean
10–19
(n 5 4,696)
%
60.5
Mean
‡20
(n 5 2,277)
%
60.1
Mean
59.5
Race
White
97.1
97.8
98.3
98.2
97.7
97.1
97.6
97.9
98.1
Black
1.3
1.1
1.0
0.8
1.5
1.3
1.4
1.4
1.4
98.2
1.0
Other/missing
1.6
1.1
0.7
1.0
0.6
1.6
1.0
0.7
0.5
0.8
Educational level
Some high school
8.0
6.4
5.6
4.8
6.1
4.8
4.6
4.7
5.1
5.7
High school graduate
20.1
16.9
15.2
15.7
11.9
32.0
32.5
32.5
30.8
28.3
Some college
24.9
25.4
24.7
28.3
29.9
29.5
31.5
32.0
33.8
38.1
College graduate or more
46.4
50.6
53.9
50.4
51.8
33.1
30.7
30.3
29.7
27.0
0.6
0.7
0.6
0.8
0.3
0.6
0.7
0.5
0.6
0.9
Never
36.9
31.5
31.8
30.7
33.0
57.3
55.3
52.9
51.8
48.2
Current
10.7
6.5
5.7
5.9
5.8
9.6
7.4
6.9
6.7
6.8
Former
51.5
61.2
61.6
62.4
60.6
31.6
35.9
38.8
40.3
43.3
Missing
Smoking status
Missing
History of diabetes
History of hypertension
0.9
0.8
0.9
1.0
0.6
1.5
1.4
1.4
1.2
1.7
5.1
9.4
11.4
12.1
14.7
2.9
5.3
7.2
8.8
9.9
28.2
40.0
44.0
46.6
48.4
23.6
33.1
39.0
41.4
42.0
Postmenopausal hormone
use in 1992
Am J Epidemiol. 2012;175(8):785–792
Never
43.1
43.3
43.2
43.0
43.0
Current
33.8
31.8
30.6
32.0
29.2
Former
17.3
19.1
20.6
19.6
21.1
5.8
5.8
5.6
5.4
Ever/unknown
Alcohol consumption, g/day
Energy intake, kcal/day
11.6
1,828
11.2
1,786
11.1
1,787
11.0
1,807
11.9
1,866
5.4
1,354
4.4
1,357
4.0
1,386
6.7
3.8
1,385
3.6
1,449
BMIb at 18 years of age
21.3
22.4
23.3
23.7
24.6
19.9
20.9
21.7
22.2
22.8
BMI in 1982
25.0
27.1
28.2
28.8
29.9
22.4
25.0
26.8
27.8
28.9
BMI in 1992
25.2
27.7
29.0
29.9
30.9
23.3
26.5
28.6
29.7
31.1
Exercise, metabolic
equivalent hours/week
13.3
13.1
13.5
13.9
13.5
12.3
11.8
11.6
11.6
12.3
Weight change from
18 years of age to 1982,
pounds
25.9
33.4
34.6
36.3
37.6
15.1
24.4
30.7
32.9
35.7
Weight change from 1982 to
1992, pounds
1.8
4.2
5.9
8.0
7.3
5.1
8.5
10.4
11.7
13.1
Abbreviation: BMI, body mass index.
a
Standardized to the age distribution of men or women in the cohort, respectively.
b
Weight (kg)/height (m)2.
