SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

While several studies have reported a positive association between overall adiposity and heart failure (HF) risk, limited and inconsistent data are available on the relation between central adiposity and incident HF in older adults. We sought to examine the association between waist circumference (WC) and incident HF and assess whether sex modifies the relation between WC and HF.

Prospective study using data on 4,861 participants of the Cardiovascular Health Study (1989–2007). HF was adjudicated by a committee using information from medical records and medications. We used Cox proportional hazard models to compute hazard ratio (HR). The mean age was 73.0 years for men and 72.3 years for women; 42.5% were men and 15.3% were African Americans. WC was positively associated with an increased risk of HF: each standard deviation of WC was associated with a 14% increased risk of HF (95% CI: 3%–26%) in a multivariable model. There was not a statistically significant sex-by-WC interaction (P = 0.081). BMI was positively associated with incident HF (HR: 1.22 (95% CI: 1.15–1.29) per standard deviation increase of BMI); however, this association was attenuated and became nonstatistically significant upon additional adjustment for WC (HR: 1.09 (95% CI: 0.99–1.21)). In conclusion, a higher WC is associated with an increased risk of HF independent of BMI in community-living older men and women.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Heart failure (HF) is a major public health concern (1), affecting ∼4.9 million Americans in 2005, with associated costs of $27.9 billion. At age 40, the lifetime risk of HF is about 20% for men and women (2,3). Despite progress in medical treatment, HF mortality remains high (4,5), ranging from 20% to 50% (6,7,8,9). The highest burden of HF, however, falls on adults aged 65 and older, for whom it is the leading cause of hospitalization (10,11). Older adults are also the fastest growing demographic group, which is bound to amplify the scope of problem.

A large proportion of HF cases is accounted for by antecedent coronary heart disease, type 2 diabetes, and hypertension (12,13,14,15,16), suggesting that predictors of coronary disease and hypertension might influence the risk of HF. To this end, adiposity is a known risk factor for coronary heart disease, hypertension, and type 2 diabetes (three major risk factors of HF). Currently, 66% of US adults are overweight or obese, and this proportion is expected to reach 75% by 2015 (17). While several studies have reported a positive association between overall adiposity (such as BMI) or central adiposity and HF risk (18,19,20) in young adults, limited data are available on the association between central adiposity (i.e., measured by waist circumference (WC)) and HF risk in older adults. Older adults tend to have more fat mass for a given BMI than younger adults (21), and the distribution of fat changes with aging (more fat accumulation in abdomen and less fat in extremities) (22).

The Heart and Soul Study (23) reported a 30% increased risk of HF per each standard deviation of waist—hip ratio in people with stable coronary disease while the Health ABC (24) reported a positive association with measures of central adiposity and HF. None of those two studies (23,24) of older adults found an independent association between BMI and HF. While some studies have reported an inverse relation between hip circumference and risk factors for HF, no previous study has examined such relation in older adults. Additional gap in the literature include the absence of data on body composition and HF subtypes (normal vs. lower left ventricular ejection fraction (LVEF)) and a potential effect modification by sex. Due to body fat redistribution with aging, the relation between adiposity and HF may become weaker at older ages (25), whereas other researchers have reported an association between measures of central adiposity in men but not women (24). This suggests that sex may modify the relation between adiposity and HF.

An earlier evaluation of predictors of HF in the Cardiovascular Health Study examined only height and weight, but did not address measures of central adiposity (15). The main purpose of this article was to test the hypothesis that central adiposity as measured by WC is positively associated with incident HF independent of BMI and that sex modifies the WC—HF relation. In a secondary analysis, we examined the association between (i) hip circumference, waist—hip ratio, and HF and (ii) WC and HF with and without normal LVEF.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Study population

