Socio-demographic correlates of multiple causes of death: Real or artifactual?

Abstract

The socio-demographic and epidemiological correlates of the total number of causes of death (TC) reported on death certificates were examined with multiple classification analysis (MCA). The data were all death certificates on white and black adult residents of Michigan who died at ages 25 or older in 1989 to 1991 (n = 222,763). TC was the sum of every morbid condition named as an underlying, intermediate, or immediate cause of death, or other medical condition contributing to death but not to its underlying cause. Autopsies performed for non-forensic reasons by physicians who are not Medical Examiners (MEs) likely yield the most accurate diagnoses of the underlying medical cause and counts of all other attendant causes, but they are extremely selective of decedents who are most closely integrated into the health care system. Thus we required a socio-demographic pattern in TCs to appear in a multivariate analysis not only for this special group of autopsied decedents but also for the non-autopsied masses before accepting the pattern as ‘real’' (not an artifact of underreporting of TC or confoundment with another socio-demographic variable).

The MCAs showed that age at death and its underlying medical cause were the most important factors related to TC, as suggested by epidemiologic transition theory. Compared with other underlying causes, diabetes, hypertension, and septicemia yielded the highest average TCs. While sex differences were generally not significant within racial groups, somewhat higher average TCs were found for blacks than whites and for lesser educated than for more educated decedents. These patterns in TC by race and education may reflect lifetime disadvantages faced by blacks and the less educated to health care or information about health practices.