OBJECTIVE

To assess the association of a “metabolically healthy obese” phenotype with mortality using five definitions of metabolic health.

RESEARCH DESIGN AND METHODS

Adults (n = 5,269; 71.7% men) aged 39–62 years in 1991 through 1993 provided data on BMI and metabolic health, defined using data from the Adult Treatment Panel-III (ATP-III); criteria from two studies; and the Matsuda and homeostasis model assessment (HOMA) indices. Cross-classification of BMI categories and metabolic status (healthy/unhealthy) created six groups. Cox proportional hazards regression models were used to analyze associations with all-cause and cardiovascular disease (CVD) mortality during a median follow-up of 17.7 years.

RESULTS

A total of 638 individuals (12.1% of the cohort) were obese, of whom 9–41% were metabolically healthy, depending on the definition. Regardless of the definition, compared with metabolically healthy, normal-weight individuals, both the metabolically healthy obese (hazard ratios [HRs] ranged from 1.81 [95% CI 1.16–2.84] for ATP-III to 2.30 [1.13–4.70] for the Matsuda index) and the metabolically abnormal obese (HRs ranged from 1.57 [1.08–2.28] for the Matsuda index to 2.05 [1.44–2.92] for criteria defined in a separate study) had an increased risk of mortality. The only exception was the lack of excess risk using the HOMA criterion for the metabolically healthy obese (1.08; 0.67–1.74). Among the obese, the risk of mortality did not vary as a function of metabolic health apart from when using the HOMA criterion (1.93; 1.15–3.22). Similar results were obtained for cardiovascular mortality.

CONCLUSIONS

For most definitions of metabolic health, both metabolically healthy and unhealthy obese patients carry an elevated risk of mortality.

Obesity is a major public health problem that has reached epidemic proportions worldwide (1). It is associated with numerous metabolic and cardiovascular disturbances such as insulin resistance, type 2 diabetes, hypertension, and dyslipidemia (25). However, these cardiometabolic abnormalities are not found in all obese people (6,7), as evidenced by the occurrence of a subset of apparently healthy obese subjects referred to as metabolically healthy obese (MHO) (8,9). Several studies have confirmed the existence of MHO individuals (1016), accounting for as much as 40% of the obese population. MHO individuals display a favorable metabolic profile, characterized by high levels of insulin sensitivity, a low prevalence of hypertension, and a favorable lipid and inflammation profile.

The long-term health consequences of obesity among those who are metabolically healthy remain unclear. Obesity is known to carry an elevated risk of mortality (17), but few studies have examined associations of the MHO phenotype with mortality, and the evidence from these studies is mixed. In general population samples from Scotland and England, MHO individuals were not at increased risk of all-cause and cardiovascular disease (CVD) mortality compared with healthy nonobese individuals (18), a finding replicated in an Italian study of obesity and insulin sensitivity (19). However, overweight and obese individuals without the metabolic syndrome had an increased risk of mortality compared with normal-weight individuals without the metabolic syndrome in a Swedish cohort of middle-aged men (20). Furthermore, in the U.S. National Health and Nutrition Examination Survey III (21), metabolically healthy and abnormal obese individuals had similar elevations in mortality risk compared with metabolically healthy, normal-weight subjects. Several factors may have contributed to these inconsistencies. The comparison group varies when estimating risk of mortality in the MHO phenotype; risk is compared either with metabolically healthy nonobese (18,19) or metabolically healthy, normal-weight people (20,21). Another difference between the studies is that metabolic health is defined in different ways, with little consensus on how best to define it. Therefore, the objective of the current study is to assess whether there is consistency in the association of the MHO phenotype with all-cause and CVD mortality using different definitions of metabolic health and reference groups.

Participants

Data were drawn from the Whitehall II cohort, established in 1985 as a longitudinal study among 10,308 (6,895 male and 3,413 female) U.K. government employees (i.e., civil servants) (22). All civil servants aged 35–55 years in 20 London-based departments were invited to participate by letter; 73% agreed. The baseline examination (phase 1) took place from 1985 to 1988 and involved a clinical examination and a self-administered questionnaire. Subsequent phases of data collection alternated between postal questionnaire alone (phases 2 [1988–1990], 4 [1995–1996], 6 [2001], and 8 [2006]) and postal questionnaire accompanied by a clinical examination (phases 3 [1991–1993], 5 [1997–1999], 7 [2002–2004] and 9 [2007–2009]). Data on metabolic factors for the current study were drawn from phase 3, considered the “baseline” for the purpose of these analyses. All participants provided written consent and the University College of London ethics committee approved the study.

