American Journal of Epidemiology
© The Author 2013. 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. 178, No. 2
DOI: 10.1093/aje/kws461
Advance Access publication:
May 3, 2013
Original Contribution
Long- and Short-term Weight Change and Incident Coronary Heart Disease and
Ischemic Stroke
The Atherosclerosis Risk in Communities Study
June Stevens*, Eva Erber, Kimberly P. Truesdale, Chin-Hua Wang, and Jianwen Cai
* Correspondence to Dr. June Stevens, Department of Nutrition, Gillings School of Global Public Health, University of North Carolina at Chapel
Hill, 2207 McGavran-Greenberg Hall, CB 7461, Chapel Hill, NC 27599 (e-mail: [email protected]).
Initially submitted June 29, 2012; accepted for publication November 20, 2012.
Weight gain increases the prevalence of obesity, a risk factor for cardiovascular disease. Nevertheless, unintentional weight loss can be a harbinger of health problems. The Atherosclerosis Risk in Communities Study (1987–
2009) included 15,792 US adults aged 45–64 years at baseline and was used to compare associations of
long-term (30 years) and short-term (3 years) weight change with the risks of coronary heart disease (CHD) and
ischemic stroke. Age-, gender-, and race-standardized incidence rates were 4.9 (95% confidence interval (CI):
4.6, 5.2) per 1,000 person-years for CHD and 2.5 (95% CI: 2.3, 2.8) per 1,000 person-years for stroke. After controlling for baseline body mass index and other covariates, long-term weight gain (since age 25 years) of more
than 2.7% was associated with elevated CHD risk, and any long-term weight gain was associated with increased
stroke risk. Among middle-aged adults, short-term (3-year) weight loss of more than 3% was associated with elevated
immediate CHD risk (hazard ratio = 1.46, 95% CI: 1.18, 1.81) and stroke risk (hazard ratio = 1.45, 95% CI: 1.10,
1.92). Risk tended to be larger in adults whose weight loss did not occur through dieting. Avoidance of weight gain
between early and middle adulthood can reduce risks of CHD and stroke, but short-term, unintentional weight loss in
middle adulthood may be an indicator of immediate elevated risk that has not previously been well recognized.
body mass index; body weight changes; coronary heart disease; ischemic stroke
Abbreviations: BMI, body mass index; CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; HR,
hazard ratio; NHANES II, Second National Health and Nutrition Examination Survey; SD, standard deviation.
Diverse findings could be due to differences in study populations, lengths of the weight-change intervals, and the
time periods between weight changes and events. Galanis
et al. (8) provided insight into this issue by examining CHD
risk in relation to 2 weight-change intervals among 6,176
Japanese-American men: 1) the “early interval” from age 25
years to the first study examination (mean = 29 years) and
2) the “late interval” between the first and third examinations
(mean = 6 years). Follow-up for incident CHD began at the
end of this later interval for both weight-change variables.
Compared with weight maintenance (±2.5 kg), the investigators found that the relative risk of CHD was increased with
weight gain (>5 kg) in the “early interval” but increased
with weight loss (>2.5 kg) in the “late interval.” These
Cardiovascular disease (CVD) caused 823,746 deaths in
the United States in 2006, with coronary heart disease
(CHD) accounting for 68% of CVD deaths and stroke for
17% (1, 2). The association between body mass index
(BMI) in adulthood and risks of CHD and ischemic stroke is
well established (3–5); however, the role of weight change is
less clear. Studies have usually (6–10), but not always (11),
found weight gain over a period of 12 or more years to be
associated with increased CHD risk. Results on weight loss
and CHD are more mixed, with 2 studies finding an association with increased risk (9, 10) and 3 others reporting no
association (6–8). One study examining ischemic stroke
found increased incidence associated with long-term weight
gain but not with weight loss (12).
239
Am J Epidemiol. 2013;178(2):239–248
240 Stevens et al.
findings may be due to a greater influence of illness on weight
change in the “late interval.” To our knowledge, this study of
Japanese-American men is the only study to have examined
CHD risk associated with early, long-term weight change and
later, short-term weight change in the same cohort.
This work, together with the inconsistencies in the literature, led us to hypothesize that risks of CHD and ischemic
stroke could differ in adults experiencing earlier, long-term
weight change compared with adults experiencing later,
short-term weight change. Specifically, we hypothesized
that earlier, long-term weight gain experienced over a period
of 20 or more years would be associated with increased
CHD and stroke risk over a long follow-up period. In contrast, we expected later, short-term weight loss to be associated with increased risk during the years immediately
following the weight change. We also explored the impact
of BMI and dieting on these relations.
was measured to the nearest centimeter with participants
wearing no shoes. Body weight was measured at all 4 examinations. These measurements were used to calculate BMI
(weight (kg)/height (m)2) at age 25 years and at each of the
4 examinations. Customary BMI categories were used (14):
underweight (<18.5), normal-weight (18.5–<25.0), overweight (25.0–<30.0), and obese (≥30.0). Long-term weight
change between age 25 years and examination 1 was categorized as weight loss (<−3%), maintenance (±3%) (15), small
gain (>3–<10%), moderate gain (≥10–<30%), and large gain
(≥30%). Short-term weight change between 2 examinations
was categorized as weight loss (<−3%), maintenance (±3%),
small gain (>3–<10%), and moderate-to-large gain (≥10%).
