Abstract
Background and objectives Body mass index and waist circumference associate with adverse health outcomes, including CKD. Studies of the association of body mass index and ESRD have been inconsistent; these adiposity measures have not been previously assessed together for ESRD risk or among postmenopausal women.
Design, settings, participants, & measurements This was prospective cohort study of 20,117 postmenopausal women enrolled in the multiethnic cohort of the Women’s Health Initiative. Body mass index and waist circumference were obtained at baseline, incident ESRD was obtained from the US Renal Data System, and all-cause death was obtained from surveillance data. A competing-risk framework was used to account for the effect of mortality before ESRD while adjusting for significant predictors and baseline kidney function. Associations of adiposity with mortality were also studied.
Results Events included 212 patients with incident ESRD and 3104 deaths for a mean follow-up of 11.6 years. Increased waist circumference and body mass index were associated with 2.59- (95% confidence interval, 1.89 to 3.53) and 1.97-fold (95% confidence interval, 1.30 to 2.98) higher hazards of ESRD as well as 1.42- (95% confidence interval, 1.32 to 1.53) and 1.21-fold (95% confidence interval, 1.11 to 1.33) higher hazards of death, respectively, compared with the lower categories in adjusted analyses. The associations of waist circumference with ESRD varied by baseline renal function (P for interaction=0.01) and were significant only among women without baseline eGFR-defined CKD (hazard ratio, 1.93; 95% confidence interval, 1.23 to 3.03).
Conclusions Central obesity was associated with an increased risk of ESRD in postmenopausal women, even among women with normal body mass index but not among women with reduced baseline kidney function, and an increased risk of death. Body mass index was associated with ESRD, and the association is likely mediated through hypertension and diabetes.
Introduction
Obesity is associated with multiple adverse health outcomes, including CKD (1–3). CKD affects approximately 14.5% of the adult United States population, with a high burden occurring among aging individuals and racial/ethnic minorities (4), including blacks (5) and Hispanics (6). CKD is associated with premature cardiovascular disease (CVD) (7) and death (4), decreased quality of life (8), and progression to ESRD. Treatment of ESRD and its complications accounts for a large percentage of United States health care use and costs (9). Older individuals now have the highest rates of both early-stage CKD and new ESRD in the United States (10) along with increased CKD-related comorbidities. CKD remains understudied in aging populations. Postmenopausal women have increased prevalence of comorbidities, including obesity (11), that are associated with both hypertension and diabetes, the most common attributable causes of CKD. Menopause transition is associated with an increased intra-abdominal adiposity independent of age and total adiposity (11), and adiposity patterns may associate with CKD. An important clinical question relevant to the care of postmenopausal women is the role of obesity (central and overall) beyond other risk factors in development and progression of CKD, because increased adiposity is amenable to lifestyle interventions. In addition, assessment of adiposity before development of ESRD is needed to better understand the reverse epidemiology observed among individuals with chronic conditions including ESRD, in whom an increased body mass index (BMI) is associated with reduced mortality (12,13).
Adipocytes have been shown to be metabolically active and are associated with inflammation, oxidative stress, and endothelial dysfunction (14). Suppression of cell autophagy has also been recently proposed as a mechanism for obesity-induced kidney dysfunction (15). Evidence from observational studies suggests that both overall and central obesity, assessed by waist circumference or waist-to-hip ratio, are associated with incident CKD in populations (16–22). However, the association between obesity and ESRD has been inconsistent (2,3), and assessment of central adiposity measures for ESRD risk has yet to be studied, particularly among postmenopausal women. A large study using health care data found positive associations between overweight and obesity with incident ESRD (2). Similar results were described in a large Chinese study (3) but not among United States veterans (23). Mechanisms related to the development of CKD and progression to ESRD are still poorly understood, especially the racial/ethnic differences in incident ESRD.
The main goal of this study was to examine the association of adiposity with ESRD in postmenopausal women. We used data from the Women’s Health Initiative (WHI) study, an ethnically diverse cohort of 161,808 postmenopausal women with extensive data on physiologic, behavioral, and clinical risk factors in addition to validated cardiovascular and renal outcomes. We assessed the risks of ESRD and death by categories of BMI and waist circumference in a sample of 20,117 African-American, white, and Hispanic WHI participants with available kidney function at baseline.