Study participants were drawn from the Cardiovascular Health Study (CHS), a prospective cohort consisting of 5,888 men and women aged 65 years and older that were randomly selected from Medicare-eligibility lists in four US communities (Forsyth County, NC; Washington County, MD; Sacramento County, CA; and Pittsburgh, PA). A detailed description of methods and procedures in the CHS has been published elsewhere (26). Briefly, participants were not institutionalized or wheelchair-dependent, did not require a proxy for consent, were not receiving treatment for cancer, and were expected to remain in their respective regions for 3 years. Between 1989 and 1990, a total of 5,201 individuals were recruited in the original cohort. Between 1992 and 1993, a total of 687 African Americans was also recruited. Baseline evaluation of study participants included standardized questionnaires, physical examination, anthropometric measurements, resting electrocardiography, and laboratory examinations. From 1989 through 1999, participants were followed up every 6 months, alternating between telephone calls and clinic visits. The institutional review board at each center approved the study, and each participant gave informed consent. Of the total 5,888, we excluded (i) 275 people because of prevalent HF, (ii) 39 people for missing data on BMI or WC, (iii) 472 participants for moderate/severe aortic or mitral regurgitation or stenosis on echocardiograms, and (iv) 241 individuals with missing covariate information. Hence, we used 4,861 participants for current analyses.

Ascertainment of HF

Self-report of a physician diagnosis of HF was validated by the CHS Events Committee as previously described (15,27). Briefly, HF validation required a constellation of symptoms (shortness of breath, fatigue, orthopnea, and paroxysmal nocturnal dyspnea); chest X-ray findings (pulmonary edema and increasing cardiomegaly); signs (edema, pulmonary rales, gallop rhythm, and displaced left ventricular apical impulse); and treatment of HF (diuretics, digitalis, or vasodilators). Incident HF was ascertained upon review of pertinent data on hospitalization or outpatient visits such as medical history, physical examination, report of chest X-ray, and medications. HF was classified as systolic or diastolic HF based on a left ventricular ejection fraction (LVEF) cut point of 50% in our analysis group. The estimated LVEF was obtained from an echocardiogram, cardiac catheterization, multiple gated cardiac pool imaging, or other modality. Records on EF were obtained by review from CHS investigators or CHS adjudication committee. We had adequate data to estimate LVEF on 730 (53%) HF events in our sample. HF with an LVEF < 50% (n = 401) and HF with an LVEF ≥ 50% (n = 329). The current analysis included validated HF through 30 June 2007.

Assessment of anthropometric measures

At baseline, trained personnel obtained anthropometric measures using a standardized protocol. Height was measured in centimeters using a stadiometer, and weight was measured using a balance beam scale in pounds while subjects were wearing examination gowns and no shoes. Waist and hip circumferences were measured on standing subjects at the level of the umbilicus and maximal protrusion of the gluteal muscles, respectively. BMI was computed as weight (kg) divided by height in meters squared. Waist—hip ratio was calculated as the ratio of WC to hip circumference.

Other variables

Information on race, education, gender, field center, education, income, prevalent chronic diseases (coronary disease, stroke, cancer, valvular or rheumatic heart disease, atrial fibrillation, and diabetes), smoking, alcohol consumption, physical activity, and current medications was obtained during clinic visits. As described previously, standard laboratory methods were used to measure serum albumin, lipids, kidney function, fasting glucose and insulin, and C-reactive protein (28). Usual dietary habits were assessed at baseline in the original cohort using a 99-item picture-sort version of the National Cancer Institute food frequency questionnaire (29). Detailed description of validity and computation of nutrients and energy intake in this cohort has been described previously (29,30).

Statistical analysis

Baseline characteristics of study participants were summarized according to sex-specific WC quartiles; continuous variables were presented as mean ± standard deviation and categorical variables as percentages. Incidence rates of HF were calculated per 10,000 person-years.

Cox proportional hazards regression was used to estimate the hazard ratio (HR) associated with HF for WC per standard deviation. Individuals were censored for death or end of follow-up. We then adjusted for other baseline risk factors and potential confounders including age, clinic site, race, education (< high school vs. high school or higher), alcohol intake (none, <0.5, 0.5–1, >1 drinks/day for women; none, <1, 1–2, and >2 drinks/day for men), smoking (never, former, and current), kilocalories of physical activity (log-transformed continuous), estimated glomerular filtration rate (using the modification of diet in renal disease (MDRD) study equation (31)), history of physician-diagnosed valvular/rheumatic heart disease, atrial fibrillation by ECG, aspirin use, and estrogen use for women (model 1). Further adjustment was made for BMI. Additionally, we adjusted for possible intermediate factors at baseline including diabetes, systolic blood pressure, hypertension medication, coronary heart disease, C-reactive protein, triglycerides, high-density and low-density lipoprotein cholesterol (model 2). We also present results separately for each sex although the interaction of WC and gender was not statistically significant.

