Abstract

Overweight and abdominal obesity increase mortality risk, although the risk may be mediated by traditional cardiac risk factors. The authors assessed the association of baseline measures, change in overall body weight and abdominal obesity (waist circumference), and weight and waist circumference cycling with total mortality among postmenopausal women with known heart disease. They used data from 2,739 US women who participated in the Heart and Estrogen/progestin Replacement Study between 1993 and 2001. Over 6.8 years of follow-up, 498 women died. In adjusted Cox models that included either baseline waist circumference or body mass index (BMI), each was associated with mortality. However, after further adjustment for diabetes, hypertension, and lipoproteins, these associations disappeared. In models including both waist circumference and BMI, larger waist circumference (hazard ratio = 1.40 per standard deviation, 95% confidence interval: 1.16, 1.68) was associated with increased risk and higher BMI (hazard ratio = 0.81 per standard deviation, 95% confidence interval: 0.67, 0.97) was associated with decreased risk of total mortality, independent of cardiac risk factors. Weight and waist circumference cycling were not associated with mortality. Results show that both BMI and waist circumference are associated with mortality among postmenopausal women with established heart disease, but waist circumference may be more important than BMI, and their effects may be largely mediated by other cardiac risk factors.

Received for publication March 24, 2003; accepted for publication June 18, 2003.

Observational studies suggest that overall obesity is an independent risk factor for coronary heart disease (CHD) and total mortality (13). However, some of the association of obesity with CHD risk can be attributed to effects of obesity on CHD risk factors such as hypertension, diabetes, and hyperlipidemia (4). US guidelines for the management of CHD and its risk factors (57) do not include a measure of body weight as an independent risk factor for CHD (8).

Upper body fat distribution was first shown to be associated with diabetes, hypertension, gout, and atherosclerosis by Jean Vague in 1947 (9). Since then, numerous studies have documented that abdominal obesity, compared with overall adiposity as estimated by body mass index (BMI), is more strongly associated with both CHD risk factors and CHD events (1018).

In epidemiologic studies, overweight and obesity are usually defined by using BMI (weight (kg)/height (m)2) as a surrogate measure of overall body fat (19). Abdominal adiposity is often quantified by waist circumference, a surrogate for intraabdominal fat (20, 21). Few studies have examined the effect of changes in adiposity and fat distribution and fluctuations in weight and waist circumference on total or CHD mortality.

We examined data from 2,739 postmenopausal women with documented CHD enrolled in the Heart and Estrogen/progestin Replacement Study (HERS) and the subsequent HERS II cohort study who were followed for a mean of 7 years between 1993 and 2001. Our goal was to determine whether baseline measures, weight changes (weight or waist circumference), or weight and waist circumference cycling were associated with CHD and total mortality.

MATERIALS AND METHODS

Study setting, participants, and design

The design, methods, baseline characteristics (22), and main findings (23) of HERS have been published previously. Briefly, this randomized, blinded, placebo-controlled trial evaluated daily use of 0.625-mg conjugated estrogens plus 2.5-mg medroxyprogesterone acetate for the prevention of coronary events in postmenopausal women with established CHD. The trial enrolled 2,763 women at 20 outpatient clinical centers in the United States and followed participants for a mean of 4.1 years. No significant differences were found in the combined primary outcome of nonfatal myocardial infarction and CHD death between women assigned to hormone therapy and those assigned to placebo (23). After completing the trial, women who were enrolled in HERS were invited to enroll in a prospective cohort study for an additional 2.7 years (HERS II). Of the 2,510 women alive at the HERS trial closeout, 2,321 (92.5 percent) enrolled in HERS II (24). After women who died between HERS closeout and HERS II enrollment were included and those with a baseline BMI of <18.5 kg/m2 were excluded, a total of 2,359 women contributed to the mortality analysis from HERS II. Contact information was available at closeout for more than 99 percent of the women enrolled in HERS II. After 6.8 years of follow-up, hormone therapy did not reduce the risk of cardiovascular events in women with established CHD (24).

Data collection

Baseline characteristics were assessed. Included were age, race or ethnicity, education, smoking habits (current, former, or never), alcohol consumption (no. of drinks per week), and exercise or walking activity (participation in an exercise program or walking activity more than three times per week).

Body weight, height, waist, hip circumference, and systolic and diastolic blood pressure were measured at baseline and at each annual visit during the HERS trial, but not thereafter. Weight was measured in kilograms by using a standard balance beam scale; height was measured in centimeters by using the height rod attached to the standard balance beam scale or a stadiometer. The scale was calibrated monthly at each clinical center. Waist and hip circumference were measured by using a flexible tape measure. Two consecutive waist measurements were taken at the level between the iliac crests and lower ribs that had the minimum circumference, usually at the level of the navel. Hip circumference was measured twice at a horizontal level over the greatest posterior extension of the buttocks. BMI and waist-hip ratio were calculated. Our main predictor variables were baseline BMI, change in weight, baseline waist circumference, and change in waist circumference. We chose to include change in weight rather than change in BMI since subsequent measurements of height loss due to bone loss may systematically differ between women in hormone therapy and placebo groups.

At baseline, participants had morning venous blood sampled after a requested 12-hour fast. The Lipoprotein Analytical Laboratory at the Johns Hopkins Hospital (Baltimore, Maryland) analyzed fasting blood for total cholesterol, high density lipoprotein cholesterol, and triglycerides. Low density lipoprotein cholesterol was estimated from the Friedewald equation (25). SmithKline Beecham Clinical Laboratory (Van Nuys, California) assayed fasting serum glucose and creatinine.

Diabetes was defined by self-report of a physician diagnosis, use of a diabetic medication, or a fasting plasma glucose level of ≥126 mg/dl at baseline or follow-up. Hypertension was classified as having a systolic blood pressure of >140 mmHg, having a diastolic blood pressure of >90 mmHg, or using an antihypertensive medication. Creatinine clearance was estimated from the Cockroft-Gault equation (26).

Outcome ascertainment

CHD mortality and total mortality were our two main outcome measures. CHD death was adjudicated by the HERS Coordinating Center if the participant experienced a fatal documented myocardial infarction, sudden death within 1 hour of onset of symptoms, an unobserved out-of-hospital death in the absence of another known cause, or death attributed to congestive heart failure or a coronary revascularization procedure. Total mortality was confirmed by questioning next of kin, searching the Social Security Death Index, and retrieving death certificates. At the end of HERS II, mortality follow-up of the entire cohort was 94.7 percent complete.

