To the Editor: Dr. Mattes has raised several issues involving both plasma concentration and dosage. As suggested in our paper, one possible reason we failed to detect a minimum therapeutic fluoxetine plus norfluoxetine plasma concentration is that the 20-mg fixed dose yielded a distribution of concentrations whose lower tail exceeded the minimum therapeutic plasma concentration. Fluoxetine dosing studies argue against the possibility that this potential problem actually affected our results.

A body of data suggests that 20 mg/day of fluoxetine does not substantially exceed a minimum therapeutic dose and that some patients require higher doses. Although some data suggest that 5 mg/day can be efficacious R5815511CHDCEHJH, two studies support the efficacy of 20 mg/day R5815511CHDDIBIG, and a fixed dose of 20 mg/day has been associated with a numerically greater response rate than a fixed dose of 5 mg/day R5815511CHDCEHJH. Further, a fixed dose of 40 mg/day has yielded a numerically higher response rate than a fixed dose of 20 mg/day R5815511CHDDIBIG. Recent findings indicate that in patients unresponsive to an initial trial of 20 mg/day (when that initial trial is of sufficient length to distinguish between time and dose effects), a dose increase to 60 mg/day results in a superior therapeutic response R5815511CHDCCGHD. Therefore, a dose of 20 mg/day should result in a distribution of plasma concentrations that would include a minimum therapeutic plasma concentration, provided one exists and can be detected.

Dr. Mattes has suggested that our formal analytical methods should be applied to that subset of data where concentrations are lower than 100 (or 150) ng/ml. Visual inspection of figures 1 through 3 indicate that the distribution of efficacy parameters associated with plasma concentrations within the range of 0–150 ng/ml is virtually random. Furthermore, 102 of the 615 (16.6%) observations included in these data and our analyses are of concentrations lower than 150 ng/ml.

We disagree with Dr. Mattes that our "data would suggest that there is no reason to increase the dose of fluoxetine above 20 mg/day." Rather, these data suggest that nonresponse to 20 mg cannot be meaningfully assessed by examining plasma concentrations. With respect to early discontinuation, data from patients who discontinued for any reason earlier than 8 weeks could potentially confound time and concentration effects; these data were, therefore, not included.

We agree with Dr. Mattes that from a logical perspective, a minimum efficacious dose and plasma concentration must exist on an individual patient basis. The results of our analyses, however, imply that these individual values vary considerably from one patient to another. From a clinical perspective, the issue is whether a common minimum therapeutic plasma concentration exists for a large group of patients and whether plasma concentration information can guide decisions about the adequacy of drug dose for individual patients. Our results indicate that this is not the case for fluoxetine.