Baseline measurements

BMI.

With the patients in only underwear, weight was measured to the nearest 0.1 kg on digital Soehnle electronic scales (Leifheit AS, Nassau, Germany). With the participant standing erect in bare feet with the head in the Frankfurt plane, height was measured to the nearest 1 mm using a stadiometer. Reproducibility of measurements over 1 month (correlation coefficient = between-subject variability/[total between + within subject variability]), undertaken for 331 participants, was 0.99 for both weight and height. BMI was calculated by dividing weight (in kilograms) by height (in meters squared) and categorized using the World Health Organization classification (23): underweight, <18.5 kg/m2; standard weight, 18.5–24.9 kg/m2; overweight, 25–29.9 kg/m2; and obese, ≥30 kg/m2, with the <18.5 category (n = 80) removed from the analysis.

Metabolic health factors.

We used standard operating protocols to measure the various components to define metabolic status. Blood pressure was measured twice in the sitting position after 5 min of rest with a Hawksley random-zero sphygmomanometer (Lynjay Services Ltd, Worthing, U.K.). The average of the two readings was considered the measured blood pressure. Venous blood was taken in the fasting state or at least 5 h after a light, fat-free breakfast before undergoing a 2-h, 75-g oral glucose tolerance test (OGTT). Serum for lipid analyses was refrigerated at −4°C and assayed within 72 h. HDL-cholesterol (HDL-c) was measured by precipitating non-HDL-c with dextran sulfate–magnesium chloride using a centrifuge and measuring cholesterol in the supernatant. Serum triglyceride was determined by enzymatic colorimetric method (glycerol-3-phosphate oxidase/phenol and aminophenazone [GPO-PAP]). The concentration of LDL-cholesterol (LDL-c) was calculated using the Friedewald formula when serum triglycerides were lower than 4.5 mmol/L. Blood glucose was measured using the glucose oxidase method (24) (YSI Model 2300 STAT PLUS Analyzer; YSI Corporation, Yellow Springs, OH) (mean coefficient of variation, 2.9–3.3%). Insulin was measured by radioimmunoassay using polyclonal guinea pig antiserum at age 49 years and by double antibody enzyme-linked immunosorbant assay at age 61 years. Homeostasis model assessment (HOMA) of insulin resistance (HOMA-IR) calculation was based on model-derived estimates (rather than linear approximations) using the HOMA2 calculator version 2.2 (http://www.dtu.ox.ac.uk/index.php?maindoc=/homa/index.php, Diabetes Trials Unit, University of Oxford, Oxford, U.K.) (25,26). C-reactive protein (CRP) was measured in serum stored at −80°C using a high-sensitivity immunonephelometric assay in a BN ProSpec nephelometer (Dade Behring, Deerfield, IL). Names of medications as provided by the participants were coded using the British National Formulary to identify antihypertensive and lipid-lowering drugs and medication for diabetes.

formula

Cross-classification of BMI and metabolic status.

We used five definitions to cross-classify individuals: the Adult Treatment Panel-III (ATP-III) definition of metabolic syndrome (27), Wildman criterion (12), Karelis criterion (28), the OGTT derived from the Matsuda index (13,29) and the HOMA index (11) (Supplementary Table 1).

The ATP-III definition (27) of metabolic normality required individuals to have one or none of the following components (waist circumference criterion was not used because of collinearity with BMI): triglycerides ≥1.7 mmol/L or use of lipid-lowering drugs, systolic blood pressure ≥130 mmHg or diastolic blood pressure ≥ 85 mmHg or use of antihypertensive drugs, glucose ≥5.6 mmol/L or use of medications for diabetes, and HDL-c <1.04 mmol/L for men and <1.29 mmol/L for women.

Participants were considered metabolically healthy by Wildman criterion (12) if they had one or none of the following components: blood pressure ≥130/85 mmHg or use of antihypertensive drugs, triglycerides ≥1.7 mmol/L or use of lipid-lowering drugs, fasting glucose ≥5.6 mmol/L or use of medications for diabetes, HOMA-IR above the 90th percentile among all participants, CRP above the 90th percentile among all participants, and HDL-c <1.3 mmol/L.