Weight maintenance was defined as weight change of ±3%,
since this cutpoint has been shown to indicate an amount of
weight change that is less than clinically relevant but more
than would be expected from measurement error or diurnal
fluctuations in fluid balance under normal conditions (15).
MATERIALS AND METHODS
Study population
We used data from the Atherosclerosis Risk in Communities Study, a study of 15,792 white and black US men and
women aged 45–64 years at baseline (examination 1: 1987–
1989) (13). Participants were invited to undergo 3 additional
examinations at approximate 3-year intervals. This study
was approved by the institutional review boards at each field
center, and all subjects gave written consent. This secondary
analysis was approved by the University of North Carolina
at Chapel Hill Non-Biomedical Institutional Review Board.
Baseline and follow-up assessments
Interviewer-administered questionnaires (13) assessed education (examination 1), smoking (all examinations), alcohol
consumption (all examinations), physical activity (examinations 1 and 3) (by means of the Baecke leisure-time physical
activity questionnaire (16)), and dieting to lose weight
(examinations 1 and 3). Participants who responded being
on a weight-loss diet either at examination 1 and/or at examination 3 were considered to have a propensity for being on a
weight-loss diet. In this paper, we will refer to participants
with or without a propensity for being on a weight-loss diet
as “dieters” and “nondieters,” respectively.
Study design
We aimed to contrast the association between earlier,
long-term weight change from early to middle adulthood
and CVD risk during a long follow-up period with the association between later, short-term weight change during midadulthood and immediate risk (Figure 1). Earlier, long-term
weight change occurred between age 25 years and examination 1 ( panel A in Figure 1). Data on events of interest were
collected from examination 1 to December 31, 2009;
however, we excluded events that occurred within 3 years
after examination 1 to avoid including events due to underlying disease and to clearly distinguish the 2 time periods of
interest. Throughout this article, we describe results derived
from this study design as “long-term.”
Later, short-term weight changes were studied over each
of the 3 intervals between the 4 examinations ( panel B in
Figure 1). After ascertaining an event, we defined short-term
weight change as the change between 2 consecutive examinations immediately prior to that event. Follow-up was censored at 3 years after the last examination so that all events
occurred within 3 years after weight change. Results from
this design will be referred to as “short-term.”
Obesity measures
During examination 1, participants recalled their weight
at age 25 years using time-associated events (13), and height
Ascertainment of incident CHD and ischemic stroke
Incident CHD was defined as definite or probable nonfatal
myocardial infarction or definite cardiac death. Inclusion of
silent myocardial infarctions and coronary procedures yielded
similar results; therefore, we present only the estimates for
cardiac death and myocardial infarction. To ascertain cases
of incident CHD and ischemic stroke, we used cohort and
community surveillance, including discharge lists from local
hospitals and local obituaries and annual vital statistics
review tapes (13, 17). CHD and ischemic stroke were classified according to the International Classification of Diseases,
Ninth Revision.
Exclusions
It was standard in this cohort study to exclude blacks from
Washington County, Maryland, or Minneapolis, Minnesota
(n = 55), and persons of ethnic groups other than black or
white (n = 48) because of small sample sizes. Participants
with prevalent or missing information on CHD (n = 1,102)
or stroke (n = 325) at examination 1 were also excluded. For
the long-term CHD analysis, we excluded participants who
had missing covariate data at examination 1 (n = 164) or at
age 25 years (n = 194) or had a CHD event within 3 years
after examination 1 (n = 176). The corresponding exclusions
for the stroke analysis were 185, 201, and 62, respectively.
Am J Epidemiol. 2013;178(2):239–248
Weight Change, Coronary Heart Disease, and Stroke 241
Figure 1. Study designs used for analysis of the associations between long-term weight change (A) and short-term weight change (B) and
incident coronary heart disease and ischemic stroke in the Atherosclerosis Risk in Communities Study, 1987–2009.
Data from 14,053 adults were available for our long-term
studies of CHD, and data from 14,916 adults were available
for stroke. For the short-term CHD analysis, we excluded
persons with missing covariate data at examination 1
(n = 157) and those who were lost to follow-up or deceased
or had an early event within 3 years after examination 1
(n = 1,081), as well as those who were missing data on BMI
and weight change (n = 213). The corresponding exclusions
for the stroke analysis were 177, 1,079, and 217, respectively. Sample sizes for the short-term analyses were 13,136
for CHD and 13,891 for stroke.
alcohol consumption, physical activity, and smoking at
examination 1. For the short-term analyses, smoking, alcohol consumption, and physical activity were used as timedependent covariates. All models were analyzed for the
entire cohort combined and by race/gender group. The interaction terms for interaction between BMI and race/gender
groups and between weight change and race/gender groups
were not significant; therefore, we present only the findings
for all race/gender groups combined. All statistical analyses
were performed using SAS, version 9.2 (SAS Institute, Inc.,
Cary, North Carolina).
