Background: In difficult-to-treat depression (DTD) the outcome metrics historically used to evaluate treatment effectiveness may be suboptimal. Metrics based on remission status and on single end-point (SEP) assessment may be problematic given infrequent symptom remission, temporal instability, and poor durability of benefit in DTD.
Methods: Self-report and clinician assessment of depression symptom severity were regularly obtained over a 2-year period in a chronic and highly treatment-resistant registry sample (N = 406) receiving treatment as usual, with or without vagus nerve stimulation. Twenty alternative metrics for characterizing symptomatic improvement were evaluated, contrasting SEP metrics with integrative (INT) metrics that aggregated information over time. Metrics were compared in effect size and discriminating power when contrasting groups that did (N = 153) and did not (N = 253) achieve a threshold level of improvement in end-point quality-of-life (QoL) scores, and in their association with continuous QoL scores.
Results: Metrics based on remission status had smaller effect size and poorer discrimination of the binary QoL outcome and weaker associations with the continuous end-point QoL scores than metrics based on partial response or response. The metrics with the strongest performance characteristics were the SEP measure of percentage change in symptom severity and the INT metric quantifying the proportion of the observation period in partial response or better. Both metrics contributed independent variance when predicting end-point QoL scores.
Conclusions: Revision is needed in the metrics used to quantify symptomatic change in DTD with consideration of INT time-based measures as primary or secondary outcomes. Metrics based on remission status may not be useful.
Keywords: difficult-to-treat depression; effect size; efficacy; intervention research; metrics; outcome measures; quality of life; remission; treatment-resistant depression.