Life Expectancy and Education

We commend the call by Montez and Zajacova for a more balanced research approach to trends in mortality-by-education gradients in the United States.¹ However, we are concerned that an attempt to force the inquiry into a dichotomy of “composition” versus “causation” may be counterproductive to our shared mission of understanding social disparities in health.

We recently raised concerns, for example, that composition cannot be considered an explanation for these patterns, but rather is an impediment to characterizing meaningful trends in subgroups whose composition has changed dramatically over time.² To illustrate the relevant issues, we used historical education and mortality data from the United States to demonstrate how the patterns of decreased life expectancy for the least educated White women could occur even if their longevity is increasing—as long as there is a large but stable underlying association between disadvantage and mortality. Nonetheless, we echo the important points that Montez and Zajacova raised—our findings do not diminish the likelihood that life has gotten harder for high school noncompleters because of the structural changes they note.

However, these structural changes raise a question to which we believe there is still no complete answer—“Is health equity getting worse? Or, is it remaining as bad as it has always been?” One alternative approach to characterizing trends has been to examine equivalent percentiles of the educational distribution over time.^3,4 Thus far, this approach has been illustrative, even though no group has experienced absolute declines in life expectancy, inequality has increased because mortality has declined more for the most educated. Work by Montez and Zajacova⁵ and others to document trends in period demographic constructs has been invaluable in starting a discussion on trends in health equity; that discussion continues to bear fruit and we caution against forcing it prematurely into the straitjacket of a false composition versus causation dichotomy.

Efforts at more accurate description of trends in equity should not be set against core tenets of social epidemiology such as the principle of “fundamental causes” of disease. This applies equally to trends in area-level mortality (because of selective migration that differed in timing by region, race/ethnicity, and gender) throughout this century. More detailed and accurate descriptions of health disparity trends are a necessary first step toward empowering science and policy to test causal hypotheses and improve equity. We look forward to the future exchanges and empirical work that these discussions stimulate.