Operational definitions of schizophrenia: what do they identify?

Several studies have shown that different systems for diagnosing schizophrenia produce diagnoses with relatively low agreement. This, however, does not necessarily imply that the different systems are identifying conceptually different diagnostic groups. They may, in fact, identify a single entity, but with different degrees of accuracy. One-hundred ninety-six inpatients were diagnosed by multiple diagnostic systems. The pattern of relationship among the diagnoses was studied with latent class analysis, a statistical methodology which has not previously been applied to this type of data. Results indicated that the diagnoses of the Research Diagnostic Criteria, Flexible 6, Schneider, and 1978 Taylor and Abrams diagnostic systems all estimated a single underlying diagnosis. The Taylor and Abrams system identified this core diagnosis with high accuracy. The results also suggest that blunted affect and the absence of an affective syndrome are related to latent class schizophrenia, while the presence of only nonspecific delusions and/or hallucinations is not related. The appropriateness of the latent class definition of schizophrenia for selecting patients for research is discussed.