The sensitivity, specificity, and positive and negative likelihood ratios were 57.7%, 72.3%, 2.08, and 0.59, respectively. (ROC curves were not plotted for case patients aged 2 years or younger because of the small numbers.)There was significant correlation of serum TARC levels with the SCORAD index (r ¼ 0.538; 95% confidence interval [CI], 0.36-0.68), QOL indices (Infants' Dermatitis QOL: r ¼ 0.823; 95% CI, 0.40-0.96; Children's Dermatology Life Quality Index: r ¼ .380; 95% CI, 0.16-0.57), peripheral eosinophils, and LDH in children older than 2 years but not in the younger age groups (Table II). The correlation with IgE was not significant. In the same age group, for SCORAD index, the correlation was highest with LDH (r ¼ 0.582; 95% CI, 0.40-0.72), followed by serum TARC (r ¼ 0.538; 95% CI, 0.36-0.68), peripheral eosinophils (r ¼ 0.397; 95% CI, 0.19-0.57), and serum IgE (r ¼ 0.331; 95% CI, 0.11-0.52). In concurrence with previous studies, our results suggest that TARC correlates with the severity of AD and QOL indices. 1,5 The sensitivity and specificity were lower than those previously reported (83%-85% and 92%-96%, respectively), possibly due to choosing control individuals with diseases mimicking AD and the use of different enzyme-linked immunosorbent assay systems. 2,4