SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

Objective

To develop and validate a mortality risk algorithm for obese black and white men and women to elucidate risk factors prognostic of short-term mortality among obese persons.

Methods

Prospective cohort study. Reasons for geographic and racial differences in stroke (REGARDS) study, is a cohort of black and white men and women aged ≥45 years. Obese (≥30 kg m−2) participants in REGARDS (n = 11 288) were randomly assigned to the derivation data set or an independent validation set.

Results

During the mean follow-up period of 4.9 years, 8.9% (n = 504) in the derivation cohort and 8.7% (n = 492) in the validation cohort died. The best-fitting model based on data from the derivation cohort included demographic (age, sex), coronary heart disease (CHD) conditions (diabetes, systolic blood pressure, history of CHD), health behaviors (smoking, physical activity, alcohol use), and socioeconomic variables (income, use of physician services). The C-statistic when the model was applied to the validation cohort was 0.80. Observed and predicted rates of mortality were similar across deciles of mortality risk by race.

Conclusions

A risk algorithm was established and validated to predict mortality among black and white obese subjects based on CHD risk factors, behavioral risk factors, and socioeconomic status.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

Obesity has reached epidemic proportions in the United States. Distinguishing the obese at highest risk for adverse outcomes is a difficult challenge given that body mass index (BMI), a surrogate measure of adiposity, is associated with mortality only at the extremes of the distribution [1-6]. Furthermore, among individuals with existing coronary heart disease (CHD), higher BMI is paradoxically protective against mortality [7]. Recent evidence is building that an obese subgroup exists that exhibit higher risks of death due to excess adiposity [8, 9]. Our recent effort to delineate this subgroup was unsuccessful using C-reactive protein (CRP) as a biomarker of mortality risk among obese subjects [5]. Strategies to predict and alter risk for the obese are lacking, though this is critical to guide interventions and education in the clinical setting.

The reasons for geographic and racial differences in stroke (REGARDS) study is a national, population-based, longitudinal study of black and white men and women aged ≥45 years [10]. In the current study, we tested the ability of CHD, behavioral and socioeconomic risk factors to predict mortality among obese black and white individuals. Furthermore, we developed and validated an algorithm to allow clinicians and patients to assess 5-year mortality risk for obese individuals.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

The REGARDS study enrolled 30,239 participants between January 2003 and October 2007 [10]. Fifty-six percent of participants resided in the stroke belt (North Carolina, South Carolina, Georgia, Alabama, Mississippi, Tennessee, Arkansas, and Louisiana), and the remaining were from the other 40 contiguous states. Individuals were recruited from a commercial list using mail and telephone. Participants were excluded from participation if they had a history of cancer or had a medical illness preventing participation in the study. Of the 30,239 participants, 11,288 were obese (≥30 kg m−2). To develop a risk algorithm, ½ of the obese REGARDS participants were randomly assigned to a derivation data set and the other half to an independent validation set. Study methods were reviewed and approved by the institutional review boards of each study institution.

Risk factor ascertainment

Demographics and medical history were obtained by telephone interview. At an in-home examination, blood pressure, anthropomorphic measures, blood samples, electrocardiogram (ECG), and medication inventory were collected. Region of residence was categorized as the stroke belt, stroke buckle, or the remainder of the contiguous states. Anthropomorphic measures were obtained using standard protocols by trained staff. BMI was calculated as weight in kilograms divided by the square of height in meters and waist circumference reported in centimeters. Waist circumference was measured over skin or lightweight clothing at the midpoint between the lowest rib on the right side and the top of the iliac crest using a cloth tape measure at the end of expiration. High waist circumference was defined as >88 cm for women and >102 cm for men. Resting blood pressure was measured two times in the seated position and the average of the two readings was used. Medication for treatment of hypertension and diabetes were determined by self-report. Hypertension was defined as systolic blood pressure 140 mm Hg or greater, diastolic pressure 90 mm Hg or greater or self-reported physician diagnosis of hypertension with use of antihypertensive medications. Diabetes was defined as elevated glucose or self-reported history with use of diabetic medications. Elevated glucose was defined as fasting value >126 mg dL−1, or if the participant was not fasting, a value >200 mg dL−1. History of coronary heart disease (CHD) was defined as myocardial infarction by self-report or ECG, or history of coronary revascularization (stenting, coronary artery bypass surgery, or percutaneous transluminal coronary angioplasty).

