NATURE REVIEWS | CARDIOLOGY ADVANCE ONLINE PUBLICATION | 1 Introduction Over the past three decades, the worldwide prevalence of obesity has increased to alarm- ing levels. In the USA, an estimated one-third of adults are obese. 1 The WHO estimates that the incidence of obesity has tripled over the past 20 years in developing countries, and projections predict that there will be 2.3 billion overweight (25–29.9 kg/m 2 ) and 700 million obese (≥30.0 kg/m 2 ) individuals worldwide by 2015. 2 The NIH, WHO, American Association of Clinical Endocrinologists, and American College of Endocrinology define obesity as an excess of total body fat, 2–6 specifically body fat that is 20–25% of total body weight in men, and 30–35% in women. 3,5,7,8 However, data to justify these limits are poor as most of these studies have not examined whether obesity correlates with cardiovascular disease (CVD) or mortality. Moreover, measuring body fat is neither commonly done nor easy to perform in clinical practice owing to the financial cost and poor availability of suitable equipment. Therefore, other mea- sures of obesity have come into use, such as total body weight, BMI, body surface area, or central adiposity assessed by waist circumference or waist-to-hip ratio. For more than 100 years, the simple calcu- lation used to measure BMI has enabled clini- cians and epidemiologists to study changes in body weight over time, and measure the impact of obesity on various diseases, parti- cularly CVD. 9 However, accumulating evi- dence indicates that BMI might not be the best measure of obesity to predict total and cardiovascular mortality. Instead, measures that account for body fat distribution, or directly assess body fat and lean mass, could improve prediction of cardiovascular risk. 8 In this article, we discuss various anthropo- metric measurements, as well as methods to determine body fat and lean mass, and their utility in assessing the risk of CVD and mortality. We also briefly discuss the techniques used to evaluate body composi- tion. Finally, we encourage the measurement of body fat and its distribution in everyday clinical practice for better risk stratification related to obesity. Obesity—a predictor of mortality Although a clear association exists between obesity and increased mortality, there is ongoing controversy regarding which meas- ure of obesity is the most appropriate to assess and predict cardiovascular risk. Various anthropometric measures of obesity and their relationship with total and cardiovascular mortality are discussed below. BMI The majority of studies assessing obesity and mortality have relied on BMI to define obesity. However, the relationship between BMI and total and cardiovascular mortal- ity follows a U-shaped or J-shaped curve, with the lowest mortality in overweight (BMI 25.0–29.9 kg/m 2 ) and mildly obese (BMI 30.0–34.9 kg/m 2 ) individuals. 10–14 Notably, this trend is evident not only in the general population, but also among indivi- duals at increased risk of mortality, such as patients with coronary artery disease (Figure 1). This unexpected phenomenon is known as the ‘obesity paradox’. 10 The largest epidemiological study demonstrating this paradox was published by the Prospective Studies Collaboration. 15 This analysis was based on 57 prospective studies that included 894,576 participants, mainly in Western Europe and North America, with a median follow-up of 8 years. The investi- gators found that mortality was lowest in the 22.5–25.0 kg/m 2 range in both sexes, confirming the U-shaped association between BMI and mortality. Above a BMI of 25.0 kg/m 2 , excess mortality was mainly associated with CVD, with a reduced median survival of 2–4 years among individuals with a BMI of 30.0–35.0 kg/m 2 , and a survival reduction of 8–10 years among individuals with a BMI of >35.0 kg/m 2 . Interestingly, a BMI below 22.5 kg/m 2 was also associated with increased mortality, mainly owing to chronic respiratory diseases. 15 Several theories have been suggested to explain the unexpected improvement in sur- vival observed in the high-normal and mildly overweight BMI ranges. The AHA released an advisory statement proposing that reverse causality, reclassification, overadjustment of confounders, and lack of statistical power could partially explain this controversy. 16 However, another possible explanation is that BMI is a poor anthropometric measure to diagnose obesity. To test this hypothesis, our group evaluated data from 13,601 individuals included in the Third National Health and Nutrition Examination Survey in whom body fat was estimated with the use of bioimpedance (a measure of resis- tance to an electric current passed through the body). 17 The diagnostic performance of BMI and its correlation with body fat OPINION Impact of obesity on total and cardiovascular mortality—fat or fiction? Beatriz Cepeda-Valery, Gregg S. Pressman, Vincent M. Figueredo and Abel Romero-Corral Abstract | Obesity is an excessive accumulation of fat that can impair health. Because the direct measurement of body fat is difficult to perform, a number of anthropometric measures have been employed as surrogates, of which BMI is the most commonly used. However, its usefulness has been questioned as a BMI in the overweight and mildly obese range is associated with improved survival and fewer cardiovascular events than a BMI in the normal range, a phenomenon known as the ‘obesity paradox’. Waist circumference, waist-to-hip ratio, and waist-to-height ratio take into consideration body-fat distribution, especially abdominal obesity, and seem to predict cardiovascular risk better than does BMI. Cepeda-Valery, B. et al. Nat. Rev. Cardiol. advance online publication 25 January 2011; doi:10.1038/nrcardio.2010.209 Competing interests The authors declare no competing interests. PERSPECTIVES © 2011 Macmillan Publishers Limited. All rights reserved