An Empirically Derived Definition of Metabolically Healthy Obesity Based on Risk of Cardiovascular and Total Mortality | Cardiology | JAMA Network Open | JAMA Network
[Skip to Navigation]
Figure 1.  Weighted Prevalence and Overlap of Metabolically Healthy Obesity (MHO) Identified by Different Definitions Within the National Health and Nutrition Examination Survey-III (NHANES-III) (n = 12 341)
Weighted Prevalence and Overlap of Metabolically Healthy Obesity (MHO) Identified by Different Definitions Within the National Health and Nutrition Examination Survey-III (NHANES-III) (n = 12 341)

ATP-III indicates Adult Treatment Panel-III; HOMA-IR, homeostatic model assessment of insulin resistance.

Figure 2.  Risk of Mortality in Subgroups of Body Mass Index and Metabolic Health Using a New Definition
Risk of Mortality in Subgroups of Body Mass Index and Metabolic Health Using a New Definition

Hazard ratios (95% CIs) adjusted for age, sex, race/ethnicity, educational level, income, marital status, smoking status, alcohol consumption, physical activity, and UK Biobank assessment center. National Health and Nutrition Examination Survey-III (NHANES-III) cohort, n = 12 341; UK Biobank cohort, n = 374 079. CVD indicates cardiovascular disease; MHNW, metabolically healthy normal weight; MHO metabolically healthy obese; MHOW, metabolically healthy overweight; MUHNW, metabolically unhealthy normal weight; MUHO, metabolically unhealthy obese; and MUHOW, metabolically unhealthy overweight.

Table 1.  Baseline Characteristics by BMI Category
Baseline Characteristics by BMI Category
Table 2.  Metabolic Markers for CVD and Total Mortality Among Participants With Obesity (NHANES-III, n = 3236)
Metabolic Markers for CVD and Total Mortality Among Participants With Obesity (NHANES-III, n = 3236)
Table 3.  Factors Associated With CVD and Total Mortality for Combinations of Factors Among Participants With Obesity (NHANES-III, n = 2511)a
Factors Associated With CVD and Total Mortality for Combinations of Factors Among Participants With Obesity (NHANES-III, n = 2511)a
Supplement.

eTable 1. Baseline Assessment Within the Third National Health and Nutrition Examination Survey (NHANES-III) and the UK Biobank

eTable 2. Summary of Missing Data for NHANES-III (n = 17 742)

eTable 3. Cutoffs at Maximal Youden Index Among Obese Participants (NHANES-III n = 2511)*

eTable 4. Baseline Characteristics of UK Biobank Participants in Groups of Metabolic Health and BMI (n = 374 079)

eTable 5. Sensitivity Analysis of CVD Mortality

eTable 6. Sensitivity Analysis of Total Mortality

eTable 7. Stratified Analysis by Sex

eTable 8. Stratified Analysis by CVD Type

eTable 9. Adjusted HRs and 95% CIs of CVD and Total Mortality in Subgroups Classified by BMI Categories Extended by Obesity Class 1 to 3 and MH Defined by Our New Definition (UK Biobank n = 374 079)

eTable 10. Adjusted HRs and 95% CIs of CVD and Total Mortality in Subgroups Classified by BMI Categories and MH Defined by a) Our New Definition, b) New Definition Without the Diabetes Criterion, c) New Definition Without the WHR Criterion, d) New Definition Without the Blood Pressure Criterion (UK Biobank n = 374 079)

eFigure 1. Adjusted HRs and 95% CIs for Risk of CVD Mortality in Subgroups Classified by BMI Categories and MH Defined by a) Our New Exploratory Definition b) New Definition, Adjusted for ATP-III Criteria, c) New Definition, Adjusted for Strict Definition d) New Definition, Adjusted for HOMA-IR (NHANES-III n = 12 341; UK Biobank Participants Fasted <6h Excluded n = 41 431).

eFigure 2. Adjusted HRs and 95% CIs for Risk of CVD Mortality in Subgroups Classified by BMI Categories and MH Defined by a) ATP-III Criteria; b) Strict Definition; c) HOMA-IR and Each Additionally Adjusted for Our New Exploratory Definition; (NHANES-III n = 12 341; UK Biobank Participants Fasted <6h Excluded n = 41 431; Definition With HOMA-IR Not Possible)

eFigure 3. Adjusted HRs and 95% CIs for Risk of Total Mortality in Subgroups Classified by BMI Categories and MH Defined by a) Our New Exploratory Definition b) New Definition, Adjusted for ATP-III Criteria, c) New Definition, Adjusted for Strict Definition d) New Definition, Adjusted for HOMA-IR (NHANES-III n = 12 341; UK Biobank Participants Fasted <6h Excluded n = 41 431)

eFigure 4. Adjusted HRs and 95% CIs for Risk of Total Mortality in Subgroups Classified by BMI Categories and MH Defined by a) ATP-III Criteria; b) Strict Definition; c) HOMA-IR and Each Additionally Adjusted for Our New Exploratory Definition; (NHANES-III n = 12 341; UK Biobank Participants Fasted <6h Excluded n = 41 431; Definition With HOMA-IR Not Possible)

