Development of youth aerobic-capacity standards using receiver operating characteristic curves

Background: Cardiovascular fitness has important implications for current and future health in children.

Purpose: In this paper, criterion-referenced standards are developed for aerobic capacity (an indicator of cardiovascular fitness) based on receiver operating characteristic (ROC) curves.

Methods: The sample was drawn from participants aged 12-18 years in the National Health and Nutrition Examination Survey (1999-2002, N=1966). Subjects completed a treadmill exercise test from which maximal oxygen uptake (VO(2)max) was estimated from heart rate response. Metabolic syndrome was classified using previously published standards based on the National Cholesterol Education Program/Adult Treatment Panel III adult values at age 20 years. Using aerobic fitness z-scores as the test and metabolic syndrome as the criterion, ROC curve analysis was used to identify aerobic-capacity thresholds.

Results: The area under the curve (AUC) value for boys (83.1%) was high, indicating good utility for detecting risk of metabolic syndrome with aerobic fitness values. The AUC for girls (77.2%) was slightly below the recommended value of 80%. Although the ROC plots identified a defensible point for classifying levels of fitness, the approach in the present study was to establish two independent thresholds, one aimed at high specificity and one aimed at high sensitivity. The resulting z values for the low- and higher-risk threshold lines were then converted back to VO(2)max estimates using published LMS (L=skewness, M=median, and S=coefficient of variation) parameters. Values at the low-risk threshold ranged from 40 to 44 mL/kg/min for boys and from 38 to 40 mL/kg/min for girls.

Conclusions: In summary, aerobic fitness can be used with moderate accuracy to differentiate between adolescents with and without metabolic syndrome. Age- and gender-specific aerobic-capacity thresholds for creating separate risk groups were identified using nationally representative growth percentiles.