Statistical analysis
Poisson regression was used to calculate incidence rates
standardized for age, education, physical activity, and height
at examination 1, smoking status at age 25 years and examination 1, and race/field center, gender, and long-term weight
change. Cox regression models (18) were used to estimate
hazard ratios and 95% confidence intervals. BMI and
weight change were analyzed in categorical and continuous
forms. Weight change in pounds and weight change in
percent yielded similar conclusions; thus, we present only
the results from the analyses using percent weight change.
We used likelihood ratio tests to fit a parsimonious model
characterizing the relationship between weight change and
incident CHD and stroke. We first evaluated 9-knot quadratic spline regression models (19) using time since examination 1 as the time scale. Testing of successively simplified
models showed that linear models were sufficient for all
analyses, except for the analysis of the association of longterm weight change with CHD risk, for which a 3-knot
model (5th, 50th, and 95th percentiles) was used. All
models were adjusted for race/field center, age, gender, education, and height at examination 1. The long-term analyses
were additionally adjusted for smoking at age 25 years and
Am J Epidemiol. 2013;178(2):239–248
RESULTS
The examination 1 descriptive characteristics shown in
Table 1 are from the data set used for the long-term analyses.
The covariate values were almost identical in the short-term
subsets and are not shown. Approximately half of the participants were female, and one-quarter were black. The mean
BMI at examination 1 was in the overweight range, but the
mean BMI at age 25 years was within the normal range.
Long-term weight change (from age 25 years to examination 1) averaged 28 kg (a 21% increase in body weight) over
a mean period of 30 years (range, 20–39 years). The mean
short-term weight change (between 2 consecutive examinations) was 1 kg (1.3%) over an average of 3 years (range,
0.6–5.7 years). Participants who were obese at age 25 years
gained less weight between age 25 years and examination 1
than those who were normal-weight at age 25 years (3.9 kg
vs. 13.7 kg; P < 0.01 in the CHD subset). The Pearson correlation coefficient for correlation between BMI at age 25
years and long-term weight change was −0.22. In contrast,
there was no association between short-term weight change
and BMI at the beginning of the weight-change interval.
These results were almost identical in the stroke subsets.
242 Stevens et al.
Table 1. Characteristics of the Study Sample at Examination
1 (1987–1989), Atherosclerosis Risk in Communities Study,
1987–2009a
Characteristic
Coronary Heart
Disease Data Set
(n = 14,053)
Mean (SD)
Age, years
%
54.0 (5.7)
Ischemic Stroke
Data Set
(n = 14,916)
Mean (SD)
%
54.1 (5.8)
Female gender
57
55
African-American
race
26
25
Less than
high school
22
23
High school
graduation
or equivalent
41
41
At least some
college
37
36
Education
BMIb
27.6 (5.4)
27.6 (5.4)
BMI at age 25
yearsc
23.0 (3.7)
23.0 (3.7)
Smoking status
Never smoker
43
42
Former smoker
31
32
Current smoker
26
26
50
51
Never/rare
drinker
43
42
Former drinker
18
19
% who smoked
at age 25
yearsd
Alcohol drinking
status
Light drinker
9
9
Moderate drinker
16
16
Heavy drinker
13
14
Low (tertile 1)
41
41
Medium (tertile 2)
30
30
High (tertile 3)
28
28
Level of physical
activity
Abbreviations: BMI, body mass index; SD, standard deviation.
a
Values shown are from the data sets used for the long-term
weight-change analyses examining coronary heart disease and
ischemic stroke as outcomes.
b
Weight (kg)/height (m)2.
c
Calculated using weight at age 25 years as recalled at
examination 1 and height as measured at examination 1.
d
Self-reported smoking at age 25 years from recall at examination 1.
The crude incidence rates over an average of 20 years of
follow-up were 5.3 (95% confidence interval (CI): 5.0, 5.6)
per 1,000 person-years and 3.0 (95% CI: 2.7, 3.2) per 1,000
person-years for CHD and ischemic stroke, respectively.
Standardizing covariates slightly attenuated both rates to 4.1
(95% CI: 3.8, 4.4) per 1,000 person-years and 2.3 (95% CI:
2.1, 2.6) per 1,000 person-years, respectively. Additionally
controlling for long-term weight change did not alter these
rates considerably.
BMI at age 25 years and during middle adulthood were
both positively associated with incident CHD and stroke
(Table 2). A BMI of ≥30 at age 25 years doubled CHD and
stroke risk, while a BMI of ≥30 during middle adulthood
elevated CHD risk by more than 50%. Additional adjustment for long-term weight change did not substantially
change these results (data not shown).
Short-term weight loss, but not long-term weight loss, elevated CHD risk in comparison with weight maintenance
(Table 3). Large long-term weight gain was also positively
associated with increased CHD risk, and this association
was strengthened after additional adjustment for BMI at age
25 years; however, short-term weight gain was not associated with CHD risk. Results for stroke were very similar.