Karelis criterion (28) required metabolically healthy subjects to have 4 of 5 of the following components: triglycerides ≤1.7 mmol/L or use of lipid-lowering drugs, HDL-c ≥1.3 mmol/L, LDL-c ≤2.6 mmol/L, HOMA ≤2.7, and CRP ≤3.0 mg/L.

The Matsuda index (13,29) was calculated as follows: Mean glucose and mean insulin were obtained using the average glucose and insulin levels, respectively, from the OGTT. The Matsuda index was divided into quartiles, and subjects were classified as metabolically healthy if they belonged to the upper quartile of this index (≥186.16 for men and ≥157.68 for women).

The HOMA index (11) was divided into quartiles, and participants were classified as metabolically healthy if they belonged to the three lower quartiles (≤1.70 for men and ≤1.52 for women) of this index.

We used these definitions alongside data on BMI to create six phenotypes: metabolically healthy, normal weight (MH-NW); metabolically healthy overweight; MHO; metabolically abnormal, normal weight; metabolically abnormal overweight; and metabolically abnormal obese (MAO).

Covariates.

Sociodemographic covariates included age, sex, occupational position, ethnicity (white vs. nonwhite), and marital status (single, married/cohabiting, divorced, widowed). Occupational position is a three-level variable representing high (administrative), intermediate (professional or executive), and low (clerical or support) grades. This measure is a comprehensive marker of socioeconomic circumstances and is related to salary, social status, and level of responsibility at work (22). Behavioral measures included smoking status (never, current, or ex-smoker); alcohol intake (assessed via questions about the number of alcoholic drinks consumed in the past 7 days and categorized as no alcohol [none or <1 unit/week], moderate drinker [1–14 units/week for women and 1–21 units/week for men], heavy drinker [>14 units for women and >21 units for men]); physical activity (categorized as active [≥2.5 h/week of moderate or ≥1 h/week of vigorous physical activity], inactive [<1 h/week of moderate and <1 h/week of vigorous physical activity], or moderately active [not active or inactive]); and consumption of fruits and vegetables (assessed using the question, “How often do you eat fresh fruits or vegetables?”; responses were given on a two-point scale: ≥1 fruit or vegetable/day, or < 1 fruit or vegetable/day). Missing data on covariates during phase 3 (1991–1993; <1%) were replaced with data from the phase immediately before or after that phase.

Mortality follow-up

Study members are linked to the National Health Services death and electronic patient records with the use of a National Health Services identification number. A total of 10,297 participants (99.9%) were traced successfully and have been followed through these registers. Mortality data, which included the cause of death, were available through the National Health Services Central Registry until 31 January 2010. We also examined CVD mortality (ICD-9: 390.0–458.9; ICD-10: I00–I99).

Statistical analysis

Analyses were undertaken using STATA 11 software. The characteristics of the sample are presented as mean ± SD or percentage, as appropriate, by BMI category and for the obese as a function of metabolic health using the ATP-III definition. We used ANOVA to test for differences in baseline characteristics between groups for continuous variables and the χ2 test for dichotomous measures.

Cox proportional-hazard regression models with follow-up period as the time scale were used to examine the relationship between BMI category/metabolic phenotypes (using the five definitions mentioned earlier) and mortality. The proportional hazards assumptions were confirmed by Schoenfeld tests. The analytic sample comprised 331 South Asian and 200 black participants. The interaction term between ethnicity and BMI/metabolic status phenotypes suggested no differences across ethnic groups (P for interaction ≥0.10 and ≤0.56). Similarly, the interaction term for sex revealed no differences between men and women (P for interaction ≥0.11 and ≤0.59). This allowed us to combine ethnic groups and men and women in the analysis. In the first set of analyses, we used the MH-NW category as the reference. Hazard ratios (HRs) and 95% CIs were adjusted sequentially for age and sex (data not shown) and then adjusted further for occupational position, ethnicity, marital status, smoking status, alcohol intake, physical activity, and fruit and vegetable consumption. We then ran a second set of analyses, stratified by BMI category, to allow us to compare the risk of mortality as a function of metabolic health status in each strata of BMI. The metabolically healthy group within each BMI category was the reference in these analyses. Both sets of analyses were repeated with CVD mortality as the outcome.

In sensitivity analyses we repeated the analyses after replacing occupational position with education and excluding all deaths that occurred in the first 5 years of follow-up to take into account possible effects of reverse causation as an explanation of our findings. We also replicated analyses using absolute rates of mortality and finally using non-CVD mortality as an outcome.

