Objectives. To elucidate why the inverse association between education level and mortality risk (the gradient) has increased markedly among White women since the mid-1980s, we identified causes of death for which the gradient increased.

Methods. We used data from the 1986 to 2006 National Health Interview Survey Linked Mortality File on non-Hispanic White women aged 45 to 84 years (n = 230 692). We examined trends in the gradient by cause of death across 4 time periods and 4 education levels using age-standardized death rates.

Results. During 1986 to 2002, the growing gradient for all-cause mortality reflected increasing mortality among low-educated women and declining mortality among college-educated women; during 2003 to 2006 it mainly reflected declining mortality among college-educated women. The gradient increased for heart disease, lung cancer, chronic lower respiratory disease, cerebrovascular disease, diabetes, and Alzheimer’s disease. Lung cancer and chronic lower respiratory disease explained 47% of the overall increase.

Conclusions. Mortality disparities among White women widened across 1986 to 2006 partially because of causes of death for which smoking is a major risk factor. A comprehensive policy framework should address the social conditions that influence smoking among disadvantaged women.

The inverse association between education level and mortality risk (the gradient) in the United States is well established.1–3 Higher education levels provide resources that tend to lower mortality risk, including higher incomes, stable jobs, salubrious social ties, healthy lifestyles, self-efficacy, and safe neighborhoods.4 The gradient has been a long-standing concern among researchers, policymakers, and those who promote public health initiatives.5,6

Despite the attempts of initiatives such as Healthy People to eliminate health disparities,6 the gradient increased over the past half century.7–12 The timing and magnitude of the increase has varied across demographic groups. Although the gradient grew during the 1960s and 1970s more among White men than among White women,8,10,12 since the mid-1980s it appears to have grown more among women than among men.13,14 Among White women in particular, this recent growth reflected declines in mortality among the higher educated alongside increases in mortality among the low educated (mortality continued to decline among low-educated men).13,14

The reasons for the growth in the gradient among women remain unclear. Understanding the reasons is critical for designing strategies to reverse the growth and for projecting future trends in women’s longevity and health care needs. We focused on non-Hispanic White (hereafter White) women because recent increases in mortality among the low-educated in this racial/ethnic group were statistically significant and substantively large.13–15 For example, during the 1990s life expectancy at age 25 years among non-Hispanic adults with 12 or fewer years of education decreased by 0.9 years among White women (P < .001) compared with 0.2 years among Black women (P < .1).13 In addition, the best insights may be gleaned by examining racial/ethnic groups separately, given historical differences in school quality, employment, immigration patterns, family structure, and cause of death distributions.

The first step toward explaining the growth is identifying causes of death for which the gradient increased. A few studies have examined such trends among White women. Although informative, they largely focused on 2 time points, dichotomized education levels, a select group of causes, and a small age range.13,15,16 They found that roughly one quarter of the gradient’s growth during the 1990s among White women aged 45 to 84 years was because of deaths from lung cancer and chronic obstructive pulmonary disease.13 In a younger group of White women, aged 25 to 64 years, deaths from accidents contributed the largest percentage to the growth during the mid to late 1990s.15 However, many questions remain unanswered. For example, has the growth been constant or has it accelerated, decelerated, or plateaued? Which education groups were responsible for the growth, over which time periods, and for which causes of death? What do the trends suggest about future disparities? We have provided insights into these largely overlooked but important questions.

We performed a comprehensive analysis of trends in mortality by cause of death and education level from 1986 to 2006 among White women. We examined 4 time periods spanning 21 years to display nonlinearities in the trends. We also analyzed 4 education levels, including a bachelor’s degree, a critical improvement given the rising importance of a college degree for access to health-enhancing resources. We included women aged 45 to 84 years, which captures the majority of deaths among White women during the study period. We assessed trends in 12 causes of death for which we could reliably estimate death rates. Last, we calculated relative and absolute measures of the gradient because they may move in different directions.17

We used the public use National Health Interview Survey Linked Mortality File (NHIS-LMF), downloaded from the Minnesota Population Center.18 It links adults in the 1986 to 2004 annual cross-sectional waves of the NHIS with death records in the National Death Index through December 31, 2006. The link is mainly derived from a probabilistic matching algorithm that correctly classifies the vital status of 98.5% of eligible survey records.19


