Obesity Indexes and Total Mortality among Elderly Subjects at High Cardiovascular Risk: The PREDIMED Study

Abstract

Background

Different indexes of regional adiposity have been proposed for identifying persons at higher risk of death. Studies specifically assessing these indexes in large cohorts are scarce. It would also be interesting to know whether a dietary intervention may counterbalance the adverse effects of adiposity on mortality.

Methods

We assessed the association of four different anthropometric indexes (waist-to-height ratio (WHtR), waist circumference (WC), body mass index (BMI) and height) with all-cause mortality in 7447 participants at high cardiovascular risk from the PREDIMED trial. Forty three percent of them were men (55 to 80 years) and 57% were women (60 to 80 years). All of them were initially free of cardiovascular disease. The recruitment took place in 11 recruiting centers between 2003 and 2009.

Results

After adjusting for age, sex, smoking, diabetes, hypertension, intervention group, family history of coronary heart disease, and leisure-time physical activity, WC and WHtR were found to be directly associated with a higher mortality after 4.8 years median follow-up. The multivariable-adjusted HRs for mortality of WHtR (cut-off points: 0.60, 0.65, 0.70) were 1.02 (0.78–1.34), 1.30 (0.97–1.75) and 1.55 (1.06–2.26). When we used WC (cut-off points: 100, 105 and 110 cm), the multivariable adjusted Hazard Ratios (HRs) for mortality were 1.18 (0.88–1.59), 1.02 (0.74–1.41) and 1.57 (1.19–2.08). In all analyses, BMI exhibited weaker associations with mortality than WC or WHtR. The direct association between WHtR and overall mortality was consistent within each of the three intervention arms of the trial.

Conclusions

Our study adds further support to a stronger association of abdominal obesity than BMI with total mortality among elderly subjects at high risk of cardiovascular disease. We did not find evidence to support that the PREDIMED intervention was able to counterbalance the harmful effects of increased adiposity on total mortality.

Trial Registration

Controlled-Trials.com ISRCTN35739639

About the Authors:

Miguel A. Martínez-González

* E-mail: [email protected]

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Preventive Medicine and Public Health, University of Navarra-Osasunbidea, Servicio Navarro de Salud, Pamplona, Spain

Ana García-Arellano

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Preventive Medicine and Public Health, University of Navarra-Osasunbidea, Servicio Navarro de Salud, Pamplona, Spain

Estefanía Toledo

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Preventive Medicine and Public Health, University of Navarra-Osasunbidea, Servicio Navarro de Salud, Pamplona, Spain

Maira Bes-Rastrollo

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Preventive Medicine and Public Health, University of Navarra-Osasunbidea, Servicio Navarro de Salud, Pamplona, Spain

Mónica Bulló

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Human Nutrition Department, Hospital Universitari Sant Joan, Institut d'Investigació Sanitaria Pere Virgili, Universitat Rovira i Virgili, Reus, Spain

Dolores Corella

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, Department of Preventive Medicine, University of Valencia, Valencia, Spain

Montserrat Fito

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, Cardiovascular and Nutrition Research Group, Institut de Recerca Hospital del Mar, Barcelona, Spain

Emilio Ros

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, Lipid Clinic, Department of Endocrinology and Nutrition, Institut d'Investigacions Biomèdiques August Pi i Sunyer, Hospital Clinic, University of Barcelona, Barcelona, Spain

Rosa Maria Lamuela-Raventós

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Nutrition and Food Science, School of Pharmacy, Xarxa de Referència en Tecnologia dels Aliments, Instituto de Investigación en Nutrición y Seguridad Alimentaria, University of Barcelona, Barcelona, Spain

Javier Rekondo

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Cardiology, University Hospital of Alava, Vitoria, Spain

Enrique Gómez-Gracia

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Preventive Medicine, University of Malaga, Malaga, Spain

Miquel Fiol

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, Institute of Health Sciences, University of Balearic Islands, and Hospital Son Espases, Palma de Mallorca, Spain

Jose Manuel Santos-Lozano

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, Department of Family Medicine, Primary Care Division of Seville, San Pablo Health Center, Seville, Spain

Lluis Serra-Majem

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Clinical Sciences, University of Las Palmas de Gran Canaria, Las Palmas, Spain

J. Alfredo Martínez

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Nutrition and Food Sciences, Physiology and Toxicology, University of Navarra, Pamplona, Spain

Sonia Eguaras

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, Department of Preventive Medicine and Public Health, University of Navarra-Osasunbidea, Servicio Navarro de Salud, Pamplona, Spain

Guillermo Sáez-Tormo

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Biochemistry and Molecular Biology-Service of Clinical Analysis- University of Valencia, Valencia, Spain

Xavier Pintó

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Lipids and Vascular Risk Unit, Internal Medicine, Hospital Universitario de Bellvitge, Hospitalet de Llobregat, Barcelona, Spain

Ramon Estruch

Affiliations Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos II, Madrid, Spain, The PREDIMED (Prevención con Dieta Mediterránea) Network (RD 06/0045) Instituto de Salud Carlos III, Madrid, Spain, Department of Internal Medicine Institut d'Investigacions Biomèdiques August Pi i Sunyer, Hospital Clinic, University of Barcelona, Barcelona, Spain

Introduction

The dose-response relationship between increased levels of overweight/obesity and all-cause mortality is controversial. A recent meta-analysis reported that overweight (defined as abody mass index [BMI] of 25 to 30 kg/m²) was associated with significantly lower all-cause mortality in comparison with the normal weight category and that grade I obesity (BMI between 30 and 35) was associated with a non-significantly lower risk of mortality [1]. Other previous meta-analyses also suggested inverse or null associations, specially among elderly subjects [2], [3]. In contrast with these reports, there is a high biological plausibility to expect a direct association between excess body weight and all-cause mortality even at moderate levels of overweight. Some potential biases may have attenuated the association. They include insufficient adjustment for confounding by smoking, reverse causation bias due to disease-related weight loss [4], effect modification by age [5], [6], with an attenuation of the detrimental effects of overweight or even a reversion to beneficial effects in the elderly [7]. In addition, studies using BMI as the single relevant measurement of adiposity and not other aspects of body composition such as visceral fat or fat distribution, may miss the true dose-response curve between adiposity and all-cause mortality. Alternative measures of abdominal obesity are known to be superior to BMI in showing a positive association with the risk of cardiovascular disease or diabetes [8], [9], [10], [11]. In addition, a recent study has clearly shown significant associations between waist circumference or other indexes of abdominal obesity and all-cause mortality in white subjects [12]. Furthermore, recent studies have suggested that the BMI should be refined by measuring additional indexes of fat distribution namely waist circumference (WC), waist-to-hip ratio (WHR) or waist-to height ratio (WHtR) to better identify higher-risk subjects [13], [14].

