Using Clinical Data to Predict Accurate ADHD Diagnoses Among Urban Children

Diagnosing attention deficit hyperactivity disorder (ADHD) requires reports of child behavior from 2 settings-most commonly home and school. Obtaining this information from teachers, however, is often challenging. We sought to determine if clinical data, supplementary to parent symptom scales, could be useful in predicting DSM-compliant diagnoses. Parents and teachers reported ADHD symptoms for 156 children using Vanderbilt scales; care managers collected clinical data; a team of specialists determined whether children met diagnostic criteria for ADHD. The ability of a parent Vanderbilt alone to predict an ADHD diagnosis was 56% (95% confidence interval = 45%, 67%). By adding child age and grade retention history to the multivariable model, the probability rose to 78% (95% confidence interval = 59%, 93%). In the maximally predictive model-which included 5 covariates-the predictive validity rose to 84% (95% confidence interval = 52%, 99%). Supplementing parent symptom reports with clinical data may be a viable alternative in certain cases when teacher reports are unavailable.