Explaining outliers: Sociodemographic and environmental determinants of unexpectedly high life expectancy in California census tracts

Although place-based characteristics play an important role in population health, few studies have quantitatively and qualitatively examined community assets that enhance population health on a fine geographic scale. We use California census tracts (CTs) to identify local assets that might explain geographic areas of unexpectedly high life expectancy in high-poverty areas. All 1999-2001 and 2009-2011 death records from California vital statistics were combined with 2000 and 2010 US Census population data, respectively, to create abridged life tables with CT-specific life expectancy (LE) values.  In Phase 1, weighted linear regression was used to model LE against CT-level poverty, education, age, gender, marital status, race, and Hispanic origin.  In Phase 2, we compared the resultant positive outliers CTs (high LE and poverty) to non-outlier CTs using weighted logistic regression, modeling outlier status against sociodemographic and environmental characteristics, including all Phase 1 variables.  For 1999-2001, outlier CTs had a higher percent of the population biking or walking to work (OR1.85, p=0.0114), lower illiteracy levels (OR 0.71, p=0.0395) and air pollutants (OR 0.76, p<0.0001), and were more rural (OR 1.23, p<0.0001).   In 2009-2011, outlier CTs were associated with being predominantly undeveloped land (OR 2.52, p=0.0034), presence of an acute care hospital (OR 2.60, p=0.0031), and lower PM2.5 concentrations (OR 0.54, p=0.0019).  Factors common to both periods included percent smokers, sexually transmitted infection rate, and ambulatory care sensitive conditions rate.  Key informant interviews will be conducted in CTs that were outliers in both periods to identify additional assets that might explain their favorable outcomes.