eReferences

1.
Bhaskaran  K, Dos-Santos-Silva  I, Leon  DA, Douglas  IJ, Smeeth  L.  Association of BMI with overall and cause-specific mortality: a population-based cohort study of 3·6 million adults in the UK.   Lancet Diabetes Endocrinol. 2018;6(12):944-953. doi:10.1016/S2213-8587(18)30288-2 PubMedGoogle ScholarCrossref
2.
Stefan  N, Häring  HU, Schulze  MB.  Metabolically healthy obesity: the low-hanging fruit in obesity treatment?   Lancet Diabetes Endocrinol. 2018;6(3):249-258. doi:10.1016/S2213-8587(17)30292-9 PubMedGoogle ScholarCrossref
3.
Schulze  MB.  Metabolic health in normal-weight and obese individuals.   Diabetologia. 2019;62(4):558-566. doi:10.1007/s00125-018-4787-8PubMedGoogle ScholarCrossref
4.
Stefan  N, Häring  HU, Hu  FB, Schulze  MB.  Metabolically healthy obesity: epidemiology, mechanisms, and clinical implications.   Lancet Diabetes Endocrinol. 2013;1(2):152-162. doi:10.1016/S2213-8587(13)70062-7 PubMedGoogle ScholarCrossref
5.
Kramer  CK, Zinman  B, Retnakaran  R.  Are metabolically healthy overweight and obesity benign conditions? a systematic review and meta-analysis.   Ann Intern Med. 2013;159(11):758-769. doi:10.7326/0003-4819-159-11-201312030-00008 PubMedGoogle ScholarCrossref
6.
Eckel  N, Meidtner  K, Kalle-Uhlmann  T, Stefan  N, Schulze  MB.  Metabolically healthy obesity and cardiovascular events: a systematic review and meta-analysis.   Eur J Prev Cardiol. 2016;23(9):956-966. doi:10.1177/2047487315623884 PubMedGoogle ScholarCrossref
7.
Lassale  C, Tzoulaki  I, Moons  KGM,  et al.  Separate and combined associations of obesity and metabolic health with coronary heart disease: a pan-European case-cohort analysis.   Eur Heart J. 2018;39(5):397-406. doi:10.1093/eurheartj/ehx448 PubMedGoogle ScholarCrossref
8.
Eckel  N, Li  Y, Kuxhaus  O, Stefan  N, Hu  FB, Schulze  MB.  Transition from metabolic healthy to unhealthy phenotypes and association with cardiovascular disease risk across BMI categories in 90 257 women (the Nurses’ Health Study): 30 year follow-up from a prospective cohort study.   Lancet Diabetes Endocrinol. 2018;6(9):714-724. doi:10.1016/S2213-8587(18)30137-2 PubMedGoogle ScholarCrossref
9.
National Center for Health Statistics.  Plan and operation of the Third National Health and Nutrition Examination Survey, 1988-94—series 1: programs and collection procedures.   Vital Health Stat 1. 1994;(32):1-407.PubMedGoogle Scholar
10.
National Center for Health Statistics. The Third National Health and Nutrition Examination Survey—NHANES III: mortality follow-up through 2006. Updated December 2, 2009. Accessed August 8, 2015. https://www.cdc.gov/nchs/data/datalinkage/matching_methodology_nhanes3_final.pdf
11.
Sudlow  C, Gallacher  J, Allen  N,  et al.  UK Biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age.   PLoS Med. 2015;12(3):e1001779. doi:10.1371/journal.pmed.1001779 PubMedGoogle Scholar
12.
Biobank  UK. Protocol for a large-scale prospective epidemiological resource. Published 2007. Accessed July 16, 2019. https://www.ukbiobank.ac.uk/wp-content/uploads/2011/11/UK-Biobank-Protocol.pdf
13.
Gunter  EW, Lewis  BG, Koncikowski  SM. Laboratory procedures used for the Third National Health and Nutrition Examination Survey (NHANES-III), 1988–1994. Published 1996. Accessed August 8, 2015. https://www.cdc.gov/nchs/data/nhanes/nhanes3/cdrom/NCHS/MANUALS/LABMAN.PDF
14.
Otten  J, Ahrén  B, Olsson  T.  Surrogate measures of insulin sensitivity vs the hyperinsulinaemic-euglycaemic clamp: a meta-analysis.   Diabetologia. 2014;57(9):1781-1788. doi:10.1007/s00125-014-3285-x PubMedGoogle ScholarCrossref
15.
National Center for Health Statistics. National Health and Nutrition Examination Survey, 2005-2006: prescription medications data documentation, codebook, and frequencies. Updated June 2009. https://wwwn.cdc.gov/Nchs/Nhanes/2005-2006/RXQ_RX_D.htm
16.
Harel  O, Zhou  XH.  Multiple imputation: review of theory, implementation and software.   Stat Med. 2007;26(16):3057-3077. doi:10.1002/sim.2787 PubMedGoogle ScholarCrossref
17.
Rubin  DR.  Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons, Inc.; 1987. doi:10.1002/9780470316696
18.
Grundy  SM, Brewer  HB  Jr, Cleeman  JI, Smith  SC  Jr, Lenfant  C; American Heart Association; National Heart, Lung, and Blood Institute.  Definition of metabolic syndrome: report of the National Heart, Lung, and Blood Institute/American Heart Association conference on scientific issues related to definition.   Circulation. 2004;109(3):433-438. doi:10.1161/01.CIR.0000111245.75752.C6 PubMedGoogle ScholarCrossref
19.
Flegal  KM, Graubard  BI, Williamson  DF, Cooper  RS.  Reverse causation and illness-related weight loss in observational studies of body weight and mortality.   Am J Epidemiol. 2011;173(1):1-9. doi:10.1093/aje/kwq341 PubMedGoogle ScholarCrossref
20.
Rey-López  JP, de Rezende  LF, Pastor-Valero  M, Tess  BH.  The prevalence of metabolically healthy obesity: a systematic review and critical evaluation of the definitions used.   Obes Rev. 2014;15(10):781-790. doi:10.1111/obr.12198 PubMedGoogle ScholarCrossref
21.
Pischon  T, Boeing  H, Hoffmann  K,  et al.  General and abdominal adiposity and risk of death in Europe.   N Engl J Med. 2008;359(20):2105-2120. doi:10.1056/NEJMoa0801891 PubMedGoogle ScholarCrossref
22.
Wormser  D, Kaptoge  S, Di Angelantonio  E,  et al; Emerging Risk Factors Collaboration.  Separate and combined associations of body-mass index and abdominal adiposity with cardiovascular disease: collaborative analysis of 58 prospective studies.   Lancet. 2011;377(9771):1085-1095. doi:10.1016/S0140-6736(11)60105-0 PubMedGoogle ScholarCrossref
23.
Stefan  N.  Causes, consequences, and treatment of metabolically unhealthy fat distribution.   Lancet Diabetes Endocrinol. 2020;8(7):616-627. doi:10.1016/S2213-8587(20)30110-8 PubMedGoogle ScholarCrossref
24.
Lotta  LA, Wittemans  LBL, Zuber  V,  et al.  Association of genetic variants related to gluteofemoral vs abdominal fat distribution with type 2 diabetes, coronary disease, and cardiovascular risk factors.   JAMA. 2018;320(24):2553-2563. doi:10.1001/jama.2018.19329PubMedGoogle ScholarCrossref
25.
Neeland  IJ, Poirier  P, Després  JP.  Cardiovascular and metabolic heterogeneity of obesity: clinical challenges and implications for management.   Circulation. 2018;137(13):1391-1406. doi:10.1161/CIRCULATIONAHA.117.029617 PubMedGoogle ScholarCrossref
26.
Hebert  PR, Gaziano  JM, Chan  KS, Hennekens  CH.  Cholesterol lowering with statin drugs, risk of stroke, and total mortality. an overview of randomized trials.   JAMA. 1997;278(4):313-321. doi:10.1001/jama.1997.03550040069040 PubMedGoogle ScholarCrossref
27.
Liu  J, Zeng  FF, Liu  ZM, Zhang  CX, Ling  WH, Chen  YM.  Effects of blood triglycerides on cardiovascular and all-cause mortality: a systematic review and meta-analysis of 61 prospective studies.   Lipids Health Dis. 2013;12(1):159. doi:10.1186/1476-511X-12-159 PubMedGoogle ScholarCrossref
28.
Lewington  S, Whitlock  G, Clarke  R,  et al; Prospective Studies Collaboration.  Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55,000 vascular deaths.   Lancet. 2007;370(9602):1829-1839. doi:10.1016/S0140-6736(07)61778-4 PubMedGoogle ScholarCrossref
29.
Song  Y, Manson  JE, Meigs  JB, Ridker  PM, Buring  JE, Liu  S.  Comparison of usefulness of body mass index versus metabolic risk factors in predicting 10-year risk of cardiovascular events in women.   Am J Cardiol. 2007;100(11):1654-1658. doi:10.1016/j.amjcard.2007.06.073 PubMedGoogle ScholarCrossref
30.
Jonsson  S, Hedblad  B, Engström  G, Nilsson  P, Berglund  G, Janzon  L.  Influence of obesity on cardiovascular risk: twenty-three-year follow-up of 22,025 men from an urban Swedish population.   Int J Obes Relat Metab Disord. 2002;26(8):1046-1053. doi:10.1038/sj.ijo.0802060 PubMedGoogle ScholarCrossref
31.
Durward  CM, Hartman  TJ, Nickols-Richardson  SM.  All-cause mortality risk of metabolically healthy obese individuals in NHANES III.   J Obes. 2012;2012:460321. doi:10.1155/2012/460321 PubMedGoogle Scholar
32.
Lear  SA, James  PT, Ko  GT, Kumanyika  S.  Appropriateness of waist circumference and waist-to-hip ratio cutoffs for different ethnic groups.   Eur J Clin Nutr. 2010;64(1):42-61. doi:10.1038/ejcn.2009.70 PubMedGoogle ScholarCrossref
33.
Carroll  JF, Chiapa  AL, Rodriquez  M,  et al.  Visceral fat, waist circumference, and BMI: impact of race/ethnicity.   Obesity (Silver Spring). 2008;16(3):600-607. doi:10.1038/oby.2007.92 PubMedGoogle ScholarCrossref
34.
Fry  A, Littlejohns  TJ, Sudlow  C,  et al.  Comparison of sociodemographic and health-related characteristics of UK Biobank participants with those of the general population.   Am J Epidemiol. 2017;186(9):1026-1034. doi:10.1093/aje/kwx246 PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    2 Comments for this article
    EXPAND ALL
    Allometric Anthropometrics
    Jesse Krakauer, MD, FACP | Associated Physicians /Endocrinology
    Colleagues and I have described allometric (mutually independent) anthropometrics including A Body Shape Index (ABSI) and Hip Index (HI), which could be compared with waist-to-hip ratio waist as a risk factor for adverse outcomes (1-6). Specifically, an overall anthropometric risk ca be determined from the sum of Log OR for BMI, ABSI, HI, and this global ARI (Anthropometric Risk Indicator) can be substituted for waist circumference in defining a metabolic syndrome score.

