An assessment of certain medical aspects of death certificate data for epidemiologic study of arteriosclerotic heart disease

Abstract

An assessment was undertaken of the adequacy of death certificate data as a means of identifying cases (and ‘non-cases’) of arteriosclerotic heart disease for use in epidemiologic studies of the disease. As a first step in such an evaluation, this study attempted to measure in a series of consecutively autopsied deaths occurring at the Albany Medical Center Hospital the agreement between two sets of data: (1) the cause of death as coded on the death certificate and (2) certain relatively welldefined lesions, myocardial infarction and coronary thrombosis as determined at autopsy. The degree of agreement found between these two sources with respect to arteriosclerotic heart disease was then compared with that for cancer. The death certificate data were found to be somewhat inadequate in sensitivity for identifying cases of arteriosclerotic heart disease, according to the criteria employed, since in only 50 per cent of the individuals with infarction or coronary thrombosis at autopsy was arteriosclerotic heart disease coded on the death certificate. The death data were, however, more specific in regard to arteriosclerotic heart disease: 82 per cent of the death certificates reporting arteriosclerotic heart disease as the cause of death were confirmed as cases by autopsy findings. On the other hand, for cancer, death certificate data were considerably more specific and sensitive. It is concluded from this comparison that death data employed in epidemiologic studies of cancer probably have a level of validity, and therefore of usefulness, which, upon demonstration of a significant association, would supply highly persuasive support to an inference. However, similar support cannot be expected from associations based upon death data relating to arteriosclerotic heart disease due to its low level of sensitivity. It is suggested that problems in disease definition, diagnostic techniques, and procedures for certifying cause of death may be chiefly responsible for this difference between the two diseases.