We defined alcohol use according to the National Institute on Alcohol Abuse and Alcoholism classifications i.e., moderate (1 drink per day for women or 2 drinks per day for men) and heavy use (>1 drink per day for women or 2 drinks per day for men) [11]. Smoking status (never, former, current), educational level (less than high school, high school, some college, college and above), and physical activity of study participants were obtained by questionnaire. Moderate to vigorous physical activity was defined by whether the physical activity was intense enough to work up a sweat (none, one to three times a week, or four or more times a week). Single-item physical activity assessments have been previously validated [12] and prognostic of mortality in a recent study in both blacks and whites [13]. Income was defined as a yearly household income (less than $20K, 20–34K, 35–74K, 75K and above, and a separate category for those who declined to disclose their income). Use of physician services was determined by participant questionnaire response, “Do you have a clinic or doctor that provides your usual medical care?”

Laboratory methods

Phlebotomy was performed in the home by trained personnel using standardized procedures. Laboratory assays for cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, and glucose were previously described [14]. Insulin was measured using the Roche Elecsys 2010 system (Roche Diagnostics, Indianapolis, IN) which utilizes an electrochemiluminescence immunoassay method. The measuring range was 0.200–1000 μU mL−1. The inter-assay CV is <5%.

Mortality outcome

Participants are followed by telephone every 6 months for surveillance of medical events. If a proxy reports a participant has died, an interview is conducted with the next of kin listed on study forms. The social security death index, National Death Index, and death certificates are used to confirm dates of death and supplement proxy information on vital status. Time to death is recorded as the time from baseline home assessment to date of death.

Statistical analysis

Characteristics of the validation and derivation datasets were compared using t tests for continuous variables and chi-square tests of association for categorical variables. The primary outcome was all-cause mortality, and Cox proportional hazards models were used to determine the best model in the derivation data set. To find the best fitting model in the derivation cohort, all exposure variables were considered in a stepwise selection process and the model with minimum Bayesian information criterion (BIC) was selected as the final model. Once the best-fitting prediction model was determined, it was prospectively tested in the validation data set using two global measures, the integrated Brier's score for survival data and Yates' slope. Predictive accuracy of the model was further assessed by examining two components of accuracy: discrimination and calibration. Discrimination was evaluated using the C-statistic modified for survival data developed by Pencina and D'Agostino [15]. To assess model calibration, the Hosmer-Lemeshow calibration statistic comparing observed and predicted risk was computed based on deciles of predicted risk. After development and validation of the multivariable model, the 5-year risk of death for each individual was estimated by using 5-year baseline hazard at the means of the risk factors and corresponding regression coefficients from the best-fitting model. Statistical significance was assessed at the 0.05 level. All data management and statistical analyses were conducted using SAS 9.2/9.3 software.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

Characteristics of the obese REGARDS participants in the derivation (n = 5,644) and validation (n = 5,644) cohorts are shown in Table 1. The derivation and validation cohorts were similar by demographic, socioeconomic, and risk factor status. Over 50% of the participants were black and 62% were women. The mean waist circumference of the obese participants was 106 ± 13 cm and 112 ± 12 cm for black women and men, respectively, and 104 ± 13 cm and 113 ± 11 cm for white women and men, respectively. During the mean follow-up period of 4.9 years, 8.9% (n = 504) in the derivation cohort and 8.7% (n = 492) in the validation cohort died.

