Objective: In Sweden there has been a continuous decrease in the autopsy rate during the last 15-20 years. The autopsy rate in the city of Malmö has declined from 81% in 1984 to 34% in 1993. The intention of this study was to improve our understanding of the causes of the decline, and to assess its consequences with regard to cause of death surveillance and case retrieval in epidemiological and clinical research.
Design: A retrospective study.
Setting: The University Hospital in Malmö, a city in southern Sweden with 230000 inhabitants.
Subjects: All deceased in 1984 (2900) and 1993 (3198). Cancer incidence 1984 and 1993.
Main outcome measures: Changes in autopsy rate in relation to age, sex, place of death and cause of death 1984 and 1993. Changes in the distribution of underlying causes of death in cases submitted and not submitted for a post mortem 1984 and 1993. Changes in cancer incidence and the number of incidentally-detected new tumours 1984 and 1993.
Results: Patients who died in nursing homes were less often sent for a post mortem than were patients who died at the hospital. From 1984 to 1993 the percentage of people dying in nursing homes increased from 1 to 29%. Care of the elderly has been reorganized, and the number of terminally ill patients who die in nursing homes has increased during the last 20 years in Sweden. This shift in place of death seems to be the single major explanation of the declining autopsy rate in Malmö. The overall autopsy rate in 1993 remained, however, lower than it was in 1984 when a shift in place of death was taken into account. During the period of study there was a decline in the autopsy rate at all departments within the hospital. The distribution of underlying causes of death according to the death certificates was similar in 1984 and in 1993. However, in cases not submitted for autopsy in 1993 there were greater proportions dying from pulmonary and circulatory diseases, respectively, than there were in 1984. The lower autopsy rate in 1993 was for several cancer sites also associated with a reduced number of new cancer cases. The percentage of tumours incidentally detected at autopsy went down in men from 40 to 19% and in women from 39 to 17%.
Conclusion: Patients submitted for autopsy represented, with regard to age, sex, cause and place of death, a selected group of all deceased. The declining autopsy rate was associated with a difference in the distribution of underlying cause of death and of the incidence of cancer. It is concluded that the changes in the autopsy rate have to be taken into account in studies dealing with time trends of causes of death and incidence of cancer.