Dr Agell raises concerns regarding the conclusions proposed in our original article (Reference Martenyi, Brown and ZhangMartenyi et al, 2002a ) that the results of our study suggest that fluoxetine is effective and well-tolerated in the prevention of PTSD relapse for up to 6 months. Dr Agell's concerns that (a) we do not discuss the results of the placebo/placebo group; (b) we do not adequately address the study results regarding SSRI discontinuation-emergent adverse events; and (c) ‘the authors mention that the reason behind the failure to show significant differences in the improvement of symptoms between the two treatment groups is the result of inconsistent patient self-rating’. We will attempt to address each of these concerns.

First, the results presented in our original article pertain to the relapse-prevention phase of a larger study. Results of the acute treatment phase (including the acute results of the placebo/placebo group) may be found in Martenyi et al (Reference Martenyi, Brown and Zhang2002b ). The primary objective of the relapse-prevention phase of our study and the focus of our original article was to assess the efficacy and tolerability of fluoxetine in the prevention of PTSD relapse. It then follows that the relevant results should come from acute phase fluoxetine responders who were continued on fluoxetine in the relapse-prevention phase or switched to placebo. The efficacy results from the placebo/placebo group would not address our question regarding the efficacy of fluoxetine in the prevention of PTSD relapse and, therefore, the full relapse-prevention efficacy results from the placebo/placebo group were not provided. We did, however, provide a breakdown of the reasons for discontinuation in the study for all treatment groups (Marteyni et al, 2002a, Fig. 1). Of the 31 patients in the placebo/placebo group (note that the sample size is small because the original randomisation was 3:1 fluoxetine:placebo), the discontinuation profile was quite similar to that of the fluoxetine/placebo group. Discontinuation profiles for the fluoxetine/placebo group v. the placebo/placebo group, respectively, were: 66.1% v. 61.3% completed the protocol; 0% v. 0% discontinued because of adverse events; 16.1% v. 16.1% discontinued because of clinical relapse; 4.8% v. 12.9% were lost to follow-up; 3.2% v. 0% discontinued because of patient decision; 9.7% v. 6.5% discontinued because of non-compliance; and 0% v. 3.2% discontinued because of lack of efficacy. These discontinuation data suggest that patients with an initial placebo response face a similar risk of recurrence of symptoms to those who had achieved an adequate pharmacological response and were then switched to placebo.

Second, it is true that approximately twice as many patients discontinued from the fluoxetine/placebo group compared with the fluoxetine/fluoxetine group. It is important, however, to note the reasons for discontinuations (Reference Martenyi, Brown and ZhangMartenyi et al, 2002a , Table 2). The protocol specified that patients meeting pre-defined criteria for clinical relapse should be discontinued, which allowed the investigators to provide follow-up care at their discretion. Only one patient in the fluoxetine/fluoxetine group discontinued because of an adverse event compared with none in the fluoxetine/placebo group, and the primary difference between the two treatment groups with regard to reason for patient discontinuation was clinical relapse (5.8% v. 16.1% for the fluoxetine/fluoxetine and fluoxetine/placebo groups, respectively). Accounting for all reasons for discontinuation with the exception of clinical relapse, 8 patients (12%) v. 11 patients (18%) discontinued early for the fluoxetine/fluoxetine and fluoxetine/placebo groups, respectively (Reference Martenyi, Brown and ZhangMartenyi et al, 2002a , Table 2). It should also be noted that there were no statistically significant differences in the numbers of patients reporting any single adverse event. The adverse events most commonly reported by patients in the fluoxetine/fluoxetine group were insomnia (15%), anxiety (6%) and headache (6%); those most commonly reported by patients in the fluoxetine/placebo group were insomnia (10%), headache (5%) and pain (5%). These data provide further support that the long half-life of fluoxetine and its active metabolite, norfluoxetine, provide benefit with regard to the minimisation of the risk of discontinuation-emergent signs and symptoms.

Third, statistically significant differences were detected between treatment groups for the a priori defined primary analysis (time to relapse, P=0.027; Reference Martenyi, Brown and ZhangMartenyi et al, 2002a , Fig. 2). In addition, using repeated-measures analysis of variance (Reference Martenyi, Brown and ZhangMartenyi et al, 2002a , Fig. 3), we can see that those patients in the fluoxetine/fluoxetine group continued to improve over time, with a statistically significant difference between groups occurring from week 28 to the study end-point (week 36), based on our primary efficacy measure, and other significant differences were detected between groups in several other illness severity measures (Reference Martenyi, Brown and ZhangMartenyi et al, 2002a , Table 3). Other patient-rated secondary measures were used in this study and, as reported, failed to show a significant difference between groups (Reference Martenyi, Brown and ZhangMartenyi et al, 2002a , Table 3).

We believe that the results of this study are robust and support our conclusions, and we maintain our opinion that the study results suggest that ‘fluoxetine is effective and well-tolerated in the prevention of PTSD relapse for up to 6 months’.