    References

    1. Krakauer NY, Krakauer JC. The New Anthropometrics and Abdominal Obesity: A BodyShape Index, Hip Index, and Anthropometric Risk Index, Chapter 2
    in Nutrition in the Prevention and Treatment of Abdominal Obesity. 2019 Elsevier Inc. https://doi.org/10.1016/B978-0-12-816093-0.00002-1free at https://nirkrakauer.net/papers/anthropometrics_chapter_2018.pdf
    2. Krakauer NY, Krakauer JC. Anthropometrics, Metabolic Syndrome, and Mortality Hazard (2018). Journal of Obesity. 2018;2018:9241904. doi:10.1155/2018/9241904. https://doi.org/10.1155/2018/9241904 free at https://nirkrakauer.net/papers/MS_2018.pdf
    3. Bertoli S, Leone A, Krakauer NY, Bedogni G, Vanzulli A, Redaelli VI, et al. (2017) Association of Body Shape Index (ABSI) with cardio-metabolic risk factors: A cross-sectional study of 6081 Caucasian adults. PLoS ONE 12(9): e0185013. https://doi.org/10.1371/journal.pone.0185013
    4. Christakoudi S, Tsilidis KK, Muller DC, et al. A Body Shape Index (ABSI) achieves better mortality risk stratification than alternative indices of abdominal obesity: results from a large Europeancohort. Sci Rep. 2020;10(1):14541. Published 2020 Sep 3. doi:10.1038/s41598-020-71302-5
    5. Cho HW, Chung W, Moon S, Ryu OH, Kim MK, Kang JG. Effect of Sarcopenia and Body Shape on Cardiovascular Disease According to Obesity Phenotypes. Diabetes Metab J. 2020 Jan;44:e38. https://doi.org/10.4093/dmj.2019.0223https://e-dmj.org/DOIx.php?id=10.4093/dmj.2019.0223
    6. Krakauer NY, Krakauer JC. Association of Body Shape Index (ABSI) with Hand Grip Strength. Int J Environ Res Public Health. 2020 Sep 17;17(18):E6797. doi: 10.3390/ijerph17186797. PMID: 32957738.
    CONFLICT OF INTEREST: None Reported
    READ MORE
    Contrast in Alcohol Consumption Association Between the US and UK
    Jessica Robbins, PhD |
    It's striking in Table 1 that while BMI is strongly and positively associated with alcohol consumption in the NHANES III cohort, the opposite is true in the UK Biobank. I would be interested to know if this is because of the different age structures of the two studies, or because of the different drinking patterns in the US vs. the UK.
    CONFLICT OF INTEREST: None Reported
    Original Investigation
    Nutrition, Obesity, and Exercise
    May 7, 2021

    An Empirically Derived Definition of Metabolically Healthy Obesity Based on Risk of Cardiovascular and Total Mortality

    Author Affiliations
    • 1Department of Molecular Epidemiology, German Institute of Human Nutrition Potsdam-Rehbruecke, Nuthetal, Germany
    • 2German Center for Diabetes Research, Neuherberg, Germany
    • 3Institute of Nutritional Science, University of Potsdam, Potsdam, Germany
    • 4Institute of Diabetes Research and Metabolic Diseases of the Helmholtz Center München, University of Tübingen, Tübingen, Germany
    • 5Department of Internal Medicine IV, University Hospital of Tübingen, Tübingen, Germany
    • 6Sorbonne Paris North University, the National Institute for Health and Medical Research, the National Research Institute for Agriculture, Food and Environment, the National Conservatory of Arts and Crafts and the Nutritional Epidemiology Research Team, Epidemiology and Statistics Research Centre, University of Paris, Bobigny, France
    JAMA Netw Open. 2021;4(5):e218505. doi:10.1001/jamanetworkopen.2021.8505
    Key Points

    Question  Can a definition of metabolic health be empirically derived that allows to stratify risk of mortality in people with obesity?

    Findings  In this study including data from 386 420 individuals across 2 cohorts, a new definition of metabolically healthy obesity was established. People with a phenotype defined as metabolically unhealthy using this definition had significantly higher hazard ratios for cardiovascular disease mortality and total mortality irrespective of body mass index category, and people with phenotypes defined as having metabolically healthy obesity displayed no increased risk.

    Meaning  A new definition of metabolic health identified in this study may allow to stratify risk of mortality not only in people with obesity but also in people with overweight and normal weight.

    Abstract

    Importance  People classified by a priori definitions as having metabolically healthy obesity have frequently been found to be at increased risk of mortality, compared with individuals with metabolically healthy normal weight, suggesting these definitions may be insufficient.

    Objectives  To systematically derive a new definition of metabolic health (MH) and investigate its association with cardiovascular disease (CVD) mortality and total mortality.

    Design, Setting, and Participants  In a cohort study using data from the third National Health and Nutrition Examination Survey (NHANES-III), a representative survey using complex multistage probability sampling, anthropometric factors, biomarkers, and blood pressure (BP) associated with total and CVD mortality among participants with obesity were identified with Cox proportional hazards regression. Area under the receiver operating characteristic was calculated to identify predictive factors for mortality to be used to define MH, cutoff levels were determined by the Youden index, and the findings were validated through comparison with the independent UK Biobank cohort, a population-based prospective study. All nonpregnant participants in the databases aged 18 to 75 years with no history of CVD, body mass index greater than or equal to 18.5, and who fasted 6 or more hours before examination in NHANES-III were included; participants in the UK Biobank cohort who did not have blood measurements were excluded. The study was conducted from 2015 to 2020.

    Exposures  Body mass index and MH were defined by the new definition and compared with 3 a priori definitions.

    Main Outcomes and Measures  Cardiovascular disease mortality and total mortality.

