Screening for depression in youth with epilepsy: Psychometric analysis of NDDI-E-Y and NDDI-E in a French population

Objectives: The objective of this study was to evaluate the Neurological Disorders Depression Inventory-Epilepsy (NDDI-E) for Youth (NDDI-E-Y) for screening for major depressive disorder (MDD) in French youth with epilepsy (YWE), in order to (1) validate this tool in a separate population; (2) determine whether the 12-item NDDI-E-Y affords advantages over the 6-item adult NDDI-E; (3) measure psychometric performance of each item. Methods: Youth with epilepsy aged 11-17 years completed a 15-item questionnaire to calculate total scores for NDDI-E-Y (12 items) and NDDI-E (6 items). Gold standard for MDD was Children's Depression Inventory (CDI). Receiver operator characteristic (ROC) analyses for total NDDI-E-Y and NDDI-E scores were compared. Psychometric properties of each item were analyzed for: floor/ceiling effect, item-internal consistency, and ROC curve. Results: Ninety-seven YWE were included; 21.6% had MDD (CDI > 15). Correlation was very high between total NDDI-E-Y and NDDI-E scores, and high between NDDI-E-Y and CDI. Cutoff point for the NDDI-E-Y maximizing both sensitivity and specificity was 23 (original study cutoff 32). The ROC analysis of the NDDI-E-Y showed an area under the curve (AUC) 0.967 (95% confidence intervals [CI] 0.909-0.992); (p <0.0001). Sensitivity, specificity, and positive (PPV) and negative predictive values (NPV) were 100% [83.9; 100], 82.9% [72.5; 90.6], 61.8 [43.6; 77.8], and 100%[94.3; 100], respectively. The NDDI-E-Y was not superior to NDDI-E according to pairwise comparison of ROC (p = 0.07). Psychometric analysis revealed marked differences between items. After eliminating items with poorer performance, a 6-item version of the NDDI-E-Y showed sensitivity, specificity, PPV, and NPV of 100% [85.5; 100], 85.5% [75.6; 92.5], 65.6[46.8; 81.4], and 100% [94.5; 100], respectively. This was significantly better than the adult NDDI-E (p = 0.03) though not NDDI-E-Y (p = 0.07). Significance: Significant difference in cutoff indicates that the NDDI-E-Y cannot yet be recommended for widespread screening of MDD in YWE. Discrepancies in psychometric performance between items suggest that further work is needed to examine both validation of the original 12-item NDDI-E-Y and comparison with a shorter version. (C) 2019 Elsevier Inc. All rights reserved.