Obesity Is an Independent Risk Factor for Heart Failure: Zona Franca Cohort Study

Authors

  • José M. Baena-Díez MD,

    Corresponding author
    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    • Cardiovascular Epidemiology and Population Genetics Program on Research on Inflammatory and Cardiovascular Disorders Carrer Dr Aiguader 88 E-08003 Barcelona, Spain
    Search for more papers by this author
  • Alice O. Byram MD,

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author
  • María Grau MD, PhD,

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author
  • Claudia Gómez-Fernández MD,

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author
  • Marc Vidal-Solsona MD,

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author
  • Gabriela Ledesma-Ulloa MD,

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author
  • Isabel González-Casafont MD,

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author
  • Javier Vasquez-Lazo MD,

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author
  • Isaac Subirana MSc,

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author
  • Helmut Schroder PhD

    1. Primary Health Care Center La Marina; Primary Care Research Institute Jordi Gol; Cardiovascular Epidemiology and Population Genetics (ULEC-EGEC) and Program on Research on Inflammatory and Cardiovascular Disorders (RICAD); Cardiovascular Risk and Nutrition Risk (ULEC-CARIN) CIBER Obesity and Nutrition (CIBEROBN), Barcelona, Spain
    Search for more papers by this author

Abstract

Background:

Heart failure (HF) is a major problem in developed countries. However, its relationship with obesity remains unclear, especially in low-risk populations. The objective of the study was to analyze the relationship between obesity and HF in a low-risk Mediterranean population.

Hypothesis:

Obesity is an independent predictor for HF.

Methods:

A prospective community-based population cohort study with 10 years' follow-up was conducted at 2 healthcare centers in the city of Barcelona, Spain. From a registered population of 35 275, the study included 932 randomly selected patients without HF, age 35–84 years. Obesity was defined as body mass index (BMI) ≥30 and HF according to European Society of Cardiology guidelines, confirmed by echocardiography. Cox proportional hazards regression was used to examine the association between obesity and heart failure.

Results:

The difference in HF incidence between obese subjects (4.7%) and nonobese subjects (1.6%) was 3.1% (95% confidence interval [CI]: 0.7–5.5). In the unadjusted model, incident HF was significantly associated with BMI: the hazard ratio [HR] was 1.09 for every 1 kg/m2 increase (95% CI: 1.05–1.14) and 3.01 for BMI ≥30 (95% CI: 1.34–6.77). After adjusting for age, sex, hypertension, ischemic heart disease, and diabetes mellitus, the results were similar: HR 1.06 (95% CI: 1.01–1.10) and HR 2.45 for BMI ≥30 (95% CI: 1.02–5.61). Overweight was not associated with HF in any of the models. The population-attributable risk of HF due to obesity was 43.0% (95% CI: 13.9–74.9).

Conclusions:

High rate differences, HRs, and attributable risk indicate that obesity is an important risk factor for incident HF. Copyright © 2010 Wiley Periodicals, Inc.

This work was supported by grants from Instituto de Salud Carlos III-FEDER0 (PI080439), and by a joint contract of the Instituto de Salud Carlos III and the Health Department of the Catalan Government.

Introduction

Heart failure (HF) morbidity and mortality is a major problem in most parts of the world.1,2 For example, in 2005 the estimated prevalence of HF in US adults age ≥20 years was 5300000.1 Prevalence increased with age, to 11%–12% in patients age >80 years.1,2 As the population ages and the prevalence of HF increases, expenditures related to the care of these patients will climb daramatically.3 Classic risk factors for HF include coronary heart disease (CHD), valvular heart disease, hypertension, and diabetes mellitus (DM).1 The identification and control of risk factors and preclinical phases of the disease is key to the prevention of HF.

In 2015, the number of obese people worldwide is expected to reach 700 million.1 Although some population-based epidemiologic studies have examined the relationship between body mass index (BMI) and HF, these were based on populations with a high incidence of cardiovascular (CV) diseases, especially CHD.

The purpose of this study was to examine the relationship of BMI and obesity with the risk of HF in a Mediterranean population with a low CV risk profile.

