Abstract

Necropsy reports were traced for 3928 of the 7049 Sheffield children who died between 1947 and 1979. Comparison of these reports with the related death certificates showed important differences in the underlying cause of death in 15%, even after allowing for variation in diagnostic terms and incorrect completion of certificates. Approximately one third of certificates had the underlying cause of death entered first instead of last in Part I, and in 251 (6%), the underlying cause of death was found in Part II of the certificate. There was an overrepresentation of infectious diseases and underrepresentation of malformations, deaths associated with perinatal factors, and cot deaths in the death certificates. Discrepancies were more frequent among very young children and among hospital deaths than home deaths. There was little variation in discrepancy frequency when analysed by parental occupation. A gradual decrease in the proportion of discrepancies occurred during the period studied, but this may be a consequence of a changing pattern of referral for necropsy rather than increasingly valid certification. The historical diagnoses 'convulsions due to an unknown but natural cause' and 'teething convulsions' may be analogous to the current diagnosis 'sudden infant death syndrome'.