Table 1. Baseline characteristics for obese derivation and validation dataset in REGARDS
Characteristic [mean (SD) or frequency (%)]Derivation Dataset; (n = 5,644, 50%)Validation dataset; (n = 5,644, 50%)
Event504 (8.9%)492 (8.7%)
Age  
45-54804 (14.3%)782 (13.9%)
55-642,437 (43.2%)2,497 (44.2%)
65-741,772 (31.4%)1,734 (30.7%)
≥75631 (11.2%)631 (11.2%)
Male gender2,132 (37.8%)2,151 (38.1%)
Black race2,861 (50.7%)2,975 (52.7%)
Region  
Stroke belt1,975 (35.0%)1,934 (34.3%)
Stroke buckle1,208 (21.4%)1,226 (21.7%)
Stroke nonbelt2,461 (43.6%)2,484 (44.0%)
Income  
<$20k1,162 (20.6%)1,201 (21.3%)
$20k-$34k1,402 (24.8%)1,387 (24.6%)
$35k-$74k1,661 (29.4%)1,653 (29.3%)
$75k and above764 (13.5%)750 (13.3%)
Refused655 (11.6%)653 (11.6%)
Education  
<High school784 (13.9%)859 (15.2%)
High school graduate1,544 (27.4%)1,562 (27.7%)
Some college1,611 (28.6%)1,565 (27.8%)
College graduate and above1,701 (30.2%)1,651 (29.3%)
Smoking  
Current689 (12.3%)642 (11.4%)
Past2,301 (40.9%)2,367 (42.1%)
Never2,632 (46.8%)2614 (46.5%)
Alcohol consumption  
Heavy159 (2.9%)137 (2.5%)
Moderate1,541 (27.9%)1,555 (28.2%)
None3,821 (69.2%)3,830 (69.4%)
History of coronary heart disease1021 (18.5%)1,016 (18.3%)
Glucose (mg dL−1)111.9 (41.8))112.8 (43.3)
Insulin (µIU mL−1)265 (4.7%)271 (4.8%)
Systolic blood pressure (mm Hg)*130.5 (16.2)130.5 (16.0)
Diastolic blood pressure (mm Hg)*78.7 (9.7)78.6 (9.8)
Hypertension4,053 (72.0%)3,990 (70.8%)
Diabetes1,814 (33.3%)1,820 (33.4%)
Triglycerides (mg dL−1)144.8 (89.7)143.0 (90.3)
HDL cholesterol (mg dL−1)48.7 (14.6)48.6 (14.3)
LDL cholesterol (mg dL−1)113.7 (35.1)113.5 (35.0)
Waist circumference (Male > 102 cm, Female > 88 cm)4,988 (88.8%)4,960 (88.3%)
Use of statin therapy1,960 (34.7%)1,976 (35.0%)
Aspirin use2,521 (44.7%)2,476 (43.9%)
Physical activity  
One to three times per week2,059 (37.0%)2,001 (36.0%)
Four or more times per week1,263 (22.7%)1,349 (24.3%)
None2,247 (40.4%)2,214 (39.8%)
Use of physician services4,145 (80.0%)4,099 (79.2%)

In the model derivation cohort, the 23 variables presented were evaluated as potential predictors of death with univariate associations are shown in the Supporting Information Table. The best-fitting model based on data from the model derivation cohort is presented in Table 2. A set of demographic (age, sex), CHD (diabetes, systolic blood pressure, history of CHD), health behaviors (smoking, physical activity, alcohol use), and socioeconomic (income, use of physician services) variables produced the lowest BIC (6426.3). The Brier's score and Yates' slope were 0.058 and 0.09, respectively.

Table 2. Final model for derivation dataset in REGARDS
VariableParameter estimate (standard error)HR (95% CI)P value
Age   
55-64 vs. 45-540.29 (0.28)1.34 (0.78, 2.31)0.29
65-74 vs. 45-541.01 (0.27)2.74 (1.60, 4.69)0.0002
≥75 vs. 45-541.64 (0.28)5.17 (2.98, 8.97)<0.0001
Gender   
male vs. female0.52 (0.11)1.68 (1.36, 2.09)<0.0001
Smoking status   
Never vs. current−0.70 (0.16)0.50 (0.37, 0.68)<0.0001
Past vs. current−0.44 (0.15)0.65 (0.48, 0.87)0.004
Alcohol consumption   
Heavy vs. none−0.20 (0.29)0.82 (0.46, 1.44)0.48
Moderate vs. none−0.56 (0.13)0.57 (0.44, 0.74)<0.0001
Diabetic status   
yes vs. no0.41 (0.10)1.50 (1.23, 1.83)<0.0001
Systolic blood pressure0.007 (0.002)1.01 (1.00, 1.01)0.008
Income   
$20k-$34k vs. <$20k−0.19 (0.13)0.83 (0.64, 1.07)0.15
$35k-$74k vs. <$20k−0.45 (0.15)0.64 (0.48, 0.85)0.002
$75k and above vs. <$20k−0.65 (0.23)0.52 (0.332, 0.816)0.004
Refused vs. <$20k0.003 (0.17)1.00 (0.72, 1.40)0.98
History of coronary heart disease   
yes vs. no0.41 (0.11)1.50 (1.21, 1.86)0.0002
Physical activity   
One to three times per week vs. none−0.40 (0.12)0.67 (0.54, 0.85)0.0006
≥4 times per week' vs. none−0.38 (0.13)0.69 (0.53, 0.89)0.004
Use of physician services   
yes v/s no−0.25 (0.11)0.78 (0.62, 0.97)0.026

