Introduction: Long-term outcomes in children with atypical psychosis have been poorly studied. Four to 6 weeks of inpatient observation and up to 11 years (mean, 4.0 +/- 1.3 years) of follow- up have afforded us some experience with this probably heterogeneous group of transiently psychotic patients commonly mislabeled as schizophrenic. Despite severe preadmission morbidity, some patients have successfully remained neuroleptic-free since discharge. Predictors of good versus poor outcome were sought.
Methods: Of roughly 150 patients admitted with the presumptive diagnosis of schizophrenia, 32 patients were discharged meeting criteria for psychosis not otherwise specified (NOS), otherwise labeled by the NIMH team as "multidimensionally impaired" (MDI). Admission and biannual follow-up data included a semistructured clinical interview with the Schedule for Affective Disorders and Schizophrenia for School Age Children (K-SADS), IQ testing, clinical rating scales (e.g., Clinical Global Impression Scale (CGI), Children's Global Assessment Scale (C-GAS), Brief Psychiatric Rating Scale (BPRS), Scales for the Assessment Negative and Positive Symptoms (SANS and SAPS), and Bunney-Hamburg (B-H)). At follow-up (as of February 2005) 38% of patients (12 of 32) met criteria for bipolar 1 disorder, 12% (4 of 23) for major depressive disorder (MDD), and 3% (1 of 32) for schizoaffective disorder. The remaining 47% of patients (15 of 32) were divided into two groups on the basis of whether they were in remission and neuroleptic-free ("good outcome," n = 5) or still severely impaired and/or psychotic regardless of pharmacotherapy ("poor outcome," n = 10) at follow-up.
Results: Good-outcome patients had a significantly higher baseline level of functioning (on admission and on medications). This was demonstrated by better scores on CGI (3.5 +/- 0.6 versus 4.8 +/- 0.8; p = 0.03) and C-GAS (66.3 +/- 6.3 versus 38.6 +/- 11.5; p = 0.01). Groups were otherwise comparable in demographic data (gender, race, socioeconomic status, age at onset), months of neuroleptic exposure, severity of psychotic symptoms, and level of premorbid functioning.
Conclusion: C-GAS (which correctly classified 85.7% of good-outcome subjects) and CGI at baseline appear to predict outcome. On other variables, MDI subgroups were remarkably similar.