To the Editor: I read the article by Richard C. Shelton, M.D., et al. (1), which reported the superior efficacy of olanzapine plus fluoxetine for treating resistant nonpsychotic unipolar depression compared with either agent alone. The investigation had two main phases: a 6-week open-label trial of fluoxetine, followed by an 8-week double-blind trial in which nonresponders to fluoxetine alone were randomly assigned to receive olanzapine alone, fluoxetine alone, or fluoxetine plus olanzapine. Patients in the group receiving olanzapine alone stopped taking fluoxetine on the day of random assignment. After 1 week of treatment, fluoxetine plus olanzapine produced a marked improvement in depression symptoms, significantly much greater than that seen with olanzapine alone and with fluoxetine alone. The marked improvement persisted during the remainder of the 8-week double-blind trial and during the following 8-week open-label extension period. This interesting finding seems to me difficult to understand, because the group receiving olanzapine alone had stopped taking fluoxetine on the day of random assignment. Because fluoxetine and its active metabolite, norfluoxetine, take many weeks to disappear from the bloodstream after discontinuation (2), the group receiving olanzapine alone continued to have significant plasma levels of fluoxetine, at least during the first weeks of the double-blind trial. Therefore, the observed marked difference in response to olanzapine alone compared with the response to olanzapine plus fluoxetine during the double-blind trial seems difficult to understand because the treatments were similar.

Furthermore, the small groups (N=10, 8, and 10) examined may have led to a significant difference among the groups that was caused by chance alone (3). Replication in much larger groups by using a selective serotonin reuptake inhibitor with a short half-life could produce clearer results.