We converted the data into a person-year file that included non-Hispanic White women aged 45 to 84 years during 1986 to 2006. We first built a person-year file that aged all match-eligible women aged 18 years and older at interview by 1 year beginning with their interview year until their year of death or 2006 if they survived. Next, we retained person-year records for women who were aged 25 to 84 years at interview, and contributed person-years during 1986 to 2006, when they were aged 45 to 84 years. The first criterion helped ensure that most women had completed their education by the time of the survey; it also accounted for the top coding of ages at 85 years starting in 1997. For the second criterion, we set the lower limit at 45 years because there are few deaths at younger than 45 years in the NHIS-LMF (3% of deaths among White women occur before 45 years)20 and because trends among adults aged 25 to 44 years contributed little to the increasing gradient during this period.13 We set the upper limit at 84 years because mortality matches are not as reliable among women aged 85 years or older.21 The final sample contained 230 692 women and 42 435 deaths.

We defined 4 time periods: 1986–1994, 1995–1998, 1999–2002, and 2003–2006. Our goal was to assess as many periods as possible while ensuring that each contained enough observations to produce reliable death rates. Nine years were required for reliable estimates from the first period, after which 4-year periods were sufficient. The first period is longer because earlier years contained fewer observations in our person-year file, as fewer NHIS surveys had been conducted (and surviving women accumulated) to that point. We defined 4 education levels: 0 to 11 years, a high school diploma or general equivalency diploma, some college or an associate’s degree, and a bachelor’s degree or higher.

We could reliably estimate trends in 12 leading causes of death among adults aged 45 years and older.22 We collapsed some related causes (e.g., septicemia with infectious diseases) with relatively few deaths, and we disaggregated the large number of cancer deaths. We recoded the causes of death from the 113-category International Classification of Diseases, 10th Revision (Geneva, Switzerland: World Health Organization; 1980) codes into the following groups: heart disease (55–68), lung cancer (27), breast cancer (29), all other cancers (20–43 except 27 and 29), cerebrovascular disease (70), chronic lower respiratory diseases (CLRD; 83–86), accidental and violent deaths (114–129), diabetes mellitus (46), influenza and pneumonia (77–78), chronic liver diseases and cirrhosis (94–95), Alzheimer’s disease (52), infectious and parasitic diseases (1–18), and all other causes.22

Supplemental data to the online version of this article (available at http://www.ajph.org) show the number of deaths and person-years in our sample for each period and education level as well as the education distribution by period. During 1986 to 1994 approximately 25% of the sample had 0 to 11 years of education, 44% had a high school credential, 16% had some college, and 14% had a bachelor’s degree or higher. During 2003 to 2006, these estimates were 12%, 41%, 24%, and 23%, respectively.


For each period, we estimated death rates by education level. We standardized the rates to the age and education distribution of the 2000 US population of non-Hispanic White women by 10-year age groups (45–54, 55–64, 65–74, 75–84) using data from the Current Population Survey. We then used the standardized rates to obtain absolute and relative measures of the gradient for each period. The standardized rate difference (SRD) is the absolute difference between the death rates of women with 0 to 11 years of education and women with at least a bachelor’s degree, and the standardized rate ratio (SRR) is their ratio.

We examined secular trends in the gradient by plotting the standardized death rates across the 4 periods and by testing for linear trends in both inequality measures (SRD and ln[SRR]). To test for linear trends, we regressed each measure on period using ordinary least squares to obtain the P value for the period.23 The x-values for period were −5.5, 0.5, 4.5, and 8.5 to center period on 1996 and to account for the longer first period. Trends with P < .05 are statistically significant. Trends with P < .1 are noteworthy, particularly because the trends contain only 4 observations.