To explore whether the elevated CHD and stroke risks
associated with short-term weight loss were related to intentional weight loss or unintentional weight loss, we examined
the associations for weight change stratified by dieting status
(data not shown). For the CHD analysis, there were 888
dieters, and for the stroke analysis, there were 936 dieters.
The average BMI and short-term weight change tended to
be higher in dieters (BMI 30.3, a 2.2% gain in weight) than
in nondieters (BMI 27.3, a 1.2% gain in weight), and over
the short term, 19% of dieters and 18% of nondieters lost
more than 3% of their weight for both outcomes. Compared
with weight maintainers, the estimate for the association
between short-term weight loss and CHD was null and
smaller in dieters (hazard ratio (HR) = 1.14, 95% CI: 0.49,
2.68) than in nondieters (HR = 1.48, 95% CI: 1.18, 1.85).
The estimates for the associations of short-term weight loss
with stroke were similar in magnitude among dieters and
nondieters; however, the 95% confidence interval was wider
in dieters (HR = 1.44, 95% CI: 0.46, 4.51) than in nondieters
(HR = 1.45, 95% CI: 1.09, 1.94). This was evidence that
dieters who lost weight over a 3-year period did not have
elevated CHD risk, but evidence for differences between
dieters and nondieters regarding stroke was weak.
Our analyses using continuous weight change elaborated
on the categorical analyses. The association between CHD
and long-term weight change was U-shaped, with increases
in risk leveling off at very high levels (at a weight gain of
approximately 65%, representing 91% and 52% increases in
CHD risk with and without adjustment for BMI at age 25
years, respectively; Figure 2). The lowest CHD risks were
seen in participants who gained 8.5% (95% CI: 5.2, 11.9)
without adjustment for BMI at age 25 years and participants
who gained 2.7% (95% CI: −4.5, 10.0) with adjustment for
BMI at age 25 years. A long-term weight change of ≥16%
was associated with a significant increase in CHD (after
adjustment for BMI at age 25 years). Although admittedly it
is from only 1 study, this result may be useful to policymakers seeking to provide long-term weight-gain cutpoints.
In addition, CHD risk was elevated by 50% (95% CI: 1.28,
1.74) at a weight gain of 36%, while a weight gain of 50%
elevated CHD risk by 73% (95% CI: 1.46, 2.05) after adjustment for BMI at age 25 years.
Am J Epidemiol. 2013;178(2):239–248
Weight Change, Coronary Heart Disease, and Stroke 243
Table 2. Associations of BMIa at Age 25 Years (Long-term Analysis) and BMI During Middle Adulthood (Short-term Analysis) With Incident
Coronary Heart Disease and Ischemic Stroke in the Atherosclerosis Risk in Communities Study, 1987–2009
BMI at Age 25 Yearsb
BMI Measure
No. of Events
Person-Years
HR
Time-dependent BMI During Middle Adulthoodc
95% CI
No. of Events
Person-Years
HR
95% CI
Coronary Heart Disease
Categorical BMI
Underweight
(<18.5)d
67
15,585
0.91
Normal-weight
(18.5–<25.0)
885
187,520
1.00
Overweight
(25.0–<30.0)
322
47,079
1.30
111
10,873
1,385
261,057
Obese (≥30.0)
Linear, continuous
BMI (per 5-unit
increase)
0.71, 1.18
3
816
114
31,859
1.00
1.13, 1.48
224
40,741
1.38
1.10, 1.73
2.03
1.63, 2.53
179
29,341
1.76
1.38, 2.25
1.33
1.24, 1.44
520
102,758
1.20
1.11, 1.31
2
844
61
33,443
1.00
Ischemic Stroke
Categorical BMI
Underweight
(<18.5)d
52
16,098
1.08
0.81, 1.44
Normal-weight
(18.5–<25.0)
543
198,446
1.00
Overweight
(25.0–<30.0)
163
51,382
1.13
0.94, 1.35
127
43,520
1.41
1.04, 1.93
Obese (≥30.0)
62
12,002
2.17
1.66, 2.85
96
31,266
1.55
1.11, 2.16
Linear, continuous
BMI (per 5-unit
increase)
820
277,928
1.25
1.14, 1.36
286
109,072
1.15
1.03, 1.29
Abbreviations: BMI, body mass index; CI, confidence interval; HR, hazard ratio.
a
Weight (kg)/height (m)2.
b
Adjusted for smoking status at age 25 years, as well as for race/field center, age, gender, education, smoking status, alcohol consumption,
physical activity, and height at examination 1.
c
Adjusted for race/field center, age, gender, education, and height at examination 1 and for time-dependent smoking status, alcohol
consumption, and physical activity.
d
The hazard ratio for the short-term analysis of BMI is not presented because of a small sample size.
