Drs. Leichsenring, Salzer, and Leibing Reply

To the Editor: HAM-A is still one of the most frequently used measures to assess anxiety, including symptoms of generalized anxiety disorder in randomized, controlled trials. This is true for studies assessing the effects of psychotherapy and those assessing the effects of psychotropic drugs. For this reason, we chose HAM-A scores as the primary outcome measure of our study. This is consistent with previous research on generalized anxiety disorder (e.g., reference 1). However, we agree with Dr. Koerner et al. that, for several reasons, HAM-A is not an optimal measure of anxiety in general and of generalized anxiety disorder in particular. This is especially true if HAM-A is applied as the only measure of anxiety.

Dr. Koerner et al. regard the Penn State Worry Questionnaire as a more appropriate measure of generalized anxiety disorder, since it utilizes the DSM-IV criterion for worry. In addition to HAM-A as the primary outcome measure, we included the Penn State Worry Questionnaire as a secondary outcome measure, a procedure that is also consistent with previous research on generalized anxiety disorder (e.g., reference 1). In our article, we reported the results for the outcome measures used, including for the Penn State Worry Questionnaire. It is true that the latter results were in favor of CBT, which we noted as well. In addition, we did use a combination of self-rated and observer-rated outcome measures, as suggested by Dr. Koerner et al. The results of HAM-A were supported, for example, by that of the Beck Anxiety Inventory. Significant differences in efficacy between short-term psychodynamic psychotherapy and CBT were not found in either the Beck Anxiety Inventory or HAM-A. In addition, we used the State-Trait Anxiety Inventory-Trait Version as another measure of anxiety. Again, we found and reported an outcome in favor of CBT. Thus, we used several measures of anxiety that appear to draw on different aspects of anxiety. As reported in our article, the Penn State Worry Questionnaire did not show significant correlations to HAM-A (r=0.16, p=0.23) or to the Beck Anxiety Inventory (r=0.16, p=0.23) in the total sample of patients with generalized anxiety disorder (N=57). In contrast, the questionnaire correlated significantly with the trait anxiety inventory (r=0.66, p<0.0001). As we noted in the article, several items of the trait anxiety inventory were related to worry. These correlations suggest that the questionnaire and, in part, the trait anxiety inventory utilize other, more cognitive aspects of anxiety than HAM-A and the Beck Anxiety Inventory. The items of HAM-A and the Beck Anxiety Inventory appear to utilize more somatic aspects of anxiety. The correlation between these two measures was 0.58 (p<0.001). These somatic symptoms are another main criterion of generalized anxiety disorder according to DSM-IV.

As already stated, the other main DSM-IV criterion of generalized anxiety disorder is extensive and uncontrollable worry. However, the specificity of pathological worry in generalized anxiety disorder has been questioned by several investigators (2, 3). The nosological controversies associated with the criterion of worry were discussed by Weisberg (4). Furthermore, worry may also be associated with other anxiety disorders and especially with depression (2). In another study conducted by our working group (5), the sensitivity and specificity of the Penn State Worry Questionnaire were assessed. Depending on the cut-off score applied, specificity was between 0.51 and 0.68. These results did not indicate a high specificity of worry or a high specificity of the Penn State Worry Questionnaire for generalized anxiety disorder.

Furthermore, Dr. Koerner et al. note that HAM-A does not adequately differentiate between generalized anxiety disorder and depression. This seems to also be the case for worry and generalized anxiety disorder on the worry questionnaire. Fresco et al. (3) reported that worry and rumination were highly correlated with each other (r=0.46) and showed equally strong relationships to both anxiety and depression. In our previous study, we found a significant and high correlation between the worry questionnaire and the Beck Depression Inventory (r=0.51) (5). After controlling for depression (using the Beck inventory), the worry questionnaire no longer differentiated between patients with generalized anxiety disorder and a group of patients suffering from depressive and/or other anxiety disorders (5). As suggested by Dr. Koerner et al., HAM-A indeed showed a significant correlation (r=0.46, p<0.001) with depression (Beck Depression Inventory) in the total sample of patients with generalized anxiety disorder in our study (N=57) (1). However, this was true for the worry questionnaire and trait anxiety inventory as well, with correlations that may be even higher (r=0.55, p<0.001 and r=0.76, p<0.001, respectively).

Thus, the superiority of CBT to short-term psychodynamic psychotherapy regarding worry (worry questionnaire and, perhaps, trait anxiety inventory), which we reported, may reflect the affinity of anxiety and depression in terms of worrying and rumination. These cognitive aspects of worrying and rumination are primarily addressed by CBT but not by short-term psychodynamic psychotherapy. This is consistent with our finding that CBT was also superior to short-term psychodynamic psychotherapy with regard to the reduction of depression as measured by the Beck inventory. In our article, we suggested that the outcome of short-term psychodynamic psychotherapy in generalized anxiety disorder may be further optimized by employing a stronger focus on the process of worrying.

More in general, it may be critically discussed whether the definition of generalized anxiety disorder as an anxiety disorder should be primarily based on cognitive aspects neglecting the emotional and somatic aspects of anxiety. Stein (6) suggested viewing generalized anxiety disorder as a set of psychobiological dysfunctions that manifest as a matrix of anxious-somatic or anxious-somatic-depressive symptoms.

Dr. Koerner et al. regard HAM-A as problematic—that it is an observer-rated instrument and may therefore be related to allegiance effects and other related factors that can influence results. However, in our study, HAM-A was applied by trained raters blind to the treatment conditions. For this reason, allegiance effects and other related factors were controlled for in the analyses.

We are pleased to see the use of HAM-A being critically discussed. However, we would be content to see this discussion not only applied to studies of psychotherapy but to studies of pharmacotherapy as well.