A total of 8,223 participants attended the clinical examination at baseline, when BMI and metabolic factors were assessed; 5,269 of these individuals had data on all the components necessary for analysis of body size phenotypes. The participants included in the analytic sample (n = 5,269) compared with those who were excluded were more likely to be men (71.7% vs. 64.9%; P < 0.001), obese (12.1% vs. 5.4%; P < 0.001), and married or cohabiting (77.5% vs. 75.2%; P = 0.001). They were less likely to be current smokers (12.3% vs. 16.1%; P < 0.001) and to have a higher occupational position (37.4% vs. 39.9%; P = 0.003). Although these differences were statistically significant because of the large sample size, the absolute differences were small in general.

Baseline characteristics of participants included in the analysis are presented in Table 1. A total of 44.8% of the participants were normal weight, 43.1% were overweight, and 12.1% were obese. Triglycerides, fasting glucose, 2-h glucose level, fasting insulin, blood pressure, CRP, and HOMA-IR were higher in MAO than in MHO participants (P < 0.001 for all these components), whereas HDL-c was higher in MHO than in MAO participants using the ATP-III definition.

Table 1

Sample characteristics at baseline (1991–1993) as a function of BMI

Sample characteristics at baseline (1991–1993) as a function of BMI
Sample characteristics at baseline (1991–1993) as a function of BMI

Supplementary Table 2 shows the prevalence of the six phenotypes created using the five definitions. Among obese participants, 236 (37.%) were classified as metabolically healthy by the ATP-III definition (27), 146 (22.9%) by the Wildman criteria (12), 119 (18.7%) by the Karelis criteria (28), 57 (9.0%) by the Matsuda index (13,29) and 260 (40.8%) using the HOMA index (11). In the entire analytic sample, using the various definitions, the prevalences of the MHO phenotype are 4.5% (ATP-III), 2.8% (Wildman), 2.3% (Karelis), 1.1% (Matsuda), and 4.9% (HOMA).

During the median follow-up of 17.7 years, 413 deaths occurred (7.8% of the total population): 76 deaths (11.9%) were among obese, 175 (7.7%) among overweight, and 162 (6.9%) among normal-weight participants. Among the 126 CVD deaths, 44 (34.9%) were among normal-weight individuals, 58 (46.0%) among overweight, and 24 (19.1%) among obese individuals.

Table 2 shows the associations of BMI categories and metabolic health factors with all-cause and CVD mortality. Compared with normal-weight individuals, the obese (HR 1.68 [95% CI 1.27–2.22]) but not the overweight (1.01 [0.82–1.26]) had an increased risk of mortality. All ATP-III (27) components and all components of the Wildman criterion (12), the HOMA index (11), and CRP as defined by Karelis (28) were associated with an increased risk of mortality, whereas the Matsuda index (13,29), LDL-c, and HOMA as defined by Karelis (28) were not. Results for CVD mortality were similar.

Table 2

The association of individual risk factors used to define metabolic health with all-cause and CVD mortality

The association of individual risk factors used to define metabolic health with all-cause and CVD mortality
The association of individual risk factors used to define metabolic health with all-cause and CVD mortality

Table 3 depicts the associations between the six BMI/metabolic categories and all-cause and CVD mortality with the MH-NW group as the reference. The MHO group did not have an increased risk of all-cause and CVD mortality, whatever the definition of metabolic health, but the MHO group had a greater risk of death (all-cause mortality) for all definitions except for the HOMA index (HR 1.08 [95% CI 0.67–1.74]). The mortality hazard in the MAO group was higher compared with the MH-NW group, whatever the definition of metabolic abnormality. Similar results were obtained for CVD mortality. Analyses using absolute rates, using deaths per 1,000 person-years (Supplementary Table 3), excluding deaths in the first 5 years (n = 63) of follow-up (Supplementary Table 4), using non-CVD mortality as an outcome (Supplementary Table 5), and adding menopausal status as a covariate (data not shown) yielded similar results to those reported in Table 3.

