RESEARCH PAPER Estimating total body fat using a skinfold prediction equation in Brazilian children Daniel J. Hoffman 1 , Tatiana Toro-Ramos 1 , Ana Lydia Sawaya 2 , Susan B. Roberts 3 & Patricia Rondo 4 1 Department of Nutritional Sciences, Rutgers, the State University of New Jersey, 26 Nichol Avenue, New Brunswick, NJ 08901, USA, 2 Department of Endocrine Physiology, Federal University of Sa ˜o Paulo School of Medicine, Rua Botucatu, 862, Sa ˜o Paulo, SP, Brazil 04023-900, 3 Energy Metabolism Laboratory, Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University, 711 Washington St., Boston, MA 02111, USA, and 4 Department of Nutrition, School of Public Health, University of Sa ˜o Paulo, Avenida Dr Arnaldo, 715, Sa ˜o Paulo, Brazil 01246-904 Background: The double burden of obesity and underweight is increasing in developing countries and simple methods for the assessment of fat mass in children are needed. Aim: To develop and validate a new anthropometric predication equation for assessment of fat mass in children. Subjects and methods: Body composition was assessed in 145 children aged 9.8 ^ 1.3 (SD) years from Sa ˜ o Paulo, Brazil using dual energy X-ray absorptiometry (DEXA) and skinfold measurements. The study sample was divided into development and validation sub-sets to develop a new prediction equation for FM (PE). Results: Using multiple linear regression analyses, the best equation for predicting FM (R 2 ¼ 0.77) included body weight, triceps skinfold, height, gender and age as independent variables. When cross-validated, the new PE was valid in this sample (R 2 ¼ 0.80), while previously published equations were not. Conclusion: The PE was more valid for Brazilian children that existing equations, but further studies are needed to assess the validity of this PE in other populations. Keywords: Anthropometry, body composition, skinfolds, stunting, developing countries INTRODUCTION The prevalence of overweight continues to increase in many developing countries, even as the prevalence of under- nutrition remains relatively high (Doak et al. 2000; Kapoor and Anand 2002; Caballero 2005). Access to body composition techniques available in wealthier countries, such as dual energy x-ray absorptiometry (DEXA) or bioelectrical impedance (BIA) (Eisenmann et al. 2004; Ramirez-Zea et al. 2006), is limited and assessment of those at risk of excess body fat generally relies on simpler and cheaper techniques, such as BMI or skinfolds. While BMI is a useful index of adiposity for populations, skinfold prediction equations have generally been developed and validated in cohorts that are not similar in terms of race or past nutritional experience to cohorts in developing countries (Goran et al. 1996; Huang et al. 2003; Hoffman et al. 2006). Thus, developing and validating new skinfold prediction equations is needed for more accurate estimates of body fatness in developing and transitional economies. Anthropometric measures are among the cheapest and most common methods available to assess human body composition, especially in developing countries (Ball et al. 2004). However, the use of skinfolds to estimate body fatness (BF) has not been without criticism (Slaughter et al. 1988; Dezenberg et al. 1999; Ellis 2001; Kapoor and Anand 2002; Walker et al. 2002; Yao et al. 2002). Much of the criticism for skinfold prediction equations is focused on the outcome variable estimated, the population in which the equations are derived and the validity of the equation in other cohorts or populations (Bernal-Orozco et al. 2010). First, fat mass (FM) is the univariate variable estimated with prediction equations and has been reported to minimize some of the complexity introduced into the prediction model process and increase the predictive power of body-composition models (Slaughter et al. 1988). Thus, equations that estimate %BF have limited predictive power and contribute to potential bias of the prediction equation. Second, the population in which the prediction equation is derived is important as equations derived from adult cohorts may over-estimate body fatness by 3 – 6% (Dezenberg et al. 1999) and few of the many skinfold prediction equations published have been found to be valid across populations Correspondence: Daniel J. Hoffman, Department of Nutritional Sciences, Rutgers, the State University of New Jersey, 26 Nichol Avenue, New Brunswick, NJ 08901, USA. E-mail: dhoffman@aesop.rutgers.edu (Received 12 August 2011; accepted 23 January 2012 ) Annals of Human Biology, March – April 2012; 39(2): 156–160 Copyright q Informa UK, Ltd. ISSN 0301-4460 print/ISSN 1464-5033 online DOI: 10.3109/03014460.2012.660989 156 Ann Hum Biol Downloaded from informahealthcare.com by Loughborough University on 04/18/12 For personal use only.