WC has been often used to appraise abdominal obesity, but WC does not take differences in height into account, and subjects with a given WC will have more abdominal fat than taller subjects with the same WC [15]. The WHtR represents a further step because it also takes height into account. A systematic review found mean areas under receiving operator characteristic curves of 0.704, 0.693 and 0.671 for WHtR, WC and BMI, respectively as predictors of diabetes or cardiovascular disease [10].

We assessed the association between several anthopometric indexes (BMI, WC, WHtR, Height) and all-cause mortality in a Mediterranean cohort of elderly subjects at high cardiovascular risk included in the PREDIMED trial. The primary objective of the PREDIMED trial was to test the effect of a Mediterranean-style diet in primary cardiovascula prevention.

Methods

The protocol for the PREDIMED trial is available as supporting information at www.predimed.es. We have previously reported the design, objectives and methods of the PREDIMED trial in a specific publication [16]. The PREDIMED study is a multicenter, randomized, primary cardiovascular prevention trial conducted in Spain (www.predimed.es). The final results were reported in 2013 [17]. We randomly assigned participants to one of three diets: a Mediterranean diet (MeDiet) supplemented with extra-virgin olive oil, a MeDiet supplemented with mixed nuts, or a control group (allocated to receive advice to reduce dietary fat). All analyses were stratified by intervention group and adjusted for potential confounding.

During the analysis (with the exception of spline models, see below) we used the lowest group of each anthopometric index as the reference category.

The protocol of this study was approved by the Institutional Review Board of the University of Navarra and the Institutional Review Board of the Hospital Clinic (Barcelona). The Data Protection Agency is the Spanish National Agency (Agencia Española de Protección de Datos), name of the file: PREDIMED, the responsible person for the file is Miguel A. Martínez-González.

The trial is registered at http://www.controlled-trials.com/ISRCTN 35739639, and all clinical investigation have been conducted according to the principles expressed in the Declaration of Helsinki. Subjects who met entry requirements agreed to participate and provided written informed consent.

Subjects

We enrolled 7,447 participants. Forty three percent of them were men (55 to 80 years) and 57% were women (60 to 80 years). All of them were initially free of cardiovascular disease, but at high cardiovascular risk because they had at least three major cardiovascular risk factors out of six candidate risk factors or, alternatively, they were type 2 diabetics. The 6 candidate risk factors considered were: overweight/obesity, hypertension, elevated low-density lipoprotein cholesterol, low high-density lipoprotein cholesterol, current smoking, or family history of premature coronary heart disease. The specific cut-off points for these factors and the exclusion criteria have been previously described [16]. Participants can be assumed to be on stable weights at the time of recruitment for the trial. Energy restriction was not part of the PREDIMED nutritional intervention.

The recruitment took place in 11 recruiting centers between 2003 and 2009. Eighty-nine percent of candidate subjects who met entry requirements agreed to participate and provided written informed consent.

Measurements

Registered nurses who had been previously trained and certified to implement the PREDIMED protocol directly measured weight, height and WC of participants as previously described [16], [18], [19]. Height (m) and weight (kg) were measured with light clothing and no shoes with calibrated scales and a wall-mounted stadiometer, respectively; BMI was calculated as the weight in kilograms divided by the square of the height in meters; WC was measured midway between the lowest rib and the iliac crest using an anthropometric tape; in other previous studies, the protocol for WC measurement specified that WC should be measured at the minimal waist (33%), midpoint (26%) and umbilicus (27%). We selected the midpoint. However, the available evidence from a 2008 meta-analysis suggests that WC measurement protocol has no substantial influence on the association between WC, all-cause and CVD mortality, CVD and diabetes [20]; the WHtR was calculated as WC divided by height, both in centimeters. Blood pressure was measured in triplicate using a validated semiautomatic oscillometer with a 5-minute interval between each measurement and the subject in a sitting position (Omron HEM-705CP, Hoofddorp, The Netherlands). Hypertension was defined as a systolic blood pressure ≥140 mm Hg, a diastolic blood pressure ≥90 mm Hg, or the use of antihypertensive therapy.

Confounders assessment

Participants underwent a baseline interview that included the evaluation of cardiovascular risk factors and physician diagnoses of hypertension, diabetes and hipercolesterolemia. At the same time we gathered information about medical, socio-demographic, anthropometric, and lifestyle variables. We used the Minnesota validated physical activity questionnaire to assess leisure-time physical activity [21], [22], Time spent in several activities in minutes per day was multiplied by its typical energy expenditure, expressed in metabolic equivalent tasks (METs), then summed over all activities to yield a METs-min/d score for each participant. Taking into account that the relationship between leisure-time physical activity (METs-min/d) and mortality was not linear in our data, we used a polynomial model to adjust for METs-min/d, adding a quadratic term to the multivariable model. The age range in the validation studies was 18–60 years. Dietary habits were collected through a semi-quantitative 137-item Food Frequency Questionnaire previously validated in Spain [23].

Statistical analysis

We examined baseline characteristics of participants in each variable of interest according to quartiles or to predefined categories of anthropometric indexes. To compare means or percentages of each variable across quartiles (or pre-defined categories) of anthropometric indexes we used one-way ANOVA and chi-squared tests, respectively.

We used Cox regression models to assess the Hazard Ratios (HR) and their 95% confidence intervals for total mortality according to quartiles or to pre-defined categories of each anthropometric index.