Table 3

The association of BMI/metabolic health status with all-cause and CVD mortality

The association of BMI/metabolic health status with all-cause and CVD mortality
The association of BMI/metabolic health status with all-cause and CVD mortality

Figure 1 and Supplementary Fig. 1 present HRs for all-cause and CVD mortality as a function of metabolic health in an analysis stratified by BMI category. Compared with their metabolically healthy counterparts, the metabolically abnormal normal-weight and overweight individuals had higher all-cause mortality risk using the ATP III definition (27). This was not the case among the obese, in whom greater mortality risk—both all-cause and CVD mortality—was observed among the metabolically abnormal compared with metabolically healthy individuals using only the HOMA index criterion (11) (HR 1.93 [95% CI 1.15–3.24] for all-cause mortality). The results for CVD mortality (Supplementary Fig. 1) were largely similar to those found for all-cause mortality, albeit with wider CIs due to the small number of deaths from CVD.

Figure 1

Association of abnormal metabolic health status with all-cause mortality in analyses stratified by BMI categories. Analyses adjusted for age, sex, occupational position, physical activity, smoking, alcohol, fruit and vegetable consumption, marital status, and ethnicity. (A high-quality color representation of this figure is available in the online issue.)

Figure 1

Association of abnormal metabolic health status with all-cause mortality in analyses stratified by BMI categories. Analyses adjusted for age, sex, occupational position, physical activity, smoking, alcohol, fruit and vegetable consumption, marital status, and ethnicity. (A high-quality color representation of this figure is available in the online issue.)

Close modal

This study assessed the association of five different definitions of the MHO phenotype with all-cause and CVD mortality. Our results show that 9–41% of the obese population qualify as being metabolically healthy, depending on the method used to ascertain metabolic health. In relation to four of the five definitions, both MHO and MAO participants had an elevated risk of mortality compared with MH-NW individuals. The metabolically healthy overweight individuals were not at increased risk of mortality regardless of the methods used; however, this was not the case for the metabolically abnormal overweight individuals. Our results show that obese individuals, irrespective of their metabolic status, carry an excess risk of mortality.

Direct comparison of our results with other studies is made difficult by the use of a different reference group to assess risk (18); often the reference group combines the normal and overweight individuals or uses a different definition of metabolic health status (21). No previous study has compared a range of definitions of metabolic status within the sample analytic framework. Arnlöv et al. (20), using the National Cholesterol Education Program criteria, and Kuk and Ardern (21), using insulin resistance by HOMA and the MetSyn criteria, also found an increased risk for all-cause mortality when comparing MHO to MH-NW individuals. However, Calori et al. (19), who estimated insulin resistance suing HOMA (insulin sensitive <2.5; insulin resistant ≥2.5), and Hamer and Stamatakis (18), who used a definition based on blood pressure, HDL-c, diabetes diagnosis, waist circumference, and low-grade inflammation, did not find greater risk of all-cause and CVD mortality in MHO compared with metabolically healthy nonobese individuals. In our study the MHO group did not carry excess risk of mortality compared with the MH-NW group only when using the HOMA index (11) to assess metabolic status. This implies that four of the five definitions used in this article show the risk of mortality in MHO individuals to be higher than MH-NW and similar to MAO individuals. The HOMA index is a measure of pancreatic β-cell function, whereas the Matsuda index is a composite measure of hepatic and muscle insulin sensitivity. Our results suggest that the HOMA index might be particularly useful in distinguishing obese individuals at a greater risk of mortality.

We used a range of methods, based on insulin resistance and clustering of metabolic abnormalities, to better understand the MHO phenotype. The five methods did not identify the same subgroups of individuals; only 23 subjects (3.6%) among the obese are classified as MHO using all five definitions. Taking the results obtained for the obese and the overweight phenotypes, our results suggest that criteria for metabolic health status using established cut-offs like those in the ATP-III criterion (27) are more reasonable because the threshold for risk has been validated on multiple occasions. However, criteria that are based on risk distribution (percentile), as the Matsuda (13,29) or HOMA indices (11), lead to less consistent results across studies. This is principally because the distribution of risk is likely to vary among different populations (obese, clinical, as a function of age, or even in the general population), and a threshold such as the top quartile may not translate well across populations. To examine obesity phenotypes we constructed six groups: metabolically healthy or metabolically abnormal normal weight, metabolically healthy or metabolically abnormal overweight, and metabolically healthy or metabolically abnormal obese, as has been done previously (20,21). This is in contrast to other studies that used only four groups: nonobese metabolically healthy or unhealthy and obese metabolically healthy or unhealthy (18,19), making the assumption that normal weight and overweight individuals have the same risk. We preferred to test this assumption and, as the results show, there is little difference between the normal weight and overweight groups. So, for mortality risk, the analyses could have been based on four categories of BMI/metabolic health phenotypes.

