*Abstract*
For decades, mortality studies have assumed a single disease causes each death. Nosologists harmonize cause-of-death data by imposing the same set of ideal-typical causal chains on millions of deaths. Both biological and social theories cast doubt on this model, favoring an approach where multiple diseases act together to produce mortality. Additionally, social theory regarding diagnosis and medicalization predict medical knowledge summarized in ideal-typical standards are stratified across race, gender, and age. This stratification will impact cause of death analyses in minority, female, and older populations. Using Bayesian Change-point models on administrative cause of death data collected by the National Center for Health Statistics from 1994 to 2003 (n=22 million), this study tests those expectations. It finds (1) a discontinuity in causes of death with use of a new coding scheme, and (2) imposing ideal-typical causes mask rising acute-to-chronic death rates for certain populations: female, black, and under 75.