    Results  Within the NHANES-III (n = 12 341) cohort, mean (SD) age was 41.6 (29.2) years, 50.7% were women, and mean follow-up was 14.5 (2.7) years. Within the UK Biobank (n = 374 079) cohort, mean (SD) age was 56.2 (8.1) years, 55.1% were women, and mean follow-up was 7.8 (1.0) years. Use of blood pressure (BP)-lowering medication (hazard ratio [HR] for CVD mortality, 2.41; 95% CI, 1.50-3.87 and total mortality, 2.05; 95% CI, 1.47-2.84), diabetes, and several continuous factors were associated with mortality. Of all significant continuous factors, the combination of systolic BP and waist-to-hip ratio showed the highest area under the receiver operating characteristic (CVD mortality: 0.775; 95% CI, 0.770-0.781; total mortality: 0.696; 95% CI, 0.694-0.699). Thus, MH was defined as systolic BP less than 130 mm Hg, no BP-lowering medication, waist-to-hip ratio less than 0.95 for women and less than 1.03 for men, and no self-reported (ie, prevalent) diabetes. In both cohorts, metabolically healthy obesity was not associated with CVD and total mortality compared with metabolically healthy normal weight. For NHANES-III, the hazard ratio was 0.68 (95% CI, 0.30-1.54) for CVD mortality and 1.03 (95% CI, 0.70-1.51) for total mortality. For UK Biobank, the hazard ratio was 1.17 (95% CI, 0.81-1.69) for CVD mortality and 0.98 (95% CI, 0.87-1.10) for total mortality. Regardless of body mass index, all metabolically unhealthy groups displayed increased risks.

    Conclusions and Relevance  This newly proposed definition of MH may identify a subgroup of people with obesity without increased risk of mortality and stratify risks in people who are overweight or normal weight.

    Introduction

    A 2018 study has shown that obesity, defined as body mass index (BMI) of 30 or more (calculated as weight in kilograms divided by height in meters squared), is associated with cardiometabolic diseases and increased mortality.1 However, not all people with obesity have an increased risk of cardiometabolic diseases. This subgroup is considered as having metabolically healthy obesity (MHO) and their counterpart, individuals with metabolically unhealthy obesity (MUHO), might have an excess risk for adverse outcomes.2 Debate on this phenotype relates to the lack of a uniform definition to identify metabolic health (MH).3 Previously used definitions related to MH are frequently based on the absence of metabolic syndrome or absence of insulin resistance (eg, assessed by homeostatic model assessment of insulin resistance [HOMA-IR]).4 In theory, cardiovascular disease (CVD) and mortality risk of individuals with MHO should be similar compared with individuals who have metabolically healthy normal weight (MHNW). However, meta-analyses5,6 and studies in large-scale cohorts7,8 observed increased risks for CVD mortality and total mortality for individuals with MHO despite the absence of metabolic syndrome. Although some studies indicate that stricter definitions of MH based on metabolic syndrome (absence of all components excluding waist circumference), may allow identification of a low-risk obesity group,6 a recent study does not support this notion.8 Using previous definitions and criteria may not be sufficient to identify an obese subgroup not at increased risk.

    An alternative approach to using a priori definitions of MH is to explore risk factors for mortality among people with obesity and empirically derive a new definition. To our knowledge, this has not been applied systematically before; therefore, we aimed to establish a new definition for MH based on the mortality follow-up of the third National Health and Nutrition Examination Survey (NHANES-III). We evaluated the association of this new definition with CVD mortality and total mortality, validated it in an independent cohort (UK Biobank), and compared it with commonly used a priori definitions in both cohorts.

    Methods

    The data analyses of the 2 cohort studies were conducted from 2015 to 2020. The NHANES-III study was approved by the National Center for Health Statistics and the UK Biobank study was approved by the North West Multicenter Research Ethics Committee. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

    NHANES-III

    NHANES-III, a representative survey with complex multistage probability sampling design, was conducted between 1988 and 1994 in the US. People aged 2 to 5 years, 60 years and older, and those who were Mexican Americans or African American were oversampled.9 Baseline data assessments included a personal interview and physical examinations in 33 199 participants. The NHANES-III protocol was approved by the National Center for Health Statistics. The participants gave written informed consent before participation.9 Our analyses included all nonpregnant participants aged 18 to 75 years with no history of CVD, BMI greater than or equal to 18.5, and fasted for 6 or more hours before the examination (n = 12 341).

    Mortality data, including the underlying causes of death until 2006, was used and classified by International Statistical Classification of Disease, 10th Revision (ICD-10) codes.10 Both total and CVD mortality (ICD-10 codes I00-I78) were considered as outcomes.

    UK Biobank

    The UK Biobank is a large, population-based prospective study with 502 506 participants aged 40 to 69 years. Baseline assessment (22 assessment centers, 2006-2010) included electronic signed informed consent, a self-completed questionnaire, personal interview, physical measurements, and blood collection.11 The North West Multicenter Research Ethics Committee approved the study.12

    In contrast to NHANES-III, UK Biobank participants were not instructed to fast before assessment.12 Because no fasting blood levels were necessary for the main analysis (n = 374 079), we made exclusions identical to those with NHANES-III plus exclusion of missing covariates and conducted a separate analysis (n = 41 431) in which we additionally excluded participants who fasted for less than 6 hours. This research was conducted using the UK Biobank resource under application number 50426. Mortality follow-up data were available via linkage with national death registries until 2016 for UK Biobank, and we used identical ICD-10 codes to define the outcomes as in NHANES-III.12

    Measurements

    Participants were categorized into BMI groups: normal weight (18.5-24.9), overweight (25.0-29.9), and obesity (≥30.0). Anthropometric and blood pressure (BP) measurements were conducted comparably in both cohorts (eTable 1 in the Supplement). Blood samples were analyzed for triglyceride, total cholesterol, high-density lipoprotein cholesterol, glucose, and hemoglobin A1c levels.12,13 In addition, we used measurements of C-reactive protein, insulin, γ-glutamyltransferase, and alanine aminotransferase levels from NHANES-III.13 Insulin sensitivity was estimated with quantitative insulin sensitivity check index, because it exhibited stronger correlations with the hyperinsulinemic-euglycemic clamp than HOMA-IR in a meta-analysis from 2014.14 Race/ethnicity and prevalent diabetes were self-reported in both cohorts; current medication was self-reported in the UK Biobank and containers were presented at the interview in NHANES-III.12,15

    Statistical Analysis

    We applied multiple imputation to missing data on outcome, exposure, and confounder variables for the NHANES-III sample (eTable 2 in the Supplement).16 We used the Markov Chain Monte Carlo method to create 10 data sets with monotone missing patterns and monotone methods to impute the remaining missing values. Rubin’s rules were used to combine the results.17

    To identify anthropometric and metabolic parameters associated with CVD or total mortality, we calculated Cox proportional hazard regression models in a subsample of NHANES-III participants with obesity (using proc surveyphreg to account for the survey design). We used age as the underlying time scale and adjusted for the following known confounders (self-reported with predefined categories to choose from): sex (male, female), race/ethnicity (non-Hispanic white, non-Hispanic black, Mexican-American, other), educational level (≤8, 9-12, ≥13 years of school completed), income (<$20 000 or≥$20 000 per year), marital status (married, previously, never), smoking status (never, former, current), alcohol consumption (0, 1-10, >10 drinks per month), and physical activity (never, 1-8, >8 times per month). Similar to definitions of metabolic syndrome,18 we aimed to identify elevated risk factor levels as components of an MH definition either when participants were receiving drug treatment or, among nontreated individuals, on measured risk factor levels. Thus, for continuous factors that showed significant associations with CVD mortality or total mortality, we calculated areas under the receiver operating characteristic (AUROC) curve in a subsample of participants with obesity without BP-lowering, glucose-lowering, and lipid-lowering medication (n = 2511) to evaluate which factors discriminate best between at risk and not at risk. We analyzed each factor separately and then in combined models, starting with the metabolic factor with the highest AUROC and extending by those with the next highest AUROC. Factors remained in the model when their inclusion led to a significant improvement in AUROC. We used Youden index to determine statistically optimal cutoff levels.