Statistical analysis

We evaluated the correlation between all measures of baseline and change in adiposity by using Pearson’s correlation coefficients. We calculated age-adjusted death rates for CHD and total mortality as death per 100 woman-years. The independent effects of total and abdominal obesity on risk of CHD death and total mortality were assessed by using multivariate Cox proportional hazards models. Hazard ratios for the baseline, change measures, and cycling variables were per standard deviation of the respective baseline obesity measure. In most analyses, we adjusted for assignment to hormone therapy because it was associated with small but statistically significant reductions in weight and waist circumference. We also adjusted for known CHD risk factors, including age, non-White race/ethnicity, and current smoking, and for seven comorbidity variables associated with mortality risk: self-report of poor/fair health, creatinine clearance of ≤40 ml/minute, hospitalization for congestive heart failure, New York Heart Association (NYHA) class I–IV congestive heart failure, two or more prior myocardial infarctions, any cancer diagnosis, and self-report of other serious medical condition at the baseline examination. We also adjusted for use of medications that could potentially confound the relation between obesity and mortality (aspirin, statins, beta blockers, and angiotensin-converting enzyme inhibitors). We did not adjust for diabetes, hypertension, and lipid levels in our primary analyses since these variables may mediate the association between obesity and mortality. In subsequent models, we evaluated these potential mediators as time-dependent covariates. Because the two baseline obesity measures were correlated, we used score residuals to assess influence and verify the stability of our results.

Changes since baseline in each obesity measure were modeled as time-dependent covariates. To avoid confounding of changes in obesity by changes in comorbidities, all co-variates except age, race/ethnicity, assignment to hormone therapy, and other serious medical conditions were modeled as time-dependent covariates. Because changes in weight and waist circumference may result from or be the consequence of an illness, the time-dependent covariates for these change measures were evaluated for the interval from baseline to the most recent annual visit that occurred at least 2 years before each outcome. This is a conservative measure of risk that does not allow the changes in weight or waist circumference that occurred closest to death (those most likely due to underlying disease processes associated with mortality) to influence the risk estimates. In contrast, other time-dependent covariates were evaluated by using the most recent available risk factor information. Although we used continuous measures of weight and waist circumference change in our multivariate models, we also categorized change as “loss” for those in the lowest quartile of the change measure, “no change” for those in the 25th–75th percentile of change, and “gain” for those in the highest quartile of change.

We also investigated the effect of weight and waist circumference cycling. All HERS participants had a baseline and as many as five annual measurements of weight and waist circumference. We measured fluctuation by using the root mean-squared error from a linear regression of each woman’s measurements over time since random assignment. Only women for whom we had at least four measurements were included in this analysis, and the number of measurements, inversely associated with risk, was included as a covariate in the analysis. Since the cycling measurements were performed during the HERS trial, we analyzed outcomes that occurred only during the HERS II follow-up period. The effects of weight and waist circumference fluctuation were evaluated in separate Cox models that adjusted for the CHD risk factors mentioned above. We examined models with and without adjustment for baseline measures and change in weight and waist circumference.

Finally, we tested whether the associations with baseline measures and change in weight and waist circumference varied across subgroups defined by baseline and time-dependent covariates. Included were race/ethnicity, current smoking, physical activity, assignment to hormone therapy, baseline BMI or waist circumference category, and presence of one or more of the seven comorbidities.

We excluded the 24 women whose BMI was <18.5 kg/m2 since they were at very high risk of mortality. One third of these women died during follow-up, and six of the eight deaths occurred during the first 2 years of follow-up.

We used SAS, version 8.2 (SAS Institute, Inc., Cary, North Carolina) and S-PLUS, version 6.1 (Insightful Corporation, Seattle, Washington) software for our analyses.

RESULTS

The mean age of the 2,739 HERS participants whose BMI was ≥18.5 kg/m2 at baseline was 67 years. Participants were mostly White and relatively sedentary (table 1). One quarter had known diabetes at baseline, and 90 percent had hypertension. According to national criteria (27), 72 percent of the women were overweight (BMI 25.0–29.9 kg/m2) or obese (BMI ≥30 kg/m2) at baseline. Mean waist circumference was 92 cm, and waist-hip ratio was 0.87; 58 percent of the women were in a high-risk category for cardiovascular events (waist circumference >88 cm). A majority of overweight (57 percent) and obese (96 percent) women had a waist circumference of >88 cm (figure 1). During HERS, on average, body weight decreased 0.7 kg (standard deviation (SD), 3.7; range, –30.1 to 23.9) and waist circumference decreased 0.5 cm (SD, 4.6; range, –28.0 to 25.0). For a majority of the women (56 percent), no significant change in weight or waist circumference occurred from baseline to year 1 (figure 2). Of 686 women who lost weight, defined by the lowest quartile of weight change, 56 percent also had a significant loss of waist circumference. Mean variability (root mean-squared error) for weight was 2.03 kg (SD, 1.52) and for waist circumference was 4.42 cm (SD, 3.05).

As expected, the two measures of overall obesity (BMI and weight) and of abdominal obesity (waist circumference and waist-hip ratio) were highly correlated (table 2). We chose to focus our analyses on baseline BMI and baseline waist circumference; results using baseline waist-hip ratio instead of waist circumference were not different. For our change measures, we analyzed change in weight instead of change in BMI since change in BMI may result from height loss.

At baseline, mean BMI and waist circumference were similar among women randomly assigned to hormone therapy and to placebo. Hormone-treated women experienced slightly, but significantly more weight loss (–0.8 kg (SD, 6.3) vs. –0.3 kg (SD, 6.3), p = 0.03; BMI: –0.1 kg/m2 (SD, 2.5) vs. 0.1 kg/m2 (SD, 2.5), p = 0.02) and decreased abdominal obesity (waist-hip ratio: –0.01 (SD, 0.07) vs. 0.0 (SD, 0.1), p = 0.01; waist circumference: –0.6 cm (SD, 7.6) vs. 0.2 cm (SD, 7.0), p = 0.003) during follow-up compared with women assigned to placebo, but the average changes in each measure were small.