Among the prognostic variables, diabetes was associated with a two times higher risk of mortality (HR 2.0, 95% CI: 1.70-2.43) compared to those without diabetes. Subjects who engaged in moderate to vigorous physical activity one to three or more than four times a week had a 41% (HR 0.59, 95% CI: 0.46-0.74) and 45% (HR 0.55, 95% CI: 0.45-0.68) lower mortality, respectively, compared to those who were not physically active. Higher family income (≥75 K vs. <20 K) was associated with lower mortality (HR 0.27, 95% CI: 0.19-0.40) as was use of physician services (HR 0.76, 95% CI: 0.62-0.93). Notably, race and region did not predict mortality in the obese derivation cohort so were not included in the final model (data not shown).

The C-statistic when the model was applied to the validation cohort was 0.80. The observed and predicted mortality across deciles of risk in the validation set was closely aligned, with a Hosmer-Lemeshow calibration statistic of 1.15, (P value = 0.99). Figure 1 illustrates the observed and predicted rates of mortality across deciles of mortality risk by race.

image

Figure 1. Absolute 5-year mortality risk deciles for Whites (TOP) and Blacks (BOTTOM).

Download figure to PowerPoint

Based on the predictive model, a 5-year mortality risk equation was developed:

5-year risk (%) = [1 − 0.94942 (exp [A-1.4533185]] × 100% where

A = 0.29455 (if age = “55–64”) + 1.00962 (if age = “65–74”) + 1.64291 (if age ≥75) + 0.52096 (if male) – 0.18536 (if income = “$20k-$34k”) – 0.45015 (if income = “$35k-$74k”) – 0.65304 (if income ≥$75k) + 0.003 (if income = “Refused”) − 0.6979 (if non-smoker) – 0.43817 (if past smoker) − 0.20358 (if “heavy” alcohol use) − 0.56367 (if “moderate” alcohol use) + 0.40706 (diabetic) + 0.40564 (history of coronary heart disease) + 0.00765*SBP –0.39612 (one to three times physical activity) −0.37635 (≥4 times physical activity) −0.25358 (if medical care)

We generated absolute 5-year risk rates of death based on the validated model to illustrate application to the clinical setting. Obese men and women with no additional risk factors at age 55, 65, and 75 years had 5-year mortality risk as follows: women: 55 years, 0.3%; 65 years, 0.6%; 75 years, 1.2%; men: 55 years, 0.5%; 65 years, 1.1%; 75 years, 2.0%. Absolute risks increased with additional risk factors (Figure 2). For example, 65-year-old obese women and men with suboptimal CHD and behavioral risk factors who were low income and did not use physician services had a 38% and 55% absolute risk of dying within 5 years, respectively (black bar).

image

Figure 2. Absolute 5-year mortality risk according to underlying risk factors among obese women and men in REGARDS. None = best risk strata for all variables (no CHD risk or history, nonsmoker, physically active, and having highest SES levels). CHD risk factors or history = individuals with elevated risk from diabetes, systolic blood pressure, or a history of CHD. + behavioral risk factors = CHD risk or history plus being a smoker, no alcohol, physically inactive. + Socioeconomic factors = Having CHD risk and behavioral factors, plus the lowest level of income, and not having a usual source of medical care. CHD = coronary heart disease. SES = socioeconomic variables.