In a secondary model, we examined HF with depressed vs. normal LVEF (for 730 HF cases with adequate echocardiographic/imaging data to assess left ventricular ejection fraction), and implemented methods for competing risks (32) while including those with unclassified HF as censored. Using this method, we stratified on HF type (normal vs. low LVEF) and estimated the associations for each outcome in the same model with a proportional hazards assumption. Further sensitivity analyses were restricted to subjects without valvular disease, no significant unintentional weight loss, never smoking, and self-reported heath status of good to excellent health. For each analysis, we checked the proportional hazards assumption with Schoenfeld residuals using log (person-time) and plots of the residuals over time; there was no meaningful violation of this assumption.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

Of the 4,861 participants included in current analyses, 42.5% were men and 15.3% were African Americans. The mean age was 73.0 ± 5.6 years (range: 65–95) for men and 72.3 ± 5.4 years (range: 65–100) for women. Table 1 shows the baseline characteristics of participants according to sex-specific quartiles of WC. During an average follow-up of 11.3 years, 1,381 incident cases of HF occurred. Pearson's correlation coefficients between WC and BMI were 0.86 in men and 0.83 in women.

Table 1.  Baseline characteristics according to sex-specific quartiles waist circumference in the Cardiovascular Health Study
inline image

Each standard deviation (13.2 cm) increase in WC was associated with a 14% (CI: 3%–26%) increased risk of HF upon adjustment for BMI, age, clinic site, gender, race, education, alcohol intake, smoking, physical activity (log-transformed kcals), estimated glomerular filtration rate, valvular/rheumatic heart disease, atrial fibrillation, estrogen use (women), and aspirin use (Table 2). Additional adjustment for hip circumference did not change the results. There was no evidence for a statistically significant interaction between sex and WC on the risk of HF (Table 2). As expected, adjustment for potential intermediate factors (systolic blood pressure, hypertension medications, prevalent diabetes and coronary disease, C-reactive protein (log-transformed), triglycerides, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol) led to attenuation of the relative risk (HR (95% CI): 1.06 (0.96–1.17)).

Table 2.  Hazard ratios (95% CI) for heart failure per standard deviation increment of adiposity measures
inline image

In a sensitivity analysis, exclusion of participants with self-reported poor health status, current or former smokers, and self-reported weight loss yielded a HR of 1.15 (95% CI: 0.95–1.38). Furthermore, exclusion of people with prevalent valvular/rheumatic disease did not alter the association between WC and HF (data not shown). When using 730 HF cases with echocardiographic data on LVEF, each standard deviation increase of WC was positively associated with HF with lower LVEF (<50% HR (95% CI): 1.28 (1.05–1.56)) but not with HF with preserved LVEF (Table 3) in the fully adjusted model. Hip circumference was not independently associated with an increased risk of incident HF after adjustment for BMI and other covariates (HR (95% CI): 1.01 (0.91–1.12) per standard deviation (10.0 cm) of hip circumference). Waist—hip ratio was positively associated with incident HF after adjustment for BMI and other covariates (HR: 1.89 (95% CI: 1.03–3.46)). Lastly, each standard deviation (4.7 kg/m2) of BMI was associated with a 22% increased risk of incident HF (Table 2); however, control for WC attenuated this relation and rendered it nonstatistically significant (HR: 1.09 (95% CI: 0.99–1.21), Table 2).

Table 3.  Hazard ratios (95% CI) for heart failure type per standard deviation increment of waist circumference
inline image

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

In this prospective study of older adults, higher WC was associated with an increased risk of HF after adjustment for BMI and other confounders. In addition, BMI was positively associated with incident HF in a multivariable model; however, after adjustment for WC, the association between BMI and incident HF was attenuated and became nonstatistically significant. While waist—hip ratio was positively associated with incident HF, hip circumference was not associated with the risk of HF.

