Trends in U.S. life expectancy gradients: the role of changing educational composition

BACKGROUND: I examined age patterns and the role of shifting educational distributions in driving trends in educational gradients in life expectancy among non-Hispanic Whites between 1991 and 2005. METHODS: Data were from the 1986-2004 National Health Interview Survey with mortality follow-up through 2006. Life expectancies were computed by sex, period and education. Age decompositions of life expectancy gradients and composition-adjusted life expectancies were computed to account for age patterns and shifting educational distributions. RESULTS: Life expectancy at age 25 among White men increased for all education groups, decreased among the least-educated White women and increased among White women with college degrees. Much of the decline in measured life expectancy for White women with less than a high school education comes from the 85+ age group. Educational gradients in life expectancy widened for White men and women. One-third of the gradient is due to ages below 50. Approximately 26% (0.7 years) and 87% (0.8 years) of the widening of the gradient in life expectancy between ages 25 and 85 for White women and men is attributable to shifting education distributions. Over half of the decline in temporary life expectancy among the least-educated White women is due to compositional change. CONCLUSIONS: Life expectancy has increased among White men for all education groups and has decreased among White women with less than a high school education, though not to the extent reported in previous studies. The fact that a large proportion of the change in education-specific life expectancy among women is due to the 85+ age group suggests changes in institutionalization may be affecting estimates. Much of the change in education-specific life expectancy and the growth in the educational gradient in life expectancy is due to the shifting distribution of individuals across education categories.