A total of 498 women died during the 6.8 years of follow-up; 254 (51 percent) died of CHD. Table 3 shows the age-adjusted incidence rates per 100 woman-years for CHD death and total mortality by baseline BMI and waist circumference strata. For BMI, the association with both outcomes was U-shaped; however, after adjustment for race/ethnicity, current smoking, assignment to hormone therapy during HERS, and seven high-risk comorbidity variables, baseline BMI was linearly and inversely associated with risk of both CHD and total mortality.

Figure 3 displays the age-adjusted rate for total mortality by baseline BMI and waist circumference category. For every category of BMI, mortality rates were higher for women whose waist circumference was >88 cm compared with those with a smaller waist circumference. For every category of waist circumference, women of normal BMI had higher mortality rates than those who were overweight or obese. Women of normal BMI who had a waist circumference of >88 cm had the highest mortality rate (18/74, 4.1 deaths per 100 woman-years).

We evaluated the relation of baseline BMI alone, waist circumference alone, and BMI and waist circumference together with each mortality outcome (table 4). In staged models that were age adjusted and then adjusted for multiple potentially confounding covariates, both higher levels of BMI alone and higher levels of waist circumference alone were significantly associated with increased risk of CHD mortality, with a similar trend regarding the associations for total mortality. In models that included both BMI and waist circumference together, higher levels of BMI were associated with a lower risk of total mortality (hazard ratio (HR) = 0.81, 95 percent confidence interval (CI): 0.67, 0.97), while higher levels of waist circumference were associated with an increased risk (HR = 1.40, 95 percent CI: 1.16, 1.68). Adding potential mediators of the association between obesity and mortality—diabetes, hypertension, lipoprotein(a), high density lipoprotein cholesterol, and low density lipoprotein cholesterol—as time-dependent covariates to the multivariable model attenuated the association between waist circumference and both outcomes but did not change the strength of the association between BMI and either outcome. In this model, higher BMI remained significantly associated with a 20 percent lower risk of total mortality, and larger waist circumference was associated with a 22 percent increased risk. Diabetes was the principal mediator of the association of larger waist circumference with CHD and total mortality.

The association of baseline BMI and waist circumference with CHD and total mortality did not vary by race/ethnicity, smoking status, physical activity, early or late mortality, presence of high-risk comorbidity, hormone therapy assignment, or baseline BMI. Exclusion of 34 total mortality outcomes that were smoking related (due to chronic obstructive pulmonary disease or lung cancer) or accidental deaths also did not substantially change the effect of baseline BMI and waist circumference on total mortality. However, the association of baseline BMI with total mortality risk did vary by baseline waist circumference (HR = 0.64, 95 percent CI: 0.47, 0.88 for a waist circumference of <80 cm; HR = 0.66, 95 percent CI: 0.37, 1.21 for a waist circumference of 80–88 cm; and HR = 1.04, 95 percent CI: 0.89, 1.21 for a waist circumference of >88 cm, p for interaction 0.04). Women whose waist circumference was <88 cm had a decreased risk of mortality with increasing BMI. We did not observe this interaction with CHD mortality.

In separate models, we evaluated the risk associated with changes in BMI and waist circumference. We found that increases in waist circumference and weight loss increased the risk for CHD and total mortality. However, these associations were evident for only those women assigned to hormone therapy, not those assigned to placebo (HR for CHD death from Δwaist circumference = 0.90 (95 percent CI: 0.45, 1.82) for women assigned to placebo and 3.06 (95 percent CI: 1.61, 5.80) for women assigned to hormone therapy; p for interaction = 0.01). Furthermore, the elevations or decreases in risk for women assigned to hormone therapy were evident only for the extreme 5 percent of subjects (figure 4). A similar interaction with hormone therapy also existed for total mortality. This apparent harmful effect of hormone therapy on weight loss and waist circumference increase for CHD and total mortality was unexplained by diabetes, hypertension, low density lipoprotein cholesterol, high density lipoprotein cholesterol, triglycerides, or lipoprotein(a).

Neither weight cycling nor waist circumference cycling during HERS was independently associated with CHD or total mortality in the subsequent 2.7 years of follow-up. After adjustment for baseline measures and change in weight and waist circumference and the potential confounding covariates, the hazard ratio of CHD mortality for weight cycling was 1.00 (95 percent CI: 0.87, 1.14), and the risk for waist circumference cycling was 1.00 (95 percent CI: 0.93, 1.08). We found similar risk estimates for total mortality with weight and waist cycling.

DISCUSSION

In this cohort of older postmenopausal women with CHD, higher levels of baseline waist circumference were associated with a 22 percent increased risk of mortality after adjusting for BMI and traditional cardiac risk factors. Additionally, after adjustment for waist circumference, higher levels of baseline BMI were associated with a 20 percent decreased risk of total mortality. Only women whose waist circumference was <88 cm had a decreased risk of mortality with increasing BMI. Weight loss and increases in waist circumference were significant predictors of death only for those women randomized to hormone therapy, an observation unexplained by variables associated with the metabolic syndrome. Fluctuations in weight and waist circumference were not associated with CHD or total mortality.

Prior investigations have found BMI to be both directly (3, 28, 29) or inversely (3033) associated with mortality, to have no independent association (3440), or to have a U- or J-shaped association (4143). Possible explanations for a lack of consistency and reproducibility of the relation between BMI and death may be due to the covariance of BMI with other cardiac risk factors, unmeasured confounders, or misclassification bias from use of surrogate markers for obesity (34, 44, 45). Two studies that examined the effects of fat mass and fat-free mass components of BMI on mortality both found that the increased risk of mortality with low BMI was explained by low fat-free mass; after adjustment for fat-free mass, there was a positive association of fat mass and BMI with mortality (46, 47). These two studies did not assess the role of traditional cardiac risk factors in explaining the association between BMI or fat mass and mortality.

Measures of central obesity, compared with total body fat, are more consistently related to CHD events (16, 48, 49). Both waist circumference and waist-hip ratio are surrogate markers for visceral adiposity, a portion of the abdominal fat depot closely related to dyslipidemia and changes in glucose and insulin homeostasis. Strong epidemiologic and pathophysiologic evidence links visceral adiposity with insulin resistance and diabetes (11, 50, 51). Dyslipidemia, hyperinsulinemia, and decreased fibrinolysis are some of the factors described that may mediate the adverse cardiac outcomes closely linked to visceral obesity (11, 52). In this study, diabetes appeared to mediate some of the association of baseline waist circumference with both CHD and total mortality.

