Published online April 2, 2007
PEDIATRICS Vol. 119 No. 4 April 2007, pp. e928-e939 (doi:10.1542/peds.2005-2181)
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow P3Rs: Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when P3Rs are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow E-mail this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My File Cabinet
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via CrossRef
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Singh, G. K.
Right arrow Articles by Kogan, M. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Singh, G. K.
Right arrow Articles by Kogan, M. D.
Related Collections
Right arrow Office Practice

ARTICLE

Persistent Socioeconomic Disparities in Infant, Neonatal, and Postneonatal Mortality Rates in the United States, 1969–2001

Gopal K. Singh, PhD and Michael D. Kogan, PhD

Maternal and Child Health Bureau, Health Resources and Services Administration, US Department of Health and Human Services, Rockville, Maryland


   ABSTRACT
TOP
 ABSTRACT
METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
OBJECTIVES. This study examines changing patterns of inequalities in US infant, neonatal, and postneonatal mortality rates between 1969 and 2001 by area deprivation and maternal education.

METHODS. A deprivation index was linked to county vital records data to derive annual infant mortality rates by deprivation quintiles from 1969 to 2000. Rates by maternal education were computed for 1986, 1991, 1996, and 2001 using national linked birth/infant death files. Log-binomial regression was used to estimate relative risks of infant mortality by deprivation and time period. Cox regression was used to model overall and birth weight–specific infant mortality risks by maternal education after adjusting for covariates. Temporal disparities were summarized by log-linear regression and inequality indices.

RESULTS. Although absolute disparities have narrowed over time, relative socioeconomic disparities in infant mortality have increased since 1985. In 1985–1989, infants in the most deprived group had, respectively, 36% and 57% higher risks of neonatal and postneonatal mortality than infants in the least deprived group. The corresponding relative risks increased to 43% and 96% in 1995–2000. The adjusted risk of infant mortality was 22% higher in 1986 for mothers with <12 years of education than for those with ≥16 years of education, with the relative risk increasing to 41% in 2001. Disparities were greatest among normal birth weight infants, with education-specific relative risks of neonatal and postneonatal mortality increasing significantly between 1986 and 2001.

CONCLUSIONS. Dramatic declines in infant mortality among all of the socioeconomic groups during 1969–2001 represent a major public health success. However, substantial socioeconomic disparities persisted in both neonatal and postneonatal mortality. Relatively larger declines in infant and postneonatal mortality among higher socioeconomic groups have contributed to the widening gap in mortality since 1985. Persistent disparities in infant mortality may reflect increasing polarization among socioeconomic groups in material and social conditions, smoking during pregnancy, and health care services.

Key Words: infant mortality • neonatal • postneonatal • deprivation • maternal education • social inequality • low birth weight • trend • United States

Abbreviations: SES—socioeconomic status • NVSS—National Vital Statistics System • RR—relative risk • NBW—normal birth weight • VLBW—very low birth weight • MLBW—moderately low birth weight • LBW—low birth weight • SIDS—sudden infant death syndrome

Infant mortality rates have declined dramatically in the United States over the past 6 decades, from a rate of 47.0 infant deaths per 1000 live births in 1940 to 6.8 in 2003.1,2 Yet, racial, ethnic, and social class disparities in infant mortality remain marked. In 2003, black infants and infants born to women with less than a high school education experienced more than twice the mortality rate of white infants and infants born to women with a college degree, respectively.2

Existence of such large social disparities in current infant mortality rates is in sharp contrast to the goals of Healthy People 2000 and 2010 that were launched a decade and a half ago.3 Their primary goal has been to reduce and ultimately eliminate health inequalities among various segments of the US population, including disparities in infant mortality among ethnic and socioeconomic groups.3–5

Although trends in health inequalities by race/ethnicity, gender, and geographical area are analyzed routinely in the United States, the analysis of changes in the extent of health inequalities over time according to socioeconomic characteristics is far less common. The studies that have examined temporal social patterning in health have shown increasing socioeconomic inequalities in overall US mortality, life expectancy, and mortality from cardiovascular diseases and certain cancers.6–11 Previous studies have also shown substantial socioeconomic inequalities in infant mortality,2,12–15 although the extent to which such disparities have changed over time has received little attention.2,16 To our knowledge, no attempt has yet been made to conduct a systematic analysis of how US infant mortality rates have changed in recent decades in relation to individual socioeconomic status (SES) and area-based deprivation. The aim of this article is to examine changes in the extent of inequalities in US infant, neonatal, and postneonatal mortality rates between 1969 and 2001 by area socioeconomic deprivation and maternal education.


