Accuracy of clinical diagnoses in a teaching hospital: a review of 997 autopsies

J Intern Med. 1993 Aug;234(2):181-7. doi: 10.1111/j.1365-2796.1993.tb00728.x.


Objectives: To analyse the concordance between clinical and autopsy diagnoses.

Design: Nine-hundred-and-ninety-seven autopsies were studied comparing the diagnoses of the autopsy requests with those of the death certificates and autopsy reports. The cases were grouped according to the 17 categories of diseases of the International Classification of Diseases (ICD) and the concordance was analysed with the kappa (kappa) coefficient of concordance.

Setting: The Hospital da Clínicas da Faculdade de Medicina de Ribeirão Preto da Universidade de São Paulo (HCFMRPUSP), Ribeirão Preto, SP, Brazil.

Subjects: The patients autopsied at HCFMRPUSP during the period between 1978 and 1980.

Main outcome measures: kappa statistics offer an alternative approach to measuring the concordance between clinical and autopsy diagnoses.

Results: The kappa-value obtained was equal to 0.601 with a variance of 1.545 x 10(-4) when comparing the clinical diagnoses and the autopsy diagnoses, and equal to 0.661 with a variance of 1.531 x 10(-4) comparing the clinical diagnoses with those obtained after the gross examination. These values are significant at the level of 5%, i.e. there is an overall statistical concordance between clinical and autopsy diagnoses although the value is not absolute (kappa = 1.00).

Conclusions: If autopsies are heeded without bias, they will continue to give important feedback concerning medical diagnosis.

MeSH terms

  • Autopsy*
  • Brazil
  • Diagnostic Errors*
  • Hospitals, Teaching / standards*
  • Humans
  • Retrospective Studies
  • Statistics as Topic