To the Editor: We are grateful to Drs. Balon and Mattes for their questions—questions and concerns that may have occurred to other readers as well.

With regard to "the high dropout rate of 51%" cited by Dr. Balon, the main reasons for study subject dropout were lack of efficacy (25%), adverse events (6%), and noncompliance (4%). Not enumerated in the article were other reasons, including lost to follow-up (4%), personal (2%), and other (9%). This experience is similar to that of recent studies R451559BCECCEFH, R451559BCEBIIDI of experimental antipsychotics in an acutely ill population, in which dropout/withdrawal rates were 58% and 59%, respectively.

Dr. Balon questioned the differences in group size reflected in tables 2 and 3. Table 2 includes all patients randomly assigned, whereas table 3 includes the intent-to-treat group (i.e., those who had a baseline rating on the Positive and Negative Syndrome Scale and at least one on-treatment rating). In addition, the group sizes for the change in efficacy scales are given in table 3. As indicated in the footnote, group sizes were slightly lower for the Scale for the Assessment of Negative Symptoms (SANS) than for the other efficacy measures. Because the initial SANS evaluation did not occur until week 2, some patients who dropped out before then did not have an on-treatment SANS evaluation.

Dr. Mattes asked if more data were available on length of hospitalization prior to random assignment. Data were analyzed and included for 217 of the 497 patients; no additional data are available.

The recruitment of patients for this and similar trials is an enormous challenge. Dr. Mattes writes of his concern that the subjects participating in these trials are "chronic and relatively refractory." Patients who consent to experimental drug trials may have experienced side effects or less than optimal results with prior drug treatments. Dr. Mattes is correct to underscore the difficulty in generalizing the current results to patients with different characteristics (e.g., highly responsive to previous treatment) or patients at a different phase of their illness (e.g., first episode). While patients who are refractory to currently available medications represent a reasonable starting point, subsequent studies on other patient populations will be invaluable.

Dr. Mattes states that "the authors reported overall equivalence of haloperidol and sertindole, with superiority for sertindole on negative symptoms." We never claimed superiority for sertindole over haloperidol in treating negative symptoms. We found that sertindole was superior to placebo for treating negative symptoms and that haloperidol was not superior to placebo in the treatment of negative symptoms.

Dr. Mattes’ desire for alternative study designs is well-taken, but all such designs have advantages and disadvantages. Ultimately, we must accept the limitations of the information derived from any single clinical trial and must conduct multiple studies to achieve the most comprehensive treatment models.