Abstract
OBJECTIVE: Because clinical and biologic research and optimal clinical practice require stability of diagnoses over time, we determined stability of ICD-10 psychotic disorder diagnoses and sought predictors of diagnostic instability. METHOD: Patients from the McLean-Harvard International First-Episode Project, conducted from 1989 to 2003, who were hospitalized for first psychotic illnesses (N = 500) were diagnosed by ICD-10 criteria at baseline and 24 months, on the basis of extensive prospective assessments, to evaluate the longitudinal stability of specific categorical diagnoses and predictors of diagnostic change. RESULTS: Diagnostic stability averaged 90.4%, ranking as follows: schizoaffective disorder (100.0%) > mania with psychosis (99.0%) > mixed affective episode (94.9%) > schizophrenia (94.6%) > delusional disorder (88.2%) > severe depressive episode with psychotic symptoms (85.2%) > acute psychosis with/without schizophrenia symptoms = unspecified psychosis (all 66.7%) >> acute schizophrenia-like psychosis (28.6%). Diagnoses changed by 24 months of follow-up to schizoaffective disorder (37.5%), bipolar disorder (25.0%), schizophrenia (16.7%), or unspecified nonorganic psychosis (8.3%), mainly through emerging affective features. By logistic regression, diagnostic change was associated with Schneiderian first-rank psychotic symptoms at intake > lack of premorbid substance use. CONCLUSIONS: We found some psychotic disorder diagnoses to be more stable by ICD-10 than DSM-IV criteria in the same patients, with implications for revisions of both diagnostic systems.