Landmark research by Raj Chetty and colleagues at Health Inequality finds wide geographic and intertemporal variation in the relationship between income and life expectancy in the U.S., using 1.4 billion administrative tax records from the IRS linked to nearly 7 million federal death records. Many questions remain about the underlying causes of these findings and their implications for policy. This study links Chetty’s income and life expectancy data with the 1998-2014 National Longitudinal Survey of Public Health Systems to examine whether the existence and strength of multi-sector population health improvement activities influence income-related disparities in life expectancy over time. Communities attaining the highest level of multi-sector population health activities increased from 24% of the sample in 1998 to 37% in 2006, but fell to 31% in 2012 and recovered modestly to 33% in 2014. Within sectors, hospitals increased their contributions to population health activities by nearly 20% between 2012 and 2014, while insurers, employers, and nonprofit community-based organizations showed smaller but significant increases in contributions (p<0.05). Residing in a community with the highest level of population health activity was associated with a 3.9 year gain in life expectancy for individuals in the bottom quartile of the income distribution after controlling for observed and unmeasured confounders (p<0.05), but no significant gain in life expectancy for the top income quartile. Differences in life expectancy between the top and bottom income quartiles declined by an estimated 2.1 years in communities with the highest level of activity (p<0.01). Community capacity to implement widely-recommended population health activities is one important contributor to geographic variation in the relationship between income and life expectancy.
- Glen P. Mays, PhD, Director, Systems for Action National Coordinating Center and F. Douglas Scutchfield Endowed Professor in Health Services and Systems Research, University of Kentucky College of Public Health
Linking NALSYS survey data with Chetty’s county-specific estimates of life expectancy by income quartile from 1999 to 2014, along with measures of community demographic, socioeconomic, and health resource characteristics from other data sources. Random-effects models with instrumental-variables are used to estimate whether changes in multi-sector public health activities lead to changes life expectancy by income quartile, while controlling for both observable and unmeasured confounders.