655. Toward a brief, sensitive tool to assess depressive symptom outcomes in research and clinical practice: lessons from the RECOVER trial

Résumé fourni par la source

Abstract Background In RECOVER, a 12-month, double-blind, randomized, controlled trial (RCT) comparing active versus sham adjunctive vagus nerve stimulation (VNS) in persons with markedly treatment-resistant depression (mTRD), total scores on the Quick Inventory of Depressive Symptomology–Clinician (QIDS-C) and –Self-Report (QIDS-SR) were more sensitive than the Montgomery-Åsberg Depression Rating Scale (MADRS) in detecting overall change and in differentiating treatment groups. Aims & Objectives Assess whether QIDS-C, QIDS-SR, and MADRS were psychometrically inadequate, or if specific items were insensitive to change in mTRD. Method The sample (n=463) included all participants with a baseline and ≥1 postbaseline symptomatic outcome assessment. Participants had >13 prior failed antidepressant treatments (lifetime) and >20 years in the current depressive episode; 75% were unemployed. The MADRS and QIDS-C, conducted by phone interview, were completed by off-site, trained raters who were masked to protocol and treatment condition. Patients completed the QIDS-SR at clinic visits cotemporaneous with MADRS and QIDS-C ratings. The Clinical Global Index of Improvement (CGI-I) was completed at the same assessment occasions by each participant’s clinician (also masked to treatment assignment). All ratings were completed at baseline and monthly from months 3 to 12. Psychometric performance was assessed using classical test theory and item-response theory (IRT) methods. Each measure was evaluated for internal consistency, factor structure, test-retest reliability, item-total correlation (ITC), and the relationship of responses to each item to total score. Differences in sensitivity to change was assessed by comparing the effect sizes for each item on each scale in detecting overall change over time, discriminating between treatments, and detecting a CGI-defined partial symptom response. Results The psychometric properties of the three measures in this mTRD sample were acceptable and consistent with results from less chronically ill and treatment-resistant patients. All three measures were unifactorial with high test-retest reliability and strong internal consistency (Cronbach’s alpha values >0.8). Total MADRS scores were the least sensitive to overall therapeutic change and between-treatment-group difference in symptom improvement. ITC and IRT analyses for QIDS-C and QIDS-SR revealed that sad mood, energy, interest, negative self-view, and concentration items were most strongly related to the total score. For MADRS, these items were apparent and reported sadness, inability to feel (reflecting interest/pleasure), inner tension, lassitude (reflecting energy), and pessimistic thoughts (which include negative self-view), followed closely by concentration. The same five items most strongly reflecting overall depression severity were identical to those that were most sensitive to detecting overall change, between-group therapeutic differences, and partial response. Discussion & Conclusions Five “core” depressive symptom domains—mood, interest (MADRS “inability to feel”), energy (MADRS “lassitude”), negative self-view (MADRS “pessimism”), and concentration—best reflected overall symptom severity, best identified partial responders, and were most sensitive to detecting change over time and between-treatment-group differences. Sleep, appetite, psychomotor function, and suicidal ideation items contributed least to detecting overall benefit and between-group differences. A brief, publicly available, five-item “core” depressive symptom measure to assess treatment outcomes is feasible and applicable in research and practice settings. Determining the optimal approach—clinician-rated, self-reported, or a combination of the two—deserves study.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.