摘要：Objectives. We investigated trends in the educational gradient of US adult mortality, which has increased at the national level since the mid-1980s, within US regions. Methods. We used data from the 1986–2006 National Health Interview Survey Linked Mortality File on non-Hispanic White and Black adults aged 45 to 84 years (n = 498 517). We examined trends in the gradient within 4 US regions by race–gender subgroup by using age-standardized death rates. Results. Trends in the gradient exhibited a few subtle regional differences. Among women, the gradient was often narrowest in the Northeast. The region’s distinction grew over time mainly because low-educated women in the Northeast did not experience a significant increase in mortality like their counterparts in other regions (particularly for White women). Among White men, the gradient narrowed to a small degree in the West. Conclusions. The subtle regional differences indicate that geographic context can accentuate or suppress trends in the gradient. Studies of smaller areas may provide insights into the specific contextual characteristics (e.g., state tax policies) that have shaped the trends, and thus help explain and reverse the widening mortality disparities among US adults. The inverse association between education level and mortality risk in the United States is pervasive and enduring. 1–3 Education provides numerous resources that tend to lower mortality, including higher incomes, stable employment, social ties, health-promoting behaviors, a sense of personal control, and safe neighborhoods. 4 The educational gradient in mortality has been a long-standing concern of researchers, policymakers, and public health organizations. 5,6 Despite initiatives such as Healthy People to reduce mortality disparities, 6 the gradient has increased over the past half-century. 7–13 The pace and size of the increase has varied across population subgroups. For example, the gradient grew during the 1960s and 1970s more among White men than women. 8,11,13 Since the mid-1980s it seems to have grown more among women than men. 14,15 Among non-Hispanic White and Black women, this recent growth reflected declining mortality among the higher-educated alongside increasing mortality among the low-educated. 14,15 Among their male peers, the recent growth reflected declining mortality across education levels, with the higher-educated experiencing the largest declines. 14,15 The reasons for the increasing gradient are poorly understood. Studies investigating the reasons have largely focused on national trends in causes of death by education level. 14,16–18 They have found that smoking-related causes played an important role, especially among women. Though informative, those studies have drawn attention to individual-level behavioral explanations. A complete explanation must incorporate contextual factors that lie upstream in the causal chain, 19 such as economic and geographic contexts. In other words, the search for explanations may benefit by moving away from methodological individualism (the notion that mortality inequalities can be explained exclusively by individual characteristics) and toward an approach that integrates the broader contexts that constrain individuals’ lives. 20 The contextual factor of interest in this study is geographic area. It is important to investigate geographic variation in how the gradient has changed because it can shed light on the causes of the trends. If the gradient widened (or narrowed) in certain geographic areas, this suggests that characteristics unique to those areas—for example, economic policy or social welfare—played an important role. If the gradient widened similarly across areas, this suggests that the underlying causes transcend areal characteristics. In addition, examining trends in the gradient within areas helps decouple contextual and compositional explanations. For instance, the mortality increase among low-educated women at the national level could simply reflect a growing proportion of low-educated women from high-mortality areas of the Deep South. Although previous studies have not examined whether geographic context shapes trends in the gradient, they have shown that geographic context shapes mortality, net of individual factors, 21–28 and mortality trends. 22,26,29,30 For instance, during the 1980s and 1990s, gains in life expectancy occurred mainly in the Northeast and West coast, and declines occurred mainly in the Deep South, in Appalachia, along the Mississippi River, and in parts of the Midwest and Texas. 22 Explanations for the geographic pattern in mortality trends remain elusive. The salient features of geographic context and their pathways to health are complex, although material infrastructure and collective social functioning are especially important. 24,25,31–34 Ezzati et al. 22 found that county-level gains in life expectancy were positively related to county-level income but not to the Gini coefficient or the percentage completing high school. Murray et al. 26 found that life expectancy gaps among 8 geographic areas could not be explained by race, income, or health care access and utilization. Despite these informative comparisons of mortality trends across areas, it remains unclear whether morality disparities have changed within areas. The geographic areas we examined in this study are the 4 US regions: Northeast, North Central and Midwest (hereafter Midwest), South, and West. 35 The Census Bureau defines regions on the basis of historical development, economic structures, political systems, topography, population composition, and other factors. 35 Regions also differ in educational development and levels. The West was a leader in the secondary education movement whereas the South has trailed behind. 36 Regional variation in economic characteristics has also emerged and may have shaped mortality disparities. For instance, whereas the Northeast has increasingly had the most progressive tax policies, the South and more recently the West have implemented regressive policies that disproportionately hurt the poor and elevate mortality. 37 The Northeast has also emerged as the region with the highest social expenditures per capita. 37 Creative occupations have become concentrated in the Northeast and West. 38 Cigarette prices have risen most sharply in the Northeast and West. 39 Because of the large variation in regional characteristics, and the spatial clustering of mortality trends, it is important to examine the gradient within regions. In this study, we addressed the following questions among US adults aged 45 to 84 years in 1986 through 2006: (1) To what extent has the educational gradient of mortality changed over time within regions?; (2) Do any regions exhibit especially wide or narrow gradients and has this changed over time?; and (3) To what extent has the increase in low-educated women’s mortality occurred across regions? Then we explored what the patterns imply about the causes of the widening gradient.
