Challenging sequential approach to treatment resistant depression: Cost-utility analysis based on the Sequenced Treatment Alternatives to Relieve Depression (STAR⁎D) trial

In major depression, when a first antidepressant does not cause remission of symptoms (60%-75%), there are several options for continuing treatment in the next step. This study is a cost-utility analysis (CUA) of different second-line approaches. In a simulated trial outpatients with MDD were treated with citalopram for 13 weeks (level 1), then based on two alternative algorithms implemented from the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study. Algorithm A: citalopram was continued until study endpoint (week 26). Algorithm B: patients who remitted during level 1 continued citalopram. Those who did not remit could opt for switching to another antidepressant (venlafaxine; sertraline) (b1) or adding bupropion to citalopram treatment (augmentation; b2). Algorithm B increased remission rate by 10.6% over Algorithm A (number needed to treat: 9.9; sensitivity range: 9.1-12.5). As a comparison, differences between active antidepressants and placebo are associated with NNT values of 6 to 8. In CUA Algorithm B was dominant with an ICER of $11,813 (sensitivity range=$1783 - $21,784), which is <1 GDP per capita cost-effectiveness threshold (USA=$47,193). Among Algorithm B options, switching (b1) dominated Algorithm A with a smaller number of responders than augmentation approach (b2) (NNT 11 vs. 7.7), whereas ICER values were similar (b1: $14,738; b2: $15,458). However we cannot exclude a bias in selecting second treatment. This cost-utility analysis shows (in line with current guidelines) a benefit in modifying antidepressant treatment if response to first-line agent does not occur within 3 months, but not a clear-cut evidence in terms of NNT.