Small-Area Life Expectancy: A Comparison of Methods, Relationship to Neighborhood Sociodemographic Factors, and Outlier Analysis

Chungfen Ren, Virginia Commonwealth University
Robert Johnson, Vanderbilt University
Steven A. Cohen, Virginia Commonwealth University
Emily Zimmerman, Virginia Commonwealth University
Allison Phillips, Virginia Commonwealth University
Amber Haley, Virginia Commonwealth University
Steven Woolf, Virginia Commonwealth University

Life expectancy (LE) varies substantially across US communities and neighborhoods. Numerous methodological challenges arise when life table methods are applied to small populations. Localized characteristics play an important role in population health. Objectives included (1.) comparing two methods of LE calculation for California, (2.) assessing associations between poverty and LE by census tract (CT), and (3.) evaluating associations between sociodemographic and environmental factors and “outlier status”, i.e. CTs with unexpectedly high or low LE given poverty level. We geocoded all deaths occurring in California (1999-2001) to CT of last residence. Life table methods were superior to Poisson modeling to create stable measures of LE in CTs. Outlier CTs (low LE and poverty) had increased domestic violence rates and unemployment, and reduced access to healthy foods. These findings have significant implications for methodology development in small area LE estimation, and in developing policies that address potentially modifiable factors driving these anomalies.