We confirmed an independent linear association of baseline waist circumference with mortality after controlling for BMI. The protective effect of increased baseline BMI for both outcomes was observed only after controlling for the effects of abdominal obesity. Since BMI measures total body mass, both fat and lean body mass, after adjusting for waist circumference, BMI may better represent the effect of lean body mass on mortality. This negative confounding may have been underappreciated in prior studies that did not adjust for measures of abdominal obesity.

We found that increases in waist circumference from baseline to the last measurement were strongly associated with mortality. However, this positive association was observed only for those women assigned to hormone therapy. This unexpected interaction was not explained by factors in the metabolic syndrome (diabetes, hypertension, and lipoproteins). This novel post hoc observation may be related to the unexpected excess of CHD in women assigned to hormone therapy (23, 53) and needs further investigation. We are not aware of other studies that have evaluated the role of changes in measures of abdominal obesity as risk factors for CHD or total mortality. Increasing waist circumference implicates an increase in the visceral fat depot, which may lead to metabolic derangements and further risk of CHD. This undesirable metabolic effect may be compounded by the proinflammatory or prothrombotic effects of hormone therapy.

Lastly, we found that weight cycling was not associated with an increased risk of CHD or total mortality. Others have reported increased mortality with weight cycling (5456). However, these studies included male subjects alone or healthy middle-aged women. We also found that greater variability in waist circumference over time was not associated with either CHD or total mortality. We are not aware of any other study that has examined the effect of central adiposity fluctuation. In this study, we observed higher variability for waist circumference than for weight. Visceral fat is mobilized more easily than peripheral or subcutaneous fat by diet and exercise and would be expected to lead to higher waist circumference fluctuation. Another possibility for higher waist circumference variability may be that measurements of waist circumference are more susceptible to error since waist circumference is more difficult than weight to measure accurately.

To our knowledge, our study is unique in that we examined the effect of baseline, an increase, a decrease, and cycling of measures of obesity on CHD and total mortality in postmenopausal women with established heart disease. We were able to control for multiple confounders and specifically examine the role of hormone therapy and diabetes in the observed associations. However, our study was limited regarding type of anthropometric measurements. Women’s weight, height, and waist and hip circumference were measured annually during the HERS trial, but they did not undergo clinical examinations during the HERS II follow-up study. Although BMI is an accepted surrogate marker for total body fat, its correlation with body fat is poorer in older adults (57). Optimally, total body fat is measured by underwater densitometry or by whole-body dual energy X-ray absorptiometry. Waist circumference has been found to be a closer surrogate marker of visceral adiposity than waist-hip ratio (20, 21), although more precise measurements of visceral fat are made by abdominal computed tomography or magnetic resonance imaging scans. It is possible that our use of surrogate markers for both overall and abdominal obesity were imprecise and led to some misclassification bias and residual confounding. Misclassification bias would be expected to obscure the associations of BMI and waist circumference with the mortality outcomes that we found.

In conclusion, among postmenopausal women with established CHD, higher levels of baseline waist circumference were associated with an increased risk of CHD and total mortality, whereas higher levels of baseline BMI were associated with a lower risk of the outcomes studied. Traditional cardiac risk factors did not fully explain the association between these baseline obesity measures and total mortality. Contrary to findings reported for men and for younger women, weight and waist circumference cycling were not associated with mortality. Our findings argue for the inclusion of both BMI and waist circumference as independent cardiac risk factors for predicting total mortality among women with established heart disease.

ACKNOWLEDGMENTS

HERS was funded by Wyeth-Ayerst Research, Radnor, Pennsylvania. The funding source had no role in the design, conduct, or analysis of this study nor a role in the decision to submit this paper for publication.

Dr. Kanaya was funded by University of California, San Francisco–Kaiser Building Interdisciplinary Research Careers in Women’s Health grant NIH 5 K12 AR47659. Drs. Grady and Vittinghoff received salary support via contracts with the University of California, San Francisco from Wyeth-Ayerst Research. Dr. Barrett-Connor has grant funding from Wyeth-Ayerst for HERS. Dr. Shlipak is funded by a Research Career Development Award from the Health Services Research and Development service of the US Department of Veterans Affairs. The research of Dr. Visser has been made possible by a fellowship of the Royal Netherlands Academy of Arts and Sciences.

Correspondence to Dr. Alka M. Kanaya, Women’s Health Clinical Research Center, 1635 Divisadero Street, Suite 600, San Francisco, CA 94115 (e-mail: alkak@itsa.ucsf.edu).

FIGURE 1. Baseline body mass index by waist circumference of women randomly assigned to placebo or hormone therapy in the Heart and Estrogen/progestin Replacement Study, United States, 1993–2001. For seven women, data for one of the two obesity variables were missing at the baseline visit.

FIGURE 1. Baseline body mass index by waist circumference of women randomly assigned to placebo or hormone therapy in the Heart and Estrogen/progestin Replacement Study, United States, 1993–2001. For seven women, data for one of the two obesity variables were missing at the baseline visit.

FIGURE 2. Numbers of women whose weight and waist circumference changed from baseline to year 1, Heart and Estrogen/progestin Replacement Study, United States, 1993–2001. Loss of weight or waist circumference was defined as being in the lowest quartile of change, no change was defined as being in the 25th–75th percentile of change, and gain of weight or in waist circumference was defined as being in the highest quartile of change. For waist circumference, loss represents –28.0 to –2.8 cm, no change represents –2.8 to 2.0 cm, and gain represents 2.0 to 25.0 cm.

FIGURE 2. Numbers of women whose weight and waist circumference changed from baseline to year 1, Heart and Estrogen/progestin Replacement Study, United States, 1993–2001. Loss of weight or waist circumference was defined as being in the lowest quartile of change, no change was defined as being in the 25th–75th percentile of change, and gain of weight or in waist circumference was defined as being in the highest quartile of change. For waist circumference, loss represents –28.0 to –2.8 cm, no change represents –2.8 to 2.0 cm, and gain represents 2.0 to 25.0 cm.

FIGURE 3. Age-adjusted total mortality rate by baseline body mass index and waist circumference, Heart and Estrogen/progestin Replacement Study, United States, 1993–2001.

