Severity at baseline and onset of improvement in depression. Meta-analysis of imipramine and moclobemide versus placebo

Abstract

SummaryMeta-analysis of several double-blind, placebo-controlled studies, including 1277 patients, has been performed in order to compare the efficacy of moclobemide, a new reversible and selective MAO-A inhibitor (RIMA) and imipramine. The main interest of the analysis was to test the time course of improvement and the impact of the pre-defined outcome criteria (50% reduction of total HAMD score) on the response to treatment, if initial severity of depression and drop-outs due to inefficacy are taken into consideration. In order to analyze the interdependence of the severity of the disease and outcome, patient sample was subdivided into three HAMD (17 items) baseline score groups: low- (score ≤ 21), medium- (22-27) and high (≥ 28) scorers. We found that reliable assessment of the point in time at which a drug begins to show therapeutic effect in each individual is a critical factor in the determination of the time course of improvement. We defined therefore the onset of improvement as the time point of a significant decrease (20%) of HAMD baseline score without subsequent deterioration. The threshold for the distinction between a significant change and spontaneous fluctuations, or error variations due to instrument or observer, was the natural variability of the HAMD score during the first two observation days. The results of the analysis replicated our earlier findings and confirmed that in treatment responders the time course of improvement is identical under placebo and antidepressants. Neither the time-points of the onset of effect, nor the time-points at which a 50% decrease of HAMD was reached, differed between the groups. In particular, no treatment- Specific time lag showed up in the onset of action. The difference in efficacy between antidepressants and placebo was evident only in the total number of responders and non-responders. Moreover, onset of improvement within the first 10 days of treatment, which was observed in 40-50% of patients, was highly predictive of the final response to treatment. The rate of correctly predicted responders in our sample was 70% for all three treatment modalities. With respect to the severity of the disease, a slight shift towards earlier onset of improvement was found for more severe cases. This finding was true for placebo and drug responders hut there was no clear-cut other interdependence with the treatment outcome. Drop-out rales due to inefficacy were in this study similar under all treatments (20-24%) anil occurred mostly during the first two weeks of the trial (64-71%.).