A comparison of 16 systems to diagnose schizophrenia

Sixteen diagnostic classification systems were compared to determine the degree of agreement on the diagnosis of schizophrenia. Data were collected by a structured interview method with 284 patients, 52 first admissions, and 196 readmissions. The kappa coefficient was used to calculate interrater reliability and concordance of the systems. Adequate reliability was found for 12 of the systems. However, concordance was generally low; 11 of 120 combinations reached kappa of .40 or higher for the total sample, 16 of 120 for first admissions, and only 6 of 120 for readmissions. These results indicate the restriction of generalization and raise questions as to the empirical validity of the construct of schizophrenia.