The negative log-linear relation of short-term weight
change with CHD risk indicated that short-term weight
losses were associated with elevated immediate risk and
weight gains with lowered risk. A short-term weight gain of
5% lowered CHD risk by 8% (95% CI: 0.86, 1.00); this
association was slightly attenuated after additional adjustment for BMI at the beginning of the weight-change interval. For stroke, a 5% long-term weight gain elevated risk by
3% (95% CI: 1.01, 1.04), but a 5% short-term weight gain
reduced risk by 16% (95% CI: 0.76, 0.93). These estimates
changed only slightly after additional adjustment for BMI
(Figure 3).
DISCUSSION
Our results generally confirmed our hypotheses that compared with weight maintenance, earlier weight gain over a
long interval was associated with increased risks of CHD
and ischemic stroke, whereas later, weight loss over a
shorter interval was associated with increased immediate
CHD and stroke risk. Although our analyses of weight
Am J Epidemiol. 2013;178(2):239–248
change in categories versus a continuous form generally supported each other, there were some differences. Given the
reliance of categories on arbitrary cutpoints, the greater
power of continuous analyses, and our extensive work to
define the shapes of the continuous associations, we are
more confident in conclusions drawn from the continuous
analyses. We found increased risk with any long-term
weight gain for stroke and with weight gain above 2.7% for
CHD. The majority of persons in our sample (83%) gained
more than 2.7% of their weight at age 25 years and were
therefore at elevated CHD risk.
Our estimates of the positive association between longterm large weight gain and CVD agree with previous studies
showing that a weight gain of >19 kg since young adulthood
(18–21 years) more than doubled the risks of CHD (6) and
ischemic stroke (12) in women and the risks of CHD and
coronary procedures in men (7), compared with maintaining
one’s weight. Associations of short-term weight change
with immediate risk have been less well studied. We found
that middle-aged adults who lost weight over a 3-year interval without intentionally dieting were more likely to have
244 Stevens et al.
Table 3. Associations of Long- and Short-term Weight Change With
Incident Coronary Heart Disease and Ischemic Stroke in the
Atherosclerosis Risk in Communities Study, 1987–2009
% Weight
Changea
No. of
Events
PersonYears
HR
Table 3. Continued
% Weight
Changea
No. of
Events
PersonYears
HR
95% CI
Short-term Weight Change Analysis
95% CI
Coronary heart
disease
Long-term Weight Change Analysis
Model 1b
Coronary heart
disease
Model 1b
Weight loss
147
19,680
1.22
0.96, 1.56
Weight
maintenance
116
22,693
1.00
Small weight
gain
169
39,899
0.84
0.66, 1.06
Moderate
weight gain
543
106,288
1.05
0.86, 1.29
Large weight
gain
410
72,496
1.26
1.02, 1.56
Model 2c
0.82, 1.35
Weight loss
133
18,534
1.46
Weight
maintenance
227
48,637
1.00
Small weight
gain
135
29,693
1.09
0.88, 1.35
25
5,894
1.06
0.70, 1.61
1.14, 1.76
Moderate-tolarge weight
gain
Model 2c
Weight loss
133
18,534
1.42
Weight
maintenance
227
48,637
1.00
Small weight
gain
135
29,693
1.10
0.89, 1.36
25
5,894
1.09
0.72, 1.66
80
19,775
1.45
1.10, 1.92
131
51,697
1.00
Small weight
gain
65
31,332
0.93
0.69, 1.25
Moderate-tolarge weight
gain
10
6,269
0.75
0.39, 1.43
80
19,775
1.43
1.08, 1.89
131
51,697
1.00
Small weight
gain
65
31,332
0.94
0.70, 1.26
Moderate-tolarge weight
gain
10
6,269
0.77
0.40, 1.47
Weight loss
147
19,680
1.06
Weight
maintenance
116
22,693
1.00
Small weight
gain
169
39,899
0.88
0.69, 1.11
Moderate
weight gain
543
106,288
1.16
0.95, 1.43
Ischemic stroke
Large weight
gain
410
72,496
1.52
1.22, 1.89
Weight loss
Moderate-tolarge weight
gain
Model 1b
Weight
maintenance
Ischemic stroke
Model 1b
Weight loss
65
21,218
1.04
0.73, 1.48
Weight
maintenance
61
24,484
1.00
Small weight
gain
100
42,577
0.98
0.71, 1.34
Moderate
weight gain
312
112,936
1.12
0.85, 1.48
Large weight
gain
282
76,713
1.38
1.04, 1.84
Model 2c
Weight loss
65
21,218
0.94
0.66, 1.33
Weight
maintenance
61
24,484
1.00
Small weight
gain
100
42,577
1.01
0.73, 1.38
Moderate
weight gain
312
112,936
1.21
0.92, 1.60
Large weight
gain
282
76,713
1.58
1.18, 2.12
Table continues
CHD or stroke in the next few years than were adults
who maintained their weight or gained weight. Weight loss
due to underlying disease and/or sarcopenia has been discussed as part of the obesity paradox, and weight loss in the
elderly and in patients suffering heart failure has been found
to predict mortality (20–22). Previous studies have also
shown involuntary weight loss to be associated with higher
1.18, 1.81
Model 2c
Weight loss
Weight
maintenance
Abbreviations: CI, confidence interval; HR, hazard ratio.