We conducted extensive additional sensitivity analyses. We have omitted the findings for parsimony, but they are available on request. In addition to the linear models for trends, we estimated models with a quadratic term for period, and we have noted when they fit the data better on the basis of the adjusted R2. We also calculated alternative indices of the gradient: the slope index of inequality (SII) and relative index of inequality (RII). We obtained these indices by regressing the death rate of each education group on the midpoint of the group’s cumulative proportionate distribution of the population from 0 to 1 (i.e., a ridit score) using weighted least squares.24,25 The SII is the difference between the predicted death rate of the lowest (ridit = 0) and highest (ridit = 1) possible education groups. The RII is their ratio. Although the SII and RII include all education levels and account for the changing education distribution over time, they impose a linear relationship between education and mortality, which may not be appropriate,26 and they can be adversely affected by highly unequal education group sizes. The SII and RII findings generally agreed with the results using SRD and SRR but were more conservative in some cases.

We have reported the SRD and SRR measures because they are intuitive, are easy to interpret, and do not impose a relationship between education and mortality. We also examined the trends using an alternative method for estimating mortality among noninstitutionalized populations in which estimates are conditional on surviving the year following the interview16; however, this adjustment had little effect on our results.

Figure 1 shows the age-standardized death rates for all-cause mortality by education level across the 4 periods. Death rates among women with 0 to 11 years of education were markedly higher than were those among other women, and the gap grew over time. It grew because mortality increased for women with 0 to 11 years of education during the first 3 periods, remained fairly constant among women with a high school credential or some college, and declined slightly among college-educated women. Table 1 lists the 2 measures of the gradient for each period and the P values for their linear trends. The tests confirm that the gradient in all-cause mortality increased. Over the 4 periods, the SRD grew significantly, from 1728 to 2267 per 100 000 (P < .05), whereas the SRR grew from 3.45 to 5.06 (P < .1).


TABLE 1— Trends in the Educational Gradient of Mortality by Cause of Death Among Non-Hispanic White Women Aged 45–84 Years: United States, 1986–2006

TABLE 1— Trends in the Educational Gradient of Mortality by Cause of Death Among Non-Hispanic White Women Aged 45–84 Years: United States, 1986–2006

Absolute Gradient, SRD
Relative Gradient, SRR
Cause of Death by Groupa1986–19941995–19981999–20022003–2006Pb1986–19941995–19981999–20022003–2006Pb
All causes1728197321032267<.0013.454.
Group 1
 Breast cancer7443033.381.081.811.591.85.17
 Other (not lung or breast) cancer212247223222.832.432.982.512.82.57
 Accidental and violent deaths12334830.331.433.103.771.96.58
 Influenza and pneumonia48424856.444.393.265.2410.12.26
 Chronic liver disease and cirrhosis15181617.573.065.372.933.27.9
 Infectious and parasitic diseases17433139.32.769.463.433.45.99
Group 2: heart disease749715667618.025.475.625.457.30.27
Group 3
 Lung cancer103160181257.032.613.273.957.09.06
 Cerebrovascular disease117125148149.053.493.833.865.51.15
Group 4
 Chronic lower respiratory disease117173193279.045.587.786.8410.33.13
 Alzheimer’s disease6123839.071.822.774.483.52.14
 Other causes261287375422.044.173.994.275.21.26

Note. SRD = standardized rate difference per 100 000; SRR = standardized rate ratio.

aGroup 1 exhibited a negligible increase in the gradient. In group 2, mortality risk declined for all women. In group 3, mortality risk increased for low-educated women and decreased for high-educated women. In group 4, mortality risk increased the most for low-educated women.

bP value for test of linear trend using ordinary least squares regression.

We next assessed trends in the gradient by cause of death. Table 1 and Figure 2 reveal that the trends differed across causes and that they could be categorized into 4 groups. In the first group, the increase in the gradient was negligible and not statistically significant. This group included nonlung cancers, accidental and violent deaths, influenza and pneumonia, chronic liver disease and cirrhosis, and infectious and parasitic diseases.

The second group contained only heart disease. Death rates from heart disease fell for all women. Table 1 shows that there was a statistically significant decrease in the absolute gradient. The linear trend for the relative gradient was not significant; however, the relative risk was much larger in the fourth period and the ancillary model with a quadratic term for period fit the data better than did the model with only a linear term.

