The reliability of case register diagnoses: a birth cohort analysis.

BACKGROUND: Few studies assess the reliability of case register diagnoses, despite their widespread use in psychiatric research. This study investigates case register diagnostic reliability in comparison to casenote derived diagnoses in a birth cohort.
METHODS: Diagnostic information from the case register and casenotes of 449 individuals was extracted. The incident and lifetime register diagnoses were compared with those derived from the casenotes.
RESULTS: Inter-rater reliability was good (kappa = 0.71). Agreement between casenote and incident register diagnosis was moderate (kappa = 0.52), as was agreement between casenote and register lifetime diagnosis (kappa = 0.58). Case register diagnoses were insufficiently accurate to stand alone. Case register diagnoses for organic disorder, schizophrenia, alcoholism, learning disability, personality disorder and transient or no psychiatric disorder were reliable enough for the case register to act as a useful screening instrument. The case register was not acceptable, even as a screening instrument, for the diagnoses of neurotic or affective disorders.
CONCLUSIONS: Studies relying only on case register diagnoses may be flawed if diagnoses are not independently verified. National statistics derived from case register data, especially for neurosis and affective disorder, may be unreliable.