Materials and Methods
Study Sample and Population
The WHI is a prospective population-based cohort study investigating postmenopausal women’s health in the United States (24). In total, 161,808 women ages 50–79 years old were recruited from 40 United States clinical centers between 1993 and 1998 to participate in the observational study and several clinical trials: postmenopausal hormone therapy (estrogen alone or estrogen plus progestin), a calcium and vitamin D supplement trial, and a dietary modification trial of reduced total fat intake to 20% of calories and increased intake of vegetables/fruits (24). Recruitment was done through mass mailing to age-eligible women obtained from voter registration, driver’s license, and Health Care Financing Administration or other insurance lists, with emphasis on recruitment of minorities and older women (24). Exclusions included participation in other randomized trials, predicted survival <3 years, alcoholism, drug dependency, mental illness, and dementia. For the clinical trials, women were ineligible if they had systolic BP >200 mmHg or diastolic BP >105 mmHg, a history of hypertriglyceridemia, or endometrial cancer.
Demographic characteristics, socioeconomic data (education), lifestyle (smoking and alcohol consumption), medical history (history of hypertension, diabetes, CVD, and kidney dialysis treatment), and self-reported medications were collected using standardized questionnaires at the screening visit. Body height, weight, waist circumference, and BP were measured at a clinical visit as described previously (25). Fasting blood samples were obtained at the baseline clinic visit. A subset of women had serum creatinine measured using an enzymatic method that was traceable to an isotope dilution mass spectrometry reference creatinine standard (coefficient of variation=3.7%), enabling eGFR using the CKD Epidemiology Collaboration (CKD-EPI) equation. Results are reported in milligrams per deciliter. Study protocols and consents were approved by the institutional review boards at all participating institutions, and the research was conducted in adherence to the Declaration of Helsinki.
A subset of 20,117 women with serum creatinine measurements at baseline was included in analyses. Women were selected for biomarker characterization if they were participants in the Single Nucleotide Polymorphism (SNP) Health Association Resource project, a randomly selected subsample of 8515 (70.1%) black and 3642 (66.6%) Hispanic women, or participants in the hormone therapy clinical trials in a subsample that reflected the age distribution of the entire white women in these trials. Within the biomarker sample, we excluded women with BMI<18.5 kg/m2 and those with missing covariate data.
Exposures.
Waist measurement was obtained in the standing position to the nearest 0.5 cm over nonbinding undergarments at the level of the umbilicus. BMI was estimated using height and weight (kilograms per meter2) in women using light clothes. Categories of BMI and waist circumference were defined using National Heart, Lung, and Blood Institute intervals (25). We also examined the risk across quintiles of the exposure distribution.
Covariates.
Diabetes mellitus was defined by self-report of physician diagnosis or use of oral hypoglycemic medications or insulin. Hypertension was defined on the basis of BP≥140/90 mmHg or the use of antihypertensive drugs (26). Prevalent CVD was on the basis of an affirmative answer to the question of have you been hospitalized for a heart attack (myocardial infarction), coronary angioplasty or stent, coronary artery bypass graft surgery, or angina. We estimated GFR using the equation developed by Levey and Stevens (27) (CKD-EPI) on the basis of age, sex, race, and serum creatinine.
Outcome measures—Incident ESRD and Mortality.
Patients with ESRD were drawn from the US Renal Data System (USRDS), a national registry of all patients with ESRD in the United States (28). USRDS files were linked to the WHI cohort data using personal identifiers. Events as of June 30, 2010, are included in these analyses. Annual (observation study) and semiannual (clinical trials) follow-up identified events, including deaths, which were classified by an expert panel of physicians on the basis of review of hospital records, death certificates, and interviews with next of kin (29).
Statistical Analyses
For descriptive analyses of baseline data, means, SDs, and frequencies were measured. We assessed differences in baseline BMI categories or waist circumference on subsequent risk of ESRD and mortality before ESRD. Because an increase in mortality in older patients may indirectly lead to an apparent reduction in ESRD, we used a competing-risk framework (30). The primary end point was the cumulative incidence of ESRD and death before ESRD, which quantifies the risks of the events over time analogously to the Kaplan–Meier estimator for data without competing risks. Person-time at risk was defined as the time between baseline visit and the ESRD event, death, or the last date of follow-up up (June 30, 2010). Nonparametric estimates were obtained within BMI categories (normal [<25 kg/m2], overweight [25–29 kg/m2], and obese [30 kg/m2 or more]) and waist circumference groups (≤88 cm versus >88 cm). Nonparametric tests for differences in these cumulative incidences were assessed using Gray’s test (31) to account for competing causes. The Fine and Gray (32) proportional subdistribution hazards model for the cumulative incidence function was fitted to obtain hazard ratios (HRs), which summarize differences in these cumulative incidence functions. A secondary end point was the cause-specific hazard of the competing–risk end points (33), which provides complementary information into the patterns of failure in the cumulative incidence functions. Log-rank tests were performed for the statistical significance of differences across BMI and waist circumference groups. Regression modeling for the cause-specific hazards was performed using the standard Cox model with joint modeling of BMI and waist circumference considered as well as the effect of each as a continuous variable. HRs from the subdistribution hazard and the cause-specific hazard models were then compared.