The third group included lung cancer and cerebrovascular disease for which the mortality risks of low- and high-educated women moved in opposite directions. Figure 2 shows that death rates from lung cancer increased sharply for women with 0 to 11 years of education, changed little among women with a high school credential, and decreased among college-educated women. The absolute divergence was statistically significant (P < .05), whereas the relative divergence was smaller but noteworthy (P < .1). The pattern for cerebrovascular disease was similar but increases in death rates among the least-educated women were less pronounced, yielding a statistically significant increase in SRD (P < .05) but not in SRR (P > .1).

The fourth group included CLRD, diabetes, Alzheimer’s disease, and the residual category. In this group, low-educated women experienced a marked increase in mortality risk, women with a high school credential experienced a smaller increase, and college-educated women experienced little to no increase. For CLRD, death rates did not decline for college-educated women as they did from lung cancer, despite their common risk factor. The gradient for CLRD increased mainly because of the sharp rise in death rates among women with 0 to 11 years of education. The growth in the gradient was significant in absolute terms and trended upward in relative terms. For the other 3 causes, growth in the absolute gradient was significant for the residual category and noteworthy for diabetes and Alzheimer’s disease. Growth in the relative gradient was not statistically significant but trended upward for Alzheimer’s disease in the linear model and for diabetes and the residual category in the ancillary quadratic models.

Table 2 shows the contribution of each cause of death to the increase in the SRD between the first and fourth periods. Causes other than heart disease widened the SRD by 670, whereas heart disease reduced it by 131, providing a net increase of 539 per 100 000. Among causes whose SRD increased (i.e., all but heart disease), lung cancer (23%) and CLRD (24%) contributed the most. Other important contributors included diabetes (6%), Alzheimer’s disease (5%), and cerebrovascular disease (5%). The residual category contributed 24%. The number of deaths for causes in this category in the NHIS-LMF was too small to analyze.


TABLE 2— Contribution of Each Cause of Death to the Change in the Standardized Rate Difference Among Non-Hispanic White Women Aged 45–84 Years: United States, 1986–1994, 2003–2006

TABLE 2— Contribution of Each Cause of Death to the Change in the Standardized Rate Difference Among Non-Hispanic White Women Aged 45–84 Years: United States, 1986–1994, 2003–2006

Cause of DeathSRD in 2003–2006 minus SRD in 1986–1994Contribution of Cause of Death, %Contribution of Cause of Death Among Causes Whose SRD Increased, %
Heart disease−131−24.3
Lung cancer15428.623.0
Breast cancer264.83.9
Other (not lung or breast) cancer101.91.5
Cerebrovascular disease325.94.8
Chronic lower respiratory disease16230.124.2
Accidental and violent deaths183.32.7
Influenza and pneumonia81.51.2
Chronic liver disease and cirrhosis20.40.3
Alzheimer’s disease336.14.9
Infectious and parasitic diseases224.13.3
Other causes16129.924.0
All causes539100.0100.0

Note. SRD = standardized rate difference per 100 000. Because of rounding, totals may not equal 100%.

From 1986 to 2006, the educational gradient in all-cause mortality grew among White women aged 45 to 84 years in relative and absolute terms. In general, it grew because mortality decreased for college-educated women, remained stable for women with a high school credential or some college, and increased for women with 0 to 11 years of education. However, the contribution of low- and high-educated women to the growing gradient varied over the 21-year period. During 1986 to 2002 the growth was mainly fueled by increasing mortality among low-educated women and slight declines in mortality among college-educated women; during 2003 to 2006, it was mainly fueled by declining mortality among college-educated women. The unique contribution of different education levels in certain portions of the 21-year period is a new and important finding that researchers can use to tailor further investigations.

Trends in the gradient by cause of death exhibited 4 patterns: (1) for nonlung cancers, accidental and violent deaths, influenza and pneumonia, chronic liver disease and cirrhosis, and infectious and parasitic diseases, the gradient showed little or no increase; (2) for heart disease, mortality decreased for all women, which decreased the absolute gradient but increased the relative gradient; (3) for lung cancer and cerebrovascular disease, mortality among women with 0 to 11 years of education increased whereas it declined for college-educated women; and (4) for chronic lower respiratory disease, diabetes, Alzheimer’s disease, and the residual category, mortality increased markedly among women with 0 to 11 years of education, whereas the increase among other women was small or negligible.