Download figure to PowerPoint

Lastly, we also explored whether the validated REGARDS obesity algorithm discriminated similarly in the nonobese REGARDS participants. The C-statistic for the model, utilizing the validated risk factors was 0.76 for the nonobese, which was statistically different from the obese population of 0.80 (P < 0.001). Furthermore, we substituted the validated risk factors in the model for a calculated Framingham Risk Score among the obese and the C-statistic was 0.73.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

We established and validated a parsimonious risk model to predict mortality among black and white obese middle aged to elderly subjects based on CHD risk factors, behavioral risk factors, and socioeconomic status. The validated risk score was a more powerful discriminator of mortality risk among the obese than non-obese in REGARDS. Our results suggest that there is a remarkable spectrum of risk within the obese population, with presence or absence of predictive factors implying an over 60-fold difference in the 5-year risk of death between low-risk obese and high risk individuals. On the basis of these results, we developed a web-based tool that may be used by clinicians and obese persons to determine absolute mortality risk based on their underlying risk factors (http://www.soph.uab.edu/riskcalculator). These findings provide useful information both for health care providers and patients for clinical decision-making, and for population health managers tasked with resource allocation decisions.

Risk algorithms for obese patients are novel. We are aware of only one other publication addressing mortality risk prediction among the obese. The Edmonton obesity staging system was developed to assess quality of life and impairment associated with being overweight or obese [16]. The scoring system included functional and psychological impairments, as well as symptoms, and may prove useful for understanding eligibility for interventions, such as bariatric surgery [17]. However, unlike the REGARDS algorithm, the staging does not provide an understanding regarding how treated risk factors, modifiable risk factors, and use of physician services play into underlying short-term risk for obese men and women. Furthermore, the REGARDS algorithm provides education for obese persons via a web-based tool regarding how existing risk factors may alter future mortality risk.

There is no association between obesity and mortality among individuals with existing CHD [7] and or diabetes (the obesity paradox) [18]. Lack of precision in measuring body composition, protective effects of fat tissue, unmeasured confounding risk factors, and biases of observational studies are several of the cited explanations [19-21]. Regardless, these findings pose a dilemma regarding how to approach weight loss or diet interventions among obese patients with existing chronic conditions. Uniquely, the REGARDS algorithm allows health professionals to address short-term mortality risk among obese patients with existing CHD or diabetes independent of the obesity paradox findings; we contend this is a clinically useful approach that provides structure for health professionals to make recommendations regarding future health risks.

Utilizing the REGARDS algorithm, the absolute 5-year risk for the obese with no other risk factors was low (0.5-1.1% for 55–65 year old men and 0.3-0.6% for 55–65 year old women), though the addition of behavioral risk factors altered mortality risk significantly. In particular, smoking, moderate use of alcohol, and lack of physical activity were important behaviors associated with death in the validated model. Smoking has consistently been reported as an important risk factor among the obese [22, 23]. Moderate alcohol use, in healthy and metabolically diseased individuals, has a demonstrated survival benefit [24, 25]. Moreover, multiple studies, including the current one, provided evidence that physical fitness can mitigate or reverse the adverse effects of obesity, fueling the fit/fat controversy [26, 27]. To our knowledge the present study is the only one to date to determine how risk might change based on the presence or absence of modifiable health behaviors among the obese. This is unique to other standard risk scores (e.g., Framingham Risk Score) which do not incorporate health behaviors such as physical activity, alcohol use or socioeconomic factors.

Disparities in health outcomes have been attributed to race, geography, socioeconomic factors such as income and educational status, as well as utilization and access to health care services [28]. Many of these factors are intertwined, making it difficult to elucidate which factors are truly associated with outcome of interest. This has been especially problematic for obese blacks, who demonstrate no increased risk in mortality, even at the highest BMI distributions [5]. Differences in disease burden and death rates among blacks as well as different BMI cutoffs and body composition make any assessment of mortality risks difficult at best [29, 30]. Despite these limitations, focusing on ways to identify disparity-related mortality risks among obese individuals is critical. We addressed this concern in the current study; blacks and whites had similar absolute risk estimates across risk deciles in models that included socioeconomic factors. Furthermore, income and use of physician services (not education or geography) provided the best-fit within a mortality model among the obese. To our knowledge, this is the first study to elucidate such simple and dramatic findings regarding health disparities and mortality outcomes among the obese.

