• bipolar disorder;
  • comorbid illness;
  • substance abuse

To the Editor:

Attempts to improve the diagnosis of bipolar illness are needed. This illness is clearly difficult to diagnose, and many patients may suffer from improper and inadequate treatment for excessively long periods of time. We examined a population of patients that was highly and chronically symptomatic, and whose symptoms were compatible with several psychiatric conditions. We utilized the Structured Clinical Interview for Diagnosis (SCID) DSM-IV criteria, which are the current research gold standard for diagnosis. We found the research diagnosis discrepant with the clinical diagnosis nearly half of the time. There are limitations with utilization of the SCID, as pointed out by Dr. Kelly (1), but there are also daunting limitations with clinical diagnosis. There are many pressures facing clinicians that can influence their decision making. For example, Medicaid does not reimburse for substance-related diagnoses, but does reimburse for a bipolar diagnosis, thus creating a subtle force for clinicians to minimize their patients’ drug-related symptomatology. Records were reviewed in approximately one-third of the patients, but in no case did the hospital records alter the SCID diagnosis. The patients were the source of information for our diagnostic evaluations. These patients had already been diagnosed with bipolar illness and presented specifically for ongoing treatment for this condition. Their level of knowledge and experience reduces concerns about the role of insight as a cause for underreporting manic or hypomanic symptoms. To the contrary, our experience was that they were eager to list their manic and hypomanic symptoms. It is important to note that our findings have been replicated by Joseph Goldberg and his colleagues utilizing a clinical diagnostic approach and a very similar design (2). The SCID is not structured to diagnose bipolar illness not otherwise specified. As Dr. Kelly points out, the diagnostic criteria for this disorder are ill defined, making their reliability (and utility) less than optimal.

Dr. Kelly’s letter highlights the dilemma of clinicians who are treating patients that do not fit into the current diagnostic framework. This creates a certain dogma that dictates that research findings are inadequate because they do not replicate the clinical experience. However, reducing the reliability of research studies to mimic the clinical reality will not advance our field. The research response to this dilemma is to adhere accurately to research criteria, to utilize accepted and validated research tools, and to properly characterize patients into the most homogenous categories possible. Creation of new, reliable diagnostic categories may be required. But utilizing state-of-art tools, bipolar illness appears to be over-diagnosed in patients with substance dependence.


  1. Top of page
  2. References