788 Stevens et al.
Table 1. Age-Standardizeda Means and Frequencies of Selected Characteristics of Men and Women, by Weight Cycling Status, Cancer Prevention Study II Nutrition Cohort, 1992
Am J Epidemiol. 2012;175(8):785–792
Table 2. Relative Risk of Death in Men and Women by Number of Weight Cycles in the Cancer Prevention Study II Nutrition Cohort, 1992–2008
Men
Cause of Death and
No. of Weight Cycles
No. of
Deaths
RRa
0
9,272
1–4
4,336
5–9
10–19
20
Women
95% CI
RRb
1.00
Referent
1.00
Referent
1.00
Referent
4,612
1.00
Referent
1.00
Referent
1.00
Referent
1.03
0.99, 1.07
0.99
0.95, 1.02
0.93
0.89, 0.97
3,547
1.05
1.00, 1.09
0.99
0.94, 1.03
0.93
0.89, 0.98
802
1.03
0.96, 1.11
0.97
0.90, 1.05
0.88
0.81, 0.94
925
1.17
1.09, 1.26
1.06
0.99, 1.14
0.95
0.88, 1.02
534
1.17
1.07, 1.28
1.09
1.00, 1.19
0.96
0.88, 1.05
661
1.22
1.12, 1.32
1.07
0.99, 1.17
0.93
0.85, 1.01
194
1.35
1.17, 1.56
1.22
1.06, 1.41
1.03
0.89, 1.19
342
1.41
1.26, 1.58
1.19
1.06, 1.33
0.99
0.88, 1.12
95% CI
RRc
No. of
Deaths
95% CI
RRa
95% CI
RR
95% CI
RRc
95% CI
All causes
P for trend
0.007
<0.0001
0.30
0.0003
<0.0001
0.65
Cardiovascular disease
0
3,165
1.00
Referent
1.00
Referent
1.00
Referent
1,394
1.00
Referent
1.00
Referent
1.00
Referent
1–4
1,556
1.12
1.05, 1.19
1.03
0.97, 1.10
0.93
0.87, 0.99
1,067
1.11
1.02, 1.20
0.99
0.91, 1.07
0.91
0.83, 0.99
5–9
310
1.24
1.10, 1.40
1.10
0.98, 1.24
0.93
0.82, 1.05
275
1.27
1.12, 1.45
1.06
0.93, 1.21
0.91
0.79, 1.04
10–19
197
1.34
1.16, 1.55
1.18
1.02, 1.37
0.97
0.83, 1.13
205
1.42
1.22, 1.64
1.14
0.98, 1.32
0.93
0.80, 1.09
1.76
1.41, 2.20
1.48
1.19, 1.86
1.14
0.91, 1.44
114
1.83
1.51, 2.22
1.40
1.16, 1.71
1.11
0.91, 1.36
20
80
P for trend
<0.0001
0.70
<0.0001
0.0002
<0.0001
0.47
Cancer
0
3,011
1.00
Referent
1.00
Referent
1.00
Referent
1,660
1.00
Referent
1.00
Referent
1.00
Referent
1–4
1,446
0.99
0.93, 1.05
1.01
0.95, 1.08
0.97
0.90, 1.03
1,310
1.00
0.93, 1.07
0.98
0.91, 1.06
0.93
0.86, 1.01
5–9
245
0.88
0.77, 1.00
0.90
0.79, 1.02
0.83
0.72, 0.95
340
1.05
0.94, 1.19
1.03
0.91, 1.16
0.94
0.83, 1.07
10–19
168
1.01
0.86, 1.18
1.02
0.87, 1.20
0.93
0.79, 1.09
236
1.03
0.90, 1.18
0.99
0.86, 1.14
0.89
0.77, 1.03
58
1.08
0.83, 1.40
1.09
0.84, 1.42
0.96
0.74, 1.26
122
1.15
0.95, 1.38
1.07
0.88, 1.29
0.94
0.78, 1.15
20
P for trend
0.86
0.85
0.17
0.13
0.55
0.34
Other
3,096
1.00
Referent
1.00
Referent
1.00
Referent
1,558
1.00
Referent
1.00
Referent
1.00
Referent
1–4
1,334
0.97
0.91, 1.04
0.91
0.86, 0.98
0.89
0.83, 0.95
1,170
1.05
0.98, 1.14
0.98
0.91, 1.06
0.93
0.86, 1.02
5–9
247
1.00
0.88, 1.14
0.92
0.80, 1.05
0.86
0.75, 0.99
310
1.23
1.08, 1.39
1.11
0.98, 1.25
0.98
0.86, 1.12
10–19
169
1.18
1.01, 1.37
1.07
0.92, 1.25
0.99
0.84, 1.16
220
1.28
1.11, 1.48
1.12
0.97, 1.30
0.96
0.82, 1.12
56
1.26
0.97, 1.64
1.11
0.85, 1.45
0.98
0.75, 1.29
106
1.41
1.15, 1.71
1.18
0.96, 1.44
0.97
0.79, 1.19
20
P for trend
0.03
0.73
0.44
<0.0001
0.02
0.84
789
Abbreviations: CI, confidence interval; RR, relative risk.