Limitations and strengths of the current study require consideration. Our risk scoring algorithm is only applicable to obese black and white men and women, not other ethnicities, body weights, or individuals <45 years of age. The study excluded those with some types of existing chronic diseases such as active cancer or a medical condition limiting participation in an epidemiological study. However, the findings should be of interest to most community-dwelling obese individuals. Importantly, not all of the factors in the risk prediction model have been assessed for causality in clinical trials. Therefore, it is possible that modifying some of the factors may not lower mortality dramatically.

Strengths of the study include the breadth of the risk factors evaluated for the derivation model in the obese population. Importantly, none of the validated risk factors among the obese were blood test results (i.e., LDL and HDL cholesterol, insulin levels) which make the model easy and inexpensive to apply. All of the factors are immediately known by the patient, making it more applicable clinically. The sample size of REGARDS implied there were over 11,000 obese individuals, and there were ∼1,000 deaths among these individuals. This large sample size permitted split sample replication approaches with ∼5,000 individuals with 500 deaths in both the derivation and validation cohorts; providing stable estimation and precise validation. The risk-prediction formula is complex without a point scoring scale, so may be difficult for a clinician to directly use; however, this concern was addressed by providing a web-based tool to perform calculations. Use of BMI to define excess adiposity was utilized rather than waist circumference or waist-to-hip ratio given the ease of use in the clinical setting. Lastly, our outcome of interest was all-cause mortality. We purposefully did not evaluate the utility of this model to predict cardiovascular or cancer-related outcomes. Obesity is associated with multiple chronic diseases which can lead to premature death. We determined that an algorithm based on risk of death had broader clinical application and higher probability to motivate patients (given death is applicable to everyone).

In summary, we provide an easily assessed, clinically relevant mortality prediction tool to enhance decision-making for health professionals regarding factors contributing importantly to mortality risk in obese persons (http://www.soph.uab.edu/riskcalculator). Policy-makers may also find these results helpful in targeting resources to manage populations at risk. Given these data, future work is now urgently needed to educate obese individuals regarding the key risk factors for heightened short-term mortality.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Neurological Disorders and Stroke or the National Institutes of Health. Representatives of the funding agency have been involved in the review of the manuscript but not directly involved in the collection, management, analysis, or interpretation of the data. The authors thank the other investigators, the staff, and the participants of the REGARDS study for their valuable contributions. A full list of participating REGARDS investigators and institutions can be found at http://www.regardsstudy.org

