In practice, ADHD is commonly assessed with parent-reports in the absence of diagnostic interviews or behavioral observations, yet little is known about how accurately rating scales can independently detect ADHD. We used receiver operating characteristic analysis to evaluate the CBCL 1.5-5 Attention-Deficit/Hyperactivity Problems scale's ability to correctly classify the presence/absence of ADHD within a sample of young children with disruptive behavior disorders (N = 44), offering a conservative test of the scale's ability to distinguish ADHD symptoms from neighboring problems (i.e., "high-end specificity"). Across cut scores, the scale accurately differentiated between children with and without co-occurring ADHD (AUC = 0.83, SE = 0.07). Applying a cut score in the range of 61-64 yielded the most favorable balance across diagnostic utility properties (i.e., sensitivity = 0.71, specificity = 0.91, positive predictive power = 0.88, negative predictive power = 0.78). Findings provide empirical support to bolster confidence regarding use of this scale to assess early child ADHD, even in the context of complex diagnostic profiles.
Keywords: ADHD; Attention-deficit/hyperactivity disorder; CBCL; High-end specificity; ROC.