For the multiple-adjusted model, the following potential confounders (all of them measured at baseline) were considered: age, sex, smoking, diabetes status, hypertensive status, intervention group and family history of CHD.

To assess the dose-response shape between adiposity and mortality we used restricted cubic splines models with 4 degrees of freedom, and considered the point associated with the lowest mortality as the reference value for each anthropometric index.

Finally, we also obtained the hazard ratios and their 95% confidence intervals for total mortality according to defined cut-off points of each anthropometric index. A p value <0.05 was considered statistically significant.

Analyses were performed using STATA version 12.1 (StataCorp, College Station, TX, USA).

Results

Table 1 shows baseline characteristics of the 7447 participants of PREDIMED according to quartiles (or pre-defined categories) of anthropometric indexes. We studied baseline characteristics by four variables of interest: WHtR, WC, BMI and height. Baseline risk factors (hypertension, overweight, type-2-diabetes) increased across increasing quartiles, as expected, with the exceptions of dyslipidemia and family history of premature CHD, that decreased with increasing quartiles. We also observed that adherence to the MeDiet tended to decrease across increasing quartiles of anthropometric indexes. Leisure-time physical activity followed the same inverse trend across adiposity indexes, but not across height quartiles, where a direct association was apparent. There were other differences in variables such as smoking or hypertension but they did not follow a consistent pattern across successive quartiles.

[Figure omitted. See PDF.]

Table 1. Baseline Characteristics of Participants according to categories of waist-to-height ratio, waist circumference, body mass index and height.

https://doi.org/10.1371/journal.pone.0103246.t001

The relationship between quartiles of anthropometric indexes and the risk of all-cause mortality is reflected in Table 2. We observed that the hazard ratio tended to be highest in the top quartiles for all anthropometric indexes.

[Figure omitted. See PDF.]

Table 2. Hazard Ratios (95% confidence intervals) for total mortality according to quartiles of the waist-to-height ratio, waist circumference, height and categories of body mass index.

https://doi.org/10.1371/journal.pone.0103246.t002

For WHtR, the hazard ratios slightly and non-significantly decreased in the second quartile and then increased in the upper quartiles. Thus, the age- and sex-adjusted HRs (95% confidence intervals) for categories of low, moderate, and high/very high WHtR compared to the lowest (reference) category were 0.88 (0.64–1.20), 1.01 (0.75–1.37) and 1.39 (1.02–1.88), respectively with a statistically significant linear trend (p = 0.027). In multivariable-adjusted models, the HRs were 0.98 (0.72–1.35), 1.01 (0.74–1.38), and 1.44 (1.05–1.97) for categories of low, moderate, and high/very high WHtR, respectively, with a significant linear trend (p = 0.026). Although the p for interaction with sex was not statistically significant (p for interaction WHtR × sex  = 0.34), when we separated men and women, no significant association was observed for men. But among women, a significant linear trend remained apparent (p = 0.045 and p = 0.026 for age-adjusted, and multivariable-adjusted models, respectively).

We observed that the hazard ratios for WC also increased across succesive quartiles. The age- and sex-adjusted HRs (95% confidence intervals) for quartiles of low, moderate, and high/very high WC compared to the lowest quartile (ref.) were 1.01 (0.72–1.42), 1.17 (0.85–1.61) and 1.42 (1.03–1.96) respectively. The linear trend was statistically significant (p = 0.020). After multivariable adjustment, the HRs were 1.03 (0.73–1.46), 1.18 (0.86–1.63) and 1.36 (0.98–1.90) with a significant linear trend (p = 0.046). Although the p for interaction with sex was not statistically significant (p for interaction WC × sex  = 0.53), when we separated men and women, the age-adjusted HRs were not significant among men. Among women the age -adjusted HRs were 1.42 (0.87–2.30), 1.43 (0.86–2.40) and 1.88 (1.13–3.15), respectively. After multivariable-adjustment, these HRs were 1.42 (0.87–2.30), 1.40 (0.84–2.34) and 2.02 (1.21–3.38). The linear trend among women was significant in both age-adjusted and multivariable-adjusted models (p = 0.018 and 0.012 respectively).

For BMI the HRs increased across successive quartiles although the results were not significant. The age- and sex-adjusted HRs (95% confidence intervals) for quartiles of low, moderate, and high/very high BMI compared to the lowest quartile (ref.) were 0.80 (0.60–1.08), 0.92 (0.68–1.24) and 1.22 (0.91– 1.65). After multivariable adjustment, these HRs were 0.80 (0.59–1.08), 0.84 (0.63–1.14) and 1.14 (0.85–1.53). The linear trend was not significant in any of both cases. The interaction with sex was not significant (p = 0.77). When we separated men and women, no significant association was observed for either category although among women the association was stronger.

When we classified the sample according to quartiles of height we also observed increased hazard ratios across successive quartiles. The p for interaction with sex was not significant (p = 0.35). When we studied each sex separately we observed that the HRs increased in both cases across successive quartiles. However among men this increase was higher, although in both cases, after multivariable adjustment, we observed no significant association between height and total mortality.

We conducted similar analyses using a priori defined cut-off points instead of quartiles. We observed in Table 3 that hazard ratios tended to be highest in the upper categories of the anthropometric indexes, in the same way as we observed in Table 2. We observed also that the hazard ratio decreased in the second category and then it increased in the upper categories.

[Figure omitted. See PDF.]

Table 3. Hazard Ratios (95% confidence intervals) for total mortality according to categories (a priori defined cut-off points) of the waist-to-height ratio and waist circumference.

https://doi.org/10.1371/journal.pone.0103246.t003

When we analyzed separately each anthropometric index, we found this same trend. In the case of WHtR (cut-off points: 0.60, 0.65, 0.70) we observed that the hazard ratio monotonically increased. The age and sex-adjusted Hazard ratios (95% confidence intervals) for each category were 1.01 (0.77–1.32), 1.28 (0.96–1.71), and 1.58 (1.11–2.26), for WHtR categories of 0.60–0.65, 0.65–0.70 and >0.70 with respect to <0.65, with a significant direct linear trend (p = 0.009). After multivariable adjustment, these HRs were 1.02 (0.78–1.34), 1.30 (0.97–1.75) and 1.55 (1.06–2.26), respectively with a significant linear trend (p = 0.013). Although the p for interaction with sex was not significant (p = 0.45), when we separated men and women, we observed no significant associations among men after multivariable adjustment. However among women, the multivariable-adjusted HRs for successive categories were 1.34 (0.81–2.22), 1.82 (1.09–3.06) and 1.84 (1.03–3.29). The linear trend among women was significant in both age-adjusted and multivariable-adjusted models (p = 0.036 and 0.016, respectively). This trend was also apparent in the spline analysis (Figure 1).

