Validity of noninvasive composite scores to assess cardiovascular risk in 10-year-old children

Abstract

Agreement between and classification accuracy of six different noninvasive composite scores and a cardiovascular disease (CVD) risk factor score were investigated in 911 (466 boys and 445 girls) 10-year-old Norwegian children. A CVD risk factor score (triglyceride, total cholesterol/HDL ratio, homeostasis model assessment of insulin resistance, systolic blood pressure (SBP), waist-to-height ratio (WHtR), and cardiorespiratory fitness) and six noninvasive risk scores (fitness+three different measurements of fatness (body mass index (BMI), WHtR, and skinfolds), with and without inclusion of SBP) were calculated (mean z-score by gender). Agreement was assessed using Bland-Altman plots. The ability of noninvasive scores to correctly classify children with clustered CVD risk was examined by receiver operating characteristic (ROC) analysis and Cohen's kappa coefficient (κ). For both sexes, the noninvasive scores without SBP showed excellent AUC values (AUC=0.93-0.94, 95% CI=0.88-0.98) and moderate kappa values (κ=0.49-0.64) and had limits of agreement of 0.0±0.78-0.89 (arbitrary unit). Inclusion of SBP increased AUC values (AUC=0.96-0.97, 95% CI=0.94-0.99), kappa values (κ=0.58-0.69), and reduced limits of agreement (0.0±0.68-0.76). Noninvasive scores that include fitness and fatness provide acceptable agreement and classification accuracy, allowing for widespread early identification of children that might be at risk for developing CVD later in life. SBP should be included in the noninvasive score to improve classification accuracy if possible.