SEVERITY OF DEPRESSION AND BENZODIAZEPINE COMEDICATION IN RELATIONSHIP TO EFFICACY OF ANTIDEPRESSANTS IN ACUTE TRIALS

The severity of depression at baseline, measured by the 17-item Hamilt on rating scale for depression (HAM-D), was analysed as a predictor of the response to treatment, with response being defined as a 50 per ce nt improvement on the HAM-D or a good/very good outcome on the clinica l global assessment over four weeks treatment. There is a good correla tion between the two outcome measures (r = 0.67), but the question of the optimal outcome measure in very mild depression remains unsolved. The meta-analysis was based on 2684 patients from a data pool of doubl e-blind trials carried out with moclobemide compared to placebo and/or standard antidepressants. The placebo response rate was high (40 per cent) in very mild depression (HAM-D < 16) and low (12 per cent) in mo derate and severe depression. The reverse was true for moclobemide and imipramine. They showed a relatively lower response rate in cases of very mild depression but an equally good response in mild, moderate an d severe depression. Co-medication with benzodiazepines increases the response to placebo treatment and decreases the power of a comparative trial considerably. Benzodiazepines were more frequently given to pat ients who had been pretreated with an antidepressant before entering a n experimental trial. When analysing the outcome of antidepressant dru g trials, it is recommended that investigators take into account pretr eatment, co-medication and baseline severity.