To the Editor:
In the paper by Stewart and El-Mallakh (1), the authors have failed to prove their case that bipolar disorder (BD) may be overdiagnosed in subjects with substance abuse. There are five major concerns regarding the paper. First, relying on a single interview with a patient as the sole source of information does not rule out BD. Hospital records were obtained but not discussed in enough detail. To truly rule out BD it is necessary to obtain additional family history (2). Individuals with BD have poor insight, similar to those who suffer from schizophrenia (3). This lack of insight makes interviewing less than reliable. Complicating the diagnostic process even further is the higher incidence of mixed episodes in individuals suffering from BD with substance abuse/dependence (4).
Second, structured interviews are of questionable reliability in diagnosing bipolar II disorder (5–8).
Third, ‘overruling’ a previous psychiatrist’s diagnosis should not be done without caution and also should not be done without examination of all prior medical records. These records may include interviews with family, family history, previous psychiatric records, and records of treatment failures and successes. Most important, these records may contain direct long-term observations of signs and symptoms—the ‘platinum standard’ for diagnosis.
Fourth, the authors made the diagnosis of substance-induced mood disorder in 100% of patients not diagnosed with bipolar I or II disorder. Historically, it can be very difficult to tell when a mood state starts to destabilize in relationship to a substance relapse even in those with good insight. Clinically, in individuals with a firm diagnosis of BD, subtle grandiosity often drives substance relapse. Here again, an interview with a family member or a review of prior records could help clarify this issue.
Fifth, and most important, is the omission of bipolar disorder not otherwise specified (BD NOS), which invalidates the study. BD NOS, though ill defined, is a valid DSM-IV diagnosis. It seems the height of frivolity if any individuals were excluded from a BD diagnosis because a hypomanic episode lasted three and a half days instead of four. There is no evidence for the requirement of four days of hypomania for a diagnosis of bipolar II disorder. In fact, studies point to hypomanic episodes lasting on average from one to three days (9).
While the authors did an excellent synopsis of the literature indicating the underdiagnosis of BD, they failed to prove that BD may be overdiagnosed in substance abuse. In so doing they have highlighted a growing dilemma that psychiatrists face. If we are to practice evidenced-based medicine, the reliance on only DSM-IV criteria for the diagnosis of BD is becoming problematic. The DSM-IV represented state-of-the-art knowledge when it was published nearly 14 years ago, but the DSM-IV-TR specifically states that it reflects an ‘evolving knowledge base’ (10). The practice of evidence-based medicine demands that we broaden our view of BD beyond that of the DSM-IV. While some may fear ‘diagnostic chaos’, the data should push clinicians and researchers alike to examine all of the literature more closely.