Rising Geographic Disparities in US Mortality
Studies of life expectancy in the United States increasingly find disparities among different groups in the population. For example, a recent report by the National Academies of Science, Engineering, and Medicine finds large and widening mortality differences based on race, ethnicity, economic status, and geography.
This paper documents and analyzes rising mortality disparities at the state level. Like the National Academies study and much of the recent research on life expectancy, the paper focuses on mortality trends for persons at midlife, defined as ages 25 through 64. Consistent with the thrust of previous research, the paper finds that state-level midlife mortality rates have become much more unequal over time.
One explanation for this divergence is that state-level disparities are driven by differences in education levels and labor market prospects. A second and possibly related explanation is that greater state-level dispersion has been driven by rising spatial inequality in income. A third possibility is that the widening divergence in mortality stems from a portmanteau of place effects that are independent of state-level income. These effects would capture both the health behaviors of individuals who live in a place as well as the evolving features of the region’s overall health environment.
The paper use data on mortality, income, health behavior, and health-care quality to test how well each of these hypotheses explains the data.
- There is no evidence that states with the most rapid income growth experienced the most rapid mortality decline. Instead, states with relatively high income levels over the past several decades have experienced the largest improvements in midlife mortality.
- The national trends in educational attainment and a rising national correlation between education and mortality ultimately explain little of the increasing importance of place in determining mortality.
- Although deaths of despair have contributed to the plateau in US life expectancy, they account for only about one-sixth of all midlife deaths. The increase in state-level midlife mortality disparities is driven largely by other causes of death.
- The differential adoption of policies—such as tobacco taxes, Medicaid expansions, and income support—in high-income but not low-income states may have led to both widening spatial disparities in mortality and to an increasingly close negative association between income and mortality. These policies are distinct from but complementary to health-related behaviors that also differ across states.
Future research on state-level disparities in health should focus on identifying the relative effects of government policies and health behaviors on state-level mortality rates. Beneficial policies and behaviors could have long-lasting effects on health that are distinct from place effects typically studied in the existing literature. Many papers in this literature evaluate the short-term impact of moving from one location to another. The causal place effects identified in those studies are conceptually different from the residual place effects that the authors measure in this paper. As the authors of this paper note, the short-term impact on health of moving from Mississippi to New York is different from the longer-term effects of growing up in Mississippi versus growing up in New York, due to the different policies and behaviors relevant for people coming into adulthood in the two states.
The 21st century has been a period of rising inequality in both income and health. In this study, we find that geographic inequality in mortality for midlife Americans increased by about 70 percent from 1992 to 2016. This was not simply because states such as New York or California benefited from having a high fraction of college-educated residents who enjoyed the largest health gains during the last several decades. Nor was higher dispersion in mortality caused entirely by the increasing importance of “deaths of despair,” or by rising spatial income inequality during the same period. Instead, over time, state-level mortality has become increasingly correlated with state-level income; in 1992 income explained only 3 percent of mortality inequality, but by 2016 state-level income explained 58 percent. These mortality patterns are consistent with the view that high-income states in 1992 were better able to enact public health strategies and adopt behaviors that, over the next quarter-century, resulted in pronounced relative declines in mortality. The substantial longevity gains in high-income states led to greater cross-state inequality in mortality.