Dr. Ross and Ms. Ellason Reply

To the Editor: We agree with most of the methodological limitations pointed out by Drs. Powell and Howell. Our study lacked controls and a standardized treatment protocol, and subjects were recruited at their most symptomatic. Some of the improvement was probably due to spontaneous remission. The efficacy of psychotherapy for dissociative identity disorder is not a scientific fact.

We disagree that a tautology occurred because of our use of outcome measures to assign subjects to integrated versus nonintegrated groups. Only the DSM-IV criteria for dissociative identity disorder from the Dissociative Disorders Interview Schedule were used in this assignment. No other outcome measures influenced this assignment.

On the basis of our data, it seems unlikely that properly conducted therapy for dissociative identity disorder causes deterioration or chronic illness. Many people who receive this diagnosis make remarkable recoveries, given their degree of baseline comorbidity. Our data support allocation of resources to more rigorous prospective treatment outcome studies. In light of the methodological limitations of our study, one can no more attribute therapeutic causality to medications than to psychotherapy or spontaneous remission.

There are many ethical, logistical, and methodological problems in designing and conducting rigorous treatment outcome studies for dissociative identity disorder. Treatment to stable integration often takes 4 or 5 years. It would be unethical to randomly assign subjects to a placebo group for that long. Therapists who believe in the specificity of psychotherapy for dissociative identity disorder would be unlikely, as a matter of personal conscience, to participate in a study that involved random assignment to 5 years of placebo treatment. Therefore, one could not recruit therapists to the study. The likelihood that patients would stay in the study in the placebo control condition for that long without dropping out is minimal.

Prospective random assignment to placebo or waiting-list control groups in a treatment study for dissociative identity disorder is impractical. We suggest that different schools of thought should be compared to each other. At present, our school of thought is the only one that has provided prospective treatment outcome data with scientific methodology of any kind. The burden is now on competing schools of thought to match or exceed this level of methodology and to demonstrate a superior outcome.

We plan to initiate a prospective treatment outcome study beginning at the time of initial diagnosis. It will include measures of health care utilization, employment status, clinical variables, and treatment modalities employed.

Another approach would be to screen a large population base, such as all individuals covered by a major managed care company. Subjects with undiagnosed dissociative identity disorder could be identified with the test battery used in our study. Half could be randomly assigned to the treatment we recommend, and half could be left undiagnosed and tracked as they receive conventional treatment for other comorbid disorders.

We hope that such studies are completed over the next decade and subjected to rational, reasonable methodological scrutiny such as that provided by Drs. Powell and Howell.