a
Weight loss (<−3%), weight maintenance (±3%), small weight
gain (>3–<10%), moderate weight gain (≥10–<30%), or large weight
gain (≥30%).
b
The long-term analysis adjusted for smoking status at age 25
years, as well as for race/field center, age, gender, education, smoking
status, alcohol consumption, physical activity, and height at examination 1; the short-term analysis adjusted for race/field center, age,
gender, education, and height at examination 1 and for time-dependent
alcohol consumption, physical activity, and smoking status.
c
Model 2 included the model 1 covariates plus body mass index
at age 25 years for the long-term analysis and time-dependent body
mass index for the short-term analysis.
mortality (23, 24), while voluntary, controlled weight loss
lowered mortality (24, 25).
Am J Epidemiol. 2013;178(2):239–248
Weight Change, Coronary Heart Disease, and Stroke 245
Figure 2. Hazard ratios for incident coronary heart disease according to long- and short-term weight change in the Atherosclerosis Risk in
Communities Study, 1987–2009. Long-term weight change was analyzed using 3-knot spline models with adjustment for smoking status at age
25 years and for race/field center, age, gender, education, smoking status, alcohol consumption, physical activity, and height at examination 1, as
well as with and without adjustment for body mass index (BMI; weight (kg)/height (m)2) at age 25 years. Short-term weight change was analyzed
using linear models with adjustment for race/field center, age, gender, education, and height at examination 1 and for time-dependent alcohol
consumption, physical activity, and smoking status, as well as with and without adjustment for BMI at the beginning of the weight-change interval.
To our knowledge, no previous study has investigated the
association between short-term weight loss and immediate
CVD risk. In fact, the literature on weight change over a
brief time period (e.g., 3–6 years) and incident CVD is
limited in general. Hamm et al. (26) used self-reported
weights at several time points between the ages of 20 and 40
years to examine patterns of 25-year fatal CHD risk over
segments of weight change averaging 5 years in duration.
Compared with men who maintained their weight, those
who consistently gained weight did not experience elevated risk; conversely, weight gain followed by weight loss
doubled the risk. Two additional studies did not find associations between weight loss over 3–10 years and CHD, possibly because the investigators did not restrict cases to those
that occurred immediately after the weight change but
included all cases arising during 16–17 years of follow-up
(8, 27). These findings illustrate the complex nature of the
relationship between timing of weight changes during adulthood and CVD risk.
Am J Epidemiol. 2013;178(2):239–248
Our study used serial measurements of body weight
during mid-adulthood to examine associations between
short-term weight change and CVD risk. Nevertheless, our
examination of long-term weight change used recall of
weight at age 25 years by participants who were middleaged. We have shown that bias in recall of past weight is
smaller than that seen in recall of current weight and that
recalled weight from the remote past (28 years prior) is
highly correlated with weight measured at that time (r > 0.8)
(28–30). In addition, we are aware that end-digit preference
and rounding tends to occur with recall. In our sample, 77%
of participants reported a weight at age 25 years ending in 0
or 5. However, a study using data from the Second National
Health and Nutrition Examination Survey (NHANES II;
1976–1980) showed that the reporting error between people
with end-digit preference and those without end-digit preference was not statistically significantly different, indicating
that rounding was random in all gender and BMI subgroups,
except that severely overweight (BMI ≥32.3) women who
246 Stevens et al.
Figure 3. Hazard ratios for incident ischemic stroke according to long- and short-term weight change in the Atherosclerosis Risk in Communities
Study, 1987–2009. Long-term weight change was analyzed using linear models with adjustment for smoking status at age 25 years and for race/
field center, age, gender, education, smoking status, alcohol consumption, physical activity, and height at examination 1, as well as with and
without adjustment for body mass index (BMI; weight (kg)/height (m)2) at age 25 years. Short-term weight change was analyzed using linear
models with adjustment for race/field center, age, gender, education, and height at examination 1 and for time-dependent alcohol consumption,
physical activity, and smoking status, as well as with and without adjustment for BMI at the beginning of the weight-change interval.
recalled their weight with a 5 or 0 digit preference seemed to
be rounding down (31). Nonetheless, severely overweight
women with end-digit preference comprised only 1% of our
study sample. In our study, the average recalled weight at
age 25 years was 165.3 pounds (75.0 kg) (standard deviation
(SD), 26.5 (12.0)) in men and 128.1 pounds (58.2 kg) (SD,
21.8 (9.9)) in women, which is comparable to the average
weight reported in NHANES II for the age groups 18–24
years (162.7 pounds (73.9 kg) [SD, 28.0 (12.7)] in men and
133.6 pounds (60.7 kg) [SD, 26.3 (11.9)] in women) and
25–34 years (173.4 pounds (78.7 kg) [SD, 30.1 (13.7)] in
men and 141.6 pounds (64.3 kg) [SD, 33.0 (15.0)] in
women) (32). Nevertheless, measured weights are preferred. Another limitation was that information on dieting
was self-reported and only available at 2 examinations. Furthermore, the potential of residual confounding through
smoking and comorbid conditions needs to be considered.