FIGURE 3. Age-adjusted total mortality rate by baseline body mass index and waist circumference, Heart and Estrogen/progestin Replacement Study, United States, 1993–2001.

FIGURE 4. Association of change in weight and change in waist circumference with coronary heart disease mortality by placebo and hormone therapy assignment, Heart and Estrogen/progestin Replacement Study (HERS), United States, 1993–2001. From left to right, <6th percentile (most loss), 6th–25th percentile (some loss), 26th–75th percentile (no change; referent group), 76th–95th percentile (some gain), and >95th percentile (most gain). These percentiles correspond to the following ranges: for weight change, <6%: –30.1 to –6.7 kg, 6–25%: –6.6 to –2.4 kg, 26–75%: –2.3 to 1.1 kg, 76–95%: 1.2 to 4.4 kg, and >95%: 4.5 to 23.9 kg. For waist circumference change, <6%: –28.0 to –8.1 cm, 6–25%: –8.0 to –2.8 cm, 26–75%: –2.7 to 2.0 cm, 76–95%: 1.9 to 6.5 cm, and >95%: 6.6 to 25.0 cm. The models were adjusted for age, non-White ethnicity, current smoking, HERS II enrollment, and seven comorbidity variables, as time-dependent covariates where possible; p for trend values are noted.

FIGURE 4. Association of change in weight and change in waist circumference with coronary heart disease mortality by placebo and hormone therapy assignment, Heart and Estrogen/progestin Replacement Study (HERS), United States, 1993–2001. From left to right, <6th percentile (most loss), 6th–25th percentile (some loss), 26th–75th percentile (no change; referent group), 76th–95th percentile (some gain), and >95th percentile (most gain). These percentiles correspond to the following ranges: for weight change, <6%: –30.1 to –6.7 kg, 6–25%: –6.6 to –2.4 kg, 26–75%: –2.3 to 1.1 kg, 76–95%: 1.2 to 4.4 kg, and >95%: 4.5 to 23.9 kg. For waist circumference change, <6%: –28.0 to –8.1 cm, 6–25%: –8.0 to –2.8 cm, 26–75%: –2.7 to 2.0 cm, 76–95%: 1.9 to 6.5 cm, and >95%: 6.6 to 25.0 cm. The models were adjusted for age, non-White ethnicity, current smoking, HERS II enrollment, and seven comorbidity variables, as time-dependent covariates where possible; p for trend values are noted.

TABLE 1.

Baseline characteristics of the Heart and Estrogen/progestin Replacement Study cohort, United States, 1993–2001*

 Value 
 (n = 2,739) 
Age in years (mean (SD†)) 66.6 (6.6) 
Race/ethnicity: White (no. (%)) 2,430 (88.7) 
Smoking (no. (%))  
Current 352 (12.8) 
Former 1,343 (49.0) 
Never 1,044 (38.1) 
Physical activity ≥3 times per week (no. (%)) 1,056 (38.5) 
Diabetes‡ (no. (%)) 729 (26.6) 
Hypertension§ (no. (%)) 2,421 (88.4) 
Congestive heart failure (NYHA† class I-III) (no. (%)) 343 (12.5) 
≥2 prior myocardial infarctions (no. (%)) 141 (5.1) 
Low density lipoprotein cholesterol (mean (SD))  
mmol/liter 3.75 (0.98) 
mg/dl 145.1 (37.8) 
High density lipoprotein cholesterol (mean (SD))  
mmol/liter 1.30 (0.34) 
mg/dl 50.2 (13.1) 
Triglycerides (mean (SD))  
mmol/liter 0.21 ( 0.72) 
mg/dl 166.4 (63.6) 
Assignment to hormone therapy (no. (%)) 1,368 (49.9) 
Weight in kg (mean (SD)) 73.0 (14.5) 
Body mass index (mean (SD)) 28.7 (5.4) 
Waist-hip ratio (mean (SD)) 0.87 (0.08) 
Waist circumference in cm (mean (SD)) 91.9 (13.3) 
 Value 
 (n = 2,739) 
Age in years (mean (SD†)) 66.6 (6.6) 
Race/ethnicity: White (no. (%)) 2,430 (88.7) 
Smoking (no. (%))  
Current 352 (12.8) 
Former 1,343 (49.0) 
Never 1,044 (38.1) 
Physical activity ≥3 times per week (no. (%)) 1,056 (38.5) 
Diabetes‡ (no. (%)) 729 (26.6) 
Hypertension§ (no. (%)) 2,421 (88.4) 
Congestive heart failure (NYHA† class I-III) (no. (%)) 343 (12.5) 
≥2 prior myocardial infarctions (no. (%)) 141 (5.1) 
Low density lipoprotein cholesterol (mean (SD))  
mmol/liter 3.75 (0.98) 
mg/dl 145.1 (37.8) 
High density lipoprotein cholesterol (mean (SD))  
mmol/liter 1.30 (0.34) 
mg/dl 50.2 (13.1) 
Triglycerides (mean (SD))  
mmol/liter 0.21 ( 0.72) 
mg/dl 166.4 (63.6) 
Assignment to hormone therapy (no. (%)) 1,368 (49.9) 
Weight in kg (mean (SD)) 73.0 (14.5) 
Body mass index (mean (SD)) 28.7 (5.4) 
Waist-hip ratio (mean (SD)) 0.87 (0.08) 
Waist circumference in cm (mean (SD)) 91.9 (13.3) 

* Women whose body mass index (weight (kg)/height (m)2) was <18.5 kg/m2 were omitted from analysis.

† SD, standard deviation; NYHA, New York Heart Association.

‡ Defined by self-report of a physician diagnosis, use of a diabetic medication, or a fasting plasma glucose level of ≥126 mg/dl.

§ Defined by a systolic blood pressure of >140 mmHg, a diastolic blood pressure of >90 mmHg, or use of an antihypertensive medication.

TABLE 2.