Methods

Data Source

A prospective community-based population cohort (Zona Franca Cohort) was selected from a cross-sectional study4 that began in 1998 in 2 primary care centers with a registered population of 35 275 inhabitants in the city of Barcelona, Spain. The study followed the Declaration of Helsinki recommendations and was approved by the local ethics committee (Jordi Gol Foundation).

The population is heterogeneous but mainly Caucasian with a middle-class socioeconomic status, and uses the participating health centers almost exclusively. From a simple random sample of the registered population, we selected 1009 patients age 35–84 years with ≥1 CV risk factor whose medical records showed no previous HF (ICD-9 428 codes) and who were willing to participate.

Data Collected and Follow-Up

Baseline data included age, sex, and history of smoking; systolic and diastolic blood pressure (BP); diagnosis of hypertension (systolic BP ≥140 mm Hg or diastolic BP ≥90 mmHg); total cholesterol and high-density lipoprotein (HDL) cholesterol in mmol/L (hypercholesterolemia defined as ≥6.4 mmol/L or taking lipid-lowering drugs); fasting glucose levels in mmol/L; type 2 DM according to the American Diabetes Association criteria;5 and CV diseases in medical records, including CHD (ICD-9 codes 410–414), cerebrovascular disease (ICD-9 codes 430–438), peripheral artery disease of the lower extremities (ICD-9 code 443.9), and valvular heart disease (ICD-9 codes 394–397).

Obesity was defined as BMI ≥30, and overweight as BMI 25–29.9. BMI was calculated as body weight (measured on a standard balance beam scale to the nearest 0.1 kg) divided by height squared (measured without shoes to the nearest 0.1 cm using a wall-mounted stadiometer).

During follow-up, HF was defined as the combination of ≥2 symptoms (dyspnea, tiredness, or ankle edema) and clinical evidence of cardiac involvement on electrocardiogram or chest x-ray, according to the European Society of Cardiology guidelines.6 Systolic dysfunction was defined by ejection fraction (EF) <40%, and nonsystolic HF by EF ≥40% and signs of ventricular dysfunction in the echocardiogram, following the criteria of the European Study Group on Diastolic Heart Failure.7

Statistical Analyses

Using the SPSS software package (version 11.0; SPSS, Inc., Chicago, IL), proportions were compared using the χ2 test and means were determined by t test or analysis of variance. Study population size was calculated accepting an α risk of 0.05 and a β risk of 0.20 in a 2-sided test. To establish statistical significance for a relative risk (RR) of HF ≥2.5, a minimum of 209 exposed and 418 nonexposed subjects (patients with and without obesity, respectively) was required. The proportion in the nonexposed group has been estimated to be 4.5%.4 We anticipated a dropout rate of 20%.

Kaplan-Meier survival curves were used to estimate cumulative incidence, and log-rank test to determine the significance of differences. Cox proportional hazards regression was used to examine the association between obesity and the risk of HF. Presentation of HF, codified dichotomously, was considered a dependent variable. The following dichotomous variables were considered independent: obesity or overweight (main variable); hypertension, DM, and ischemic heart disease (associated with HF in prior epidemiological studies); and other variables if P < 0.05. The magnitude of association was determined using hazard ratio (HR) together with its 95% confidence interval (CI). The population attributable risk (PAR) of HF was calculated using probability of disease, and also conditional probability of disease in the nonexposed, adjusted for confounding factors:

equation image

Where:

PAR: Population attributable risk.

P(D) = average probability of disease (D) in the population (containing both exposed and unexposed individuals).

P(D|Ē,C) = marginal conditional probability of disease (D) given no exposure (Ē), averaged over strata of potential confounders (C).

The CIs of the PAR estimates obtained were estimated by the bootstrap method.8

A 2-sided P value of <0.05 was adopted in all cases.

Results

Follow-up was completed in 932 cases: 245 (26.3%) patients were overweight and 362 (38.8%) were obese (Table 1); 77 cases (7.6% of the original sample) were lost. Hypertension, systolic and diastolic BP, HDL, type 2 DM, and fasting glucose measurements in obese and overweight patients showed a more unfavorable CV risk profile (Table 1). Median follow-up was 119.8 months.

