• Life expectancy is going down for many

    Given my interest in life expectancy, and my desire to overcome the myth that its “meteoric rise” requires trimming benefits for the elderly, I simply cannot believe I missed this study from Health Affairs last summer. “Differences In Life Expectancy Due To Race And Educational Differences Are Widening, And Many May Not Catch Up“:

    It has long been known that despite well-documented improvements in longevity for most Americans, alarming disparities persist among racial groups and between the well-educated and those with less education. In this article we update estimates of the impact of race and education on past and present life expectancy, examine trends in disparities from 1990 through 2008, and place observed disparities in the context of a rapidly aging society that is emerging at a time of optimism about the next revolution in longevity. We found that in 2008 US adult men and women with fewer than twelve years of education had life expectancies not much better than those of all adults in the 1950s and 1960s. When race and education are combined, the disparity is even more striking. In 2008 white US men and women with 16 years or more of schooling had life expectancies far greater than black Americans with fewer than 12 years of education—14.2 years more for white men than black men, and 10.3 years more for white women than black women. These gaps have widened over time and have led to at least two “Americas,” if not multiple others, in terms of life expectancy, demarcated by level of education and racial-group membership. The message for policy makers is clear: implement educational enhancements at young, middle, and older ages for people of all races, to reduce the large gap in health and longevity that persists today.

    I’m not sure if the last sentence is true, as I’m not sure there’s a direct causal link between education and life expectancy. I think it’s more likely that education is a marker for socio-economic status and resources available to people, which are far more likely to impact health and life expectancy directly.

    But this study does show that not only aren’t all people seeing the same gains in life expectancy, but that many are actually losing ground. Check this out:

    LE Birth by education


    What you’re seeing is life expectancy by years of education for a white woman  at age 25 in three different years. In general, the more educated you are, the longer you can expect to live. But I’m more interested in the trends over time. The blue line is your expected life expectancy for these women in 1990, the green line in 2000, and the red line in 2008. If you’ve been to graduate school, your life expectancy has been going up quite nicely. If you’ve finished high school or been to college, the gains are less pronounced – but they’re there. The horror story is if you didn’t finish high school. Then, in just this 18 year period, your life expectancy dropped quite sharply.

    The trends for white males  with less than a high school education were similarly bad. Ironically, Hispanic and Black people of both sexes showed that even in this under-educated group, life expectancy increased over time. The decrease? It’s a white thing. Remember that the next time someone tries to tell you that it’s minority groups bringing our population level statistics down.

    There’s a huge disparity between the most and least educated among us. In 2008, 25 year old males in the former category had a life expectancy that was 14.2 years longer than those in the latter. For females, the difference was 10.3 years.

    Not all of us are living longer. For many US citizens, life expectancy is going down. Remember that the next time someone tells you that we need to raise the retirement age, and that it’s not a regressive idea.


    (h/t Kevin Drum)

    UPDATE: Sometimes it’s like you guys purposely forget that the Internet (and this blog) are forever. For everyone who disputes that life expectancy is going down for many, here’s this. From the past.

    • I agree that the life expectancy trend in the left tail of the education distribution is troubling, but certainly the left tail of the education distribution in 2008 differed significantly from that in 1990.

      According to http://www.census.gov/prod/2012pubs/p20-566.pdf, 85 percent of individuals 25 and older had at least a high school diploma in 2009. The corresponding figure from 1990 looks to be around 75 percent. Thus simply looking at these bar charts likely overstates the decline for low-education individuals. I wonder if it would be possible to apply something akin to the Dinardo-Fortin-Lemieux semiparametric approach (http://www.uh.edu/~adkugler/DiNardoetal.pdf) to estimate the distribution of life expectancies by year holding the education distribution constant over time.

      • I agree with Dan. When posting these kinds of charts where data is broken down by groups, it is vital to understand how the groups themselves might be changing. From Dan’s post, it seems likely that the marginal high school dropouts from 1990 would be graduates in 2009. If their life expectancy was higher, then that would explain this graph.

        Another interesting idea would be to use a difference in difference to determine if there’s a discrete jump in life expectancy for those who gain a diploma.

        • No, they are already 25 year olds. If they were to finish high school, they almost all already would have,

    • Be careful to avoid Simpson’s Paradox in interpreting data like these. It is mathematically possible for all individuals to be better off but have groups show decline based on composition changes. A warning bell should go off in your head whenever you see individual-level interpretations on a series of cross-sectional data points.

    • You missed it to your credit because it is a spurious finding…

      but now you wrote :

      But this study does show that not only aren’t all people seeing the same gains in life expectancy, but that many are actually losing ground

      From my post on this poorly done study:

      So since so many fewer people fail to graduate high school today those who still do not graduate are from a lower group than than in the past. To clarify what he is actually saying is: Since the bottom 5% today are living less long than the bottom 25% lived 25 years ago life expectancy is not increasing for all groups.

    • Bedsides if it were true it would be an indictment of the health care system and medicaid in particular because access to healthcare has expanded for this group. Perhaps the only group for which access has expanded, the others already had access.

    • It’s worth noting that the high-school graduation rate has been increasing over time, and thus the number of people who fall into the “dropped out of high school and dying young” category is actually shrinking in relative terms.

      In order for the claim about white high-school dropouts bringing US longevity stats down to be substantiated, one would have to compensate for changes in the underlying population’s racial and educational characteristics over time.

      That could very well be the case – but it’s not clear from the data, and the magnitude of the longevity loss in a group that represents an ever-shrinking percentage of the population would have to be larger than the effect of changes in ethnic composition over the same time period.

      Here’s a recent paper on race and mortality that touches on these questions:

      State-level variations in racial disparities in life expectancy.
      Bharmal N, Tseng CH, Kaplan R, Wong MD.

      Division of General Internal Medicine and Health Services Research, UCLA David Geffen School of Medicine, University of California-Los Angeles, CA, USA.

      States with small racial differences are due to higher-than-expected life expectancy for blacks or lower-than-expected for whites. States with large disparity are explained by higher-than-average life expectancy among whites or lower-than-average life expectancy among blacks.

      Heterogeneous state patterns in racial disparity in life expectancy exist. Eliminating disparity in states with large black populations would make the greatest impact nationally.”


    • You’re right in pointing out that there’s not a lot of evidence showing the direct causal linkage between education and health. But Paula Braveman has a great illustration theorizing the multiple ways in which educational attainment impacts health. She’s presented it in multiple venues, including on p. 74 of the IOM report “How Far Have We Come in Reducing Health Disparities?” )

    • Come on man, this is a methodologically terrible analysis that no one should take seriously. I’m surprised Health Affairs published it, well maybe not…

      A much more rigorous study is required before drawing any meaningful conculsions.

    • Is there a study that examines life expectancy in relation to rates of being insured or uninsured during someone’s lifetime?