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information
  • 1
    Adams KF, Schatzkin A, Harris TB, et al. Overweight, obesity, and mortality in a large prospective cohort of persons 50 to 71 years old. N Engl J Med 2006;355:763-778.
  • 2
    Stevens J, Cai J, Pamuk ER, et al. The effect of age on the association between body-mass index and mortality. N Engl J Med 1998;338:1-7.
  • 3
    Pischon T, Boeing H, Hoffmann K, et al. General and abdominal adiposity and risk of death in Europe. N Engl J Med 2008;359:2105-2120.
  • 4
    Durazo-Arvizu R, Cooper RS, Luke A, et al. Relative weight and mortality in U.S. blacks and whites: findings from representative national population samples. Ann Epidemiol 1997;7:383-395.
  • 5
    Lakoski SG, Le AH, Muntner P, et al. Adiposity, inflammation, and risk for death in black and white men and women in the United States: the reasons for geographic and racial differences in stroke (REGARDS) study. J Clin Endocrinol Metab 2011;96:1805-1814.
  • 6
    Berrington de Gonzalez A, Hartge P, Cerhan JR, et al. Body-mass index and mortality among 1.46 million white adults. N Engl J Med 2010;363:2211-2219.
  • 7
    Romero-Corral A, Montori VM, Somers VK, et al. Association of bodyweight with total mortality and with cardiovascular events in coronary artery disease: a systematic review of cohort studies. Lancet 2006;368:666-678.
  • 8
    Ogorodnikova AD, Kim M, McGinn AP, et al. Incident cardiovascular disease events in metabolically benign obese individuals. Obesity (Silver Spring) 2012;20:651-659.
  • 9
    Stefan N, Kantartzis K, Machann J, et al. Identification and characterization of metabolically benign obesity in humans. Arch Inter Med 2008;168:1609-1616.
  • 10
    Howard VJ, Cushman M, Pulley L, et al. The reasons for geographic and racial differences in stroke study: objectives and design. Neuroepidemiology 2005;25:135-143.
  • 11
    National Institute on Alcohol Abuse and Alcoholism. Alcoholism NIoAAa: FAQ for the General Public, 2010. Available at: http://www.niaaa.nih.gov/FAQS/General-English/default.htm (accessed on 5 November 2012).
  • 12
    Schechtman KB, Barzilai B, Rost K, et al. Measuring physical activity with a single question. Am J Pub Health 1991;81:771-773.
  • 13
    Mathieu RA, Powell-Wiley TM, Ayers CR, et al. Physical activity participation, health perceptions, and cardiovascular disease mortality in a multiethnic population: the Dallas Heart Study. Am Heart J 2012;163:1037-1040.
  • 14
    Cushman M, McClure LA, Howard VJ, et al. Implications of increased C-reactive protein for cardiovascular risk stratification in black and white men and women in the US. Clin Chem 2009;55:1627-1636.
  • 15
    Pencina MJ, D'Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;23:2109-2123.
  • 16
    Padwal RS, Pajewski NM, Allison DB, et al. Using the Edmonton obesity staging system to predict mortality in a population-representative cohort of people with overweight and obesity. Can Med Assoc J 2011;183:E1059-E1066.
  • 17
    Mechanick JI, Apovian CM. Rethinking our approach to the obese patient: time to create the obesity treatment score. Curr Opin Endocrinol Diabet Obes 2012;19:339-340.
  • 18
    Carnethon MR, De Chavez PJ, Biggs ML, et al. Association of weight status with mortality in adults with incident diabetes. JAMA 2012;308:581-590.
  • 19
    von Haehling S, Hartmann O, Anker SD. Does obesity make it better or worse: insights into cardiovascular illnesses. Eur Heart J 2012;34:330332.
  • 20
    Florez H, Castillo-Florez S. Beyond the obesity paradox in diabetes: fitness, fatness, and mortality. JAMA 2012;308:619-620.
  • 21
    Lavie CJ, Milani RV, Ventura HO. Obesity and the “obesity paradox” in cardiovascular diseases. Clin Pharmacol Therap 2011;90:23-25.
  • 22
    Freedman DM, Sigurdson AJ, Rajaraman P, et al. The mortality risk of smoking and obesity combined. Am J Prevent Med 2006;31:355-362.
  • 23
    Wei M, Kampert JB, Barlow CE, et al. Relationship between low cardiorespiratory fitness and mortality in normal-weight, overweight, and obese men. JAMA 1999;282:1547-1553.
  • 24
    Thun MJ, Peto R, Lopez AD, et al. Alcohol consumption and mortality among middle-aged and elderly US adults. N Engl J Med 1997;337:1705-1714.
  • 25
    Beulens JW, Algra A, Soedamah-Muthu SS, et al. Alcohol consumption and risk of recurrent cardiovascular events and mortality in patients with clinically manifest vascular disease and diabetes mellitus: the Second Manifestations of ARTerial (SMART) disease study. Atherosclerosis 2010;212:281-286.
  • 26
    Stevens J, Cai J, Evenson KR, et al. Fitness and fatness as predictors of mortality from all causes and from cardiovascular disease in men and women in the lipid research clinics study. Am J Epidemiol 2002;156:832-841.
  • 27
    Ortega FB, Lee DC, Katzmarzyk PT, et al. The intriguing metabolically healthy but obese phenotype: cardiovascular prognosis and role of fitness. Eur Heart J 2012;34:389397.
  • 28
    Braveman P. Health disparities and health equity: concepts and measurement. Annu Rev Public Health 2006;27:167-194.
  • 29
    Kumanyika S. Obesity, health disparities, and prevention paradigms: hard questions and hardchoices. Prevent Chronic Dis 2005;2:A02.
  • 30
    Stevens J, Juhaeri J, Cai J, et al. The effect of decision rules on the choice of a body mass index cutoff for obesity: examples from African American and white women. Am J Clin Nutr 2002;75:986-992.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
oby20622-sup-0001-suppinfo01.docx21KSupplementary Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.