In analyses comparing individuals within the same BMI category and using the metabolically healthy as the reference group, we found that in normal and overweight groups metabolic abnormality carried some excess risk when the criteria for metabolic health were ATP-III (27), Wildman (12), or Karelis (28), but no excess risk was observed for the Matsuda (13,29) or the HOMA indices (11). In the obese group, the risk of mortality did not differ as a function of metabolic health except for that based on the HOMA index (11). These results can be understood in light of results showing a favorable metabolic profile for a range of measures when the HOMA index was used to identify MHO individuals (15). The study in question compared different methods (hyperinsulinemic-euglycemic clamp, Matsuda index, HOMA index, Karelis and Wildman criteria) to identify MHO individuals; those identified using the HOMA index had lower levels of BMI, central fat mass, fasting glucose, and insulin. This suggests that the HOMA index might be a more efficient measure of health status in the obese population.

Given the obesity epidemic, a better understanding of the extent to which the MHO phenotype carries risks for health is important. Identifying mechanisms of this favorable metabolic profile would improve the treatment of obese patients to reduce cardiometabolic risk (16). The mechanisms underlying the favorable metabolic profile of MHO individuals remain unclear. It has been suggested that adipose cell size could be a link between obesity and insulin resistance. Comparing the size of adipose cells between MHO and MAO individuals, McLaughlin et al. (30) found that the MAO group had a higher proportion of small adipose cells, which were seen to have adverse effects, compared with MHO individuals. Meigs et al. (11) reported that MHO individuals also had a lower amount of visceral fat content. There is some evidence to suggest that MHO people might be less responsive to lifestyle interventions for improvement of insulin sensitivity (31,32), but these findings are based on small-scale studies. Our findings and those of several other studies (20,21) suggest that the adverse effects of obesity on survival cannot be eliminated by targeting only metabolic health.

The main strengths of this study include the large sample size and the nearly 18-year follow-up for mortality. We also were able to replicate the findings for CVD mortality. Furthermore, to our knowledge, this is the first study that evaluated the risk of all-cause and CVD mortality associated with BMI/metabolic phenotypes using five different definitions to identify the MHO phenotype that have been published. We adjusted for several covariates, including occupational position, ethnic group, alcohol consumption, marital status, physical activity, smoking, and dietary behavior, which allowed us to minimize the potential for confounding by these factors. Replacing occupational position with education in the analysis did not change the results. The most important limitation of the study is that participants of the Whitehall II study are mainly office-based civil servants and are not fully representative of the British population because the study does not include unemployed or individuals in “blue-collar” professions. However, they cover a wide socioeconomic range, with >10-fold difference between the highest and lowest salaries. Another limitation is the lack of ethnic diversity in the study sample, although tests suggested no heterogeneity in associations as a function of ethnicity, which could be due to the limited sample size of racial/ethnic minorities. We were unable to assess the change in metabolic status during the follow-up period; it is possible that some participants who were healthy at baseline developed metabolic risk factors.

Our study focuses on mortality as an outcome, but findings for morbidity are mixed. One study showed MHO to have a cardiovascular risk profile between that of healthy nonobese and MAO (33). A longitudinal study (11) showed that the MHO phenotype was associated with a reduced risk of developing CVD, whereas others studies have shown that obese individuals without metabolic syndrome had an increased risk for cardiovascular events (20). Further large-scale studies are required to assess the impact of the MHO phenotype on morbidity related to obesity for example, CVD, cancers, respiratory diseases, and depression.

In summary, our data suggest that the MHO phenotype exists even though there are large variations in prevalence rates as a function of criteria used to define this phenotype. However, MHO individuals do not seem to be at a lower risk for all-cause and CVD mortality compared with MH-NW individuals—in effect their risk of mortality was similar to that of the MAO group. This risk was independent of the definition used except for the HOMA index, for which MHO did not carry excess risk. Our analyses suggest that thresholds to define metabolic health elaborated on the basis of the distribution of the risk factor may not be useful because they cannot be generalized across populations.

A.S.-M. is supported by a European Young Investigator Award from the European Science Foundation. M.K. is supported by an Economic and Social Research Council professorship and the Academy of Finland. The Whitehall II study has been supported by grants from the British Medical Research Council (MRC); the British Heart Foundation; the British Health and Safety Executive; the British Department of Health; the National Heart, Lung, and Blood Institute, National Institutes of Health (NIH) (R01HL036310); and the National Institute on Aging, NIH (R01AG013196, R01AG034454). G.D.B. is a Wellcome Trust Fellow.