Age-adjusted model.
b
Multivariate-adjusted model stratified by age that included alcohol consumption, race, smoking status, educational level, physical activity level, history of high blood pressure, history of
diabetes, and total energy intake.
c
Multivariate-adjusted model stratified by age that included alcohol consumption, race, smoking status, educational level, physical activity level, body mass index in 1982, weight change
from 18 years of age to 1982, history of high blood pressure, history of diabetes, and total energy intake.
a
Weight Cycling and Mortality
0
790 Stevens et al.
A)
Hazard Ratio
2.0
1.0
0.8
0.6
<25
25–29.9
30
Body Mass Index in 1982
Hazard Ratio
B)
2.0
1.0
0 8.
0.6
<25
25–29.9
30
Body Mass Index in 1982
Figure 1. Risk of death from all causes and weight cycling stratified
by body mass index (weight (kg)/height (m)2) in Cancer Prevention
Study II Nutrition Cohort participants in 1982. The associations of 1–4
(open circles), 5–9 (filled circles), 10–19 (open diamonds), and 20 or
more (filled diamonds) cycles with all-cause mortality relative to participants who did not experience weight cycling is shown among men
(A) and women (B). The hazard ratios and 95% confidence intervals
were determined using a multivariable-adjusted model that was stratified by age and included alcohol use, race, smoking status, educational
level, physical activity level, body mass index in 1982, weight change
from 18 years of age to 1982, history of high blood pressure, history of
diabetes, and total energy intake. Bars, 95% confidence interval.
DISCUSSION
In the present large prospective study, high levels of weight
cycling were not associated with increased all-cause mortality rates in either men or women after adjustment for
BMI and other confounding factors. However, low levels
of weight cycling (1–4 or 5–9 weight cycles in men and 1–4
weight cycles in women) were inversely associated with allcause mortality. The importance of adequately controlling
for body weight and weight change at middle age, which can
influence mortality (39, 40), is illustrated by the differences
in the age-adjusted associations after adjustment for these and
other covariates. These findings suggest that, when considered independently of the BMI and weight change, weight
cycling does not increase the risk of premature death.
No association was found between weight cycling and
death from cardiovascular disease for any category of weight
cycling except the lowest. Most categories of weight cycling
were also not associated with death from cancer or other
causes. The statistically significant inverse association of
5–9 weight cycles with cancer death in men was likely due
to chance because there was no evidence of a significantly
reduced risk of death from cancer for either higher or lower
levels of weight cycling. The significant associations of 1–4
and 5–9 weight cycles with death from other causes in men
may reflect a true reduced risk, although the broad nature of
this endpoint precludes a meaningful interpretation of this
finding. A large study is needed to fully assess the specific
causes of death.
The findings presented here are consistent with those of
the only other study that considered the intentionality of
weight loss when defining weight cycling. Field et al. (16)
reported that weight cycling was not associated with allcause or cardiovascular disease mortality for women in
the Nurses’ Health Study. However, similar to our findings
for men and women with low numbers of weight cycles,
Field et al. (16) found that intentionally losing at least 4.5 kg
(10 pounds) at least 3 times was associated with a statistically significant 17% lower risk of all-cause mortality compared with noncyclers (16). Why low levels of weight cycling
might be associated with a reduced risk of premature death
is not readily apparent.