    We then compared relative mortality risks in subgroups stratified by BMI and MH using aforementioned adjustments. We defined MH using our new definition, as well as 3 a priori definitions:

    1. Absence of metabolic syndrome according to National Cholesterol Education Adult Treatment Panel-III criteria. Individuals are considered healthy if 2 or fewer criteria are present: waist circumference, 102 cm (men) or greater than 88 cm (women); BP greater than or equal to 130/85 mm Hg or using BP-lowering medication; triglyceride level greater than or equal to 150 mg/dL or using lipid-lowering medication; high-density lipoprotein cholesterol level less than 40 mg/dL (men) or less than 50 mg/dL (women); and fasting glucose level greater than or equal to 110 mg/dL or prevalent diabetes.18

    2. HOMA-IR (healthy if <2.5 [arbitrary unit of measure]).

    3. Strict definition. Individuals are considered healthy if all of the following criteria are absent: BP greater than or equal to 130/85 mm Hg or using BP-lowering medication; fasting glucose level greater than or equal to 100 mg/dL and hemoglobin A1c level greater than or equal to 5.7% (39 mmol/mol) or using glucose-lowering medication; triglyceride level greater than or equal to 150 mg/dL, total cholesterol level greater than or equal to 240 mg/dL, or high-density lipoprotein cholesterol level less than 40 mg/dL (men) or less than 50 mg/dL (women) or using lipid-lowering medication.

    To compare our new definition with these definitions, we used models with mutual adjustments for our and the other definitions, assuming that a definition that is superior in distinguishing between at-risk and not-at-risk individuals would identify other population subgroups and its associations with mortality would remain unaffected when adjusted for other definitions. To address typical sources of bias in obesity/mortality analyses19 we performed 4 sensitivity analyses: exclusion of ever smokers, deaths within the first 2 years of follow-up, participants older than 60 years, and changing the reference BMI category to 20.0 to 22.49. In addition, we repeated the analysis stratified by sex and CVD mortality subtype (coronary heart disease and stroke mortality).

    We applied our new definition in the UK Biobank, using identical Cox proportional hazards regression models of associations between subgroups of BMI and MH status and CVD and total mortality and adjustments as used in NHANES-III, with additional adjustment for assessment center. We attempted to group covariates into categories as similar as possible to the NHANES-III analysis: sex (male, female), race/ethnicity (White, mixed, Asian, Black, Chinese, other), educational level (university/college [UK degree, similar to US graduate school], A levels [UK degree, similar to US high school diploma], O levels/GCSE [UK degree, similar to US grade 10], CSE [UK degree awarded until 1987 equivalent to GSCE], NVQ [technical training]), income (<£18 000 or≥£18 000 per year [approximately $25 000]), marital status (married and living together, living with another family member, living alone), smoking status (never, former, current), alcohol consumption (never, special occasions, 1-3 times per month, 1-2 times per week, 3-4 times per week, daily or almost daily), and physical activity (never, ≤20 minutes, and >20 minutes of vigorous activity per week). We performed the following analyses with identical models as conducted in NHANES-III: comparison of MH defined by our new definition with MH defined by National Cholesterol Education Adult Treatment Panel-III criteria and the strict definition, sensitivity analyses, and stratified analyses as described above. Because insulin levels were not measured in the UK Biobank it was not possible to compare MH defined by HOMA-IR with the other definitions.

    All analyses were performed in SAS, version 9.4, Enterprise Guide 7.1 (SAS Institute Inc). Significance tests were 2-tailed and P values <.05 were considered statistically significant. Analysis within NHANES-III was conducted in 2015 and the UK Biobank was conducted in 2020.

    Results

    The NHANES-III participants (n = 12 341) had a mean (SD) age of 41.6 (29.2) years and mean BMI of 27.2; 50.7% of the participants were women and 49.3% were men. The UK Biobank participants (n = 374 079) had a mean age of 56.2 (8.1) years and a mean BMI of 27.4; 55.1% were women and 44.9% men. During a mean (SD) follow-up of 14.5 (2.7) years in the NHANES-III cohort, 1758 deaths from all causes and 594 deaths associated with CVD occurred; during the mean (SD) 7.8 (1.0)-year follow-up in the UK Biobank cohort, 12 950 deaths from all causes and 2162 deaths associated with CVD occurred. In both cohorts, participants with obesity were less likely to have higher educational levels, more likely to be categorized in the lowest income group, and more likely to be physically inactive, compared with normal-weight and overweight participants. The UK Biobank cohort consisted of older, less ethnically diverse participants, and the proportion of never drinkers and current smokers was considerably smaller compared with the NHANES-III cohort. Generally, the confounder structure was similar in both cohorts (Table 1).

    Anthropometric and metabolic factors significantly associated with both CVD and total mortality in NHANES-III were waist circumference (hazard ratio [HR] per SD, CVD mortality: 1.40; 95% CI, 1.18-1.66; total mortality: 1.35; 95% CI, 1.19-1.52), waist-to-hip ratio (WHR) (HR per SD, CVD mortality: 1.37; 95% CI, 1.08-1.75; total mortality: 1.29; 95% CI, 1.10-1.51), quantitative insulin sensitivity check index (HR per SD, CVD mortality: 0.74; 95% CI, 0.58-0.94; total mortality: 0.74; 95% CI, 0.64-0.86), use of BP-lowering medication (HR per SD, CVD mortality: 2.41; 95% CI, 1.50-3.87; total mortality: 2.05; 95% CI, 1.47-2.84), and prevalent diabetes (HR per SD, CVD mortality: 1.80; 95% CI, 1.02-3.18; total mortality: 1.88; 95% CI, 1.27-2.78). In addition, systolic BP was significantly associated with CVD mortality (HR per SD, 1.32; 95% CI, 1.07-1.62), and BMI (HR per SD, 1.20; 95% CI, 1.06-1.36), fasting glucose (HR per SD, 1.15; 95% CI, 1.02-1.30), hemoglobin A1c, (HR per SD, 1.13; 95% CI, 1.01-1.27), C-reactive protein (HR per SD, 1.18; 95% CI, 1.04-1.34), and γ-glutamyltransferase (HR per SD, 1.14; 95% CI, 1.06-1.23) levels were significantly associated with total mortality (Table 2). We retained the categorical factors prevalent diabetes and use of BP-lowering medication and subsequently calculated AUROCs for all significant continuous factors (Table 3) within NHANES-III. Systolic BP showed the highest predictive ability for both outcomes (CVD mortality: AUROC, 0.775; 95% CI, 0.770-0.781; total mortality: AUROC, 0.696; 95% CI, 0.694-0.699). Of the remaining variables, WHR improved the AUROC significantly, but only for total mortality (0.640; 95% CI, 0.639-0.642). Statistically significant optimal cutoffs for systolic BP were 125 (Youden index, 0.460) and 124 mm Hg (Youden index, 0.320) for CVD and total mortality (eTable 3 in the Supplement). However, owing to the proximity of these levels to the well-established cutoff of 130 mm Hg for pre-hypertension, we decided to adhere to that established cutoff. When evaluating WHR, AUROCs were higher for CVD mortality than for total mortality; hence, we used the WHR cutoff for CVD mortality. Accordingly, we constructed the following criteria for MH:

    1. systolic BP less than 130 mm Hg and no use of BP-lowering medication,

    2. WHR less than 0.95 (women) and less than 1.03 (men), and

    3. no prevalent diabetes.

