Creation of a Markov chain model of National Center for Health Statistics 2002 mortality data: Examining life expectancies, hypothetical cure scenarios, and mortality disparities.

摘要

The lack of a single, cohesive examination of cause-specific mortality in the United States has been recognized. Creating such a comprehensive, nation-wide picture of the mortality experience within the United States is crucial for identifying targets for public health interventions, including mortality disparities between sexes and races. To that end, a Markov chain model was created from the National Center for Health Statistics 2002 Compressed Mortality File (CMF). The model stratified each death in the United States by cause, age, sex, race (black or white), and region of residence in the United States (South, Northeast, Midwest, West) at the time of death. A total of 16 chains were created and used to describe the mortality experience of the United States by examining age-conditional life expectancies, the age-conditional probabilities of ultimately dying from specified causes, the long-term mortality impact of hypothetical cure scenarios, and overall mortality disparities between the stratification levels. In addition, the eventual causes of death of the 2002 US population were predicted. The models revealed consistent and persistent sex and race based mortality disparities. Within sex, whites had higher life expectancies than blacks at most ages, and within race, women generally had higher life expectancies than men. The causes of death identified as leading to the highest increases in life expectancy if cured were: for white men, heart disease, lung cancer, stroke, COPD, and suicide; for white women, heart disease, stroke, lung cancer, COPD, and breast cancer; for black men, heart disease, homicide, lung cancer, HIV/AIDS, and stroke; for black women, heart disease, stroke, diabetes, lung cancer, and breast cancer. Future public health interventions should particularly focus on causes of death identified as involving disparities and thus impacting one race or sex over the other, particularly causes of death significantly impacting life expectancy such as suicide, HIV/AIDS, homicide, and diabetes.