We fitted three models. Model 1 adjusted for age and baseline eGFR. Model 2 adjusted additionally for race/ethnicity, education (<12 years versus ≥12 years), ever smoking, and indicator variables for participation in the observational versus clinical trials and geographic region. Model 3 adjusted additionally for history of diabetes mellitus, systolic and diastolic BPs, and use of antihypertensive drugs (including angiotensin-converting enzyme (ACE) inhibitors and/or angiotensin receptor blockers). We tested for interactions between adiposity measures and race/ethnicity and between adiposity measures and CKD, which was defined by an eGFR<60 ml/min per 1.73 m2. We also tested interactions between BMI and waist circumference on the risk of ESRD or death. Subdistribution, cause-specific HRs, and 95% confidence intervals (95% CIs) are reported. We used an α=0.05 in a two-sided test. All analyses were performed using STATA 11.0 (Stata Corp LP, College Station, TX).
Results
The mean age was comparable among the biomarker sample and the overall WHI cohort for women of the same ethnic/racial background (Table 1). The prevalence of African-American and Hispanic women was higher in the subsample compared with the WHI cohort. In addition, women in the biomarker sample had higher mean BMI and waist circumference and higher prevalence of diabetes, hypertension, and treated hypertension compared with those in the WHI cohort. None of the participants in the WHI reported being on dialysis at enrollment.
Baseline characteristic of participants in the biomarker sample and the whole cohort: Women’s Health Initiative, 1993–1998
Six percent of women met criteria for (stages 3–5) CKD on the basis of eGFR (Table 1). We identified 212 ESRD events and 3104 deaths over a mean follow-up of 11.6 years (SD=3.0). Traditional ESRD predictors, including diabetes, hypertension, and baseline eGFR, were significantly associated with incident ESRD (P<0.05). Obese women as well as those with an increased waist circumference had higher cumulative incident ESRD compared with women in the lower categories (Figure 1). Obesity and increased waist circumference were associated with 2- and 2.6-fold increases in ESRD, respectively, when adjusting for age and baseline eGFR (model 1, Table 2). When accounting for other risk factors, estimates for BMI association with ESRD were attenuated but remained significant, except when adjusting for diabetes and hypertension-related variables (models 2 and 3, Table 2). Waist circumference was associated with ESRD in all models (Table 2). For comparison, we also show the estimates for incident ESRD without accounting for competing risks in Table 2 (cause-specific HR obtained from the conventional Cox proportional hazard models). The increased risk of ESRD was also observed across quintiles of the waist circumference distribution starting near the clinically defined threshold and vary by baseline eGFR (Figure 2). BMI and waist circumference were highly correlated in our study (coefficient=0.81; P<0.001). However, continuous waist circumference was independently and significantly associated with ESRD after additional adjustment for BMI in model 2 (HR, 1.02; 95% CI, 1.02 to 1.03).
Cumulative incidence function estimates and 95% confidence limits for ESRD by categories of (A) body mass index (BMI) and (B) waist circumference. The test by Gray (31) for equality of cumulative incidence curves among categories of BMI (kilograms per meter2) or waist circumference (centimeters) was significant (P<0.001 for both).
Association of body mass index and waist circumference categories with ESRD and mortality
Age-adjusted hazard ratio and 95% confidence intervals of associations with ESRD by quintiles of waist circumference for (A) the biomarker sample, (B) women without eGFR-defined CKD, and (C) women with CKD. Numbers on the x axis are the quintiles and below the mean values for each quintile. 95% CI, 95% confidence interval.
We then tested interactions of race/ethnicity with BMI or waist circumference on the risk of ESRD, which were not significant (P=0.21 and P=0.51, respectively). There were also no significant interactions among BMI and waist circumference categories on the risk of ESRD (P=0.55). Because decreased eGFR is a strong predictor of ESRD, we also tested its interactions with BMI and waist circumference on the risk of ESRD; this was significant only for waist circumference (P=0.01). In a stratified analysis, waist circumference higher was significantly associated with incident ESRD among women with an eGFR≥60 ml/min per 1.73 m2 (HR, 1.93; 95% CI, 1.23 to 3.03; n=18,963; P<0.01) but not among women with reduced eGFR, for whom the point estimate was close to 1 (HR, 0.96; 95% CI, 0.57 to 1.60; n=1154; P=0.87).
