Influence of baseline severity on the effects of SSRIs in depression: an item-based, patient-level post-hoc analysis

Background Reports claiming that antidepressants are effective only in patients with severe depression have affected treatment guidelines but these reports usually use a disputed measure of improvement, a decrease in the sum-score of the 17-item Hamilton Depression Rating Scale (HDRS-17), and are based on group-level rather than patient-level data. Method In this item-based, patient-level, post-hoc analysis, we pooled data from all completed, acute-phase, placebo-controlled, industry-sponsored, H DRS-based trials of the SSRIs citalopram, paroxetine, or sertraline in adult major depression. Patient-level data were pooled and subjected to item-based post-hoc analyses to assess the effect of baseline severity of depression on the response to treatment as assessed with HDRS-17 sum score, the depressed mood item of the HDRS, a six-item HDRS subscale (HDRS-6), and the remaining 11 HDRS items not included in this subscale (non-HDRS-6). Patients were defined as having non-severe depression if they had a baseline HDRS-17 sum score of 18 points or less and as having severe depression if they had a score of 27 points or more. Findings Our study population consisted of 8262 patients from 28 placebo-controlled SSRI trials. Participants were treated with either citalopram (n=744), paroxetine (n=2981), sertraline (n=1202), fluoxetine (active-control group; n=754), or placebo (n=2581). 654 patients were defined as having non-severe depression and 1377 as having severe depression. Patients with non-severe and severe depression did not differ with respect to SSRI-induced decrease in depressed mood and other HDRS symptoms belonging to the HDRS-6 subscale. However, after exclusion of patients with rare extreme baseline values, a positive association was seen between severity and efficacy when using HDRS-17 sum score as the effect parameter. This result was largely due to a more pronounced response to treatment with respect to non-HDRS-6 items in patients with severe depression than in those with non-severe depression. This outcome could be explained by non-HDRS-6 items, more so than HDRS-6 items, being more severe and prevalent at baseline in severe than in non-severe cases; hence, less room was left for improvement in these areas in patients with non-severe depression. Interpretation The use of an outcome measure that includes symptoms that rate low at baseline in patients with non-severe depression might result in the interpretation that SSRIs are ineffective in these patients. With respect to alleviation of HDRS-6 items, SSRIs appear to be as effective in patients with non-severe depression as in those with severe depression. Copyright (C) 2019 Elsevier Ltd. All rights reserved.