However, we carefully adjusted our long-term models for
smoking status at age 25 years and at examination 1 and
adjusted our short-term models for time-dependent smoking
status. In addition, we excluded events that occurred within
3 years after examination 1 in the long-term weight-change
analysis to avoid inclusion of events due to preexisting
illness (33).
Strengths of our study include the use of carefully adjudicated assessments of CHD and ischemic stroke. We adjusted
our analyses for BMI because it has been shown that weight
change is related to initial weight (27, 34). Further, we
attempted to distinguish intentional weight loss from unintentional weight loss over the short term.
Our results support the avoidance of weight gains greater
than 2.7% between early and middle adulthood, since such
gains are associated with increased risks of CHD and stroke.
Because the majority of adults in the United States gain
more than 10% of their body weight during this period (35),
there is an urgent need for effective interventions to prevent
weight gain during this high-risk period. Additional clinical
implications from this work point to a need for increased
Am J Epidemiol. 2013;178(2):239–248
Weight Change, Coronary Heart Disease, and Stroke 247
attention to unexplained, short-term weight loss. In middleaged adults, this weight loss could be a warning sign of an
impending cardiovascular event. Studies confirming these
estimates and elucidating potential mechanisms are needed.
ACKNOWLEDGMENTS
Author affiliations: Department of Nutrition, Gillings
School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina (June
Stevens, Eva Erber, Kimberly P. Truesdale, Chin-Hua
Wang); Department of Epidemiology, Gillings School of
Global Public Health, University of North Carolina at
Chapel Hill, Chapel Hill, North Carolina (June Stevens);
and Department of Biostatistics, Gillings School of Global
Public Health, University of North Carolina at Chapel Hill,
Chapel Hill, North Carolina (Jianwen Cai).
The Atherosclerosis Risk in Communities Study is carried
out as a collaborative study supported by National Heart,
Lung, and Blood Institute grants HHSN268201100005C,
HHSN268201100006C, HHSN268201100007C, HHSN
268201100008C,
HHSN268201100009C,
HHSN
268201100010C, HHSN268201100011C, and HHSN
268201100012C. This secondary data analysis was supported by a grant from the National Heart, Lung, and Blood
Institute (grant 1 RC1 HL099429Z).
We thank the staff of the Atherosclerosis Risk in Communities Study for their important contributions.
The results on ischemic stroke were presented in poster
form at the American Heart Association Epidemiology and
Prevention/Nutrition, Physical Activity and Metabolism 2011
Scientific Sessions in Atlanta, Georgia, on March 22–25, 2011.
The results on coronary heart disease were presented orally
and in poster form at the American Society for Nutrition
Scientific Sessions and Annual Meeting, held at the Experimental Biology 2011 conference in Washington, DC, on
April 9–13, 2011.
All authors had full access to all of the data in the study
and take responsibility for the integrity of the data and the
accuracy of the data analysis.
Conflict of interest: none declared.
REFERENCES
1. Heron M, Hoyert DL, Murphy SL, et al. Deaths: final data for
2006. Natl Vital Stat Rep. 2009;57(14):1–134.
2. Brown JR, O’Connor GT. Coronary heart disease and
prevention in the United States. N Engl J Med. 2010;362(23):
2150–2153.
3. Bogers RP, Bemelmans WJ, Hoogenveen RT, et al.
Association of overweight with increased risk of coronary
heart disease partly independent of blood pressure and
cholesterol levels: a meta-analysis of 21 cohort studies
including more than 300 000 persons. Arch Intern Med.
2007;167(16):1720–1728.
Am J Epidemiol. 2013;178(2):239–248
4. Strazzullo P, D’Elia L, Cairella G, et al. Excess body weight
and incidence of stroke: meta-analysis of prospective studies
with 2 million participants. Stroke. 2010;41(5):e418–e426.
5. Owen CG, Whincup PH, Orfei L, et al. Is body mass index
before middle age related to coronary heart disease risk in later
life? Evidence from observational studies. Int J Obes (Lond).
2009;33(8):866–877.
6. Willett WC, Manson JE, Stampfer MJ, et al. Weight, weight
change, and coronary heart disease in women. Risk within the
‘normal’ range. JAMA. 1995;273(6):461–465.
7. Rimm EB. Body size and fat distribution as predictors of
coronary heart disease among middle-aged and older US men.
Am J Epidemiol. 1995;141(12):1117–1127.
8. Galanis DJ, Harris T, Sharp DS, et al. Relative weight, weight
change, and risk of coronary heart disease in the Honolulu
Heart Program. Am J Epidemiol. 1998;147(4):379–386.
9. Rosengren A, Wedel H, Wilhelmsen L. Body weight and
weight gain during adult life in men in relation to coronary
heart disease and mortality: a prospective population study.
Eur Heart J. 1999;20(4):269–277.
