Dr. Lenze and Colleagues Reply

To the Editor: We are pleased that Dr. Kruszewski took interest in our study, but we respectfully disagree with his assertion that our study conclusions are misleading or inconsistent with our results. Our conclusion was that citalopram was efficacious and well-tolerated in elderly persons with anxiety disorders. We are happy to respond to some of his specific concerns. Dr. Kruszewski notes that the placebo and citalopram groups had unequal gender proportions, with more men in the placebo arm. We commented that this was a limitation, but we also mentioned on page 148 that we controlled for gender and found no change in the significantly higher rate of response to citalopram compared to placebo. Thus, the gender proportions did not appear to account for our efficacy finding.

Dr. Kruszewski also notes that the subjects in both groups also received lorazepam. However, there are two important reasons why it is unlikely that lorazepam co-administration could have accounted for our efficacy finding. First, the subjects were taking low doses (the median dose was 0.75 mg/day for the subjects in the citalopram arm). Second, the subjects were required to have been taking a fixed dose of this medication for at least 2 weeks before their random assignment, with no changes in their dosage during the study, and no subjects were administered benzodiazepines during the trial (they kept taking the medication if they were already taking it to avoid the added confounder of benzodiazepine withdrawal during the treatment study). Thus, the subjects still met entry criteria for significant anxiety symptoms despite taking a low dose of lorazepam.

Dr. Kruszewski states that the citalopram group experienced more adverse side effects and had a higher dropout rate than the placebo group. This is not really correct. In fact, a majority of the subjects in both the citalopram and placebo arms mentioned at least one side effect, and the difference in the proportions who reported any side effects was small and statistically insignificant (χ²=1.12, df=1, N=24, exact p=0.48; effect size: φ=0.18). The difference in dropout rates was also small and insignificant (χ²=1.13, df=1, N=24, exact p=0.29; effect size: φ=0.18). Figure 2 on page 148, with the Udvalg for Kliniske Undersøgelser Side Effect Rating Scale, shows that the subjects who were randomly assigned to citalopram tended to have lower side effect scores during treatment. As a whole, these data support the tolerability of citalopram in this population.

Finally, Dr. Kruszewski notes that the Hamilton Anxiety Rating Scale score and Hamilton Depression Rating Scale score were higher at baseline in the placebo groups, but he fails to state that these differences were not statistically reliable. Moreover, the sizes of the effects were small (for the Hamilton depression scale, effect size: d=0.37) to extremely small (the Hamilton anxiety scale, effect size: d=0.06). In practical terms, a difference of 1.7 points on the Hamilton anxiety scale and 1.1 points on the Hamilton depression scale are not clinically meaningful. Thus, despite Dr. Kruszewski’s assertion, we would have been remiss to conclude that the placebo group was more symptomatic than the treatment group before the initiation of treatment. He also notes that the baseline Hamilton depression scale and Hamilton anxiety scale scores were not the same as the week-0 scores in Figure 1 on page 148, which shows the course of symptoms over 8 weeks of treatment. This difference is because the subjects’ baseline assessment was not on the same day as their week-0 random assignment. The subjects were assessed at baseline to determine inclusion into the study. As is typical of medication studies, they were again assessed with the outcome measure at week 0 (the day of random assignment). The week-0 Hamilton anxiety scale scores shown in Figure 1 demonstrate that the groups were similar in anxiety severity at the point of random assignment.

Thus, we believe that our conclusions were neither misleading nor inconsistent with the results. Despite the study’s limitations owing to the small group size, this is, to our knowledge, the first prospective randomized, controlled study that demonstrates the efficacy of a serotonergic antidepressant medication for late-life anxiety disorders. We are currently confirming and extending the results in a larger clinical trial funded by the National Institute of Mental Health that focuses on late-life generalized anxiety disorder.