   METHODS
TOP
 ABSTRACT
 METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
To analyze temporal inequalities in US infant mortality by area deprivation, we used birth and infant death data from the National Vital Statistics System (NVSS) and the decennial census.1,17,18 Area socioeconomic patterns in infant mortality were derived indirectly by linking county-level SES data from the 1990 decennial census with the NVSS data via the common county Federal Information Processing Standards geocodes.1,17,18 We used a factor-based deprivation index that consisted of 17 census-based social indicators, which may be viewed as broadly representing educational opportunities, labor force skills, and economic and housing conditions in a given county. Selected indicators of education, occupation, wealth, income distribution, unemployment rate, poverty rate, and housing quality were used to construct the index.8–11 The factor loadings (correlations of indicators with the index) ranged from 0.92 for 150% of the poverty rate to 0.45 for household plumbing.8,10 The US deprivation index was constructed at the county level for the 1970, 1980, and 1990 censuses.8,10,11 Substantive and methodologic details underlying the construction of the US deprivation index are provided elsewhere.8,10

To analyze trends in infant mortality by deprivation, we used the weighted population quintile distribution of the 1990 deprivation index that classified all of the US counties into 5 groups of approximately equal population size. The groups thus created ranged from being the most deprived (first quintile) to the least disadvantaged (fifth quintile) population groups. The 1990 index was used to compute annual rates from 1969 through 2000. Specifically, county-specific infant, neonatal, and postneonatal deaths from 1969 through 2000 were obtained using the national mortality database,1 whereas county-specific live births, obtained from the natality component of NVSS, served as denominators for computing infant mortality rates.17 Each of the 3097 counties in the NVSS database was assigned 1 of the 5 deprivation categories. In the case of Alaska and Hawaii, state- rather than county-level data were used.

To estimate SES trends at the individual level, we computed infant mortality rates by maternal education using the national linked birth and infant death files for the 1986, 1991, 1996, and 2001 birth cohorts.19–22 Availability of the linked file since 1983 permits the analysis of educational trends in infant mortality over time. Maternal education was measured by the number of years of school completed and was grouped into 4 categories: <12, 12, 13 to 15, and ≥16 years. Other measures of individual SES, such as income, occupation, or employment status, are not available in the linked files.2,19–22

Cox proportional hazards regression model, fitted by the SAS PHREG procedure (SAS Institute, Inc, Cary, NC), was used to estimate relative risks (RRs; hazard ratios) of infant, neonatal, and postneontal mortality associated with maternal education before and after adjusting for such covariates as maternal age (≤19, 20–34, or ≥35 years), race/ethnicity (non-Hispanic white, black, Hispanic, Asian/Pacific Islander, or American Indian/Alaska Native), marital status (married or unmarried), live birth order (1, 2–3, or ≥4), infant sex, plurality (singleton or twin/multiple birth), prenatal care use (first trimester, second trimester, third trimester, or no care), smoking during pregnancy (unavailable for the 1986 cohort), gestational age (<33, 33–36, or ≥37 weeks), and birth weight (<1500, 1500–2499, or ≥2500 g).14,23,24 We preferred the Cox model to the logistic model, because the latter fails to take into account the varying survival times of infants dying in the first year of life.14 Hazards proportionality assumption of the Cox model was tested and confirmed by inspecting the plots of log-log survivor functions against survival time for various covariate categories, including those for maternal education.14,23,24 Survival times were measured in days. In estimating the risk of infant mortality, all of the live births surviving beyond the first year of life were treated as right-censored observations. Analyses for the 1986, 1991, 1996, and 2001 linked birth cohorts were based on 2906066, 4111059, 3892133, and 4026323 live births, respectively.19–22

Log-binomial regression, estimated by the SAS GENMOD procedure (SAS Institute, Inc, Cary, NC), was used to estimate RRs of infant, neonatal, and postneonatal mortality for each deprivation group and time period.25 Although rates for each deprivation group were computed annually from 1969 to 2000, we modeled infant mortality as a function of area deprivation for four 5-year and two 6-year time periods, 1969–1974, 1975–1979, 1980–1984, 1985–1989, 1990–1994, and 1995–2000, to reduce variability associated with annual rates and to provide more stable RR estimates. Other than RRs, disparities in infant mortality were also measured by the absolute difference in rates between the least deprived group (or the highest education group) and each of the other deprivation or education groups. Log-linear regression was used to calculate average annual exponential rates of decline in infant mortality for each deprivation group.12 An index of disparity, which approximated in relative terms the average deviation of the rates from the rate for the highest SES group, was used to summarize disparities over time across all of the deprivation and education groups.26,27 This relative mean deviation index of disparity was calculated as follows:

Formula
, where "Qri" is the rate for the ith quintile (i = 1,2,3,4,5), "Qr5" is the rate for the 5th quintile, and "I" is the number of groups or quintiles being compared. In addition, a measure of excess mortality, defined as the difference between the rates for the total population and the highest SES group, was computed to assess the extent of improvement in infant mortality if all of the groups were to have the rate of the highest SES group.28


   RESULTS
TOP
 ABSTRACT
 METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Trends in Area Socioeconomic Disparities in Infant, Neonatal, and Postneonatal Mortality
Figure 1 shows county socioeconomic gradients in annual infant mortality rates between 1969 and 2000. Although rates declined over time for all of the deprivation groups, more deprived groups had higher infant mortality than less deprived groups each year, and the gradients (rate ratios) seemed to be greater in the 1990s than in the 1980s. Infant mortality in the least through the most deprived groups declined at average annual rates of 3.50%, 3.39%, 3.24%, 3.21%, and 3.37%, respectively, between 1969 and 2000.


Figure 1
View larger version (52K):
[in this window]
[in a new window]

  		FIGURE 1 Trends in US infant, neonatal, and postneonatal mortality according to area (county) socioeconomic deprivation, 1969–2000. A, infant mortality rates; B, RR of infant mortality; C, neonatal mortality rates; D, RR of neonatal mortality; E, postneonatal mortality rates; F, RR of postneonatal mortality.

 
Figure 1 also presents trends in neonatal and postneonatal mortality rates. Neonatal mortality refers to infant deaths occurring in the first 27 days of life, whereas postneonatal mortality refers to infant deaths between 28 days and 1 year of age. All of the deprivation groups showed a decreasing trend in neonatal and postneonatal mortality between 1969 and 2000, although county socioeconomic gradients (rate ratios) were larger for postneonatal mortality than for neonatal mortality. Postneonatal mortality declined faster in the least deprived group than in the other deprivation groups, thus contributing to increasing rate ratios, particularly since 1985. In the least deprived group, the postneonatal mortality rate declined by 4.62% per year between 1985 and 2000, whereas it declined by just 2.64% per year in the most deprived group over the same time period. Neonatal mortality in all of the deprivation groups showed similar declines, with the rates declining at ~3.8% per year during 1969–2000.

Tables 1 and 2 summarize changing county socioeconomic differentials in infant mortality across different time periods. Between 1969 and 1984, both absolute and relative socioeconomic disparities in infant, neonatal, and postneonatal mortality declined. For example, the interquintile differences (Q1–Q5) in infant, neonatal, and postneonatal mortality rates declined from 8.3, 5.3, and 3.0 deaths per 1000 live births in 1969–1974 to 3.7, 1.9, and 1.8 in 1980–1984, respectively (Table 1). The RRs of infant, neonatal, and postneonatal mortality between the least and most deprived groups dropped significantly from 1.56, 1.46, and 1.86 in 1969–1974 to 1.38, 1.28, and 1.62 in 1980–1984, respectively (Table 2).


View this table:
[in this window]
[in a new window]

  		TABLE 1 Infant, Neonatal, and Postneontal Mortality Rates per 1000 Live Births by Area (County) Socioeconomic Deprivation, Interquintile (Qi–Q5) Rate Differences, Excess Mortality (%), and Index of Disparity: United States, 1969–2000

 

View this table:
[in this window]
[in a new window]

  		TABLE 2 RRs of Infant, Neonatal, and Postneonatal Mortality by Area (County) Socioeconomic Deprivation (SES) Index, (Derived From Log-Binomial Regression Models): United States, 1969–2000

 
Beginning with the 1985–1989 period, however, the RRs of infant, neonatal, and postneonatal mortality associated with area deprivation generally increased. In 1985–1989, infants in the most deprived group had, respectively, 36% and 57% higher risks of neonatal and postneonatal mortality than infants in the least deprived group. In 1995–2000, the deprivation gradient in infant mortality widened significantly, with infants in the most deprived group experiencing 43% and 96% higher RRs of neonatal and postneonatal mortality, respectively, than their least deprived counterparts (Table 2). The summary index of inequality in Table 1, which quantifies the magnitude of disparities across all of the deprivation groups, also indicates generally increasing socioeconomic disparities in infant mortality, particularly in postneonatal mortality since 1985. The relative overall disparity in infant and postneonatal mortality across deprivation groups, for example, widened from 23% and 33% in 1985–1989 to 31% and 52% in 1995–2000, respectively. The excess mortality estimates in Table 1 indicate that infant and postneonatal mortality rates in 1995–2000 would have declined by 23% and 34% if infants in the more deprived groups experienced mortality rates similar to those of the least deprived group.