    The proportion of participants with obesity who did not fulfill criteria was 42.2% for criterion 1, 33.9% for criterion 2, and 6.58% for criterion 3 in NHANES-III and, in the UK Biobank, was 77.7% for criterion 1, 14.1% for criterion 2, and 9.9% for criterion 3. According to the new definition, 41.2% of participants with obesity were metabolically healthy within NHANES-III and 19.3% of participants with obesity were metabolically healthy within the UK Biobank. The proportion of MHO by the other 3 definitions in NHANES-III was 9.9% (strict definition), 31.7% (HOMA-IR), and 46.7% (National Cholesterol Education Adult Treatment Panel-III criteria). Only 5.7% of NHANES-III participants had MHO according to all 4 definitions (Figure 1). More details on the frequency of covariates in each group of MH and BMI can be found in eTable 4 in the Supplement.

    Cross-classifying participants by BMI and MH defined by our new definition yielded increased adjusted risks for CVD mortality for all individuals classified as unhealthy, compared with individuals with MHNW independently of BMI categories in both cohorts. Conversely, CVD mortality of individuals with MHO was not significantly increased compared with MHNW participants (NHANES-III: hazard ratio, 0.68; 95% CI, 0.30-1.54; UKB: hazard ratio, 1.17; 95% CI, 0.81-1.69) (Figure 2). These associations remained stable when we adjusted for the other 3 definitions (eFigure 1 in the Supplement).

    When MH was defined by the other definitions, similar associations could be observed only in some instances (eFigure 2 in the Supplement). Metabolic health defined by National Cholesterol Education Adult Treatment Panel-III criteria revealed differing risk groups within the UK Biobank; however, almost none of the groups displayed increased risks in NHANES-III, especially when adjusted for the new definition. When MH was defined by the strict definition, most participants were classified as unhealthy within both cohorts, which led to the healthy groups being small and, in the MHO group in the UKB, even too small for analyses (no cases observed). Defining MH according to HOMA-IR was only possible within NHANES-III. Increased risks could be observed in the MUHNW and the MUHO groups for both total and CVD mortality, but not in the MUHOW group. For all 3 definitions, the risk estimates were lower when adjusted for our new definition.

    When total mortality was investigated, similar risk patterns were observed in both cohorts: individuals classified as MHO by the new definition were not at increased risk compared with MHNW participants (NHANES-III: HR, 1.03; 95% CI, 0.70-1.51; UK Biobank: HR, 0.98; 95% CI, 0.87-1.10), but individuals classified as metabolically unhealthy had an increased risk, irrespective of BMI category (Figure 2). However, the risk estimates were considerably smaller in the UK Biobank. These associations remained when adjusted for the other definitions (eFigure 3 in the Supplement). When using the other 3 methods to define MH, results similar to those for CVD mortality were observed (eFigure 4 in the Supplement).

    All sensitivity analyses yielded qualitatively similar results in both cohorts (eTable 5 and eTable 6 in the Supplement). We observed similar risks for women and men with MHO (eTable 7 in the Supplement). Individuals with MHO were not at increased risk for both coronary heart disease and stroke mortality compared with MHNW participants (eTable 8 in the Supplement). Furthermore, we subdivided participants with obesity into obesity class 1 (BMI, 30.0-34.9), 2 (35.0-39.9), and 3 (≥40) and repeated the main analysis within the UK Biobank (eTable 9 in the Supplement). The analysis confirmed higher risks for both outcomes in unhealthy groups. Metabolically healthy obesity was not associated with increased risks compared with MHNW for obesity classes 1 and 2 (CVD mortality, obesity class 1: HR, 1.00; 95% CI, 0.65-1.52; obesity class 2: HR, 0.83; 95% CI, 0.31-2.26). However, for obesity class 3, MHO was associated with higher risks for CVD mortality compared with MHNW (HR, 6.34; 95% CI, 3.43-6.16). When defining MH with only 2 of the 3 criteria of our definition, omitting 1 criterion at a time, we observed relatively similar increases in HRs for CVD mortality for MHO (without diabetes criterion: HR, 1.31; 95% CI, 0.92-1.84; without WHR criterion: HR, 1.28; 95% CI 0.92-1.80; without BP criterion: HR, 1.32; 95% CI, 1.15-1.51), supporting the importance of all 3 criteria (eTable 10 in the Supplement).

    Discussion

    This systematic assessment of mortality data and various anthropometric and metabolic factors yielded a simple definition to categorize participants with obesity as metabolically healthy or unhealthy. This definition classified more than 40% of individuals with obesity as MHO within the NHANES-III population, a representative sample of the US adult population, and 20% within the UK Biobank; this group was not at increased risk for CVD and total mortality compared with MHNW individuals. Conversely, individuals classified as metabolically unhealthy were at increased risk compared with MHNW participants, independent of BMI category in both cohorts. The results were robust after accounting for typical sources of bias for the associations between obesity and mortality. The risks of both outcomes were almost equally increased in participants with MUHNW as in MUHO in both cohorts, indicating that our new definition may also be helpful to detect an increased risk in lean people.

    Prospective studies on CVD and mortality risks of MHO provided inconsistent results.3,4 The use of varying definitions for MH might be responsible for the observed heterogeneity.20 Often, the absence of metabolic syndrome or insulin resistance has been used to identify MHO, however, with differing criteria and cutoffs. When the results from individual studies using a variety of definitions based on metabolic syndrome or HOMA-IR and risk for CVD events and total mortality were summarized in meta-analyses, individuals with MHO were still at increased risk compared with individuals with MHNW, especially in the long term.5,6 In the present study, MHO defined by either of these 2 definitions showed inconsistent results among both cohorts, and the associations were not independent of our new definition; however, our definition was associated with the risk of mortality independently of other definitions. This finding suggests that our definition better distinguishes between at-risk and not-at-risk individuals. Our results, however, also indicate that our definition may only be able to identify a low-risk MHO phenotype among individuals with a BMI less than 40. Higher mortality was observed for people with severe obesity (BMI ≥40), irrespective of MH.

    Our definition has substantial differences from common versions of metabolic syndrome definitions. Waist circumference as a marker for abdominal obesity was replaced by WHR, which has the advantage of less-pronounced correlations with BMI21,22 and the additional consideration of hip circumference. Although waist circumference estimates visceral fat, which is a risk factor for metabolic disorders and CVD, waist circumference also estimates subcutaneous abdominal fat, which is considered to have less-detrimental effects on metabolism.23 In contrast, hip circumference is a proxy of lower body fat, which might have protective effects on metabolism.23-25 In support of stronger predictive power of WHR compared with waist circumference, WHR was more associated with mortality than waist circumference in individuals with high BMI.21

    Because measures of dyslipidemia were not significantly associated with mortality in our study, our new definition does not include criteria for dyslipidemia. Although this lack of inclusion seems surprising considering the well-known association between dyslipidemia and increased risk of death and CVD,26,27 associations between total cholesterol level and vascular mortality are weaker among participants with obesity compared with other BMI categories.28

    Furthermore, our definition only classifies individuals as MH who fulfill all 3 criteria, whereas using the absence of metabolic syndrome to define MH allows up to 2 risk factors to be present. Our study supports findings from 2 studies that found that the absence of all metabolic syndrome criteria (without the waist circumference criterion) identifies people with obesity who are not at increased risk for total and CVD mortality.29,30 Nevertheless, in our analysis the association of this strict definition with total mortality was attenuated when adjusted for our new definition. The simultaneous absence of all metabolic syndrome criteria classifies a small fraction of individuals with obesity as MH, whereas our exploratory definition identified a considerably larger proportion of people with obesity not at increased risk for total and CVD mortality.