The association between weight cycling and all-cause
mortality did not vary by BMI at middle age (in 1982) or
by weight change between 18 years of age and 1982. To our
knowledge, this fact has never been investigated before and
indicates that potential effects of weight cycling relevant to
mortality do not differ in lean, overweight, or obese individuals
and are not differentially affected by weight changes during
adulthood. The association between weight cycling and mortality was also not confounded by smoking, as indicated by the
sensitivity analysis in which only nonsmokers were included.
The present study has a number of strengths, including its
prospective nature and the large number of weight cyclers of
both sexes. To our knowledge, it includes the largest number
of weight cyclers of any study reported to date and is the first
to investigate the association between weight cycling initiated by intentional weight loss and mortality in men. The
large study population allowed us to examine the associations between 4 levels of this behavior and mortality in men
and women separately. These levels covered a broader range
of number of weight cycles than other studies, with the top
category of 20 or more weight cycles, which was twice that
of the maximum category of any previous report (23). Because information was collected on both purposeful weight
loss and subsequent regain, we were able to define weight
cycles based on both phases rather than just assuming regain, as has been done in many studies (5, 16, 41–44). The
availability of data on BMI at middle age, weight change,
and other covariates for the CPS-II Nutrition Cohort participants was critical for separating the influence of these factors on mortality from that of weight cycling.
An important limitation of the present study is the lack of
information on the timing, magnitude, and duration of each
phase of the weight cycles, and it is unclear whether this
behavior began during adolescence or adulthood. Therefore,
whether the association of weight cycling with mortality
differs during various phases of life could not be assessed.
Similarly, without knowing how much weight was lost in
each cycle, we were unable to determine whether the associations of mortality with larger weight loss cycles differed
from those with smaller weight loss cycles. In the present
Am J Epidemiol. 2012;175(8):785–792
Weight Cycling and Mortality
study, weight cycling history was self-reported. Although
this may have resulted in some misclassification, there is
no apparent reason why it would be differential with respect
to mortality. In addition, the weight cycling responses have
not been validated. However, 75% of the weight cyclers reported equal numbers of times they purposely lost weight and
regained weight, which would be expected given that most
people who lose weight do not maintain that weight loss (45).
In summary, the results of the present study indicate that
weight cycling, independent of BMI and weight gain, does
not increase the overall risk of mortality. These results apply
to participants with varying numbers of weight cycles and to
both men and women of all body weights. Thus, weight loss
attempts by overweight and obese individuals should be
encouraged even though weight lost may often be regained.
ACKNOWLEDGMENTS
Author affiliation: Epidemiology Research Program,
American Cancer Society, Atlanta, Georgia.
Conflict of interest: none declared.
REFERENCES
1. Flegal KM, Carroll MD, Ogden CL, et al. Prevalence and
trends in obesity among US adults, 1999–2008. JAMA. 2010;
303(3):235–241.
2. Yaemsiri S, Slining MM, Agarwal SK. Perceived weight
status, overweight diagnosis, and weight control among US
adults: the NHANES 2003–2008 Study. Int J Obes (Lond).
2011;35(8):1063–1070.
3. Katan MB. Weight-loss diets for the prevention and treatment
of obesity. N Engl J Med. 2009;360(9):923–925.
4. Lahti-Koski M, Männistö S, Pietinen P, et al. Prevalence of
weight cycling and its relation to health indicators in Finland.
Obes Res. 2005;13(2):333–341.
5. Tsai CJ, Leitzmann MF, Willett WC, et al. Weight cycling and
risk of gallstone disease in men. Arch Intern Med. 2006;166(21):
2369–2374.
6. Syngal S, Coakley EH, Willett WC, et al. Long-term weight
patterns and risk for cholecystectomy in women. Ann Intern
Med. 1999;130(6):471–477.
