Assessing symptom severity in patients with markedly treatment-resistant depression: Insights from the RECOVER trial

The performances of the Montgomery-Åsberg Depression Rating Scale (MADRS) and the Quick Inventory of Depressive Symptomatology-Clinician (QIDS-C) and Self-Report (QIDS-SR) ratings have never been evaluated in markedly treatment-resistant depression (TRD). RECOVER, a 12-month randomized, controlled trial, compared adjunctive active to adjunctive sham vagus nerve stimulation, provided data to address this knowledge gap, and to examine the basis for the differential sensitivities of these scales in detecting overall and between-treatment-group symptom changes.

We employed classical test and item response theory methods to characterize each scale. Effect sizes identified items that were most sensitive to change over time across treatments, best discriminated between the two treatments, and best detected at least a partial symptom response in clinician global ratings.

All three inventories were unidimensional, with high retest reliability and strong internal consistency. MADRS total scores were less sensitive in detecting overall and between-treatment-group symptom differences. On each scale, five symptom domains-mood, energy/lassitude, interest/inability to feel, negative self-view/pessimism, and concentration-covaried most strongly with overall depression severity and had the largest effect sizes in detecting overall therapeutic change and between-group difference in efficacy.

These findings may not generalize to less treatment-resistant populations or to different treatments. The relative insensitivity of suicidal ideation, sleep, and appetite items may be due to study design or concomitant medications.

The MADRS, QIDS-C, and QIDS-SR are suitable for use in markedly TRD. Whether sadness, energy, interest, guilt, and concentration are "core" symptoms that are especially sensitive to change deserves further study.