    Our results for MH and total mortality, using metabolic syndrome criteria and HOMA-IR to define MH, are in agreement in direction of effect with a previous analysis in the NHANES-III population, using the same duration of follow-up; however, effect estimates differ quantitatively.31

    Strengths and Limitations

    To our knowledge, this is the first study that systematically assessed multiple cardiometabolic risk factors for a possible definition for the MHO phenotype. Strengths of this study are the prospective design, long follow-up, large sample sizes, and validation in an independent cohort. Limitations are lack of data on changes in body weight and metabolic factors owing to the original design of NHANES-III. Thus, we were unable to investigate the outcomes associated with changes in MH status, which have been reported to substantially alter CVD risk.3,8 Furthermore, body fat distribution might differ according to race/ethnicity and our proposed WHR cutoff might not be suitable for all populations, especially Asian populations, which were underrepresented in both cohorts.32,33 In addition, although NHANES-III is a representative survey, the UK Biobank consists of volunteers not representative of the UK population34 and has a higher mean age, which might explain the difference in proportions of MHO participants between the 2 cohorts. Another limitation is that, although the C index might be better suited for survival data than area under the receiver operating characteristic, the C index was not available for the surveyphreg procedure. Also, we cannot rule out that alternative variable selection methods may have resulted in different criteria. Still, we validated the utility of our definition by evaluating associations with mortality, also in comparison with other definitions and by repeating the analyses in an independent cohort.

    Conclusions

    In this study, we developed a new definition to identify individuals with MHO, based on self-reported diabetes, use of BP medication, systolic BP, and WHR, which can be used in both clinical and research settings. Our results suggest that people with MHO classified by this definition are not at increased risk for CVD or total mortality. Metabolically unhealthy individuals have a substantially higher risk, which is not explained by conventional definitions of MH. Thus, our new definition may be important not only to stratify risk of mortality in people with obesity, but also in people with overweight and normal weight.

    Back to top
    Article Information

    Accepted for Publication: March 4, 2021.

    Published: May 7, 2021. doi:10.1001/jamanetworkopen.2021.8505

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Zembic A et al. JAMA Network Open.

    Corresponding Author: Matthias B. Schulze, DrPH, Department of Molecular Epidemiology, German Institute of Human Nutrition Potsdam-Rehbruecke, Arthur-Scheunert-Allee 114-116, 14558 Nuthetal, Germany (mschulze@dife.de).

    Author Contributions: Ms Zembic and Dr Eckel contributed equally to this work, had full access to all of the data in the study, and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Zembic, Stefan, Schulze.

    Acquisition, analysis, or interpretation of data: Zembic, Eckel, Baudry, Schulze.

    Drafting of the manuscript: Zembic, Eckel.

    Critical revision of the manuscript for important intellectual content: Zembic, Stefan, Baudry, Schulze.

    Statistical analysis: Zembic, Eckel, Stefan.

    Obtained funding: Schulze.

    Administrative, technical, or material support: Zembic.

    Supervision: Stefan, Baudry, Schulze.

    Conflict of Interest Disclosures: Dr Zembic reported grants from German Federal Ministry of Education and Research (BMBF) DZD grant 82DZD00302 during the conduct of the study. Dr Eckel reported grants from German Federal Ministry of Education and Research (BMBF) DZD grant 82DZD00302 during the conduct of the study. Dr Schulze reported grants from German Federal Ministry of Education and Research during the conduct of the study. No other disclosures were reported.

    Funding/Support: The study was supported by grant 82DZD00302 from the German Federal Ministry of Education and Research to the German Center for Diabetes Research.

    Role of the Funder/Sponsor: The German Federal Ministry of Education and Research had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Additional Contributions: We thank all National Health and Nutrition Examination-III Survey and UK Biobank study participants, data managers, and staff for contributing to this study.