No potential conflicts of interest relevant to this article were reported.

G.-M.H. performed statistical analyses and drafted the manuscript. S.C. and A.S.-M. contributed to the interpretation of results and revised the manuscript. A.D. helped with the statistical analyses. G.D.B. and M.K. edited and reviewed the manuscript. G.-M.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank all of the participating civil service departments and their welfare, personnel, and establishment officers; the British Occupational Health and Safety Agency; the British Council of Civil Service Unions; all participating civil servants in the Whitehall II study; and all members of the Whitehall II study team.

1.
Finucane
MM
,
Stevens
GA
,
Cowan
MJ
, et al
Global Burden of Metabolic Risk Factors of Chronic Diseases Collaborating Group (Body Mass Index)
.
National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9·1 million participants
.
Lancet
2011
;
377
:
557
567
[PubMed]
2.
Hedley
AA
,
Ogden
CL
,
Johnson
CL
,
Carroll
MD
,
Curtin
LR
,
Flegal
KM
.
Prevalence of overweight and obesity among US children, adolescents, and adults, 1999-2002
.
JAMA
2004
;
291
:
2847
2850
[PubMed]
3.
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
[PubMed]
4.
Eckel
RH
,
Kahn
R
,
Robertson
RM
,
Rizza
RA
.
Preventing cardiovascular disease and diabetes: a call to action from the American Diabetes Association and the American Heart Association
.
Diabetes Care
2006
;
29
:
1697
1699
[PubMed]
5.
Hossain
P
,
Kawar
B
,
El Nahas
M
.
Obesity and diabetes in the developing world—a growing challenge
.
N Engl J Med
2007
;
356
:
213
215
[PubMed]
6.
Ferrannini
E
,
Natali
A
,
Bell
P
,
Cavallo-Perin
P
,
Lalic
N
,
Mingrone
G
European Group for the Study of Insulin Resistance (EGIR)
.
Insulin resistance and hypersecretion in obesity
.
J Clin Invest
1997
;
100
:
1166
1173
[PubMed]
7.
Bonora
E
,
Kiechl
S
,
Willeit
J
, et al
.
Prevalence of insulin resistance in metabolic disorders: the Bruneck Study
.
Diabetes
1998
;
47
:
1643
1649
[PubMed]
8.
Brochu
M
,
Tchernof
A
,
Dionne
IJ
, et al
.
What are the physical characteristics associated with a normal metabolic profile despite a high level of obesity in postmenopausal women?
J Clin Endocrinol Metab
2001
;
86
:
1020
1025
[PubMed]
9.
Sims
EA
.
Are there persons who are obese, but metabolically healthy?
Metabolism
2001
;
50
:
1499
1504
[PubMed]
10.
Karelis
AD
,
Brochu
M
,
Rabasa-Lhoret
R
.
Can we identify metabolically healthy but obese individuals (MHO)?
Diabetes Metab
2004
;
30
:
569
572
[PubMed]
11.
Meigs
JB
,
Wilson
PW
,
Fox
CS
, et al
.
Body mass index, metabolic syndrome, and risk of type 2 diabetes or cardiovascular disease
.
J Clin Endocrinol Metab
2006
;
91
:
2906
2912
[PubMed]
12.
Wildman
RP
,
Muntner
P
,
Reynolds
K
, et al
.
The obese without cardiometabolic risk factor clustering and the normal weight with cardiometabolic risk factor clustering: prevalence and correlates of 2 phenotypes among the US population (NHANES 1999-2004)
.
Arch Intern Med
2008
;
168
:
1617
1624
[PubMed]
13.
Stefan
N
,
Kantartzis
K
,
Machann
J
, et al
.
Identification and characterization of metabolically benign obesity in humans
.
Arch Intern Med
2008
;
168
:
1609
1616
[PubMed]
14.
Karelis
AD
.
Metabolically healthy but obese individuals
.
Lancet
2008
;
372
:
1281
1283
[PubMed]
15.
Messier
V
,
Karelis
AD
,
Prud’homme
D
,
Primeau
V
,
Brochu
M
,
Rabasa-Lhoret
R
.
Identifying metabolically healthy but obese individuals in sedentary postmenopausal women
