Clinical Features of Psychotic Disorders: Comparing Categorical and Dimensional Models.

Objective: Despite research demonstrating the value of dimensional approaches, standard systems for classifying psychotic disorders rely primarily on categorization of patients into distinct diagnoses. We present the first study comparing analyses of dimensional features, categories, and standard diagnoses, all derived from the same sample. Methods: Using symptom ratings from 934 patients hospitalized for psychosis, we examined dimensional models, fit using factor analysis, categorical models, fit to factor-based scores from the dimensional model, and their correspondence with DSM-defined diagnoses. We compared the ability of each model to discriminate patients' assignment to medication regimen as a clinical validator. Results: Dimensional modeling identified four factors (manic, depressive, negative symptoms, and positive symptoms), which corresponded to factors in prior studies and appeared robust to statistical approach. Scores based on these factors overlapped substantially among DSM diagnoses. Patients assigned to clusters had less overlap in factor-based scores. However, categorical models were sensitive to statistical approach. The addition of DSM diagnoses, but not cluster assignments, improved the fits of models with dimensional scores alone as the clinical predictors for some medication classes. Conclusions: The results highlight the variability of symptom presentation within DSM-defined diagnostic categories, the utility of symptom dimensions or factors, and a potential lack of robustness of data-driven categorical approaches. Findings support initiatives to develop updated diagnostic systems that complement categorical classification of psychotic illness with factors representing dimensional ratings of symptoms.