It is interesting that the absolute gradient decreased for heart disease, the leading cause of death in the United States. Similarly, absolute inequalities in cardiovascular mortality have decreased among women in New Zealand23 and Norway.27 The trend may reflect improved medical treatments for heart disease. Between 1980 and 2000, 47% of the reduction in mortality from coronary heart disease in the United States was because of better treatment.28 The remainder was because of changes in risk factors, in which positive changes in the population (e.g., lower cholesterol) outweighed the negative (e.g., higher body mass index).

Our results indicate that smoking patterns played an important role in the gradient’s overall growth. Two causes of death for which smoking is a major risk factor—lung cancer and CLRD—explained 47% of overall growth between 1986 to 1994 and 2003 to 2006. This percentage is higher than previously reported. Meara et al.13 found that among White women aged 45 to 84 years, lung cancer and CLRD explained roughly one quarter of the growing mortality gap from 1990 to 2000. Our estimate may be higher because we studied a longer period and compared more extreme education levels. Miech et al.,29 analyzing a younger group of White women aged 40 to 64 years, also found that CLRD and cancers of the trachea, lung, and bronchus were among the largest contributors to the increase in the gradient between 1999 and 2007, in addition to accidental poisoning.

Our findings have several policy implications. The substantial mortality disadvantage of women with 0 to 11 years of education suggests that 1 priority should be raising the high school graduation rate to minimize the size of the most disadvantaged group. This goal is underscored by recent evidence that graduation rates (receiving a diploma, not a general equivalency diploma) are lower than commonly reported and have not increased over the past 40 years among women.30

Continued policy efforts to reduce smoking prevalence, such as Healthy People 2020,6 are also needed. However, these efforts must go beyond conventional tobacco controls focused on modifying individual behaviors.31,32 The prevalence of smoking has increased among low-educated women despite these strategies.33 The concentration of smoking among low-educated women argues for tailored tobacco controls in a comprehensive policy framework that addresses the social conditions that shape smoking behavior.31,34,35 Nuanced and ethical tobacco control strategies that consider the constellation of adverse conditions faced by low-educated women are needed.32 Furthermore, social and economic policies (e.g., housing, education, welfare, labor) should be leveraged to modify those conditions.31,32 For instance, disadvantaged women state that they smoke to relieve stress from the daily hassles of poverty, single parenting, and conflict-ridden relationships; because they are lonely and it provides a rare opportunity to socialize; and because they often have nothing to do outside the home such as employment or affordable recreational activities.34 As Graham et al. stated, “Social policies are tobacco control policies.”31(p11)

More broadly, social and economic policies are crucial for tackling the distal origins of mortality disparities and for preventing disparities in 1 proximal risk factor from shifting to another.36,37 Distal origins, such as employment,38 income,39 and marriage,39 have also diverged across education levels among women. Inequalities in proximal risk factors, such as obesity, have also grown.40 Social and economic policies may help narrow the growing mortality gap in cerebrovascular disease, diabetes, and Alzheimer’s disease, which reflect multiple social and behavioral risk factors and hinder disparities from developing in other causes of death.

If recent trends portend near future trends, the gradient will likely continue to grow in large part because of mortality declines among college-educated women. A college education has become increasingly important for health-enhancing resources such as income and marriage, especially among women.39 This may challenge efforts to close the longevity gap. Specifically, improving social conditions for the most disadvantaged may be necessary but insufficient if college-educated women’s longevity gains continue to outpace others’ gains.