7. National Task Force on the Prevention and Treatment of
Obesity. Weight cycling. JAMA. 1994;272(15):1196–1202.
8. Hamm P, Shekelle RB, Stamler J. Large fluctuations in body
weight during young adulthood and twenty-five-year risk of
coronary death in men. Am J Epidemiol. 1989;129(2):312–318.
9. Lissner L, Odell PM, D’Agostino RB, et al. Variability of body
weight and health outcomes in the Framingham population.
N Engl J Med. 1991;324(26):1839–1844.
10. Blair SN, Shaten J, Brownell K, et al. Body weight change,
all-cause mortality, and cause-specific mortality in the
Multiple Risk Factor Intervention Trial. Ann Intern Med. 1993;
119(7 pt 2):749–757.
11. Diaz VA, Mainous AG III, Everett CJ. The association between
weight fluctuation and mortality: results from a populationbased cohort study. J Community Health. 2005;30(3):153–165.
12. Rzehak P, Meisinger C, Woelke G, et al. Weight change,
weight cycling and mortality in the ERFORT Male Cohort
Study. Eur J Epidemiol. 2007;22(10):665–673.
Am J Epidemiol. 2012;175(8):785–792
791
13. Dyer AR, Stamler J, Greenland P. Associations of weight
change and weight variability with cardiovascular and
all-cause mortality in the Chicago Western Electric Company
Study. Am J Epidemiol. 2000;152(4):324–333.
14. Lissner L, Andres R, Muller DC, et al. Body weight variability
in men: metabolic rate, health and longevity. Int J Obes. 1990;
14(4):373–383.
15. Folsom AR, French SA, Zheng W, et al. Weight variability and
mortality: the Iowa Women’s Health Study. Int J Obes Relat
Metab Disord. 1996;20(8):704–709.
16. Field AE, Malspeis S, Willett WC. Weight cycling and mortality among middle-aged or older women. Arch Intern Med.
2009;169(9):881–886.
17. Wannamethee SG, Shaper AG, Walker M. Weight change,
weight fluctuation, and mortality. Arch Intern Med. 2002;
162(22):2575–2580.
18. French SA, Folsom AR, Jeffery RW, et al. Weight variability
and incident disease in older women: the Iowa Women’s
Health Study. Int J Obes Relat Metab Disord. 1997;21(3):
217–223.
19. Field AE, Byers T, Hunter DJ, et al. Weight cycling, weight
gain, and risk of hypertension in women. Am J Epidemiol.
1999;150(6):573–579.
20. Guagnano MT, Ballone E, Pace-Palitti V, et al. Risk factors for
hypertension in obese women: the role of weight cycling. Eur
J Clin Nutr. 2000;54(4):356–360.
21. Waring ME, Eaton CB, Lasater TM, et al. Incident diabetes in
relation to weight patterns during middle age. Am J Epidemiol.
2010;171(5):550–556.
22. Søgaard AJ, Meyer HE, Tonstad S, et al. Weight cycling and
risk of forearm fractures: a 28-year follow-up of men in the
Oslo Study. Am J Epidemiol. 2008;167(8):1005–1013.
23. Luo J, Margolis KL, Adami HO, et al. Body size, weight
cycling, and risk of renal cell carcinoma among postmenopausal women: the Women’s Health Initiative (United States).
Am J Epidemiol. 2007;166(7):752–759.
24. Trentham-Dietz A, Newcomb PA, Egan KM, et al. Weight
change and risk of postmenopausal breast cancer (United
States). Cancer Causes Control. 2000;11(6):533–542.
25. De Roos AJ, Ulrich CM, Ray RM, et al. Intentional weight loss
and risk of lymphohematopoietic cancers. Cancer Causes
Control. 2010;21(2):223–236.