Correlation between baseline and change measures of obesity, Heart and Estrogen/progestin Replacement Study, United States, 1993–2001*

 Weight Waist circumference Waist-hip ratio ΔBody mass index ΔWeight ΔWaist circumference ΔWaist-hip ratio 
Body mass index        
r value 0.92 0.84 0.27 –0.07 –0.07 0.01 0.04 
p value <0.001 <0.001 <0.001 <0.001 <0.001 0.70 0.03 
Weight        
r value  0.85 0.27 –0.04 –0.05 0.01 0.04 
p value  <0.001 <0.001 0.02 0.005 0.62 0.03 
Waist circumference        
r value   0.64 –0.04 –0.05 –0.14 –0.09 
p value   <0.001 0.05 0.006 <0.001 <0.001 
Waist-hip ratio        
r value    –0.03 –0.04 –0.21 –0.32 
p value    0.13 0.05 <0.001 <0.001 
ΔBody mass index        
r value     0.94 0.48 0.10 
p value     <0.001 <0.001 <0.001 
ΔWeight        
r value      0.49 0.09 
p value      <0.001 <0.001 
ΔWaist circumference        
r value       0.66 
p value       <0.001 
 Weight Waist circumference Waist-hip ratio ΔBody mass index ΔWeight ΔWaist circumference ΔWaist-hip ratio 
Body mass index        
r value 0.92 0.84 0.27 –0.07 –0.07 0.01 0.04 
p value <0.001 <0.001 <0.001 <0.001 <0.001 0.70 0.03 
Weight        
r value  0.85 0.27 –0.04 –0.05 0.01 0.04 
p value  <0.001 <0.001 0.02 0.005 0.62 0.03 
Waist circumference        
r value   0.64 –0.04 –0.05 –0.14 –0.09 
p value   <0.001 0.05 0.006 <0.001 <0.001 
Waist-hip ratio        
r value    –0.03 –0.04 –0.21 –0.32 
p value    0.13 0.05 <0.001 <0.001 
ΔBody mass index        
r value     0.94 0.48 0.10 
p value     <0.001 <0.001 <0.001 
ΔWeight        
r value      0.49 0.09 
p value      <0.001 <0.001 
ΔWaist circumference        
r value       0.66 
p value       <0.001 

* Pearson’s correlation coefficients were used.

TABLE 3.

Age-adjusted rates (and 95% confidence intervals) of coronary heart disease death and total mortality by baseline measures of overall and abdominal obesity, Heart and Estrogen/progestin Replacement Study, United States, 1993–2001*

 Coronary heart disease death(n = 254)  Total mortality(n = 498) 
  
 n/N Age-adjusted rate 95% CI†  n/N Age-adjusted rate 95% CI 
Body mass index         
18.5–21.9 20/224 1.4 0.4, 2.5  53/224 3.8 2.0, 5.5 
22.0–24.9 37/517 1.1 0.2, 2.0  85/517 2.5 1.2, 3.8 
25.0–29.9 90/1,051 1.4 0.4, 2.4  177/1,051 2.7 1.3, 4.2 
≥30 101/942 1.8 0.7, 2.9  175/942 3.1 1.6, 4.6 
Waist circumference (cm)        
≤80 34/543 0.9 0.1, 1.8  84/543 2.3 1.0, 3.7 
81–88 55/611 1.4 0.5, 2.4  102/611 2.7 1.2, 4.1 
>88 160/1,583 1.7 0.6, 2.8  305/583 3.2 1.7, 4.7 
 Coronary heart disease death(n = 254)  Total mortality(n = 498) 
  
 n/N Age-adjusted rate 95% CI†  n/N Age-adjusted rate 95% CI 
Body mass index         
18.5–21.9 20/224 1.4 0.4, 2.5  53/224 3.8 2.0, 5.5 
22.0–24.9 37/517 1.1 0.2, 2.0  85/517 2.5 1.2, 3.8 
25.0–29.9 90/1,051 1.4 0.4, 2.4  177/1,051 2.7 1.3, 4.2 
≥30 101/942 1.8 0.7, 2.9  175/942 3.1 1.6, 4.6 
Waist circumference (cm)        
≤80 34/543 0.9 0.1, 1.8  84/543 2.3 1.0, 3.7 
81–88 55/611 1.4 0.5, 2.4  102/611 2.7 1.2, 4.1 
>88 160/1,583 1.7 0.6, 2.8  305/583 3.2 1.7, 4.7 

* Rates are expressed as deaths/100 woman-years; the analysis excluded 24 women whose body mass index (weight (kg)/height (m)2) was <18.5 kg/m2.

† CI, confidence interval.

TABLE 4.

Risk of coronary heart disease mortality and total mortality by baseline body mass index and waist circumference, separately or together in models, Heart and Estrogen/progestin Replacement Study, United States, 1993–2001

 Body mass index* alone  Waist circumference (cm) alone  Both body mass index and waist circumference together 
 HR† (per SD†) 95% CI†  HR (per SD) 95% CI  HR (per SD) 95% CI 
CHD† mortality         
Age-adjusted 1.19 1.06, 1.35  1.32 1.16, 1.50  Body mass index: 0.86 0.68, 1.10 
       Waist circumference: 1.49 1.16, 1.91 
Multivariate model 1‡ 1.17 1.03, 1.33  1.28 1.12, 1.47  Body mass index: 0.86 0.67, 1.10 
       Waist circumference: 1.47 1.14, 1.89 
Multivariate model 2§ 1.01 0.88, 1.15  1.07 0.93, 1.24  Body mass index: 0.85 0.66, 1.10 
       Waist circumference: 1.23 0.95, 1.60 
Total mortality         
Age-adjusted 1.05 0.95, 1.15  1.14 1.04, 1.25  Body mass index: 0.78 0.65, 0.94 
       Waist circumference: 1.41 1.17, 1.68 
Multivariate model 1‡ 1.07 0.97, 1.18  1.16 1.05, 1.28  Body mass index: 0.81  0.67, 0.97 
       Waist circumference: 1.40 1.16, 1.68 
Multivariate model 2§ 0.94 0.85, 1.04  1.01 0.91, 1.12  Body mass index: 0.80,  0.66, 0.96 
       Waist circumference: 1.22 1.01, 1.48 
 Body mass index* alone  Waist circumference (cm) alone  Both body mass index and waist circumference together 
 HR† (per SD†) 95% CI†  HR (per SD) 95% CI  HR (per SD) 95% CI 
CHD† mortality         
Age-adjusted 1.19 1.06, 1.35  1.32 1.16, 1.50  Body mass index: 0.86 0.68, 1.10 
       Waist circumference: 1.49 1.16, 1.91 
Multivariate model 1‡ 1.17 1.03, 1.33  1.28 1.12, 1.47  Body mass index: 0.86 0.67, 1.10 
       Waist circumference: 1.47 1.14, 1.89 
Multivariate model 2§ 1.01 0.88, 1.15  1.07 0.93, 1.24  Body mass index: 0.85 0.66, 1.10 
       Waist circumference: 1.23 0.95, 1.60 
Total mortality         
Age-adjusted 1.05 0.95, 1.15  1.14 1.04, 1.25  Body mass index: 0.78 0.65, 0.94 
       Waist circumference: 1.41 1.17, 1.68 
Multivariate model 1‡ 1.07 0.97, 1.18  1.16 1.05, 1.28  Body mass index: 0.81  0.67, 0.97 
       Waist circumference: 1.40 1.16, 1.68 
Multivariate model 2§ 0.94 0.85, 1.04  1.01 0.91, 1.12  Body mass index: 0.80,  0.66, 0.96 
       Waist circumference: 1.22 1.01, 1.48 

* Weight (kg)/height (m)2.