[Figure omitted. See PDF.]

Figure 1. Relative risk of death according to Waist-to-height ratio.

Restricted cubic spline model adjusted for age, smoking, diabetes status, hypertensive status, intervention group and family history of CHD.

https://doi.org/10.1371/journal.pone.0103246.g001

In the case of WC (cut-off points: 100, 105 and 110 cm) we observed that the hazard ratios increased across successively increasing categories with the exception of one of the intermediate groups (105 to 110 cm). The age-, and sex-adjusted HRs (95% confidence intervals) were 1.14 (0.85–1.53), 1.04 (0.76–1.43) and 1.69 (1.29–2.23) with a significant linear trend (p = 0.002). After multivariable adjustment, these HRs were 1.18 (0.88–1.59), 1.02 (0.74–1.41) and 1.57 (1.19–2.08) with a significant linear trend (p = 0.008). The p for interaction with sex was not significant (p = 0.73). When we divided the sample into men and women, no significant association was observed after multivariable adjustment among men. But the association remained apparent among women (cut-off points: 95,100, 105), with multivariable-adjusted HRs of 1.40 (0.85–2.31), 1.08 (0.60–1.94) and 1.93 (1.21–3.07), with a significant linear trend (p = 0.012).

The risk of total mortality across the different categories of BMI (<25, 25–30, 30–35, >35) is shown in the Table 4. We observed that the Hazard ratios decreased in the second group and then they increased in the upper categories. This trend was also apparent in the spline analysis (Figure 2). The p for interaction with sex was not significant (p = 0.43). The adjusted HRs obtained were not significant in men nor women when we stratified the sample by sex. We repeated these analyses after excluding deaths in the first follow-up year and also after excluding ever smokers. In these analyses we found slightly stronger direct associations between BMI and risk of death.

[Figure omitted. See PDF.]

Figure 2. Relative risk of death according to BMI.

Restricted cubic spline model adjusted for age, smoking, diabetes status, hypertensive status, intervention group and family history of CHD.

https://doi.org/10.1371/journal.pone.0103246.g002

[Figure omitted. See PDF.]

Table 4. Hazard Ratios (95% confidence intervals) for total mortality according to categories of body mass index.

https://doi.org/10.1371/journal.pone.0103246.t004

When we assessed height, we found increased risk of death in the taller groups with the exception of the second group for men where the HR decreased. The age-sex-adjusted HRs (95% confidence intervals) for each group (cut-off points: 1.65, 1.70, 1.75 cm) were 0.99 (0.77–1.28), 1.10 (0.80–1.52) and 1.57 (1.07–2.31). The linear trend was significant (p = 0.038). But after multivariable adjustment the HRs were not significant. Although the p for interaction was not significant (p = 0.33), when we separated men and women, the age-adjusted HRs were not significant for women (cut-off points: 1.55, 1.60, 1.65). Among men the age -adjusted HRs were 0.96 (0.69–1.34), 1.11 (0.76–1.62) and 1.73 (1.13–2.63), respectively, with a significant linear trend (p = 0.046). However, after multivariable adjustment this association became non-significant.

Almost fifty percent of our participants were diabetics. We investigated the relationship between WHtR and risk of death within the group of diabetic participants (table 5). We found a significant direct linear trend after multivariable adjustment in the diabetic group (p = 0.031). However no statistically significant interaction with diabetes was found (p = 0.85).

[Figure omitted. See PDF.]

Table 5. Hazard Ratios (95% confidence intervals) for total mortality according to quartiles of the waist-to-height ratio within diabetics or no diabetics persons.

https://doi.org/10.1371/journal.pone.0103246.t005

We also studied the relationship between the WHtR index and the risk of all-cause mortality separating the sample according to the randomly allocated group of dietary intervention (arm of the trial) in the Predimed trial: MeDiet with EVOO, MeDiet with nuts, and control diet (Table 6). We found a consistent direct association between WHtR and overall mortality within each of the three intervention arms of the trial. However, the p value for interaction was close to statistical significance (p = 0.056). A significant linear trend after multivariable adjustment suggesting a direct, dose-response relationship between baseline WHtR and mortality was apparent only within the group allocated to a MeDiet with nuts (p = 0.034). In any case, we did not find evidence to support that the PREDIMED intervention was able to counterbalance the harmful effects of increased adiposity on total mortality. Reassuringly, no significant increase in total mortality associated with the active intervention was apparent within the subgroup of participants with the highest levels of WHtR (p = 0.72 for MeDiet + EVOO and p = 0.64 for MeDiet + mixed nuts).

[Figure omitted. See PDF.]

Table 6. Hazard Ratios (95% confidence intervals) for total mortality according to quartiles of the waist-to-height ratio within each of the three intervention groups.

https://doi.org/10.1371/journal.pone.0103246.t006

We fitted restricted cubic spline models to show graphically the relationship between the WHtR (Figure 1) or BMI (Figure 2) and relative risk of death. In figure 1 we present the dose-response relationship between WHtR and total mortality. We found the highest risk of death in participants with the highest levels of abdominal obesity, as measured by the WHtR, and the lowest risk for a WHtR  = 0.6. When we assessed the shape of the dose-response association between BMI and total mortality (Figure 2), we observed that the risk of death was highest for the lowest category of BMI, then it decreased to the lowest risk for values of BMI = 28 kg/m², and then it tended to increase again for upper values of BMI. The comparison of both trends suggested that the upper levels of WHtR are associated with a higher risk of death than the upper levels of BMI.

