ORIGINAL ARTICLE Metabolically healthy obesity: different prevalences using different criteria S Velho 1 , F Paccaud 2 , G Waeber 3 , P Vollenweider 3 and P Marques-Vidal 2,4 1 Department of Nutrition and Dietetics, Portuguese Institute of Oncology Francisco Gentil, Lisbon, Portugal; 2 Institute of Social and Preventive Medicine (IUMSP), CHUV and Faculty of biology and medicine, Lausanne, Switzerland; 3 Department of Medicine, Internal Medicine, CHUV and Faculty of biology and medicine, Lausanne, Switzerland and 4 Cardiomet, CHUV, Lausanne, Switzerland Objective: To estimate the prevalence of metabolically healthy obesity (MHO) according to different definitions. Methods: Population-based sample of 2803 women and 2557 men participated in the study. Metabolic abnormalities were defined using six sets of criteria, which included different combinations of the following: waist; blood pressure; total, high-density lipoprotein or low-density lipoprotein-cholesterol; triglycerides; fasting glucose; homeostasis model assessment; high-sensitivity C-reactive protein; personal history of cardiovascular, respiratory or metabolic diseases. For each set, prevalence of MHO was assessed for body mass index (BMI); waist or percent body fat. Results: Among obese (BMI X30 kg/m 2 ) participants, prevalence of MHO ranged between 3.3 and 32.1% in men and between 11.4 and 43.3% in women according to the criteria used. Using abdominal obesity, prevalence of MHO ranged between 5.7 and 36.7% (men) and 12.2 and 57.5% (women). Using percent body fat led to a prevalence of MHO ranging between 6.4 and 43.1% (men) and 12.0 and 55.5% (women). MHO participants had a lower odd of presenting a family history of type 2 diabetes. After multivariate adjustment, the odds of presenting with MHO decreased with increasing age, whereas no relationship was found with gender, alcohol consumption or tobacco smoking using most sets of criteria. Physical activity was positively related, whereas increased waist was negatively related with BMI-defined MHO. Conclusion: MHO prevalence varies considerably according to the criteria used, underscoring the need for a standard definition of this metabolic entity. Physical activity increases the likelihood of presenting with MHO, and MHO is associated with a lower prevalence of family history of type 2 diabetes. European Journal of Clinical Nutrition (2010) 64, 1043–1051; doi:10.1038/ejcn.2010.114; published online 14 July 2010 Keywords: cross-sectional studies; obesity; risk factors; insulin resistance; body mass index; waist Introduction In the last decade, the worldwide prevalence of obesity has increased considerably, and so has the incidence of type 2 diabetes and cardiovascular disease (CVD) (Seidell, 2000; Kelly et al., 2008). Several studies have shown that the obese phenotype may be present without metabolic abnormalities (Karelis, 2008; Karelis and Rabasa-Lhoret, 2008; Stefan et al., 2008; Wildman et al., 2008). This clinical condition, termed benign obesity or metabolically healthy obesity (MHO), is restricted to a unique subset of the obese population which, despite excessive body mass index (BMI) or body fat, are insulin sensitive and have a normal blood pressure, lipid, in- flammation and hormonal profile (Karelis and Rabasa-Lhoret, 2008; Karelis, 2008; Stefan et al., 2008; Wildman et al., 2008). Among US obese adults, 29.2% of men and 35.4% of women (a total of approximately 19.5 million adults) are metabolically healthy. On the other hand, apparently 30.1% of normal- weight men and 21.1% of normal-weight women (a total of approximately 16.3 million adults) are metabolically abnormal (Wildman et al., 2008). There are currently no unique criteria for MHO, and several definitions have been proposed (Meigs et al., 2006; Aguilar-Salinas et al., 2008; Karelis, 2008; Wildman et al., 2008; Lynch et al., 2009). The criteria used to define MHO are summarized in Table 1. Most authors use blood pressure, lipids (mainly high-density lipoprotein-cholesterol and/or triglycerides) and glycaemic status (fasting blood glucose or homeostasis model assessment) to define MHO, although other clinical or biological data are also used. Still, to our Received 19 January 2010; revised 20 May 2010; accepted 24 May 2010; published online 14 July 2010 Correspondence: Dr P Marques-Vidal, Institut Universitaire de Me ´decine Sociale et Pre ´ventive, 17, rue du Bugnon, Lausanne, Vaud CH 1005, Switzerland. E-mail: Pedro-Manuel.Marques-Vidal@chuv.ch European Journal of Clinical Nutrition (2010) 64, 1043–1051 & 2010 Macmillan Publishers Limited All rights reserved 0954-3007/10 www.nature.com/ejcn