† HR, hazard ratio; SD, standard deviation; CI, confidence interval; CHD, coronary heart disease.

‡ Model 1: adjusted for age, non-White race/ethnicity, current smoking, assignment to hormone therapy, self-report of poor/fair health, creatinine clearance of ≤40 ml/minute, hospitalization for congestive heart failure, New York Heart Association class I–IV congestive heart failure, ≥2 prior myocardial infarctions, any cancer diagnosis, self-report of other serious medical condition, and medication use: aspirin, statins, beta blockers, and angiotensin-converting enzyme inhibitors.

§ Model 2: adjusted for all covariates in model 1 and for potential mediators: diabetes, hypertension, lipoprotein(a), high density lipoprotein cholesterol, and low density lipoprotein cholesterol as time-dependent covariates.

References

1.
Hubert HB, Feinleib M, McNamara PM, et al. Obesity as an independent risk factor for cardiovascular disease: a 26-year follow-up of participants in the Framingham Heart Study.
Circulation
 
1983
;
67
:
968
–77.
2.
Manson JE, Willett WC, Stampfer MJ, et al. Body weight and mortality among women.
N Engl J Med
 
1995
;
333
:
677
–85.
3.
Lee IM, Manson JE, Hennekens CH, et al. Body weight and mortality: a 27-year follow-up of middle-aged men.
JAMA
 
1993
;
270
:
2823
–8.
4.
Van Itallie TB. Obesity: adverse effects on health and longevity.
Am J Clin Nutr
 
1979
;
32
:
2723
–33.
5.
Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III).
JAMA
 
2001
;
285
:
2486
–97.
6.
Arauz-Pacheco C, Parrott MA, Raskin P, et al. Treatment of hypertension in adults with diabetes.
Diabetes Care
 
2003
;
26(suppl 1)
:
S80
–2.
7.
Haffner SM, American Diabetes Association. Management of dyslipidemia in adults with diabetes.
Diabetes Care
 
2003
;
26(suppl 1)
:
S83
–6.
8.
Krauss RM, Winston M, Fletcher RN, et al. Obesity: impact of cardiovascular disease.
Circulation
 
1998
;
98
:
1472
–6.
9.
Vague J. La differenciation sexualle, facteur determinant des formes de l‘obesite. (In French).
Presse Med
 
1947
;
55
:
339
–40.
10.
Peiris AN, Sothmann MS, Hoffmann RG, et al. Adiposity, fat distribution, and cardiovascular risk.
Ann Intern Med
 
1989
;
110
:
867
–72.
11.
Kissebah AH. Intra-abdominal fat: is it a major factor in developing diabetes and coronary artery disease?
Diabetes Res Clin Pract
 
1996
;
30(suppl)
:
25
–30.
12.
Bertrais S, Balkau B, Vol S, et al. Relationships between abdominal body fat distribution and cardiovascular risk factors: an explanation for women’s healthier cardiovascular risk profile. The D.E.S.I.R. Study.
Int J Obes Relat Metab Disord
 
1999
;
23
:
1085
–94.
13.
Gray RS, Fabsitz RR, Cowan LD, et al. Relation of generalized and central obesity to cardiovascular risk factors and prevalent coronary heart disease in a sample of American Indians: the Strong Heart Study.
Int J Obes Relat Metab Disord
 
2000
;
24
:
849
–60.
14.
Ross SJ, Poehlman ET, Johnson RK, et al. Body fat distribution predicts cardiac risk factors in older female coronary patients.
J Cardiopulm Rehabil
 
1997
;
17
:
419
–27.
15.
Matsuzawa Y, Nakamura T, Shimomura I, et al. Visceral fat accumulation and cardiovascular disease.
Obes Res
 
1995
;
3(suppl 5)
:
645S
–7S.
16.
Lapidus L, Bengtsson C, Larsson B, et al. Distribution of adipose tissue and risk of cardiovascular disease and death: a 12 year follow up of participants in the population study of women in Gothenburg, Sweden.
BMJ (Clin Res Ed)
 
1984
;
289
:
1257
–61.
17.
Rexrode KM, Carey VJ, Hennekens CH, et al. Abdominal adiposity and coronary heart disease in women.
JAMA
 
1998
;
280
:
1843
–8.
18.
Fujimoto WY, Bergstrom RW, Boyko EJ, et al. Visceral adiposity and incident coronary heart disease in Japanese-American men: the 10-year follow-up results of the Seattle Japanese-American Community Diabetes Study.
Diabetes Care
 
1999
;
22
:
1808
–12.
19.
Obesity: preventing and managing the global epidemic. Report of a WHO consultation. Geneva, Switzerland: World Health Organization, 2000. (WHO technical report series, no. 894).
20.
Lean ME, Han TS, Morrison CE. Waist circumference as a measure for indicating need for weight management.
BMJ
 
1995
;
311
:
158
–61.
21.
Han TS, McNeill G, Seidell JC, et al. Predicting intra-abdominal fatness from anthropometric measures: the influence of stature.
Int J Obes Relat Metab Disord
 
1997
;
21
:
587
–93.
22.
Grady D, Applegate W, Bush T, et al. Heart and Estrogen/progestin Replacement Study (HERS): design, methods, and baseline characteristics.
Control Clin Trials
 
1998
;
19
:
314
–35.
23.
Hulley S, Grady D, Bush T, et al. Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women. Heart and Estrogen/progestin Replacement Study (HERS) Research Group.
JAMA
 
