Defining low aerobic fitness in children: a post-hoc analysis of fitness and clustering of cardiovascular risk factors

Background: Low aerobic fitness among children has been frequently defined as the lowest quartile for sex- and age-specific distribution. Although this cut-off is correlated to health outcomes, correlation analysis cannot describe the nature and extent of misclassifications, when the purpose is to identify individuals at risk. Therefore, the accuracy of the diagnostic test to distinguish between the presence and absence of a certain trait will depend on the cut-offs chosen and not only the degree of closeness between the predictor and the predicted variable.

Aim: To define the optimal cut-off of aerobic fitness and to evaluate its ability to predict clustering of risk-factors for cardiovascular disease in children and adolescents.

Methods: This study used cross-sectional data from the European Youth Heart Study (EYHS). 2827 children in the 3rd (8–11 years) and 9th (14–17 years) grade were randomly selected from Denmark, Estonia and Portugal. Aerobic fitness (VO2max) was assessed from a maximal cycle test, and expressed as maximum power output relative to bodyweight (watts/kg). Risk-factors included in the composite risk-factor score (mean of Z-scores) were systolic and diastolic blood pressure, triglyceride, total cholesterol, HDL-cholesterol, insulin resistance and sum of four skin-folds. Children with a risk score >1SD of the composite variable were defined as being at risk. Receiver operating characteristic (ROC) analysis was used to define optimal cut-off for sex-, age and country-specific distribution. Cut-offs referring to the 10th and 25th percentile for the sex-, age and country-specific distribution were also considered. The diagnostic accuracy, measured by the area under the ROC curve (AUC), of each cut-off was compared.

Results: Optimal cutoff was slightly above the 25th percentile for the sex-, age and country-specific distribution. Using optimal cutoff and cut-offs referring to the 10th and 25th percentile 75%, 83.5% and 75.5% children were correctly classified as being at risk, respectively. Optimal cut-off (AUC: 0.69; 95%CI 0.67-0.71) performed better than the 10th (AUC: 0.62; 95%CI 0.59-0.64; p <0.001) and 25th percentile (AUC: 0.66; 95%CI 0.64-0.68; p <0.001) as a diagnostic test.

Discussion: Previous studies from the EYHS have shown that fitness level is inversely associated with metabolic risk. This study builds on these earlier findings by examining the predictive ability of different cut-offs of fitness in detecting children at risk. In conclusion, fitness is easier and less costly to measure than most of blood parameters, and yet it is an accurate tool for screening children with clustering of cardiovascular risk-factors.