Predictive Utility of Four Methods of Incorporating Parent and Teacher Symptom Ratings of ADHD for Longitudinal Outcomes.

Despite robust evidence that parents and teachers provide incremental validity in the assessment of attention-deficit/hyperactivity disorder (ADHD), far less is known about the predictive utility of various strategies for incorporating these multi-informant data. Thus, we compared the 2-year predictive validity of four common assessment methods (i.e., algorithms) for ADHD symptoms--(a) parent only, (b) teacher only, (c) parent or teacher ("or rule"), and (d) parent and teacher ("and rule")--with respect to psychopathology and multidomain functional outcomes. At baseline, separate parent and teacher ratings of ADHD were obtained from an ethnically diverse (53% non-White) sample of 195 6- to 10-year-old children (30% female) to classify children according to the 4 algorithms. We then evaluated the predictive validity of each baseline ADHD algorithm with respect to its prediction of separate measures of internalizing and externalizing psychopathology, as well as normed ratings of academic and social impairment obtained at a 2-year follow-up. The "or rule" algorithm, based on symptoms being endorsed by either informant, optimally predicted psychopathology and functional outcomes relative to the other algorithms. These findings converge with previous evidence that incorporating data from multiple informants, and more sensitive approaches in particular, provide incremental validity in the assessment of ADHD.