1998
;
280
:
605
–13.
24.
Grady D, Herrington D, Bittner V, et al. Cardiovascular disease outcomes during 6.8 years of hormone therapy: Heart and Estrogen/progestin Replacement Study follow-up (HERS II).
JAMA
 
2002
;
288
:
49
–57.
25.
Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge.
Clin Chem
 
1972
;
18
:
499
–502.
26.
Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine.
Nephron
 
1976
;
16
:
31
–41.
27.
Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults—The Evidence Report. National Institutes of Health.
Obes Res
 
1998
;
Sep(6 suppl)
:
51S
–209S.
28.
Willett WC, Manson JE, Stampfer MJ, et al. Weight, weight change, and coronary heart disease in women: risk within the ‘normal’ weight range.
JAMA
 
1995
;
273
:
461
–5.
29.
Rabkin SW, Mathewson FA, Hsu PH. Relation of body weight to development of ischemic heart disease in a cohort of young North American men after a 26 year observation period: the Manitoba Study.
Am J Cardiol
 
1977
;
39
:
452
–8.
30.
Grabowski DC, Ellis JE. High body mass index does not predict mortality in older people: analysis of the Longitudinal Study of Aging.
J Am Geriatr Soc
 
2001
;
49
:
968
–79.
31.
Dey DK, Rothenberg E, Sundh V, et al. Body mass index, weight change and mortality in the elderly. A 15 y longitudinal population study of 70 y olds.
Eur J Clin Nutr
 
2001
;
55
:
482
–92.
32.
Diehr P, Bild DE, Harris TB, et al. Body mass index and mortality in nonsmoking older adults: the Cardiovascular Health Study.
Am J Public Health
 
1998
;
88
:
623
–9.
33.
Takala JK, Mattila KJ, Ryynanen OP. Overweight, underweight and mortality among the aged.
Scand J Prim Health Care
 
1994
;
12
:
244
–8.
34.
Barrett-Connor EL. Obesity, atherosclerosis, and coronary artery disease.
Ann Intern Med
 
1985
;
103
:
1010
–19.
35.
Carmelli D, Halpern J, Swan GE, et al. 27-year mortality in the Western Collaborative Group Study: construction of risk groups by recursive partitioning.
J Clin Epidemiol
 
1991
;
44
:
1341
–51.
36.
Keil JE, Sutherland SE, Knapp RG, et al. Mortality rates and risk factors for coronary disease in black as compared with white men and women.
N Engl J Med
 
1993
;
329
:
73
–8.
37.
Menotti A, Descovich GC, Lanti M, et al. Indexes of obesity and all-causes mortality in Italian epidemiological data.
Prev Med
 
1993
;
22
:
293
–303.
38.
Seidell JC, Andres R, Sorkin JD, et al. The sagittal waist diameter and mortality in men: the Baltimore Longitudinal Study on Aging.
Int J Obes Relat Metab Disord
 
1994
;
18
:
61
–7.
39.
Stevens J, Plankey MW, Williamson DF, et al. The body mass index-mortality relationship in white and African American women.
Obes Res
 
1998
;
6
:
268
–77.
40.
Stevens J, Keil JE, Rust PF, et al. Body mass index and body girths as predictors of mortality in black and white women.
Arch Intern Med
 
1992
;
152
:
1257
–62.
41.
Allison DB, Saunders SE. Obesity in North America: an overview.
Med Clin North Am
 
2000
;
84
:
305
–32,
v
.
42.
Troiano RP, Frongillo EA Jr, Sobal J, et al. The relationship between body weight and mortality: a quantitative analysis of combined information from existing studies.
Int J Obes Relat Metab Disord
 
1996
;
20
:
63
–75.
43.
Effect of smoking on the body mass index-mortality relation: empirical evidence from 15 studies. The BMI in Diverse Populations Collaborative Group.
Am J Epidemiol
 
1999
;
150
:
1297
–308.
44.
Gaesser GA. Thinness and weight loss: beneficial or detrimental to longevity?
Med Sci Sports Exerc
 
1999
;
31
:
1118
–28.
45.
Stallones RA. Epidemiologic studies of obesity.
Ann Intern Med
 
1985
;
103
:
1003
–5.
46.
Heitmann BL, Erikson H, Ellsinger BM, et al. Mortality associated with body fat, fat-free mass and body mass index among 60-year-old Swedish men—a 22-year follow-up. The study of men born in 1913.
Int J Obes Relat Metab Disord
 
2000
;
24
:
33
–7.
47.
Allison DB, Zhu SK, Plankey M, et al. Differential associations of body mass index and adiposity with all-cause mortality among men in the First and Second National Health and Nutrition Examination Surveys (NHANES I and NHANES II) follow-up studies.
Int J Obes Relat Metab Disord
 
2002
;
26
:
410
–16.
48.
Prineas RJ, Folsom AR, Kaye SA. Central adiposity and increased risk of coronary artery disease mortality in older women.
Ann Epidemiol
 
1993
;
3
:
35
–41.
49.
McNeely MJ, Boyko EJ, Weigle DS, et al. Association between baseline plasma leptin levels and subsequent development of diabetes in Japanese Americans.
Diabetes Care
 
1999
;
22
:
65
–70.
50.
Rios S. Relationship between obesity and the increased risk of major complications in non-insulin-dependent diabetes mellitus.
Eur J Clin Invest
 
1998
;
28
:
14
–18.
51.
Montague CT, O’Rahilly S. The perils of portliness: causes and consequences of visceral adiposity.
Diabetes
 
2000
;
49
:
883
–8.
52.
Peiris AN, Sothmann MS, Hennes MI, et al. Relative contribution of obesity and body fat distribution to alterations in glucose insulin homeostasis: predictive values of selected indices in premenopausal women.
Am J Clin Nutr
 
1989
;
49
:
758
–64.
53.
Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women’s Health Initiative randomized controlled trial.
JAMA
 
2002
;
288
:
321
–33.
54.
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
:
1839
–44.
55.
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
:
749
–57.
56.
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
:
704
–9.
57.
Harris TB, Visser M, Everhart J, et al. Waist circumference and sagittal diameter reflect total body fat better than visceral fat in older men and women. The Health, Aging and Body Composition Study.
Ann N Y Acad Sci
 
2000
;
904
:
462
–73.