10. Chei CL, Iso H, Yamagishi K, et al. Body mass index and
weight change since 20 years of age and risk of coronary heart
disease among Japanese: the Japan Public Health CenterBased Study. Int J Obes (Lond). 2008;32(1):144–151.
11. Yarnell JW, Patterson CC, Thomas HF, et al. Comparison of
weight in middle age, weight at 18 years, and weight change
between, in predicting subsequent 14 year mortality and
coronary events: Caerphilly Prospective Study. J Epidemiol
Community Health. 2000;54(5):344–348.
12. Rexrode KM, Hennekens CH, Willett WC, et al. A prospective
study of body mass index, weight change, and risk of stroke in
women. JAMA. 1997;277(19):1539–1545.
13. The ARIC Investigators. The Atherosclerosis Risk in
Communities (ARIC) Study: design and objectives. Am J
Epidemiol. 1989;129(4):687–702.
14. National Heart, Lung, and Blood Institute. Assessing Your Weight
and Health Risk. Bethesda, MD: National Heart, Lung, and
Blood Institute; 2012. (http://www.nhlbi.nih.gov/health/public/
heart/obesity/lose_wt/risk.htm). (Accessed November 13, 2012).
15. Stevens J, Truesdale K, McClain J, et al. The definition of
weight maintenance. Int J Obes. 2006;30(3):391–399.
16. Baecke J, Burema J, Frijters J. A short questionnaire for the
measurement of habitual physical activity in epidemiological
studies. Am J Clin Nutr. 1982;36(5):936–942.
17. White AD, Folsom AR, Chambless LE, et al. Community
surveillance of coronary heart disease in the Atherosclerosis
Risk in Communities (ARIC) Study: methods and initial two
years’ experience. J Clin Epidemiol. 1996;49(2):223–233.
18. Cox DR. Regression models and life tables [with discussion].
J R Stat Soc B. 1972;34(2):187–220.
19. Witte JS, Greenland S. A nested approach to evaluating
dose-response and trend. Ann Epidemiol. 1997;7(3):
188–193.
20. Strandberg TE, Strandberg AY, Salomaa VV, et al. Explaining
the obesity paradox: cardiovascular risk, weight change, and
mortality during long-term follow-up in men. Eur Heart J.
2009;30(14):1720–1727.
21. Oreopoulos A, Padwal R, Kalantar-Zadeh K, et al. Body mass
index and mortality in heart failure: a meta-analysis. Am Heart
J. 2008;156(1):13–22.
22. Miller SL, Wolfe RR. The danger of weight loss in the elderly.
J Nutr Health Aging. 2008;12(7):487–491.
23. Yaari S, Goldbourt U. Voluntary and involuntary weight
loss: associations with long term mortality in 9,228
middle-aged and elderly men. Am J Epidemiol. 1998;148(6):
546–555.
248 Stevens et al.
24. Gregg EW, Gerzoff RB, Thompson TJ, et al. Intentional
weight loss and death in overweight and obese U.S. adults
35 years of age and older. Ann Intern Med. 2003;138(5):
383–389.
25. Williamson DF, Thompson TJ, Thun M, et al. Intentional
weight loss and mortality among overweight individuals with
diabetes. Diabetes Care. 2000;23(10):1499–1504.
26. 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.
27. Iribarren C, Sharp DS, Burchfiel CM, et al. Association of
weight loss and weight fluctuation with mortality among
Japanese American men. N Engl J Med. 1995;333(11):
686–692.
28. Stevens J, Keil JE, Waid LR, et al. Accuracy of current, 4-year,
and 28-year self-reported body weight in an elderly
population. Am J Epidemiol. 1990;132(6):1156–1163.
29. Troy LM, Hunter DJ, Manson JE, et al. The validity of
recalled weight among younger women. Int J Obes Relat
Metab Disord. 1995;19(8):570–572.
30. Kyulo NL, Knutsen SF, Tonstad S, et al. Validation of recall
of body weight over a 26-year period in cohort members of
the Adventist Health Study 2. Ann Epidemiol. 2012;22
(10):744–746.
31. Rowland ML. Self-reported weight and height. Am J Clin
Nutr. 1990;52(6):1125–1133.
32. Najjar MF, Rowland M. Anthropometric reference data and
prevalence of overweight, United States, 1976–80. Vital
Health Stat 11. 1987;(238):1–73.
33. Stevens J, Juhaeri, Cai J. Changes in body mass index prior to
baseline among participants who are ill or who die during the
early years of follow-up. Am J Epidemiol. 2001;153(10):
946–953.
34. Glymour MM, Weuve J, Berkman LF, et al. When is baseline
adjustment useful in analyses of change? An example with
education and cognitive change. Am J Epidemiol. 2005;
162(3):267–278.
35. Truesdale KP, Stevens J, Lewis CE, et al. Changes in risk
factors for cardiovascular disease by baseline weight status in
young adults who maintain or gain weight over 15 years: the
CARDIA study. Int J Obes. 2006;30(9):1397–1407.
Am J Epidemiol. 2013;178(2):239–248