In analysis of the mortality outcome, women with increased waist circumference had a 42% higher risk of death compared with those in the reference category (model 2, Table 2). Obesity was associated with mortality in models adjusted for age, baseline eGFR, and other risk factors but not models adjusted for diabetes and hypertension (Table 2). Because mortality can vary by race/ethnicity, we also tested the interactions of BMI or waist circumference with race, which were of borderline significance only for waist circumference (P<0.10). Stratified analysis by race/ethnicity showed a stronger risk of death for an increased waist circumference among Hispanic women followed by African Americans and whites (Table 3). Although the interaction by race was not significant, obesity was significantly associated with mortality in Hispanic women only in fully adjusted models.
Hazard ratios and 95% confidence intervals of mortality by race/ethnicity subgroups and adiposity categories
Discussion
Obesity has reached epidemic proportions and is associated with metabolic syndrome, diabetes, CVD complications, and short life expectancy. Our study showed strong associations of waist circumference in addition to BMI categories with ESRD when adjusting for multiple confounders, baseline kidney function, and competing risk of death. Notably, we have shown a 2.6-fold higher hazard of incident ESRD and 42% higher hazard of death for women with increased compared with normal waist circumference. ESRD risk increased in women with waist circumference above the clinically defined threshold; this increased risk was independent of BMI, and it was stronger among women with preserved eGFR and not significant in women with low eGFR.
Cross-sectional and longitudinal studies have shown associations of obesity measures (principally BMI) with CKD and ESRD (2,3,20,22,34,35). Few population studies, however, have tested the association of central obesity with incident ESRD, and none have focused in women. A prospective population-based study from Iran reported an association of waist circumference rather than BMI with the development of CKD, which was defined by an eGFR<60 ml/min per 1.73 m2 (20). A study using data from white and African-American participants of the Atherosclerosis Risk in Communities and Cardiovascular Health Study also showed significant associations of waist circumference but not BMI with CKD outcomes (defined as an increase of 0.4 mg serum creatinine or decrease in eGFR of at least 15 ml/min per 1.73 m2 to a final value below 60 ml/min per 1.73m2) (36). This study did not examine sex-specific associations. A case-control study of moderately severe CKD, which used serial BMI calculated using self-reported weight at ages 20, 40, and 60 years old, showed significant associations of obesity and overweight at earlier ages with subsequent CKD but no association of current BMI with CKD (37). Our findings were similar to these findings but provide important new data supporting increased waist circumference as a risk factor for progression to ESRD. Our study showed strong associations of waist circumference in addition to BMI categories with ESRD when adjusting for multiple confounders, baseline kidney function, and competing risk of death. Our findings expand the current knowledge on adiposity risk factors for ESRD among postmenopausal women, who have increased intra-abdominal obesity compared with premenopausal women (11), and across race/ethnicity by studying a multiethnic cohort of white, African-American, and Hispanic individuals. They suggest that adiposity distribution is an important clinical measure to be assessed in healthy postmenopausal women and that it should be targeted in prevention and lifestyle interventions.
Waist circumference is a commonly used surrogate of central adiposity in epidemiologic studies, and its assessment is currently recommended only for adults with a BMI of 25.0–34.9 kg/m2 by clinical expert panels (38). Our study did not find significant interactions of waist circumference and BMI, suggesting similar effects among women with normal and increased BMI. Waist circumference is better correlated with visceral fat and simpler to interpret than other measures of central adiposity, such as waist-to-hip ratio (39–41). Prior studies have shown a role of central obesity in cardiovascular and metabolic diseases, including type 2 diabetes, coronary heart disease, and stroke (42–47). Interestingly, we did not see differences in the association of adiposity measures with ESRD by race/ethnicity, although there are known differences in obesity prevalence, fat distribution, and cardiometabolic risk among some racial/ethnic subgroups.