26. Calle EE, Rodriguez C, Jacobs EJ, et al. The American Cancer
Society Cancer Prevention Study II Nutrition Cohort: rationale,
study design, and baseline characteristics. Cancer. 2002;94(9):
2490–2501.
27. Block G, Hartman AM, Naughton D. A reduced dietary
questionnaire: development and validation. Epidemiology.
1990;1(1):58–64.
28. Block G, Coyl L, Smucker R, et al. Health Habits and History
Questionnaire: Diet History and Other Risk Factors. Personal
Computer System Documentation. Bethesda, MD: National
Cancer Institute; 1989.
29. Calle EE, Terrell DD. Utility of the National Death Index
for ascertainment of mortality among Cancer Prevention
Study II participants. Am J Epidemiol. 1993;137(2):
235–241.
30. Casey VA, Dwyer JT, Berkey CS, et al. Long-term memory
of body weight and past weight satisfaction: a longitudinal
follow-up study. Am J Clin Nutr. 1991;53(6):1493–1498.
31. Must A, Willett WC, Dietz WH. Remote recall of childhood
height, weight, and body build by elderly subjects. Am J
Epidemiol. 1993;138(1):56–64.
32. Cox D. Regression models and life tables. J R Stat Soc B. 1972;
34(2):187–202.
792 Stevens et al.
33. Kleinbaum GG, Kupper LL, Morgenstern H. Epidemiologic
Research: Principles and Quantitative Methods. New York,
NY: Van Nostrand Reinhold; 1982.
34. Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of
physical activities: classification of energy costs of human
physical activities. Med Sci Sports Exerc. 1993;25(1):71–80.
35. Hu FB. Obesity and Mortality. In: Hu FB, ed. Obesity Epidemiology. New York, NY: Oxford University Press; 2008:216–233.
36. Calle EE, Thun MJ, Petrelli JM, et al. Body-mass index and
mortality in a prospective cohort of U.S. adults. N Engl J Med.
1999;341(15):1097–1105.
37. Wack JT, Rodin J. Smoking and its effects on body weight and
the systems of caloric regulation. Am J Clin Nutr. 1982;35(2):
366–380.
38. US Department of Health and Human Services. The Health
Consequences of Smoking: A Report of the Surgeon General
Surgeon General’s Reports. Atlanta, GA: Centers for Disease
Control and Prevention; 2004.
39. Berrington de Gonzalez A, Hartge P, Cerhan JR, et al. Bodymass index and mortality among 1.46 million white adults.
N Engl J Med. 2010;363(23):2211–2219.
40. Lee IM, Paffenbarger RS Jr. Change in body weight and
longevity. JAMA. 1992;268(15):2045–2049.
41. Shade ED, Ulrich CM, Wener MH, et al. Frequent intentional
weight loss is associated with lower natural killer cell cytotoxicity in postmenopausal women: possible long-term immune
effects. J Am Diet Assoc. 2004;104(6):903–912.
42. Hooper LE, Foster-Schubert KE, Weigle DS, et al. Frequent
intentional weight loss is associated with higher ghrelin and
lower glucose and androgen levels in postmenopausal women.
Nutr Res. 2010;30(3):163–170.
43. Strychar I, Lavoie ME, Messier L, et al. Anthropometric,
metabolic, psychosocial, and dietary characteristics of overweight/obese postmenopausal women with a history of weight
cycling: a MONET (Montreal Ottawa New Emerging Team)
Study. J Am Diet Assoc. 2009;109(4):718–724.
44. Van Wye G, Dubin JA, Blair SN, et al. Weight cycling and 6-year
weight change in healthy adults: the Aerobics Center Longitudinal Study. Obesity (Silver Spring). 2007;15(3):731–739.
45. Kraschnewski JL, Boan J, Esposito J, et al. Long-term weight
loss maintenance in the United States. Int J Obes (Lond). 2010;
34(11):1644–1654.
Am J Epidemiol. 2012;175(8):785–792