Abstracts

Rationale: The Child Behavior Checklist (CBCL) is a parent-completed questionnaire that is widely used to screen for affective and behavioral problems in children with epilepsy. While the CBCL DSM-Oriented subscales are used to make inferences regarding behavioral complications and DSM-IV diagnoses, its comparability to psychiatric diagnoses generated by standardized research procedures has rarely been investigated. This study investigated the diagnostic performance and discriminant validity of the selected DSM-Oriented subscales and provided preliminary suggestion for optimal cutoff scores for their increased clinical utility. Methods: 69 participants with epilepsy ranging from 8 to18 years old (mean age of 14.3 years; mean epilepsy age of onset of 11.3 years) and their parents were independently assessed using the Schedule for Affective Disorders and Schizophrenia (K-SADS) interview and parents completed the CBCL. The empirical and normative based cutoff T scores (55, 60, 63 and 65) of the DSM-Oriented subscales (i.e., Anxiety, Affective and ADHD Problems) were compared to the K-SADS current comparable diagnoses (i.e., anxiety, depression, and ADHD). Additionally, the Total Problems subscale of the CBCL was compared to the presence of any Axis I disorder based on the K-SADS. The diagnostic utility of these scales were examined in regard to sensitivity, specificity, positive and negative predictive power (PPP and NPP, respectively), overall correct classification (OCC) and Kappa value (see Table 1).Results: Overall, the diagnostic efficiency statistics demonstrated that the DSM-Oriented subscales exhibit lower sensitivity (correct identification of diagnosis) with higher specificity (correct identification of absence of diagnosis) compared to the K-SADS (see Table 2). On the other hand, they exhibited higher PPP (correctly predicting the presence of diagnosis) than NPP (ability to predict the absence of symptoms). To derive clinically optimal cutoff scores of the subscales, the OCC and Kappa values (level of agreement between two instruments) were evaluated. To detect anxiety symptoms, the cutoff score of 65 for the Anxiety Problems subscale had the highest OCC at 69% (k = 0.04). To detect depressive symptoms, the Affective Problems subscale had the highest OCC at 81% (k = 0.28) using a cutoff score of 65. The ADHD Problems subscale had the highest OCC at 84% (k = 0.49) using a cutoff score of 65. Finally, when examining the presence of any current DSM-IV Axis I psychiatric disorder, a cutoff score of 55 for the Total Problems subscale had the highest OCC at 64% (k = 0.28). Conclusions: The OCC values at the optimal cutoff scores for the DSM-Oriented subscales were within an acceptable range (64-84%). However, the Kappa values indicate a poor agreement between the CBCL and K-SADS. The DSM-Oriented subscales do not provide a shortcut to K-SADS diagnoses of anxiety, depression and ADHD in children with epilepsy. Further research is warranted to estimate optimal cutoff scores in a larger sample of children with epilepsy.