Comparison between register and structured interview diagnoses of schizophrenia: a case for longitudinal diagnostic profiles

Objective: Mental health registers contain diagnoses from serial contacts with mental health facilities over many years. This study examines the relationship between longitudinal diagnostic profiles and structured interview diagnoses. The aim is to improve the definition of diagnoses drawn from clinical case registers. Method: The Tasmanian Mental Health Case Register includes 1922 individuals, each with at least one diagnosis of schizophrenia between 1965 and 1990. A representative subsample of 29 individuals were assessed by the structured diagnostic interview for DSM-III-R (SCID). Diagnostic agreement between Register and SCID diagnoses was compared. Results: Twenty-four subjects (82.8%) received a lifetime diagnosis of schizophrenia on the SCID. For each subject, 'schizophrenia diagnostic dominance', the percentage of register entries with schizophrenia diagnoses over total entries, was calculated. Agreement between register and SCID correlated positively with schizophrenia diagnostic dominance and negatively with register mood diagnoses. Conclusions: Longitudinal diagnostic profiles on databases may be superior to cross-sectional clinical diagnoses in predicting structured interview diagnoses, and may be useful in defining caseness in epidemiological studies using register diagnoses.