Objective: The authors systematically examined a sample of patients who were referred to an ongoing National Institute of Mental Health (NIMH) study of childhood-onset schizophrenia (COS), but who received diagnoses of mood disorders at the NIMH, to analyze the reliability of these research-setting diagnoses and to characterize the patients clinically. Pilot data regarding the clinical course of these patients over a 2- to 7-year follow-up period were also obtained.
Method: Thirty-three cases were selected from the 215 pediatric patients who had been screened in person from 1991 to 1999 for admission to the COS study. These 33 patients had been excluded from the COS study on the basis of a day-long evaluation, including a structured diagnostic interview, which yielded a diagnosis of a mood disorder rather than schizophrenia. This subgroup, together with six COS subjects (for a total N= 39), were included in a diagnostic reliability study in which they were reevaluated by three psychiatrists who were blind to the initial research diagnosis. In addition, pilot follow-up data regarding current function and treatment status were obtained for 25 of the 33 patients with mood disorders.
Results: Overall, the interrater reliability of the three raters was excellent (kappa = 0.90). Global reliability between these raters and the NIMH research diagnoses was good (average kappa across diagnoses = 0.61), and agreement for those patients who had mood disorders was good (86% agreement; kappa = 0.60). Pilot follow-up data indicate that none of the subjects with a diagnosed mood disorder developed a clinical course resembling schizophrenia.
Conclusions: Many of the patients referred to the NIMH COS study with clinical diagnoses of schizophrenia had psychotic mood disorders diagnosed on the basis of a comprehensive research evaluation including structured diagnostic interviews, and these research diagnoses were reliable. The diagnosis of COS is difficult and requires a time-consuming evaluation process.