Competing causes of death: an analysis using multiple-cause-of-death data from The Netherlands.

The standard methodology for cause-elimination life tables assumes that the various causes of death are statistically unrelated to one another, so that the mortality risks of those who are saved from an eliminated cause equal the risks of dying from other causes which are observed for the general population. In the analysis reported in this paper, data on multiple causes of death for the year 1990 in the Netherlands were used to investigate whether this is likely to be a valid assumption. For four groups of underlying causes of death (malignant neoplasms, cardiovascular diseases, and external causes), the age-standardized prevalence at death of other conditions was calculated. Two series of calculations were performed: one with all other coded conditions present at death and one with a selection of conditions that were eligible to become the new underlying cause of death after the present underlying cause had been eliminated. The results suggested that there are major differences between underlying causes of death in the prevalence at death of other conditions. According to the second series of calculations, the prevalence at death of other conditions. According to the second series of calculations, the prevalence of possible new underlying causes of death was relatively high among persons who died from cardiovascular diseases; about average for persons who died from respiratory diseases; and relatively low for persons who died from malignant neoplasms and external causes. Although studies validating the multiple-cause-of-death data as they appear in the official statistics are necessary, these results reconfirm that this is a potentially rich source of information and that the assumption made in conventional cause-elimination life tables is unlikely to be valid.