Dissociative Identity Disorder in Psychiatric Inpatients

We conducted this study between July 1996 and February 1997 at the Hillside Hospital Division of Long Island Jewish Medical Center in Queens, N.Y. Hillside Hospital, a teaching hospital, serves mainly Queens and Long Island. Almost all inpatients come from the local area. We randomly selected female subjects between the ages of 16 and 50 years, choosing this group because dissociative identity disorder occurs most frequently in this age range. We excluded patients receiving ECT because the treatment might interfere with subjects' ability to provide accurate information. Third- and fourth-year psychiatric residents conducted the initial interviews with the SCID-D (10) between the fifth and 10th days of hospitalization. If a subject refused to cooperate or appeared too symptomatic to reasonably provide information, the rater returned a few days later to determine if the interview seemed feasible. Raters discussed the subject with ward staff to obtain the description, if any, of different personalities.

Over 50% of all initial interviews were done by one rater (D.G.). If the initial interview with the SCID-D indicated definite or probable dissociative identity disorder, a senior clinician (A.R.) interviewed the subject again with the SCID-D.

The SCID-D contains questions about the symptoms required by DSM-IV for dissociative identity disorder. We consider this a more direct means of ascertaining the diagnosis than using rating scales or an interview schedule that is not clearly pegged to the DSM-IV criteria and leaves no room for inquiry. The SCID-D not only provides questions for all relevant symptoms for dissociative identity disorder and other dissociative disorders, but, as a semistructured interview schedule, it also requires the interviewer to expand on the suggested wording of a question, as a clinician would do during a standard diagnostic interview. The SCID-D assesses not only the presence of symptoms, but also the degree of distress and cause of functional impairment, which are requirements necessary for assigning the diagnosis. The SCID-D is a guide to a clinical interview, not a rating scale. We believe the accuracy of our diagnoses comes from 1) the assurance that in following the SCID-D we have not omitted any relevant questions and have considered all criteria necessary for the diagnosis; and 2) our clinical skills in conducting the interview. We hope this provides the closest to a “gold standard” for diagnosis: two interviews, one by an experienced clinician, using a semistructured interview guide. Previous studies of the prevalence of dissociative identity disorder in inpatients did not use this procedure.

We made no attempt to establish a diagnosis other than a dissociative disorder. The interviewers read the subject's chart before the interview and knew the working diagnosis and clinical information about the subject. The interviewers carefully looked for symptoms of psychosis, mood disorders, anxiety disorder, or cognitive impairment that might resemble symptoms of dissociation. After complete description of the study to the subjects, written informed consent was obtained.

We interviewed 100 subjects (63%) of 158 approached. We could not interview 58 subjects for the following reasons: 30 patients were discharged before the fifth day, nine were too psychotic to be interviewed, nine refused, nine could not be interviewed for administrative reasons, and one patient was transferred to a medical ward. We found one subject (1%) with dissociative identity disorder. Both interviewers concurred in the diagnosis.

By having two clinicians use a semistructured interview guide, rather than a self-rating screening instrument or a fixed-question scale, we think our methodology most closely approaches the gold standard of diagnosis. Our finding of a 1% occurrence rate is lower than what previous research found. We attribute this to our improved methodology, but further research should answer the question more definitively.

The major limitation of this study is that we could not interview 37% of the subjects. Even if a substantial proportion of these subjects had dissociative identity disorder, we would still not approach the figures reported elsewhere, such as 12% by Latz et al. (8). We found that dissociative identity disorder rarely occurred among a randomly selected sample of female psychiatric inpatients. This conclusion should be softened by the fact that other investigators have found a higher rate. Clearly, more studies are needed.

Received June 12, 1997; revision received Oct. 27, 1997; accepted Dec. 22, 1997. From Hillside Hospital Division of Long Island Jewish Medical Center. Address reprint requests to Dr. Rifkin, Hillside Hospital, Glen Oaks, NY 11004.