Performance of Clinician Versus Self-Report Measures to Identify Obsessive-Compulsive Disorder in Children and Adolescents

Abstract

The aim of this study was to study agreement between clinician-rated measures and self-report measures previously used in epidemiologic studies to identify obsessive-compulsive disorder (OCD) in youth and to determine the adequacy of self-report measures as screening instruments. Leyton Obsessional Inventory-Child Version (LOI-CV) survey form self-report and Children's Yale-Brown Obsessive Compulsive Scale (CY-BOCS) clinician-report measures were compared in a sample of 81 OCD-affected youths diagnosed using structured diagnostic interviews and the best-estimate method. Sensitivities and agreement between tests are determined for different cutoff scores. The LOI-CV survey form total (item+interference) score is correlated with KSADS-E OCD severity (r=0.37, p=0.001) but not clinician-rated GAF scores (r=-0.18, p=0.12). LOI-CV survey form sensitivities at cutoff scores of 15, 20, 25, and 30 are 55%, 36%, 28%, and 19%, respectively. CY-BOCS scores are correlated with subject-rated interview severity (p<0.001) and clinician-rated GAF scores (p<or=0.05). CY-BOCS sensitivities at cutoff scores of 8, 10, 12, and 14 are 100%, 99%, 98%, and 90%. The cutoff scores required for greater than 95% sensitivity of the LOI-CV survey form and the CY-BOCS, respectively, are 2 and 12. LOI-CV survey form total scores do not correlate with CY-BOCS total or obsession subscores (p>or=0.03), but correlate with the CY-BOCS compulsion subscore. The clinician-rated CY-BOCS measure using parent and child reporting performs superiorly to the subject-rated LOI-CV measure using child reporting to identify pediatric OCD in a clinically referred population. Because self-report measures have been used in epidemiological studies, youth OCD prevalence rates may be higher than previously reported.