Waist circumference has also been associated with mobility disabilities and death (48,49). In the European Prospective Investigation into Cancer and Nutrition, both BMI and waist circumference were associated with increased risk of death (48). A recent analysis of postmenopausal WHI participants also showed an increase in mortality before 85 years among women with waist circumference >88 cm (49). A BMI of ≥30 kg/m2 was also associated with mortality in our data, even when accounting for effects on potential mediators, but only among obese Hispanic women. We noted a nonsignificant protective effect among overweight women on mortality across race/ethnicity. This reverse epidemiology has been reported previously for BMI and mortality among individuals with chronic conditions, such as CKD (12) and heart failure (13), and its mechanisms are still not well understood. Our study, however, is composed of postmenopausal women with higher socioeconomic status compared with the general population, and the WHI study has excluded women with severe hypertension, hyperlipidemia, cancer, drug addiction, and expected poor survival.
The prevalence of CKD on the basis of eGFR was low in our study (<6%) among women for whom serum creatinine was available, a sample that overrepresented racial/ethnic minorities, and women with comorbidities compared with the overall WHI sample. After adjustments for BMI, comorbidities, and other risk factors, waist circumference was still significantly associated with both ESRD and mortality, although the estimates were attenuated by these adjustments. These findings suggest that some risk factors (e.g., diabetes and hypertension) may mediate these associations. Potential mediation may also explain the lack of association of BMI with incident ESRD in the fully adjusted models.
A recent analysis of the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study also showed an attenuation of the association of metabolic risk factors, including BMI, with CKD when adjusting for baseline eGFR and albuminuria (50). The REGARDS study was restricted to individuals with an eGFR<60 ml/min per 1.73 m2, which is in contrast with the low prevalence of eGFR-defined CKD in our study. We identified a significant interaction of waist circumference and eGFR. The increased ESRD risk attributable to a large waist circumference was significant only among women without reduced kidney function. We could not account for albuminuria in our analysis, because the WHI study lacks these data. Universal screening for CKD (albuminuria and kidney function) in postmenopausal women is currently not recommended by clinical guidelines, except for individuals with hypertension and diabetes. However, BMI is routinely evaluated at primary care visits.
Our study suggests that assessment of central obesity among postmenopausal women with normal kidney function may contribute additional important clinical information to estimation of ESRD risk. The absolute risk of ESRD among women with low prevalence of eGFR-defined CKD was 1.5% for women with an increased waist circumference compared with 0.6% for women with normal waist circumference. Waist circumference is a simple, easy to measure, low-cost, and feasible measure to implement in clinical care with potential added prognostic value for ESRD risk. Although interventions to reduce central adiposity are difficult to implement in clinical practice, increased awareness and prevention of central obesity in postmenopausal women could be important public health targets to reduce ESRD risk in aging populations.
Strengths of our study are the large sample size of postmenopausal women, the multiethnic cohort, the standardized measures of adiposity and other risk factors, the prospective data, and the large number of ESRD events. Some of the limitations of our study are that the waist circumference cannot distinguish between subcutaneous and intra-abdominal fat and that we relied on a single measure of adiposity obtained at screening visit; future research accounting for temporal changes of BMI and waist circumference is needed. Our findings cannot be generalized to men or premenopausal women who were not included in the WHI cohort. Power to detect interactions may be limited in our study. In addition, women with low BMI were excluded because of the small number of events and sample size.
In summary, our study has shown an important association of waist circumference, a measure of fat distribution, with the risk of ESRD among postmenopausal women, which was stronger among those women with preserved baseline eGFR and racial/ethnic minorities. Increased waist circumference was also associated with death, particularly among Hispanic women, suggesting that central obesity measures may have a clinical role in efforts to prevent and treat obesity-related CKD and its complications. Our overall findings suggest that accounting for waist circumference will be important in future studies of adiposity and CVD outcomes and mortality.
Disclosures
None.
Acknowledgments
The Women’s Health Initiative program is funded by National Heart, Lung, and Blood Institute (NHLBI), National Institutes of Health (NIH), US Department of Health and Human Services Contracts HHSN268201100046C, HHSN268201100001C, HHSN268201100002C, HHSN268201100003C, HHSN268201100004C, and HHSN271201100004C. Funding for the linkage to the US Renal Data System was provided by NIH Contract HHSN268201100004C NHLBI Control Number N01-WH-04354 (to N.F.). N.F. is supported by R21HL123677-01.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
- Received March 19, 2014.
- Accepted October 20, 2014.
- Copyright © 2015 by the American Society of Nephrology
References
In this issue
Jump to section
Recent Tweets about the Article
-
RT @tonysly50: FACT CHECK: Did Google Remove Photos https://t.co/heWgpeYE23
-
FACT CHECK: Did Google Remove Photos https://t.co/heWgpeYE23