Table 1. Participant Baseline Characteristics According to BMI
 BMI <25, n = 325BMI 25–29.9, n = 245BMI ≥30, n = 362P Value
  1. Abbreviations: BMI, body mass index; BP, blood pressure; F, female; HDL, high-density lipoprotein; T2DM, type 2 diabetes mellitus.

Age56.4 ± 13.158.8 ± 12.458.8 ± 11.70.020
F (%)199 (61.2)125 (51.0)233 (64.4)0.004
Current smoking (%)124 (38.2)95 (38.8)109 (30.1)0.035
Hypertension (%)101 (31.1)88 (35.9)194 (53.6)<0.001
Systolic BP128.1 ± 18.6130.3 ± 16.5136.6 ± 18.4<0.001
Diastolic BP77.9 ± 10.578.8 ± 10.082.3 ± 10.1<0.001
Hypercholesterolemia (%)77 (22.7)67 (27.3)101 (27.9)0.415
Total cholesterol (mmol/L)5.6 ± 1.05.7 ± 0.95.7 ± 1.00.980
HDL cholesterol (mmol/L)1.5 ± 0.41.4 ± 0.41.3 ± 0.30.002
T2DM (%)37 (11.4)41 (16.7)66 (18.2)0.038
Fasting glucose (mmol/L)5.7 ± 1.75.9 ± 1.96.0 ± 1.90.031
Ischemic heart disease (%)18 (5.5)15 (6.1)19 (5.2)0.899
Cerebrovascular disease (%)18 (5.5)9 (3.7)12 (3.3)0.312
Peripheral artery disease (%)6 (1.8)4 (1.6)7 (1.9)0.963
Valvular heart disease (%)2 (0.6)3 (1.2)4 (1.1)0.718
BMI24.2 ± 2.128.4 ± 0.933.7 ± 4.1<0.001

Heart failure developed in 26 participants (2.8%); 3 of these patients died of HF during the follow-up. Fourteen patients had systolic HF (EF <40%), and 12 had nonsystolic HF (EF ≥40% with ventricular dysfunction).

The difference in HF rate between obese subjects (4.7% incidence) and nonobese subjects (1.6% incidence) was 3.1% (95% CI: 0.7–5.5). The survival analysis (Figure 1) shows a superior risk of HF in obese patients (log-rank test P = 0.005).

Figure 1.

BMI and the risk of incident heart failure. Abbreviation: BMI, body mass index.

In the unadjusted model (Table 2), incident HF was significantly associated with BMI. The HR was 1.09 for every 1 kg/m2 increase (95% CI: 1.05–1.14) and 3.01 for BMI ≥30 (95% CI: 1.34–6.77). Overweight was not associated with HF. In the model adjusted by age and sex (Table 2), the results were similar. HR was 1.08 for every 1 kg/m2 increase (95% CI: 1.04–1.14) and 2.92 for BMI ≥30 (95% CI: 1.08–7.92). Similarly, overweight was not associated with HF. The model adjusted by age, sex, hypertension, ischemic heart disease, and DM (Table 2) also produced results similar to the unadjusted model. HR was 1.06 for every 1 kg/m2 increase (95% CI: 1.01–1.10) and 2.45 for BMI ≥30 (95% CI: 1.02–5.61). Again, overweight was not associated with HF.

Table 2. Hazard Ratio Models for Incident Heart Failure
 Cases (%)Adjusted HR (95% CI)P Value
  • Abbreviations: BMI, body mass index; CI, confidence interval; DM, diabetes mellitus; HR, hazard ratio.

  • a

    Hypertension, ischemic heart disease, and DM.