One limitation of our study is the small number of deaths for individual causes in the residual category. Because they explained 24% of the gradient’s overall growth, future studies should use alternative data sources to disaggregate them. Another potential limitation is the accuracy of cause of death information on death certificates.41–43 The trends for lung cancer and CLRD, in particular, are so sharp that it is unlikely that such inaccuracies explain the trend. Another consideration is that mortality rates from the NHIS-LMF are somewhat lower than are rates from vital statistics because the survey excludes institutionalized adults. We tested the extent to which this might influence our results16 but found a negligible impact. A major strength of the NHIS-LMF is its respondent-provided education and link of respondents to death records. Most recent studies on trends by cause of death used vital statistics, in which an informant provides education level at time of death,13,15,29 with known inaccuracies.44

Another limitation of our study is the generalizability of results. Trends for other population subgroups should be examined separately because social conditions and cause of death distributions vary across subgroups. Finally, we note that compositional changes among the low-educated group (e.g., they may have become more negatively select) might have occurred and contributed to the gradient’s growth. However, other research has found that compositional changes contributed little to diverging trends in well-being and mortality.13,16,45,46 The fact that the widening gradient during 2003–2006 was driven mostly by the declining mortality of college-educated women is additional evidence against compositional changes as a primary explanation in our study.


Disparities in longevity among US White women continued to grow across 1986–2006 partially because of causes of death for which smoking is a major risk factor. The growing disparities occurred despite initiatives to eliminate them such as Healthy People. This does not imply that such initiatives were ineffective—only that forces working against them were stronger.13 Future research should identify these forces, such as economic policy, social integration, and labor market trends, and why they led to diverging smoking patterns and other deleterious health consequences across education levels.


The Robert Wood Johnson Foundation Health & Society Scholars program supported this study (to J. K. M.).

A previous version of this article was presented at the 2012 annual meeting of the Population Association of America, May 3–5, San Francisco, CA.

The authors are grateful for the suggestions of 3 reviewers.

Human Participant Protection

This study is exempt from institutional review board approval because data were publicly available and deidentified.


