The predictive validity of the Leonhardean classification of endogenous psychoses: a 21-33-year follow-up of a prospective study ("BUDAPEST 2000").

To our knowledge, no previous long-term studies of the Leonhardean classification in the whole spectrum of endogenous psychoses have been conducted. This prospective study (n = 276; female patients n = 222; normal control persons n = 54) started in 1967-1976. The same population was followed-up by participation of a "blinded control" psychiatrist in 1997-2002 [patients available at follow-up = 125 (56.3%); available controls = 38 (70.4%)]. Patients for this investigation were selected by two independent diagnosticians from eight nosological groups based on full diagnostic agreement. Diagnostic agreement at follow-up (weighted-kappa) was 0.87. Predictive validity of the diagnostic categories was measured empirically and using a stochastic (Markovian) model, thus combining validity and reliability. Hebephrenias, group of normal persons and of schizophrenias proved to be valid categories, with diagnostic stabilities of 0.94, 0.91, and 0.93, for the three groups, respectively. In addition, bipolar manic-depressive psychoses and cycloid psychoses were also valid (diagnostic stability of 0.77 and 0.76, respectively). Unipolar depression was valid (diagnostic stability = 0.84) only by forming a "nosological family" based on diagnostic stability and on current status and clinical presentation during the period preceding the follow-up with regard to other mood-congruent disorders and outcome-diagnosis "normal control". Validity of systematic paraphrenias (diagnostic stability = 0.69) was in the moderate range. Division of schizophrenias in "systematic versus non-systematic" nosological categories was inconclusive; the categories of affect-laden paraphrenia, periodic catatonia and systematic catatonias could not be confirmed reliably in this study.