[The usefulness of death certificates in epidemiological studies].

An analysis of a total of 189 deaths recorded between 1953 and 1990 from two companies in Hamburg was performed to examine the question whether and if so to what extent differences existed between the causes of death (primary disease) stated in the death certificates on the one hand, and the diagnoses based on subsequent medical (clinical and pathologico-anatomical) reports, on the other. Postmortems were conducted in 94 cases; the diagnoses resulting therefrom had also been taken into consideration in 58 of the pertaining death certificates. Considerable differences were seen for the remaining 36 postmortems where only clinical diagnoses were recorded in the death certificates. In 10 cases (= 27.8%) the statements made in the death certificates did not agree with the PM diagnoses. The overall rate of deviations was 12.7% the figures for the individual fatal diseases varied, the greatest difference occurring with diseases of the heart and with chronic bronchitis and the least difference with acute apoplexy, accidents, suicides and malignant tumours. It is noted that death certificates and hence also the official statistics derived therefrom are but an inaccurate reflection of the actual distribution of causes of death in case of unicausal evaluations. The quota of diagnoses based on PM and actually taken into consideration, is an important factor in this regard. However, the data supplied by the death certificates are reasonably useful for epidemiological studies to estimate the risks involved in certain diseases.