1. Braveman PA, Cubbin C, Egerter S, Williams DR, Pamuk E. Socioeconomic disparities in health in the United States: what the patterns tell us. Am J Public Health. 2010;100(suppl 1):S186S196. LinkGoogle Scholar
2. Hummer RA, Lariscy JT. Educational attainment and adult mortality. In: Rogers RG, Crimmins EM, eds. International Handbook of Adult Mortality. New York: Springer; 2011:241261. CrossrefGoogle Scholar
3. Mackenbach JP, Kunst AE, Groenhof F, et al. Socioeconomic inequalities in mortality among women and among men: an international study. Am J Public Health. 1999;89(12):18001806. LinkGoogle Scholar
4. Mirowsky J, Ross CE. Education, Social Status, and Health. New York: Aldine de Gruyter; 2003. Google Scholar
5. Kitagawa EM, Hauser PM. Differential Mortality in the United States: A Study in Socioeconomic Epidemiology. Cambridge, MA: Harvard University Press; 1973. CrossrefGoogle Scholar
6. US Department of Health and Human Services. Healthy People 2020. Available at: http://www.hhs.gov/opa/title-x-family-planning/initiatives-and-resources/healthy-people-2020. Accessed August 31, 2012. Google Scholar
7. Crimmins EM, Saito Y. Trends in healthy life expectancy in the United States, 1970–1990: gender, racial, and educational differences. Soc Sci Med. 2001;52(11):16291641. Crossref, MedlineGoogle Scholar
8. Feldman JJ, Makuc DM, Kleinman JC, Cornoni-Huntley J. National trends in educational differentials in mortality. Am J Epidemiol. 1989;129(5):919933. Crossref, MedlineGoogle Scholar
9. Lauderdale DS. Education and survival: birth cohort, period, and age effects. Demography. 2001;38(4):551561. Crossref, MedlineGoogle Scholar
10. Pappas G, Queen S, Hadden W, Fisher G. The increasing disparity in mortality between socioeconomic groups in the United States, 1960 and 1986. N Engl J Med. 1993;329(2):103109. Crossref, MedlineGoogle Scholar
11. Preston SH, Elo IT. Are educational differentials in adult mortality increasing in the United States?J Aging Health. 1995;7(4):476496. Crossref, MedlineGoogle Scholar
12. Rogot E, Sorlie PD, Johnson NJ. Life expectancy by employment status, income, and education in the National Longitudinal Mortality Study. Public Health Rep. 1992;107(4):457461. MedlineGoogle Scholar
13. Meara ER, Richards S, Cutler DM. The gap gets bigger: changes in mortality and life expectancy, by education, 1981–2000. Health Aff (Millwood). 2008;27(2):350360. Crossref, MedlineGoogle Scholar
14. Montez JK, Hummer RA, Hayward MD, Woo H, Rogers RG. Trends in the educational gradient of U.S. adult mortality from 1986 through 2006 by race, gender, and age group. Res Aging. 2011;33(2):145171. Crossref, MedlineGoogle Scholar
15. Jemal A, Ward E, Anderson RN, Murray T, Thun MJ. Widening of socioeconomic inequalities in U.S. death rates, 1993–2001. PLoS ONE. 2008;3(5):e2181. Crossref, MedlineGoogle Scholar
16. Cutler DM, Lange F, Meara E, Richards-Shubik S, Ruhme CJ. Rising educational gradients in mortality: the role of behavioral risk factors. J Health Econ. 2011;30(6):11741187. Crossref, MedlineGoogle Scholar
17. Mackenbach JP, Kunst AE. Measuring the magnitude of socio-economic inequalities in health: an overview of available measures illustrated with two examples from Europe. Soc Sci Med. 1997;44(6):757771. Crossref, MedlineGoogle Scholar
18. Minnesota Population Center. Integrated Health Interview Series: Version 5.0. Minneapolis: University of Minnesota; 2012. Available at: http://www.ihis.us. Accessed October 15, 2011. Google Scholar
19. National Center for Health Statistics, Office of Analysis and Epidemiology. The National Health Interview Survey (1986–2004) Linked Mortality Files, Mortality Follow-Up Through 2006: Matching Methodology. Hyattsville, MD; May 2009. Available at: http://www.cdc.gov/nchs/data/datalink/matching_methodology_nhis_final.pdf. Accessed April 25, 2010. Google Scholar
20. Anderson RN. United States life tables, 1997. Natl Vital Stat Rep. 1999;47(28):137. MedlineGoogle Scholar
21. Ingram DD, Lochner KA, Cox CS. Mortality experiences of the 1986–2000 National Health Interview Survey Linked Mortality Files participants. Vital Health Stat 2. 2008;(147):137. Google Scholar
22. Kochanek KD, Xu J, Murphy SL, Miniño AM, Kung H- C. Deaths: preliminary data for 2009. Natl Vital Stat Rep. 2011;59(4). DHHS publication PHS 2011–1120. MedlineGoogle Scholar
23. Blakely T, Tobias M, Atkinson J. Inequalities in mortality during and after restructuring of the New Zealand economy: repeated cohort studies. BMJ. 2008;336(7640):371375. Crossref, MedlineGoogle Scholar
24. Schalick LM, Hadden WC, Pamuk E, Navarro V, Pappas G. The widening gap in death rates among income groups in the United States from 1967 to 1986. Int J Health Serv. 2000;30(1):1326. Crossref, MedlineGoogle Scholar
