In Reply: We appreciate Dr. Robbins' input. We agree with all his comments. Our statement that clinicians may not distinguish between bipolar and unipolar depression was not intended to reflect badly on clinicians' diagnostic skills. Rather, it derives from our observation that in a great many published reports, outcome data from the two groups are often combined.
Our comments about minimizing the use of antidepressants for patients with bipolar illness or a family history of bipolar illness are important. Clinical research is conducted in populations taking a particular medication in order to determine how patients will respond to it. Because it is impossible to predict the response of an individual patient on the basis of such research, one is forced to make generalizations that serve to guide treatment decisions. This is the essence of data-based management. There are always individual exceptions that must be addressed individually. But generalizations are important because they apply to most members of a population.
Antidepressants are useful agents in the management of bipolar depression, as we stated in our review. However, their use may be problematic. Because there is no evidence that antidepressants can prevent patients' relapse to bipolar depression and there is abundant evidence that long-term use of antidepressants may be problematic in a significant fraction of individuals, we believe that minimizing the use of antidepressants for bipolar patients is desirable unless individual evidence suggests otherwise.