Efficacy of antidepressant drugs in MDD: Meta-analysis of the effect of trial characteristics on placebo and antidepressant response

Background: The increasing rate of failure of antidepressant trials in major depressive disorder (MDD) has led to the assumption that antidepressants do not have meaningful clinical benefits. Varying results of antidepressant trials may be related to trial characteristics (such as: baseline depression severity, duration of study, probability of receiving placebo, the number of visits and year of publication) and not to the antidepressant drugs themselves. Little prior studies investigated trial methodology and contradicting results were demonstrated. Objectives: This report aims to determine the effects of trial characteristics on placebo and antidepressant response, and thereby trial outcome, in adult subjects with MDD by means of a (random effects model) meta-analysis. Search methods: Database Pubmed/MEDLINE and registers of the Food and Drug Administration (FDA) database were searched for published and unpublished studies (reported in English) concerning randomized controlled trials (RCTs) of antidepressant efficacy between January 1¬st 1980 and May 13th 2011. Data collection and analysis: Data was extracted by the investigator (J.C.K) and risk of bias was assessed with the “Cochrane Collaboration’s tool for assessing risk of bias”. Main results: 3256 published and 36 unpublished studies of randomized controlled antidepressants trials in MDD were identified. 51 trials regarding SSRIs and SNRIs (serotonergic/noradrenergic reuptake inhibitors) were included in the analysis of which 6429 patients were randomized to placebo and 5989 to antidepressants. Meta-analyses demonstrated large effect sizes for both antidepressant (d=1.33) and placebo (d=0.95) treatment. Although antidepressants showed greater effect size than placebo, it remained below recommended criteria for clinical significance. Antidepressants demonstrated to be effective in severely depressed patients (ddiff=0.55) due to a decrease in placebo response. Older trials also show significant antidepressant efficacy (ddiff=0.50) because of an associated decrease in baseline depression for both placebo (coefficient -0.43; p=0.00) and antidepressant treatment (coefficient -0.34; p=0.01) over time. Since study duration was negatively related to antidepressant response (coefficient -0.09; p=0.03), short trials (4-6 weeks) demonstrated antidepressant efficacy. Author’s conclusions: The present study suggests that the various, and often poor, results of antidepressant trials may be related to the design of the antidepressant trials or patient characteristics and not to antidepressant medication itself. With this in mind, current antidepressant trial designs could be improved so that the possibility of a positive trial outcome increases.