Concern has been expressed as to the reliability of clinical ICD-10 diagnosis of schizophrenia. This study was designed to assess the diagnostic reliability of the clinical ICD-10 diagnosis of schizophrenia in a random sample of Danish in- and outpatients with a history of psychosis. A sample of 100 subjects was assessed using the operational criteria OPCRIT checklist for psychotic and affective illness. The most recent principal and clinical ICD-10 diagnosis was compared with diagnoses generated by the OPCRIT instrument. Data documented very high sensitivity (93%) and positive predictive value (87%) of ICD-10 schizophrenia and an overall good agreement between clinical and OPCRIT-derived diagnoses (kappa=0.60). An even higher positive predictive value was obtained when diagnoses were amalgamated into a diagnostic entity of schizophrenia-spectrum disorders (98%). Near perfect agreement was seen between OPCRIT-derived ICD-10 and DSM-IV diagnoses (kappa=0.87). Thus, this study demonstrates high reliability of the clinical diagnosis of schizophrenia and even more so of the diagnosis of schizophrenia-spectrum disorder.

Background: To investigate whether diagnostic agreement across different diagnostic systems improves in a sample of chronic patients suffering from functional psychosis compared to first-admitted patients. Sampling and Methods: Among 353 patients with a history of functional psychosis, a subset of 100 individuals (35 women and 65 men) were randomly sampled and assessed using the Operational Criteria Checklist for Psychotic Illness and Affective Illness (OPCRIT). Based on the OPCRIT diagnoses the subjects suffering from schizophrenia and schizophrenia spectrum disorders according to seven diagnostic systems were identified. Diagnostic agreement was assessed using unweighted ĸ-statistics and pairwise concordance rates (CR). Results: High diagnostic agreement of schizophrenia was observed across the ICD-10 and DSM systems (CR >0.70, ĸ >0.70), which all had a significantly lower concordance to the St. Louis Criteria (SLC), research diagnostic criteria and Schneider’s first rank symptoms (FRS) (0.32< CR <0.66; –0.10< ĸ <0.51). Agreement on schizophrenia across all systems was observed for one fourth of the subjects. Elimination of the diagnostic impact of ‘co-occurrence of psychotic and affective symptoms’ excluded FRS standalone individuals from the sample, increased overall homogeneity and resulted in a dichotomized sample according to SLC (46 positive vs. 47 negative). SLC status could be predicted in 78% of cases by four items relating to family history and psychosocial function previous to the onset of illness. Similarly high pairwise CR were observed for schizophrenia spectrum disorders across all diagnostic systems. Conclusions: This study demonstrates that diagnostic agreement is higher among chronic patients than that observed in subjects with a recent onset of psychosis, although considerable discordance is also observed in this chronic sample. However, the discordance among chronic patients with functional psychosis largely derives from the different emphasis that diagnostic systems place on co-occurrence of psychotic and affective symptoms. This may have serious epistemological consequences, thus underlining the conventional nature of the present schizophrenia diagnoses and the need for biologically founded diagnostic criteria.