Dr. Reimherr Replies

To the Editor: The focus of our article was on the optimal length of therapy required to reduce the risk of relapse, which, as Dr. Travers points out, is not the same as the optimal length of therapy, more broadly defined. I strongly agree that determining the optimal length of treatment for depression involves balancing risks and benefits. Indeed, because the issue of safety and tolerability during long-term fluoxe­tine administration is so important, we have addressed these issues in great detail at several scientific meetings (unpublished presentations by Michelsen, 1997, and Beasley, 1998) and in publications (1–3).

Additionally, Eli Lilly and Company made the complete data set available to each of the five principal investigators (J.Z., J.D.A., F.M.Q., F.W.R., and J.F.R.) for further analysis. The data set could be used for whatever analysis they decided to pursue. This degree of openness is highly unusual in industry-sponsored studies.

This article focused on a comparison of the three points in time in which patients were randomly crossed over from drug treatment to placebo. This narrow comparison helped produce a significant and succinct set of study data. A very pertinent question that might have been posed by Dr. Travers was whether there were unique side effects or risks associated with each of these three crossover points. So far none has been identified.

Finally, to summarize our findings with respect to risk, the data strongly suggest that long-term therapy with fluoxetine is well tolerated and that the risks of early discontinuation of therapy (i.e., the consequences of depressive relapse) far outweigh the risks related to adverse events. All common adverse events specific to fluoxetine treatment during the initiation of therapy declined significantly over time, and previously uncommon categories of adverse events did not become common late in treatment (1).