Background: Mood disorders are common diagnoses in youth (Merikangas et al, 2010). Mood disorder diagnoses tend to show poor agreement across interviewers (Regier et al., 2013). Questionnaires are recommended to improve the reliability of diagnosis. Screening can result in false positives. Anxiety disorders are likely to increase false positives for mood due to shared negative affect. Therefore, the purpose of the current study is to determine (a) clinical cut-offs and (b) what diagnoses cause false positives.
Method: Youth and their caregivers (N=620; ages 4-18) completed the K-SADS. Mood disorders (depressive and bipolar disorders) were common (n=225, 36%). Caregivers and adolescents completed the Achenbach CBCL and YSR.
Results: Receiver Operating Characteristic (ROC) analyses indicated that caregiver-report discriminated between children with and without mood disorder, AUC=.71, p < .001. Low scores (T-score < 61) decreased risk of mood disorder (DLR=.14). T-score>=71 increased risk slightly (DLR=1.08). Children with ODD (OR=1.97), adjustment disorder (OR=2.80) and adjustment disorders with mood (OR=4.06) were more likely to be false positives. ADHD, CD, anxiety, history of mood disorder and suicide were not associated with false positives.
ROC indicated that caregiver report and self-report discriminated between adolescents with and without mood disorders, CBCL AUC=.75, YSR AUC=.67, p < .001. Caregiver T-scores >=70 and YSR T-scores>=67 doubled the risk for mood disorder. Both adolescent and caregiver-report false positives increased due to adjustment disorder (OR=4.62, 5.34) or adjustment disorder with mood (OR=6.20, 7.02). Diagnoses of ADHD, ODD, CD, anxiety disorders, history of mood disorder, and suicide were not associated with false positives.
ODD was significantly more common in children false positives than adolescents. Among adolescents, all caregiver false positives were self-report false positives. Caregivers of false positives tended to have high scores across many CBCL scales (Child t(65.34)=11.99, p < .001; Adolescent t(43.08)=7.10, p < .001), but not adolescents t(79.87)=1.58, p=.12.
Discussion: Both caregiver and adolescent self-report Internalizing scores had diagnostic utility in distinguishing youth with and without mood disorders. Like all screening measures, false positives occurred. Child false positives were more likely to have ODD and both child and adolescent false positives were more likely to have adjustment disorders. When interpreting the CBCL Internalizing subscale, clinicians should consider these alternative diagnoses and be warier about contending diagnoses when caregivers endorse multiple scales at clinically elevated levels.
Yen-Ling Chen– Graduate Research Assistant, University of Nevada, Las Vegas, Las Vegas, Nevada
Eric Youngstrom– Professor and ABCT Fellow, University of North Carolina, Chapel Hill, North Carolina
Robert Findling– Johns Hopkins University/Kennedy Krieger Institute
Andrew Freeman– Assistant Professor of Psychology, University of Nevada, Las Vegas
Graduate Research Assistant
University of Nevada, Las Vegas
Las Vegas, Nevada
Professor and ABCT Fellow
University of North Carolina
Chapel Hill, North Carolina