To the Editor: The article by Jay D. Amsterdam, M.D., and his colleagues R5715511CHDBJDEH on fluoxetine plasma levels was impressive and interesting. The authors, however, state that "it is possible . . . that a lower dose would have shown a threshold effect" (p. 967). I think it would be more accurate to say, on the basis of elementary pharmacology, that a lower dose would have to show a threshold effect if the medication is, in fact, effective; an ineffective dose might be 1 mg or 1 molecule, but there has to be a dose that is too small to be effective. This is not an empirical issue, and one cannot question the fact that an ineffective dose would have been found had the authors used lower doses.

In fact, the scattergrams the authors present suggest that they may have found evidence of an ineffective dose, but this was obscured by the fact that the vast majority of patients were above the threshold needed for effectiveness. The authors might evaluate the relationship between plasma levels and benefit by looking only at patients with plasma levels below 100 (or 150) ng/ml.

Without any evidence of a threshold effect, the authors’ data would suggest that there is no reason to increase the dose of fluoxetine above 20 mg/day. As the authors indicate, however, there is some evidence that higher doses are helpful for some patients; I think it’s fair to say that even if only five patients out of 100 would benefit from an increased dose of fluoxetine, this would be of some clinical (although perhaps not statistical) significance. Of note, the authors did not specify whether any patients dropped out of the study before completing the 8 weeks on a regimen of 20 mg/day because of an inadequate response.