Trends in Educational Disparities in Overall Infant, Neonatal, and Postneonatal Mortality
Table 3 shows changes in individual-level educational disparities in infant mortality between 1986 and 2001. Although area deprivation quintiles are substantively not equivalent to the 4 maternal educational strata, the absolute and relative socioeconomic disparities in infant mortality are indeed greater at the individual level than at the area level. All of the education groups showed substantial declines in mortality, with the mortality rate for infants born to mothers with <12 years of education decreasing by 45% from 15.32 in 1986 to 8.49 deaths per 1000 live births in 2001. The rate for infants born to mothers with ≥16 years of education fell by 38% from 6.85 in 1986 to 4.24 deaths per 1000 live births in 2001. Between 1986 and 2001, the neonatal mortality rates for infants born to mothers with <12 and ≥16 years of education decreased by 44% and 35% respectively, whereas the corresponding postneonatal mortality rates decreased by 46%.


View this table:
[in this window]
[in a new window]

  		TABLE 3 Infant, Neonatal, and Postneonatal Mortality Rates per 1000 Live Births by Maternal Education (Individual Socioeconomic Position), Excess Mortality, and Index of Disparity: United States, 1986–2001

 
The index of inequality, summarizing the magnitude of disparities across all 4 of the education groups, shows an increasing overall disparity in postneonatal mortality between 1986 and 2001, with the index score increasing from 90% in 1996% to 98% in 1996% to 123% in 2001. However, the index shows stable educational disparities in neonatal mortality over the same time period. The excess mortality estimates in Table 3 indicate that neonatal and postneonatal mortality in 2001 would have declined by 29% and 55% if the mortality rate for all of the educational groups were the same as the rate for infants born to mothers with a college degree.

Table 4 shows crude and adjusted RRs (hazard ratios) of mortality associated with maternal education for each birth cohort in 1986, 1991, 1996, and 2001. In 1986, the crude RRs indicate 125%, 48%, and 21% higher mortality risks for infants born to mothers with <12, 12, and 13 to 15 years of education compared with infants born to mothers with a college degree (model 1). Crude RRs of infant mortality in 2001 decreased for mothers with <12 years of education (RR: 2.00) but increased significantly for mothers with 12 and 13 to 15 years of education (RR: 1.75 and 1.44, respectively). Adjusting for maternal age, race/ethnicity, marital status, birth order, infant sex, plurality, and prenatal care in model 2 decreased the RRs in all 4 of the birth cohorts, but the adjusted relative educational disparities were greater in 2001 than in 1986. The inclusion of prenatal smoking, gestational age, and birth weight, along with all of the other covariates in model 2 further decreased the RRs associated with education, but they all remained statistically significant (model 3). Compared with ≥16 years of education, the fully adjusted risk of infant mortality associated with <12 and 12 years of education was 22% and 10% higher in 1986 and 41% and 28% higher in 2001, respectively, with the increase in the RRs being statistically significantly during 1986–2001.


View this table:
[in this window]
[in a new window]

  		TABLE 4 Multivariate Cox Regressions Showing Crude and Adjusted Educational Differentials in Infant, Neonatal, and Postneonatal Mortality: United States, 1986–2001

 
Although the crude RRs show a consistent narrowing of the gap in neonatal mortality between 1986 and 2001 (model 1), the adjustment for all of the covariates, including gestational age and birth weight in model 3, show no statistically significant difference between educational groups in 1986 and 1991 but a slightly increased RR of neonatal death associated with lower education in 1996 and 2001. The adjusted RRs associated with education increased significantly between 1986 and 2001 (model 3).

The crude RRs show particularly marked differentials in postneonatal mortality by maternal education, with the risk being ≥3.3 times greater in each cohort for mothers with <12 years of education than for those with ≥16 years of education (model 1). Even after the adjustment for all of the covariates, lower maternal education was associated with substantially increased postneonatal mortality, with the RRs associated with high school education or less significantly higher in 2001 than in 1986 (model 3).