    References
    1.
    Bhaskaran  K, Dos-Santos-Silva  I, Leon  DA, Douglas  IJ, Smeeth  L.  Association of BMI with overall and cause-specific mortality: a population-based cohort study of 3·6 million adults in the UK.   Lancet Diabetes Endocrinol. 2018;6(12):944-953. doi:10.1016/S2213-8587(18)30288-2 PubMedGoogle ScholarCrossref
    2.
    Stefan  N, Häring  HU, Schulze  MB.  Metabolically healthy obesity: the low-hanging fruit in obesity treatment?   Lancet Diabetes Endocrinol. 2018;6(3):249-258. doi:10.1016/S2213-8587(17)30292-9 PubMedGoogle ScholarCrossref
    3.
    Schulze  MB.  Metabolic health in normal-weight and obese individuals.   Diabetologia. 2019;62(4):558-566. doi:10.1007/s00125-018-4787-8PubMedGoogle ScholarCrossref
    4.
    Stefan  N, Häring  HU, Hu  FB, Schulze  MB.  Metabolically healthy obesity: epidemiology, mechanisms, and clinical implications.   Lancet Diabetes Endocrinol. 2013;1(2):152-162. doi:10.1016/S2213-8587(13)70062-7 PubMedGoogle ScholarCrossref
    5.
    Kramer  CK, Zinman  B, Retnakaran  R.  Are metabolically healthy overweight and obesity benign conditions? a systematic review and meta-analysis.   Ann Intern Med. 2013;159(11):758-769. doi:10.7326/0003-4819-159-11-201312030-00008 PubMedGoogle ScholarCrossref
    6.
    Eckel  N, Meidtner  K, Kalle-Uhlmann  T, Stefan  N, Schulze  MB.  Metabolically healthy obesity and cardiovascular events: a systematic review and meta-analysis.   Eur J Prev Cardiol. 2016;23(9):956-966. doi:10.1177/2047487315623884 PubMedGoogle ScholarCrossref
    7.
    Lassale  C, Tzoulaki  I, Moons  KGM,  et al.  Separate and combined associations of obesity and metabolic health with coronary heart disease: a pan-European case-cohort analysis.   Eur Heart J. 2018;39(5):397-406. doi:10.1093/eurheartj/ehx448 PubMedGoogle ScholarCrossref
    8.
    Eckel  N, Li  Y, Kuxhaus  O, Stefan  N, Hu  FB, Schulze  MB.  Transition from metabolic healthy to unhealthy phenotypes and association with cardiovascular disease risk across BMI categories in 90 257 women (the Nurses’ Health Study): 30 year follow-up from a prospective cohort study.   Lancet Diabetes Endocrinol. 2018;6(9):714-724. doi:10.1016/S2213-8587(18)30137-2 PubMedGoogle ScholarCrossref
    9.
    National Center for Health Statistics.  Plan and operation of the Third National Health and Nutrition Examination Survey, 1988-94—series 1: programs and collection procedures.   Vital Health Stat 1. 1994;(32):1-407.PubMedGoogle Scholar
    10.
    National Center for Health Statistics. The Third National Health and Nutrition Examination Survey—NHANES III: mortality follow-up through 2006. Updated December 2, 2009. Accessed August 8, 2015. https://www.cdc.gov/nchs/data/datalinkage/matching_methodology_nhanes3_final.pdf
    11.
    Sudlow  C, Gallacher  J, Allen  N,  et al.  UK Biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age.   PLoS Med. 2015;12(3):e1001779. doi:10.1371/journal.pmed.1001779 PubMedGoogle Scholar
    12.
    Biobank  UK. Protocol for a large-scale prospective epidemiological resource. Published 2007. Accessed July 16, 2019. https://www.ukbiobank.ac.uk/wp-content/uploads/2011/11/UK-Biobank-Protocol.pdf
    13.
    Gunter  EW, Lewis  BG, Koncikowski  SM. Laboratory procedures used for the Third National Health and Nutrition Examination Survey (NHANES-III), 1988–1994. Published 1996. Accessed August 8, 2015. https://www.cdc.gov/nchs/data/nhanes/nhanes3/cdrom/NCHS/MANUALS/LABMAN.PDF
    14.
    Otten  J, Ahrén  B, Olsson  T.  Surrogate measures of insulin sensitivity vs the hyperinsulinaemic-euglycaemic clamp: a meta-analysis.   Diabetologia. 2014;57(9):1781-1788. doi:10.1007/s00125-014-3285-x PubMedGoogle ScholarCrossref
    15.
    National Center for Health Statistics. National Health and Nutrition Examination Survey, 2005-2006: prescription medications data documentation, codebook, and frequencies. Updated June 2009. https://wwwn.cdc.gov/Nchs/Nhanes/2005-2006/RXQ_RX_D.htm
    16.
    Harel  O, Zhou  XH.  Multiple imputation: review of theory, implementation and software.   Stat Med. 2007;26(16):3057-3077. doi:10.1002/sim.2787 PubMedGoogle ScholarCrossref
    17.
    Rubin  DR.  Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons, Inc.; 1987. doi:10.1002/9780470316696
    18.
    Grundy  SM, Brewer  HB  Jr, Cleeman  JI, Smith  SC  Jr, Lenfant  C; American Heart Association; National Heart, Lung, and Blood Institute.  Definition of metabolic syndrome: report of the National Heart, Lung, and Blood Institute/American Heart Association conference on scientific issues related to definition.   Circulation. 2004;109(3):433-438. doi:10.1161/01.CIR.0000111245.75752.C6 PubMedGoogle ScholarCrossref
    19.
    Flegal  KM, Graubard  BI, Williamson  DF, Cooper  RS.  Reverse causation and illness-related weight loss in observational studies of body weight and mortality.   Am J Epidemiol. 2011;173(1):1-9. doi:10.1093/aje/kwq341 PubMedGoogle ScholarCrossref
    20.
    Rey-López  JP, de Rezende  LF, Pastor-Valero  M, Tess  BH.  The prevalence of metabolically healthy obesity: a systematic review and critical evaluation of the definitions used.   Obes Rev. 2014;15(10):781-790. doi:10.1111/obr.12198 PubMedGoogle ScholarCrossref
    21.
    Pischon  T, Boeing  H, Hoffmann  K,  et al.  General and abdominal adiposity and risk of death in Europe.   N Engl J Med. 2008;359(20):2105-2120. doi:10.1056/NEJMoa0801891 PubMedGoogle ScholarCrossref
    22.
    Wormser  D, Kaptoge  S, Di Angelantonio  E,  et al; Emerging Risk Factors Collaboration.  Separate and combined associations of body-mass index and abdominal adiposity with cardiovascular disease: collaborative analysis of 58 prospective studies.   Lancet. 2011;377(9771):1085-1095. doi:10.1016/S0140-6736(11)60105-0 PubMedGoogle ScholarCrossref
    23.
    Stefan  N.  Causes, consequences, and treatment of metabolically unhealthy fat distribution.   Lancet Diabetes Endocrinol. 2020;8(7):616-627. doi:10.1016/S2213-8587(20)30110-8 PubMedGoogle ScholarCrossref
    24.
    Lotta  LA, Wittemans  LBL, Zuber  V,  et al.  Association of genetic variants related to gluteofemoral vs abdominal fat distribution with type 2 diabetes, coronary disease, and cardiovascular risk factors.   JAMA. 2018;320(24):2553-2563. doi:10.1001/jama.2018.19329PubMedGoogle ScholarCrossref
    25.
    Neeland  IJ, Poirier  P, Després  JP.  Cardiovascular and metabolic heterogeneity of obesity: clinical challenges and implications for management.   Circulation. 2018;137(13):1391-1406. doi:10.1161/CIRCULATIONAHA.117.029617 PubMedGoogle ScholarCrossref
    26.
    Hebert  PR, Gaziano  JM, Chan  KS, Hennekens  CH.  Cholesterol lowering with statin drugs, risk of stroke, and total mortality. an overview of randomized trials.   JAMA. 1997;278(4):313-321. doi:10.1001/jama.1997.03550040069040 PubMedGoogle ScholarCrossref
    27.
    Liu  J, Zeng  FF, Liu  ZM, Zhang  CX, Ling  WH, Chen  YM.  Effects of blood triglycerides on cardiovascular and all-cause mortality: a systematic review and meta-analysis of 61 prospective studies.   Lipids Health Dis. 2013;12(1):159. doi:10.1186/1476-511X-12-159 PubMedGoogle ScholarCrossref
    28.
    Lewington  S, Whitlock  G, Clarke  R,  et al; Prospective Studies Collaboration.  Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55,000 vascular deaths.   Lancet. 2007;370(9602):1829-1839. doi:10.1016/S0140-6736(07)61778-4 PubMedGoogle ScholarCrossref
    29.
    Song  Y, Manson  JE, Meigs  JB, Ridker  PM, Buring  JE, Liu  S.  Comparison of usefulness of body mass index versus metabolic risk factors in predicting 10-year risk of cardiovascular events in women.   Am J Cardiol. 2007;100(11):1654-1658. doi:10.1016/j.amjcard.2007.06.073 PubMedGoogle ScholarCrossref
    30.
    Jonsson  S, Hedblad  B, Engström  G, Nilsson  P, Berglund  G, Janzon  L.  Influence of obesity on cardiovascular risk: twenty-three-year follow-up of 22,025 men from an urban Swedish population.   Int J Obes Relat Metab Disord. 2002;26(8):1046-1053. doi:10.1038/sj.ijo.0802060 PubMedGoogle ScholarCrossref
    31.
    Durward  CM, Hartman  TJ, Nickols-Richardson  SM.  All-cause mortality risk of metabolically healthy obese individuals in NHANES III.   J Obes. 2012;2012:460321. doi:10.1155/2012/460321 PubMedGoogle Scholar
    32.
    Lear  SA, James  PT, Ko  GT, Kumanyika  S.  Appropriateness of waist circumference and waist-to-hip ratio cutoffs for different ethnic groups.   Eur J Clin Nutr. 2010;64(1):42-61. doi:10.1038/ejcn.2009.70 PubMedGoogle ScholarCrossref
    33.
    Carroll  JF, Chiapa  AL, Rodriquez  M,  et al.  Visceral fat, waist circumference, and BMI: impact of race/ethnicity.   Obesity (Silver Spring). 2008;16(3):600-607. doi:10.1038/oby.2007.92 PubMedGoogle ScholarCrossref
    34.
    Fry  A, Littlejohns  TJ, Sudlow  C,  et al.  Comparison of sociodemographic and health-related characteristics of UK Biobank participants with those of the general population.   Am J Epidemiol. 2017;186(9):1026-1034. doi:10.1093/aje/kwx246 PubMedGoogle ScholarCrossref
    ×