Discussion

In this trial of 7447 elderly participants at high cardiovascular risk but initially free of cardiovascular disease, we observed a direct relationship between some adiposity measurements (WC, WHtR, BMI, height) and all cause mortality after 4.8 years of follow-up. However, this relationship was not significant for men and it was not linear in most cases.

A high biological plausibility exists to expect a direct association between excess body weight and all-cause mortality. However, we did not observe any linear trend for body mass index and a curvilinear shaped dose-response association was apparent for indexes of abdominal obesity. Importantly, we observed a lower risk in persons who had moderate levels of overweight compared with normal persons. Thereby, the risk of mortality was sometimes higher for the lowest category of adiposity indexes, specially for BMI, then it decreased and finally it increased again in the upper categories of each anthropometric index. This was in parcial agreement with the results of a recent meta-analysis on BMI and total mortality which concluded that overweight ([BMI] of 25 to 30 kg/m²) was associated with significantly lower all-cause mortality in comparison with the normal weight category and that grade I obesity (BMI between 30 and 35) was associated with a non-significantly lower risk of mortality [1]. Other previous metaanalyses also suggested inverse or null associations, specially among elderly subjects. [2], [3]. In any case, methodological aspects deserve caution in the interpretation of some studies included in these meta-analyses. A potential cause of concern is reverse causation bias due to pre-existent disease [24]. In addition, BMI may not be the most appropriate anthropometric index to assess the risks of adiposity.

However, a recent meta-analysis [25] concluded that the existence of a “healthy” pattern of increased weight is not likely to be tenable. On the contrary, any increase in adiposity is associated with increases in levels of cardiovascular risk factors and also in cardiovascular events that will, very probably, lead to increases in total mortality. Excess weight is associated with subclinical metabolic and vascular dysfunction that with the passage of time leads to an increased risk of CV events and mortality. Nevertheless, normal weight persons could have high risk for having metabolic abnormalities as it has been reported recently by De Larochellière et al. [13]. In this context, the routine measurement of WC and the use of WHtR>0.6 as a criterion for excess adiposity would improve the identification of higher-risk subjects. These individuals with normal weight nd increased levels of abdominal obesity could be genetically predisposed to CV disease and could have events even without experiencing increases in body weight or general adiposity. Therefore, it is necessary to consider not only body weight but also metabolic abnormalities. In this context, the use of WC or WHtR should be recommended, instead of using only the BMI.

In this line of thought, one of our contributions in the present work was the use of different indexes of adiposity, and not only BMI. Many previous studies have also used WC. But the use of the WHtR is still scarce. BMI has been widely used by the scientific community to assess obesity. However, BMI cannot distinguish between the accumulation of lean or fat mass, it cannot assess the distribution of fat and the change of this distribution with age nor take into account the changes that occurs in people who begin a diet accompanied or not by exercise.

Interestingly, we found that WC presented a stronger association with mortality than BMI, observing significant results, specially among women. This finding was consistent with other studies [26], [27] showing stronger correlations with visceral fat for WC than for BMI, particulary among elderly participants such as those included in our study.

Recent studies [26], [ 28]–[30] suggest that WHtR is likely to be the best anthropometric index to assess the association of adiposity with cardiovascular risk or overall mortality. This is partially in line with our investigation in which we observed in the restricted cubic spline models that the highest risk of mortality were present in participants from the upper categories of WHtR in contrast with the dose-response trend observed for BMI in which the increased risk of death was less attenuated and the highest relative risk was observed for the leanest subjects. Therefore this is one of the main contributions of our investigation, because we used this anthopometric index (WHtR) and almost no previous studies have longitudinally related this parameter to the risk of death in elderly subjects. In our study we observed that WHtR and WC were the best measures of obesity because using both indexes we obtained significant direct associations, specially for WC. However, WC might underestimate the relative amount of abdominal fat in short persons and overestimate it in tall persons [9]–[11], [15]. This was the reason why we also assessed the role of height. Taking into account that height showed also some associations with mortality, and can introduce confounding in the assessment of the association between WC and mortality, and that the results for WC and WHtR were similar, we consider that WHtR may be preferable and can be used when evaluating CVD mortality risk.

The association between the WHtR and overall mortality was slightly stronger among diabetics. Our study is unique in terms of the large percentage of diabetics (almost fifty percent) of our cohort. However the statistical interaction with diabetes was not significant and further studies on this issue are needed.

Our sample was composed of elderly subjects at high risk of cardiovascular disease, for whom being underweight has been also identified as a predictor of diabetes [31]. However, the precise mechanisms behind the association between low BMI (underweight) and diabetes among the elderly are not yet clarified. A possible explanation could be that protein-calorie malnutrition and magnesium deficiency may cause low insulin secretion and a low pancreatic insuline store. A recent study showed that low dietary magnesium and low mean serum albumin levels were associated with higher risk of type 2 diabetes [32].

However, it is true that abdominal adipose tissue can play an important role in the development of diabetes mellitus and it can also increase the risk of death [33], [34].

Interestingly, after taking into account the dietary intervention conducted in PREDIMED trial, we did not obtain a significant interaction. Therefore we cannot provide evidence that our dietary intervention was able to counterbalance the higher mortality risk associated with higher levels of adiposity. However, in the main results of the PREDIMED trial [17], we found strong evidence for the prevention of major cardiovascular events, but not for total mortality, using a MeDiet supplemented with either extra-virgin olive oil or mixed nuts. A posible explanation could be that when total mortality is used as outcome, we consider all causes of mortality not only cardiovascular causes and there are other important causes of mortality (such as cancer or respiratory causes) where subclinical disease is associated with weight loss. Another aspect to be considered is that in the PREDIMED trial the average WC of participants was greater than 100 cm and the average BMI was greater than 30 kg/m². In addition, these participants were at high cardiovascular risk because they had other cardiovascular risk factors and fifty percent of them were type-2 diabetics. Nevertheless the protocol of the PREDIMED intervention did not include specific caloric restriction or targeted goals for weight loss. In this context, and given the present results, a trial using an energy-restricted MeDiet would be likely to obtain even greater cardiovascular benefits. The new on-going trial referred to as PREDIMEDPLUS (www.predimedplus.com) will assess the effect of this combined intervention on major cardiovascular events and total mortality.