Unadjusted model   
 BMI as a continuous variable (1 kg/m2 increase)26 (2.8)1.09 (1.05–1.14)<0.001
 BMI 3 categories   
 BMI <255 (1.5)(Reference) 
 BMI 25–29.94 (1.6)1.07 (0.29–3.97)0.925
 BMI ≥3017 (4.7)3.01 (1.34–6.77)0.007
Model adjusted for age and sex   
 BMI as a continuous variable (1 kg/m2 increase)26 (2.8)1.08 (1.04–1.14)<0.001
 BMI categories   
 BMI <255 (1.5)(Reference) 
 BMI 25–29.94 (1.6)0.96 (0.26–3.59)0.952
 BMI ≥3017 (4.7)2.92 (1.08–7.92)0.036
Model adjusted for age, sex, and other variablesa   
 BMI as a continuous variable (1 kg/m2 increase)26 (2.8)1.06 (1.01–1.10)0.012
 BMI categories   
 BMI <255 (1.5)(Reference) 
 BMI 25–29.94 (1.6)0.79 (0.21–3.00)0.721
 BMI ≥3017 (4.7)2.45 (1.02–5.61)0.048

The PAR of HF due to obesity, hypertension, and ischemic heart disease was 43.0% (95% CI: 13.9–74.9), 50.0% (95% CI: 11.5–80.8), and 18.6% (95% CI: 0.2–37.0), respectively.

Discussion

In the present study, obesity was a strong predictor of incident HF. The magnitude of the effect was maintained after adjustment for other risk factors. Population attributable risk of HF due to obesity is high, and a growing epidemic of obesity in developed countries will increase this risk in the near future. Our results are consistent with previous studies in populations at high risk for CV diseases.

For example, Chen et al9 associated BMI ≥28 in elderly patients with incident HF. In the National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study,10 overweight (BMI ≥27.8 in men and ≥27.3 in women) was an independent risk factor. Kenchaiah et al11 reported a gradual increase in the risk of HF across BMI categories (overweight, obesity). Although obesity yielded statistical differences in both men and women, overweight was an independent risk factor only in women.11 In the Reykjavik Study,12 overweight and obesity were associated (P < 0.05) with incident HF. For obesity, the magnitude of risk was similar to that of other studies.11,12 The results of a nested case-control study13 are similar to those of the present study: a strong association between obesity and HF (RR: 2.1), and lack of association with overweight. Body-fat distribution may be a stronger risk factor for HF than obesity in general.14 In a multivariate model including waist circumference and BMI, waist circumference was associated with incident HF (HR: 1.27, 95% CI: 1.04–1.54 per SD); BMI was not.14 Paradoxically, higher BMI was associated with a better prognosis, independently of other variables, in patients with HF.15 Furthermore, in patients with established HF, a high percentage of body fat appears to be protective.16

Several plausible mechanisms explain this association between obesity and HF. Severe obesity has long been recognized as causing a form of cardiomyopathy, and BMI is a risk factor for hypertension, type 2 DM, and dyslipidemia, all of which are risk factors for myocardial infarction (a precursor of HF in many cases).15 BMI also has been associated with left ventricular remodeling17 and neurohormonal alterations.18 A direct effect of obesity on the myocardium has been reported in animal models (cardiac steatosis and lipoapoptosis).19 Therefore, evidence of the role of obesity in the diagnosis, pathophysiology, and prognosis of HF is very strong at present.20

The main strengths of this study are the random selection of subjects, long-term follow-up, and use of standardized objective diagnostic criteria, including echocardiography, for a diagnosis that is not generally based on an objective evaluation of cardiac function.21,22 A limitation is the low statistical power for detecting the significance of valvular heart disease and other variables (Table 2). Likewise, the small number of events limited the number of variables to include in the models (Table 2) and suggests that the HR and PAR results must be interpreted with caution; the calculation of PAR required wide confidence intervals.

Conclusion

In this study, obesity is an important risk factor for incident HF, due to attributable risk and high magnitude of effect. Improved control of hypertension, primary prevention of myocardial infarction, and secondary prevention measures are essential for the prevention of HF.23 Promotion of healthy body weight may reduce the risk of HF and other CV events.15,24 Alternatively, the epidemic of obesity1,2 in developed countries provides cause for concern about increased CV risk in the near future.

Acknowledgements

The authors are grateful to Dr. Jaume Marrugat for reviewing and commenting of the final version of this manuscript, and Elaine Lilly, PhD, for revision of the English text.

Ancillary