Although several studies have examined the association between adiposity and incident HF, few have assessed whether measures of central adiposity predict HF independent of BMI. Our observation that WC but not BMI was independently associated with incident HF is consistent with findings of the Health, Aging and Body Composition study (24), in which each standard deviation increase of WC (but not BMI) was associated with a 27% (95% CI: 4%–54%) increased risk of HF in a model adjusted for BMI. In a study (33) of 1,187 men aged 70+ years, both BMI and WC were positively associated with HF hospitalization; however, investigators in that study did not adjust for BMI when assessing the effects of WC on HF and vice versa. In the HOPE study (34), BMI, waist—hip ratio, and WC were individually associated with an increased risk of HF in people aged 66 years on average; however, upon additional control for potential mediators including hypertension, diabetes, total cholesterol, and high-density lipoprotein cholesterol, only WC remained positively associated with incident HF. When stratified by sex, WC and waist—hip ratio but not BMI were positively associated with incident HF in women but not in men (34). Since the HOPE study (34) did not provide sex-specific data that were unadjusted for intermediate factors, it is difficult to contrast the sex-by-central adiposity interaction observed by Dagenais et al. (34) with our results indicating no meaningful sex-by-adiposity interaction on HF risk. Our findings of a stronger association for a central than a general measure of adiposity are consistent with observations in generally younger populations (25). In a prospective study of women aged 48–83 years, WC was associated with higher risk of HF at all levels of BMI (25), yet BMI was not associated with increased risk of HF across strata of WC in women (25). Among men aged 45–79 years, WC was positively associated with HF risk in all BMI levels whereas BMI conferred only a modest increased risk of HF across categories of WC (25). The absence of an association for BMI independent of WC in our cohort may relate to the reported weakening of its association with older age.

Several biological mechanisms might help explain the observed stronger relation between measures of central adiposity (i.e., WC) and incident HF. Adipose tissue expresses several hormones including resistin or adiponectin that have been associated with impaired glucose metabolism and/or HF risk. Impaired glucose metabolism has been related to left ventricular systolic and diastolic dysfunction (35). Hyperinsulinemia can lead to sodium retention (36) and activation of the sympathetic nervous system (37), factors that can foster HF development. In addition, central obesity is associated with hypertension, dyslipidemia, coronary heart disease, inflammatory state, etc and could influence the risk of HF via those mediators. The attenuation of the HR upon additional adjustment for blood pressure, hypertension, prevalent diabetes and coronary disease, C-reactive protein, triglycerides, high-density lipoprotein, and low-density lipoprotein lends support to this hypothesis. Visceral fat (including epicardial fat) expresses several hormones including fatty acid—binding proteins with known negative inotropic effects in vitro (38), suggesting that central adiposity may be more important in the development of HF with poor LVEF. Such hypothesis is consistent with our data: each standard deviation of WC was associated with a 28% increased risk of HF with LVEF but only with an 11% (nonstatistically significant increased risk of HF with normal LVEF, Table 3). Of note, our secondary analysis on HF with and without normal LVEF was underpowered as we only had adequate information on LVEF on slightly more than half of the HF cases.

Our study has some limitations. As an observational study, we cannot exclude unmeasured confounding as partial explanation to our findings. The inability to classify half of HF cases according to LVEF reduced our statistical power to examine the association between adiposity and HF with preserved or depressed left ventricular systolic function. We were unable to examine the effects of WC across standard categories of BMI due to sparse number of cases in some cells. The generalizability of our findings is limited as our participants were mostly whites. However, our paper has numerous strengths including the large number of male and female subjects with corresponding incident events; the availability of data on numerous covariates including diet, comorbidity, metabolic, and lifestyle factors; the use of standardized protocol to collect data; the validation of HF cases by an Endpoint Committee; and the 10+ years of follow-up.

In summary, our data showed that a higher WC but not BMI is associated with an increased risk of HF after mutual adjustment, whereas the same was not true of BMI after adjustment for WC. If confirmed in other populations, this suggests that a measure of central adiposity might be a stronger risk factor of HF than overall adiposity.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES

We are indebted to the participants and staff of the Cardiovascular Health Study. A full list of participating CHS investigators and institutions can be found at http:www.chs-nhlbi.org.

The research reported in this article was supported by the National Heart Lung Blood Institute (R01HL094555) to Drs Djousse, Ix, Mukamal, Zieman, and Kizer. Additional support was provided by contracts N01-HC-85079 through N01-HC-85086, N01-HC-35129, N01 HC-15103, N01 HC-55222; N01-HC-75150, N01-HC-45133, N01-HC-85239, N01-HC-55222, U01 HL080295, and with additional contribution from the National Institute of Neurological Disorders and Stroke.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Disclosure
  9. REFERENCES