There are several strengths in our research. The prospective nature of our study with a long follow-up period allowed us to detect a sufficient number of deaths. In addition we were able to control for multiple potential confounders using multiple-adjusted models. The results could suggest causal associations because in Cox regression we controlled for a high number of potential confounders. However, residual confounding by unmeasured factors is always a possibility in observational research.

In contrast, the older age of our participants together with their high cardiovascular risk could contribute to modify the association between adiposity measures and all-cause mortality. Obesity has been associated with improved survival in patients with existing chronic diseases, including chronic obstructive pulmonary disease, chronic kidney disease, or heart failure. This fact has been termed the “obesity paradox” [24], [35]. This paradox may attenuate the detrimental effects of adiposity on mortality. In addition, total body mass increases across the life span until about age 60 and then it tends to decrease due to the process of sarcopenia which involves a decrease in muscle mass and on the other hand a smaller increase in fat mass. Moreover, at that age a fat re-distribution occurs, and, as a result of this re-distribution, the fat that was out of the trunk tend to be displaced to abdominal areas, thus increasing the metabolic risk [36]. Previous studies have suggested that BMI do not capture the increase in abdominal adipose tissue that tends to occur with age and this could explain the weakened association between BMI and risk of death that is usually observed with age [29]. All these reasons support the preferential use of the WHtR in elderly subjects.

Previous studies suggested a weak relationship of WC and BMI with cardiovascular risk in older adults [3], [37], [38]. Stevens et al. [7] described that the relative risk of all-cause and cardiovascular mortality associated with an excess body mass is greater for younger and middle-aged individuals than for older individuals. However, this effect modification by age has been recently challenged because the reduction in the magnitude of the association obesity–mortality with age maybe confounded by age at survey and cohort effects [39].

For these reasons, the present research gained more importance. Notwithstanding that in our study all participants were elderly subjects, we found a significant direct relationship between some adiposity measures (specially WC and WHtR) and all-cause mortality. But as in other studies, we found a weaker relationship for BMI than for these other measures.

A possible explanation for the non-significant relationships or the weaker associations found among men than among women, is the well-known longer survival of women. There could be a selective survival effect of healthier, older men because the unhealthy men with higher levels of adiposity at high cardiovascular risk may have already died from the detrimental effects of obesity. In addition, older individuals tend to decrease their intake of food and may not consume energy-rich diets that are related to a higher risk of death.

In conclusion, our study showed a direct association between adiposity measures reflecting visceral accumulation of fat (WC, WHtR) and all-cause mortality in an elderly population at high cardiovascular risk. The highest risk of death was observed in participants who initally were in the highest quartiles of these anthropometric indexes (WC, WHtR) that better capture abdominal obesity. For BMI, instead, we observed an U-shaped dose-response pattern, with the highest mortality in participants with the initial lowest values of BMI. However, further research is warranted to confirm these findings and to extend our findings to other elderly populations with a lower cardiovascular risk.

Acknowledgments

We thank all the participants in the trial for their enthusiastic and maintained collaboration.

Other PREDIMED Investigators

University of Navarra, Primary Care Centres, Pamplona, Spain: I Zazpe, B. Sanjulian, A. Marti, P. Buil-Cosiales, M. Serrano-Martinez, J. Diez-Espino, J. Basterra-Gortari, E.H. Martinez-Lapiscina, N. Ortuno, N. Berrade, V. Extremera-Urabayen, C. Arroyo-Azpa, L. Garcia-Perez, J. Villanueva Telleria, F. Cortes Ugalde, T. Sagredo Arce, Ma D. Garcia de la Noceda Montoy, Ma D. Vigata Lopez, Ma T. Arceiz Campo, A. Urtasun Samper, Ma V. Gueto Rubio, and B. Churio Beraza.

Hospital Clinic, Institut d'Investigacions Biomediques August Pi i Sunyer, Barcelona, Spain: M. Serra, A. Perez-Heras, C. Vinas, R. Casas, L. de Santamaria, S. Romero, J.M. Baena, M. Garcia, M. Oller, J. Amat, I. Duaso, Y. Garcia, C. Iglesias, C. Simon, Ll. Quinzavos, Ll. Parra, M. Liroz, J. Benavent, J. Clos, I. Pla, M. Amoros, M.T. Bonet, M.T. Martin, M.S. Sanchez, J. Altirruba, E. Manzano, A. Altes, M. Cofan, C. Valls-Pedret, A. Sala-Vila, and M. Domenech.

University Rovira i Virgili, Reus, Spain: J Salas-Salvado, J. Basora, R. Gonzalez, C. Molina, F. Marquez, N. Babio, P, Martinez, N. Ibarrola, R. Balanza, M. Sorli, J. Garcia Roselló, F. Martin, R. Tort, A. Isach, B. Costa, J.J. Cabre, and J. Fernandez-Ballart.

Institute de Recerca Hospital del Mar, Barcelona, Spain: M.I. Covas, S. Tello, J. Vila, M. Fito, R. de la Torre, D. Munoz-Aguayo, R. Elosua, J. Marrugat, and M. Ferrer.

University of Valencia, Valencia, Spain: P. Carrasco, R. Osma, M. Guillen, P. Guillem-Saiz, O. Portoles, V. Pascual, C. Riera, J. Valderrama, A. Serrano, E. Lazaro, A. Sanmartin, A. Girbes, V. Santamaria, C. Sanchez, Z. Pla, and E.

University Hospital of Alava, Vitoria, Spain: I. Salaverria, T. del Hierro, J. Algorta, S. Francisco, A. Alonso, J. San Vicente, A. Casi, E. Sanz, I. Felipe, F. Aros and A. Loma-Osorio.

University of Malaga, Malaga, Spain: J. Wärnberg, R. Benitez Pont, M. Bianchi Alba, J. Fernandez-Crehuet Navajas, J. Warnberg, R. Gomez-Huelgas, J. Martinez-Gonzalez, V. Velasco Garcia, J. de Diego Salas, A. Baca Osorio, J. Gil Zarzosa, J.J. Sanchez Luque, and E. Vargas Lopez.

