Reformulating Attention-Deficit/Hyperactivity Disorder According to Signal Detection Theory

Abstract

The current diagnostic algorithm for attention-deficit/hyperactivity disorder (ADHD) is not highly predictive of impairment and yields low interrater agreement. The objective of this study was to enhance the diagnostic accuracy of ADHD by identifying the symptoms that are associated with impairment. Semistructured interviews and impairment rating scales were administered to parents and teachers of 218 children. Combinations of ADHD symptoms were examined according to a receiver operating characteristic (ROC) based procedure to create diagnostic algorithms that predict impairment. In comparison with the DSM-IV, the ROC-based algorithms were 2 to 3 times more efficient in discriminating impaired from nonimpaired children. Parent and teacher agreement was also 3 times higher. Limiting the diagnosis to symptoms that predict impairment can increase the validity of the ADHD diagnosis. Results also support the use of ROC analyses in developing better diagnostic algorithms.