.
Obesity (Silver Spring)
2010
;
18
:
911
917
[PubMed]
16.
Primeau
V
,
Coderre
L
,
Karelis
AD
, et al
.
Characterizing the profile of obese patients who are metabolically healthy
.
Int J Obes (Lond)
2011
;
35
:
971
981
[PubMed]
17.
Czernichow
S
,
Kengne
AP
,
Stamatakis
E
,
Hamer
M
,
Batty
GD
.
Body mass index, waist circumference and waist-hip ratio: which is the better discriminator of cardiovascular disease mortality risk?: evidence from an individual-participant meta-analysis of 82 864 participants from nine cohort studies
.
Obes Rev
2011
;
12
:
680
687
[PubMed]
18.
Hamer
M
,
Stamatakis
E
.
Metabolically healthy obesity and risk of all-cause and cardiovascular disease mortality
.
J Clin Endocrinol Metab
2012
;
97
:
2482
2488
[PubMed]
19.
Calori
G
,
Lattuada
G
,
Piemonti
L
, et al
.
Prevalence, metabolic features, and prognosis of metabolically healthy obese Italian individuals: the Cremona Study
.
Diabetes Care
2011
;
34
:
210
215
[PubMed]
20.
Arnlöv
J
,
Ingelsson
E
,
Sundström
J
,
Lind
L
.
Impact of body mass index and the metabolic syndrome on the risk of cardiovascular disease and death in middle-aged men
.
Circulation
2010
;
121
:
230
236
[PubMed]
21.
Kuk
JL
,
Ardern
CI
.
Are metabolically normal but obese individuals at lower risk for all-cause mortality?
Diabetes Care
2009
;
32
:
2297
2299
[PubMed]
22.
Marmot
M
,
Brunner
E
.
Cohort Profile: the Whitehall II study
.
Int J Epidemiol
2005
;
34
:
251
256
[PubMed]
23.
Obesity: preventing and managing the global epidemic. Report of a WHO consultation
.
World Health Organ Tech Rep Ser
2000
;
894
:
i
xii
,
1
253
24.
Cooper
GR
.
Methods for determining the amount of glucose in blood
.
CRC Crit Rev Clin Lab Sci
1973
;
4
:
101
145
[PubMed]
25.
Matthews
DR
,
Hosker
JP
,
Rudenski
AS
,
Naylor
BA
,
Treacher
DF
,
Turner
RC
.
Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man
.
Diabetologia
1985
;
28
:
412
419
[PubMed]
26.
Wallace
TM
,
Levy
JC
,
Matthews
DR
.
Use and abuse of HOMA modeling
.
Diabetes Care
2004
;
27
:
1487
1495
[PubMed]
27.
Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
.
Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III)
.
JAMA
2001
;
285
:
2486
2497
[PubMed]
28.
Karelis
AD
,
Rabasa-Lhoret
R
.
Inclusion of C-reactive protein in the identification of metabolically healthy but obese (MHO) individuals
.
Diabetes Metab
2008
;
34
:
183
184
[PubMed]
29.
Matsuda
M
,
DeFronzo
RA
.
Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp
.
Diabetes Care
1999
;
22
:
1462
1470
[PubMed]
30.
McLaughlin
T
,
Sherman
A
,
Tsao
P
, et al
.
Enhanced proportion of small adipose cells in insulin-resistant vs insulin-sensitive obese individuals implicates impaired adipogenesis
.
Diabetologia
2007
;
50
:
1707
1715
[PubMed]
31.
Karelis
AD
,
Messier
V
,
Brochu
M
,
Rabasa-Lhoret
R
.
Metabolically healthy but obese women: effect of an energy-restricted diet
.
Diabetologia
2008
;
51
:
1752
1754
[PubMed]
32.
De Filippis
E
,
Alvarez
G
,
Berria
R
, et al
.
Insulin-resistant muscle is exercise resistant: evidence for reduced response of nuclear-encoded mitochondrial genes to exercise
.
Am J Physiol Endocrinol Metab
2008
;
294
:
E607
E614
[PubMed]
33.
Marini
MA
,
Succurro
E
,
Frontoni
S
, et al
.
Metabolically healthy but obese women have an intermediate cardiovascular risk profile between healthy nonobese women and obese insulin-resistant women
.
Diabetes Care
2007
;
30
:
2145
2147
[PubMed]
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

Supplementary data