Predicting MDD with a Semistructured Depression Symptom Severity Interview in Type 2 Diabetes

Few studies in T2D have examined the accuracy of clinician-rated depression severity interviews, the gold-standard assessment. We assessed the factor structure and item-performance of the Montgomery-Asberg Depression Rating Scale (MADRS) in a sample of adults with T2D. Patients from primary and specialty care clinics of an urban academic medical center (N= 213, M (SD) A1c 8.2 (1.6), age 56 (9.5); 55% male) were screened for a trial of cognitive behavioral therapy for adherence and depression. Assessments included the MADRS and a structured interview (MINI) for diagnosis of MDD. Principal Component Analysis (PCA) with oblique rotation evaluated the number and character of MADRS components. Sensitivity, specificity, and area under the curve (AUC) of determined subscales and MADRS total score in predicting an MDD were calculated. Fifty-four percent (n=118) had current MDD and the mean MADRS score (M=21.2, SD=10.5) indicated moderate severity. PCA with extraction based on an eigenvalue > 1 yielded a single component, supporting use of the MADRS total score. However, in forced multi-component solutions, each of the somatic items (reduced appetite and sleep) loaded as distinct factors, and “concentration difficulties” failed to load highly onto any factor. AUC was 89.2% [95% CI: 84.9-93.5%] when excluding the three problematic items, and 88.9% [95% CI: 84.7-93.2%] for the total MADRS, demonstrating similar performance. Appetite and sleep yielded the smallest AUCs (.59 and .71, respectively). Sensitivity and specificity of the MADRS total score were 81.2% and 80.2%, respectively, with a cutoff score of 20.5. The MADRS is a strong instrument in predicting MDD and possibly provides a more accurate alternative to self-report measures. Results indicate that 7 of the 10 items could be administrated to T2D adults with depressive symptoms, shortening administration time. Disclosure A. Shapira: None. V. Zemon: None. S. Safren: None. J.S. Gonzalez: None.