Paroxetine versus placebo: a double-blind comparison in depressed patients

Paroxetine is a potent and selective serotonin reuptake inhibitor (SSRI). The present study assessed the efficacy and tolerability of paroxetine against placebo in depressed outpatients.

A double-blind, parallel-group study was undertaken in four stand-alone centers. Patients aged 18-65 years, meeting DSM-III criteria for major depression, and having a Hamilton Rating Scale for Depression (HAM-D) score > or = 18 on the first 17 items of the HAM-D-21 were randomized to paroxetine or placebo for 6 weeks of treatment. Efficacy outcome variables included the HAM-D, the Montgomery-Asberg Depression Rating Scale, the Clinical Global Impressions Scale (CGI), and the Covi Anxiety Scale. Tolerability was assessed by asking a non-leading question. Routine laboratory safety and vital sign data from all four centers were pooled. The primary analysis used the intention-to-treat sample and for efficacy variables the last-observation-carried-forward data set was employed. Statistical methods included one-way analysis of variance for parametric and Fisher exact test for nonparametric variables.

Significant differences (p < or = .05) were found between paroxetine and placebo on the HAM-D and CGI by Week 2 and on all efficacy outcome variables by Week 4. Improvement on the HAM-D sleep factor occurred 2 weeks prior to that seen on the retardation factor. Similar results were obtained when an adequate treatment group (therapy for > or = 28 days) was considered. A full clinical response (CGI-severity of illness score 1 or 2) was seen in over 40% of subjects. Adverse events were more common for paroxetine compared with placebo (p < or = .01). Somnolence was twice more common than nervousness. Dropout due to adverse events was similar between therapies. Paroxetine had no clinically significant effect on laboratory safety data or vital signs.

Paroxetine was an effective, well tolerated, and safe antidepressant. Side effects were typical of the SSRI class of drugs. Symptoms indicative of a nonalerting profile were more common than those associated with alerting effects.