Dr. Rickels and Colleagues Reply

To the Editor: We thank Drs. Gupta, Gupta, and Avasthi for commenting on our data. Dr. Gupta et al. point out that the different attrition rates of the patients who were and were not taking benzodiazepines 3 months after the taper may have affected our success rates at 12 months. Indeed, this was true for the patients taking benzodiazepines but not for the patients who were free of benzodiazepines at 3 months. When we carried forward 3-month data to 12 months for the patients who were not available at the 12-month follow-up, our rate of successful tapers remained at 83% for the patients who were not taking benzodiazepines at 3 months, but it decreased from 41.2% to 25.0% for the patients who were taking benzodiazepines at 3 months. The use of a data set from the last observation carried forward thus leads to a lower rate of success at 12 months than was originally reported for the patients who were unsuccessful in their taper attempt, making our rates of successful benzodiazepine taper even more remarkable.

We apologize for not having used the phrase “random assignment” when describing our double-blind treatment methods. Indeed, the patients were assigned to the three treatments according to a random-assignment schedule, which resulted in three treatment groups of equal sizes. In terms of potential predictors of a successful taper, Dr. Gupta et al. wonder why we did not use the level of baseline symptoms as a potential predictor. We did not do so because depressive symptoms at baseline did not correlate significantly (p<0.10) with treatment outcome and, thus, as we stated in the article, did not fulfill our criteria for variables entered into the logistic regression analysis. We believe that the other comments made by Dr. Gutpa and colleagues, as they relate to the study group, the generalizability of the results, benzodiazepine dependence, long-term treatment for generalized anxiety disorder, resistance to benzodiazepine therapy, and the effect of treatment group on benzodiazepine status at 12 months, were adequately covered in the article.