Instituto de la Grasa, Consejo Superior de Investigaciones Cientificas, Sevilla, Spain: J. Sanchez Perona, E. Montero Romero, M. Garcia-Garcia and E. Jurado-Ruiz.

Institute of Health Sciences IUNICS, University of Balearic Islands, and Hospital Son Espases, Palma de Mallorca, Spain: M. Garcia-Valdueza, M. Monino, A. Proenza, R. Prieto, G. Frontera, M. Ginard, F. Fiol, A. Jover, D. Romaguera and J. Garcia.

Department of Family Medicine, Primary Care Division of Sevilla, Sevilla, Spain: M. Leal, E. Martinez, J. Lapetra, M. Ortega-Calvo, P. Roman, F. Jose Garcia, P. Iglesias, Y. Corchado, E. Mayoral, L. Mellado, L. Miró, JM. Lozano and C. Lama.

School of Pharmacy, University of Barcelona, Barcelona, Spain: M.C. Lopez- Sabater, A.I. Castellote-Bargallo, A. Medina-Remon, and A. Tresserra-Rimbau.

University of Las Palmas de Gran Canaria, Las Palmas, Spain: J. Alvarez-Perez, E. Diez Benitez, I. Bautista Castaño, I. Maldonado Diaz, A. Sanchez-Villegas, M.J. Férnandez-Rodríguez, F. Sarmiendo de la Fe, C. Simon Garcia, I. Falcon Sanabria, B. Macias Gutierrez, and A.J. Santana Santana.

Hospital Universitario de Bellvitge, Hospitalet de Llobregat, Barcelona, Spain: E. de la Cruz, A. Galera, Y. Soler, F. Trias, I. Sarasa, E. Padres, and E. Corbella.

Primary Care Division, Catalan Institute of Health, Barcelona, Spain: C. Cabezas, E. Vinyoles, M.A. Rovira, L. Garcia, G. Flores, J.M. Verdu, P. Baby, A. Ramos, L. Mengual, P. Roura, M.C. Yuste, A. Guarner, A. Rovira, M.I. Santamaria, M. Mata, C. de Juan, and A. Brau.

Other investigators of the PREDIMED network: J.A. Tur (University of Balearic Islands), M.P. Portillo (University of Basque Country) and G. T. Saez, Carmen Tormos, Antonio Iradi, Concha Cerdá, Benjamín Climent, Carlos Sánchez (Department of Biochemistry and Molecular Biology. General University Hospital. University of Valencia).

Author Contributions

Conceived and designed the experiments: MAM-G ET. Performed the experiments: MAM-G AG-A JR MB DC M. Fito ER EG-G M. Fiol LS-M XP. Analyzed the data: MAM-G SE. Contributed reagents/materials/analysis tools: MAM-G AG-A MB DC M. Fito ER JR EG-G M. Fiol JMS-L LS-M XP RE. Wrote the paper: MAM-G SE. Critically reviewed the manuscript and have approved the final version submitted for publication: MAM-G AG-A ET MB-R MB DC M. Fito ER RML-R JR EG-G M. Fiol JMS-L LS-M JAM SE GS-T XP RE.

Citation: Martínez-González MA, García-Arellano A, Toledo E, Bes-Rastrollo M, Bulló M, Corella D, et al. (2014) Obesity Indexes and Total Mortality among Elderly Subjects at High Cardiovascular Risk: The PREDIMED Study. PLoS ONE 9(7): e103246. https://doi.org/10.1371/journal.pone.0103246

References

1. Flegal KM, Kit BK, Orpana H, Graubard BI (2013) Association of all-cause mortality with overweight and obesity using standard body mass index categories: a systematic review and meta-analysis. JAMA 309: 71–82.

2. McGee DL (2005) Body mass index and mortality: a meta-analysis based on person-level data from twenty-six observational studies. Ann Epidemiol 15: 87–97.

3. Janssen I, Mark AE (2007) Elevated body mass index and mortality risk in the elderly. Obes Rev 8: 41–59.

4. Flanders WD, Augestad LB (2008) Adjusting for reverse causality in the relationship between obesity and mortality. Int J Obes 3: S42–46.

5. Stevens J, Cai J, Pamuk ER, Williamson DF, Thun MJ, et al. (1998) The effect of age on the association between body-mass index and mortality. N Engl J Med 338: 1–7.

6. Kuk JL, Ardern CI (2009) Influence of age on the association between various measures of obesity and all-cause mortality. J Am Geriatr Soc 57: 2077–2084.

7. Canning KL, Brown RE, Jamnik VK, Kuk JL (2013) Relationship between obesity and obesity-related morbidities weakens with aging. J Gerontol A Biol Sci Med Sci Mar 22. [Epub ahead of print]

8. Schneider HJ, Klotsche J, Silber S, Stalla GK, Wittchen HU (2011) Measuring abdominal obesity: effects of height on distribution of cardiometabolic risk factors using waist circumference and waist-to-height ratio. Diabetes Care 34: e7.

9. Browning LM, Hsieh SD, Ashwell M (2010) A systematic review of waist-to-height ratio as a screening tool for the prediction of cardiovascular disease and diabetes: 0·5 could be a suitable global boundary value. Nutr Res Rev 23: 247–269.

10. Lee CM, Huxley RR, Wildman RP, Woodward M (2008) Indices of abdominal obesity are better discriminators of cardiovascular risk factors than BMI: a meta-analysis. J Clin Epidemiol 61: 646–653.

11. Mombelli G, Zanaboni AM, Gaito S, Sirtori CR (2009) Waist-to-height ratio is a highly sensitive index for the metabolic syndrome in a Mediterranean population. Metab Syndr Relat Disord 7: 477–484.

12. Katzmarzyk PT, Mire E, Bray GA, Greenway FL, Heymsfield SB, et al. (2013) Anthropometric markers of obesity and mortality in white and African American adults: the Pennington center longitudinal study. Obesity (Silver Spring) 21: 1070–5.

13. De Larochellière E, Côté J, Gilbert G, Bibeau K, Ross MK, et al. (2014) Visceral/epicardial adiposity in nonobese and apparently healthy young adults: association with the cardiometabolic profile. Atherosclerosis 234: 23–9.

