Original Research Article Cross Validation of ROC Generated Thresholds for Field Assessed Aerobic Fitness related to Weight Status and Cardiovascular Disease Risk in Portuguese Young People MICHAEL J. DUNCAN, 1 * SUSANA VALE, 2 MARIA PAULA SANTOS, 2 JOS  E CARLOS RIBEIRO, 2 AND JORGE MOTA 2 1 Sport and Exercise Applied Research Group, Faculty of Health and Life Sciences, Coventry University, Coventry, United Kingdom 2 Research Centre in Physical Activity Health and Leisure, Faculty of Sports Sciences and Physical Education, University of Porto, 4200 450 Porto, Portugal Objectives: To examine the efficacy of aerobic fitness thresholds in predicting weight status and cardiovascular dis- ease risk (CVD) in young people. Methods: A cross-sectional school-based study was conducted on 414 Portuguese young people (235 girls and 179 boys) aged 10–16 years (Mean age 6 SD 5 13.6 6 1. 8 years). Height and mass were assessed to determine body mass index (BMI). The 20m multistage shuttle-fitness test (MSFT) was used as an estimate of aerobic fitness. Capillary blood sampling was used to determine: total cholesterol, triglycerides, high-, and low-density lipoprotein. These were com- bined with measures of systolic blood pressure as z-scores and summed to create a CVD risk score. Results: Analysis of covariance, controlling for sexual maturation, indicated a significant main effect for BMI as a result of fitness category (P 5 0.0001). When applied to CVD risk data, there was no difference between “fit” and “unfit” grioups (P 5 0.136). Subsequent receiver operating curve (ROC) analysis indicated signifciant diagnostic accuracy of 20mMSFT performance for boys and girls (both P 5 0.0001) with subsequent cut-offs of estimated VO 2 peak of 49.5 ml kg 21 min 21 for girls and 47.7 ml kg 21 min 21 for boys. When applied to BMI and CVD risk data, there was a significant main effect as a result of fitness category for BMI (P 5 0.0001) and CVD risk score (P 5 0.0001). Conclusions: Recently established cut-points proposed by Boddy et al. (Boddy et al. [2012]: PLoS One 7(9): e45755) show validity in distinguishing between weight status but not CVD risk in Portuguese young people. Alter- native ROC generated cut points significantly predicted BMI and CVD risk in this sample. Am. J. Hum. Biol. 00:000– 000, 2013. V C 2013 Wiley Periodicals, Inc. Increasing incidence of type 2 diabetes, metabolic abnormality and obesity in young people is a worldwide public health concern (Weiss and Caprio, 2005). Studies have consistently identified that children and adolescents with greater aerobic fitness present lesser risk for cardio- vascular disease (CVD) (Martins et al., 2010) and that aerobic fitness is a key independent determinant of health (Ortega et al., 2008). Of greater health concern is the com- bination of obesity and poor aerobic fitness which confer a significant health risk to youth (Eisenmann et al., 2007). This is particularly so given that the prevalence of obesity in both adults and young people in Europe has stabilised (Boddy et al., 2010; Rockholm et al., 2010) but that levels of cardiorespiratory fitness have declined independent of changes in body mass index (BMI) (Boddy et al., 2010). Studies have established thresholds for directly assessed aerobic fitness that are related to CVD risk in 9–10 year olds (Ruiz et al., 2007) and 12–18 year olds (Welk et al., 2011). More recently, authors have employed similar methodology to generate cut-points based on results from the 20 m multi-stage fitness test (Boddy et al., 2012). Such an approach seems sensible given that field based measures such as the 20 m multi-stage fitness test are often used in schools and large scale studies as they provide a pragmatic alternative to direct assessment of aerobic fitness at the population level. In their study, Boddy et al. proposed cut-offs of 46.6 ml kg 21 min 21 and 41.9 ml kg 21 min 21 for boys and girls, respectively which were related to healthy weight in a sample of over 16,000, 9–10 year olds and associated with clustered CVD risk in a separate sample of 300, 11–14 year olds. In Portugal, the 20 m multistage fitness test is widely used to assess cardiorespiratory fitness in the school setting (Aires et al., 2012; Moreira, Santos, Ruiz et al., 2011; Mota et al., 2002, 2006) and prior research studies have demonstrated that aerobic fitness, predicted from the 20 m multistage fitness test, is associated with a number of health related varia- bles in Portuguese young people. This has included meta- bolic risk (Moreira, Santos, Ruiz et al., 2011; Ribeiro et al., 2004), overweight and obesity (Mota et al., 2006) and physical activity patterns (Aires et al., 2012). Data identifying thresholds for those young people who may be at increased risk of poorer cardiovascular health based on the 20 m multistage fitness test are important and useful. This is particularly because they could be com- pleted within school Physical Education and thus may be more likely to result in wider reach and public health ben- efit. Despite this, Boddy et al. (2012) noted a need to apply their thresholds to a wider age range due to the narrow (9–10 years) range employed in their study. Moreover, as use of the 20 m multistage fitness test is also prevalent in many countries, including Portugal, it may be useful for researchers to identify the efficacy of the Boddy et al. (2012) cut-points, generated on a British sample, in other populations. Boddy et al. (2012) also identified a need to consider maturation in any subsequent analysis as ROC *Correspondence to: Michael Duncan; Department of Biomolecular and Sports Sciences, Coventry University, Priory Street, Coventry CV11 5FB, UK. E-mail: michael.duncan@coventry.ac.uk Received 16 June 2013; Revision received 17 July 2013; Accepted 2 August 2013 DOI: 10.1002/ajhb.22443 Published online 00 Month 2013 in Wiley Online Library (wileyonlinelibrary.com). V C 2013 Wiley Periodicals, Inc. AMERICAN JOURNAL OF HUMAN BIOLOGY 00:00–00 (2013)