25. Wagstaff A, Paci P, Doorslaer EV. On the measurement of inequalities in health. Soc Sci Med. 1991;33(5):545557. Crossref, MedlineGoogle Scholar
26. Montez JK, Hummer RA, Hayward MD. Educational attainment and adult mortality in the United States: a systematic assessment of functional form. Demography. 2012;49(1):315336. Crossref, MedlineGoogle Scholar
27. Strand BH, Grøholt E- K, Steingrímsdóttir ÓA, Blakely T, Graff-Iversen S, Næss Ø. Educational inequalities in mortality over four decades in Norway: prospective study of middle aged men and women followed for cause specific mortality, 1960–2000. BMJ. 2010;340:c654. Crossref, MedlineGoogle Scholar
28. Ford ES, Ajani UA, Croft JB, et al. Explaining the decrease in U.S. deaths from coronary disease, 1980–2000. N Engl J Med. 2007;356(23):23882398. Crossref, MedlineGoogle Scholar
29. Miech R, Pampel F, Kim J, Rogers R. The enduring association of education and mortality: the role of emergent and growing disparities. Am Sociol Rev. 2011;76(6):913934. Crossref, MedlineGoogle Scholar
30. Heckman JJ, LaFontaine PA. The American high school graduation rate: trends and levels. Rev Econ Stat. 2010;92(2):244262. Crossref, MedlineGoogle Scholar
31. Graham H, Inskip HM, Francis B, Harman J. Pathways of disadvantage and smoking careers: evidence and policy implications. J Epidemiol Community Health. 2006;60(suppl 2):712. MedlineGoogle Scholar
32. Greaves L, Jategaonkar N. Tobacco policies and vulnerable girls and women: toward a framework for gender sensitive policy development. J Epidemiol Community Health. 2006;60(suppl 2):5765. MedlineGoogle Scholar
33. Escobedo LG, Peddicord JP. Smoking prevalence in U.S. birth cohorts: the influence of gender and education. Am J Public Health. 1996;86(2):231236. LinkGoogle Scholar
34. Stewart MJ, Greaves L, Kushner KE, Letourneau NL, Spitzer DL, Boscoe M. Where there is smoke, there is stress: low-income women identify support needs and preferences for smoking reduction. Health Care Women Int. 2011;32(5):359383. Crossref, MedlineGoogle Scholar
35. Sorensen G, Barbeau E, Hunt MK, Emmons K. Reducing social disparities in tobacco use: a social-contextual model for reducing tobacco use among blue-collar workers. Am J Public Health. 2004;94(2):230239. LinkGoogle Scholar
36. House JS, Schoeni RF, Kaplan GA, Pollack H. The health effects of social and economic policy: the promise and challenge for research and policy. In: Schoeni RF, House JS, Kaplan GA, Pollack H, eds. Making Americans Healthier: Social and Economic Policy as Health Policy. New York: Russell Sage Foundation; 2008:126. Google Scholar
37. Link BG, Phelan JC. McKeown and the idea that social conditions are fundamental causes of disease. Am J Public Health. 2002;92(5):730732. LinkGoogle Scholar
38. Spain D, Bianchi SM. Balancing Act: Motherhood, Marriage, and Employment Among American Women. New York: Russell Sage Foundation; 1996. Google Scholar
39. DiPrete TA, Buchmann C. Gender-specific trends in the value of education and the emerging gender gap in college completion. Demography. 2006;43(1):124. Crossref, MedlineGoogle Scholar
40. Ogden CL, Lamb MM, Carroll MD, Flegal KM. Obesity and Socioeconomic Status in Adults: United States 1988–1994 and 2005–2008. Hyattsville, MD: National Center for Health Statistics; 2010. NCHS data brief no. 50. Google Scholar
41. Johansson LA, Westerling R, Rosenberg HM. Methodology of studies evaluating death certificate accuracy were flawed. J Clin Epidemiol. 2006;59(2):125131. Crossref, MedlineGoogle Scholar
42. Messite J, Stellman SD. Accuracy of death certificate completion: the need for formalized physician training. JAMA. 1996;275(10):794796. Crossref, MedlineGoogle Scholar
43. Moriyama IM. Problems in measurement of accuracy of cause-of-death statistics. Am J Public Health. 1989;79(10):13491350. LinkGoogle Scholar
44. Sorlie PD, Johnson NJ. Validity of education information on the death certificate. Epidemiology. 1996;7(4):437439. Crossref, MedlineGoogle Scholar
45. Blau FD. Trends in the well-being of American women, 1970–1995. J Econ Lit. 1998;36(1):112165. Google Scholar
46. Martikainen P, Blomgren J, Valkonen T. Change in the total and independent effects of education and occupational social class on mortality: analyses of all Finnish men and women in the period 1971–2000. J Epidemiol Community Health. 2007;61(6):499505. Crossref, MedlineGoogle Scholar


No related items




Jennifer Karas Montez, PhD, and Anna Zajacova, PhDJennifer Karas Montez is with the Harvard Center for Population and Development Studies, Harvard University, Cambridge MA. Anna Zajacova is with the Department of Sociology, University of Wyoming, Laramie. “Trends in Mortality Risk by Education Level and Cause of Death Among US White Women From 1986 to 2006”, American Journal of Public Health 103, no. 3 (March 1, 2013): pp. 473-479.


PMID: 23327260