To the Editor: We thank Drs. Allen and Chessick for their comments. Although there are similarities in the clinical presentation of bipolar depression and unipolar depression, neurobiological evidence from neuroimaging (1) and genetic studies (2) lends credence to the notion that unipolar and bipolar disorders are not the same. Thus, their data and ours are difficult to compare.

Nonetheless, they raise interesting points. They suggest an interaction between attempter status and predictors of suicidal ideation, with "anxiety, depression, and neuroticism" predicting suicidal ideation in nonattempters and "temperamental behavior, poor role functioning, and negative life events" predicting suicidal ideation among past attempters. This may well be the case, but is a different question from the one we examined in our article, namely, whether depressive symptoms are similar within subjects but across different episodes. The prediction of suicidal behavior is the focus of a recently published report in which we found that a history of suicide attempts, which was present in half of that group of depressed patients, predicts future suicidal behavior, as do pessimism and aggression/impulsivity (3).

The role of anxiety and depression in suicidal behavior appears to be complex. Some investigators found an association between anxiety and suicidal behavior in depressed individuals (4). We have previously reported that anxiety symptoms during a major depressive episode, and the presence of panic disorder that was comorbid with a major depressive episode in particular, appear to protect individuals against suicidal behavior (5). This is in agreement with the clinical presentation of anxious or panic disorder patients who often are fearful that their anxiety symptoms signal impending death. These patients hardly appear to embrace their own demise as a solution to their problems. However, anxiety disorders cover a wide spectrum of clinical presentations, and we have also reported that posttraumatic stress disorder (PTSD) increases the risk of suicidal behavior (6). Indeed, in a recent work describing a mostly independent sample (7), we replicated this finding but found that the association between PTSD and suicidal acts was related to the frequent comorbidity of cluster B personality disorders in depressed patients with PTSD. Thus, not all anxiety disorders are the same in terms of their relationship with suicidal acts.

For the current article, we examined our data in detail to determine whether anxiety and suicidal ideation were inversely related within or across episodes of depression in patients with major depressive disorder. We could find no evidence for this or for the opposite relationship, although, clearly, our measures of both anxiety and suicidal ideation were limited. We compared only the ratings on the Hamilton Depression Rating Scale, as was the goal of our article. Thus, an examination of anxiety and suicidal behavior, such as the ones we previously reported, may have yielded different results. Moreover, the relationship between suicidal behavior and anxiety may differ in unipolar and bipolar depression. This would not be totally surprising, given the fact that suicidal behavior is almost twice as common in bipolar disorder as in major depressive disorder (8–11), further underscoring the fact that these two disorders are different entities that share some commonalities.