14. Bastien M, Poirier P, Lemieux I, Després JP (2014) Overview of epidemiology and contribution of obesity to cardiovascular disease. Prog Cardiovasc Dis 56: 369–81.

15. Schneider HJ, Klotsche J, Silber S, Stalla GK, Wittchen HU (2011) Measuring abdominal obesity: effects of height on distribution of cardiometabolic risk factors risk using waist circumference and waist-to-height ratio. Diabetes Care 34: e7.

16. Martínez-González MA, Corella D, Salas-Salvado J, Ros E, Covas MI, et al. (2012) Cohort profile: design and methods of the PREDIMED study. Int J Epidemiol 41: 377–385.

17. Estruch R, Ros E, Salas-Salvadó J, Covas MI, Corella D, et al. (2013) Primary prevention of cardiovascular disease with a Mediterranean diet. N Engl J Med 368: 1279–1290.

18. Zazpe I, Sanchez-Tainta A, Estruch R, Lamuela-Raventos RM, Schröder H, et al. (2008) A large randomized individual and group intervention conducted by registered dietitians increased adherence to Mediterranean-type diets: the PREDIMED study. J Am Diet Assoc 108: 1134–1144.

19. Estruch R, Martínez-González MA, Corella D, Salas-Salvadó J, Ruiz-Gutiérrez V, et al. (2006) Effects of a Mediterranean-style diet on cardiovascular risk factors: a randomized trial. Ann Intern Med 145: 1–11.

20. Ross R, Berentzen T, Bradshaw AJ, Janssen I, Kahn HS, et al. (2008) Does the relationship between waist circumference, morbidity and mortality depend on measurement protocol for waist circumference? Obes Rev 9: 312–25.

21. Elosua R, Marrugat J, Molina L, Pons S, Pujol E (1994) Validation of the Minnesota Leisure Time Physical Activity Questionnaire in Spanish men. The MARATHOM Investigators. Am J Epidemiol 139: 1197–1209.

22. Elosua R, Garcia M, Aguilar A, Molina L, Covas MI, et al. (2000) Validation of the Minnesota Leisure Time Physical Activity Questionnaire In Spanish Women. Investigators of the MARATDOM Group. Med Sci Sports Exerc 32: 1431–1437.

23. Fernández-Ballart JD, Piñol JL, Zazpe I, Corella D, Carrasco P, et al. (2010) Relative validity of a semi-quantitative food-frequency questionnaire in an elderly Mediterranean population of Spain. Br J Nutr 103: 1808–1816.

24. Tobías DK, Hu FB (2013) Does being overweight really reduce mortality? Obesity 21: 1746–1749.

25. Kramer CK, Zinman B, Retnakaran R (2013) Are metabolically healthy overweight and obesity benign conditions? Ann Intern Med 159: 758–769.

26. Gelber RP, Gaziano JM, Orav EJ, Manson JE, Buring JE, et al. (2008) Measures of obesity and cardiovascular risk among men and women. J Am Coll Cardiol 52: 605–615.

27. De Hollander EL, Bemelmans WJ, Boshuizen HC, Friedrich N, Wallaschofski H, et al. (2012) The association between waist circumference and risk of mortality considering body mass index in 65-to 74-year-olds: a meta-analysis of 29 cohorts involving more than 58000 elderly persons. Int J Epidemiol 41: 805–817.

28. Petursoon H, Sigurdsson JA, Bengtsson C, Nilsen TI, Getz L (2011) Body configuration as a predictor of mortality: comparison of five anthropometric measures in a 12 year follow-up of the Norwegian HUNT 2 study. PLoS One 6: e26621.

29. Kahn HS, Bullard KM, Barker LE, Imperatore G (2012) Differences between adiposity indicators for predicting all-cause mortality in a representative sample of United States non-elderly adults. PLoS One 7: e50428.

30. Schneider HJ, Friedrich N, Klotsche J, Pieper L, Nauck M, et al. (2010) The predictive value of different measures of obesity for incident cardiovascular events and mortality. J Clin Endocrinol Metab 95: 1777–85.

31. Kodama S, Horikawa C, Fujihara K, Heianza Y, Hirasawa R, et al. (2012) Comparisons of the strength of associations with future type 2 diabetes risk among anthropometric obesity indicators, including waist-to-height ratio: a meta-analysis. Am J Epidemiol 176: 959–969.

32. Sairenchi T, Iso H, Irie F, Fukasawa N, Ota H, Muto T (2008) Underweight as a predictor of diabetes in older adults: a large cohort study. Diabetes Care 31: 583–584.

33. Cox BD, Whichelow MJ, Prevost AT (1998) The development of cardiovascular disease in relation to anthropometric indices and hypertension in British adults. Int J Obes Relat Metab Disord 22: 966–973.

34. Taylor AE, Ebrahim S, Ben-Shlomo Y, Martin RM, Whincup PH, et al. (2010) Comparison of the associations of body mass index and measures of central adiposity and fat mass with coronary heart disease, diabetes, and all-cause mortality: a study using data from 4 UK cohorts. Am J Clin Nutr 91: 547–556.

35. Hu FB (2008) Obesity Epidemiology. New York: Oxford University Press.

36. Mauriège P, Imbeault P, Doucet E, Lacaille M, Langin D, et al. (2013) Weight loss and regain in obese individuals: a link with adipose tissue metabolism indices? J Physiol Biochem 69: 497–505.

37. Reis JP, Macera CA, Araneta MR, Lindsay SP, Marshall SJ, et al. (2009) Comparison of overall obesity and body fat distribution in predicting risk of mortality. Obesity (Silver Spring) 17: 1232–1239.

38. Reis JP, Araneta MR, Wingard DL, Macera CA, Lindsay SP, et al. (2009) Overall obesity and abdominal adiposity as predictors of mortality in U.S. White and black adults. Ann Epidemiol 19: 134–142.

39. Masters RK, Powers DA, Link BG (2013) Obesity and U.S. mortality risk over the adult life course. Am J Epidemiol 177: 431–442.