Trends in Educational Disparities in Birth Weight–specific Infant Mortality
Although the RRs in Table 4 adjust for educational differences in the birth weight composition, they do not reflect the extent to which educational differences in infant mortality vary across different birth weight strata. Birth weight–specific analyses should reveal whether educational disparities in infant mortality are more pronounced among the low or normal birth weight (NBW) infants and whether such disparities have changed over time. Infant, neonatal, and postneonatal mortality rates show dramatic declines between 1986 and 2001 for the 3 birth weight strata (Table 5). The rate of decline in neonatal mortality during 1986–2001, however, was greater for very low birth weight (VLBW), moderately low birth weight (MLBW), and NBW infants in the highest educational category than in the lower educational categories. The adjusted RRs indicate significant educational differences in neonatal mortality risks of VLBW infants in 2001 but not in 1986. However, the adjusted RRs of neonatal mortality associated with education did not increase significantly between 1986 and 2001. Lower education was associated with a significantly decreased neonatal mortality risk for MLBW infants in 1986, which is consistent with sometimes higher survival of socially disadvantaged infants at low birth weights (LBWs). However, this survival advantage disappeared by 2001. Among the NBW infants, the adjusted RRs of neonatal mortality associated with <12 years of education increased from 1.30 in 1986 to 1.62 in 2001 (Table 5).


View this table:
[in this window]
[in a new window]

  		TABLE 5 Birth Weight–Specific Mortality Rates and Cox Regressions Showing Crude and Adjusted Educational Differentials in US Infant, Neonatal, and Postneonatal Mortality, 1986 and 2001

 
The rate of decline in postneonatal mortality during 1986–2001 was markedly greater for VLBW and MLBW infants in the highest educational category than in the lower educational categories. Although maternal education, adjusted for covariates, was not significantly related to postneonatal mortality among VLBW and MLBW infants in 1986, it was significantly associated with increased mortality in 2001 (RR: 1.53 and 1.46 for <12 years of education, respectively). Among the NBW infants, the adjusted RRs of postneonatal mortality associated with <12 and 12 years of education increased from 1.97 and 1.35 in 1986 to 2.65 and 2.04, respectively, in 2001. The adjusted infant mortality risk associated lower education increased significantly between 1986 and 2001 for each birth weight stratum.


   DISCUSSION
TOP
 ABSTRACT
 METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
In this study, using a comprehensive area-based deprivation index and an individual SES measure of maternal education, we have analyzed the extent to which socioeconomic patterns in US infant, neonatal, and postneonatal mortality have changed in the last 3 decades. Infant mortality declined dramatically among all of the SES groups, and the absolute socioeconomic disparities generally narrowed in both neonatal and postneonatal mortality during 1969–2001, which should be viewed as a major public health success. To illustrate the public health impact of such declines, consider the following scenario based on the data in Table 1. If all US infants experienced the mortality rate of infants in the highest SES group, there would have been ~87434 fewer infant deaths during 1969–1974, compared with 39867 fewer deaths during 1995–2000. Moreover, the absolute change during 1969–2000 also favored the most deprived group (a decline of 14 infant deaths per 1000 live births) compared with the least deprived group (a reduction of 9 deaths). A similar calculation shows more infants saved per 1000 live births during 1986–2001 for mothers with <12 years of education than for mothers with ≥16 years of education.

The success in reducing absolute disparities in infant mortality must be weighed against changes in the relative measures of disparity, which are particularly important for assessing the progress of various social groups relative to the best-off group toward reaching a health policy goal, such as the elimination of health disparities.5,26,27,29 Despite the impressive overall reductions in mortality over the long term, large socioeconomic inequalities in infant mortality persisted throughout the study period. Interestingly, socioeconomic disparities in infant mortality declined markedly between 1969 and 1984 both in absolute and relative terms. However, the relative socioeconomic disparities in infant and postneonatal mortality, as measured by the summary inequality indices and crude and adjusted RRs, increased substantially since 1985. Improvements in infant mortality would be substantial if infants in the lower SES groups experienced mortality rates similar to those of the highest SES group.

The pattern of continuing socioeconomic disparities in US infant mortality over the past 3 decades is consistent with the persistent occupational social class inequalities in infant mortality in the United Kingdom between 1975 and 2001.30–32 Moreover, the recent pattern of increasing socioeconomic inequalities in US infant mortality coincides with increasing inequalities in US life expectancy and mortality among working-age adults and the elderly.8–10 However, the recent patterns in infant mortality differed from those for the United States during 1930–1960.33 Kitagawa and Hauser33 showed substantial gradients in infant mortality by census tract SES for the Chicago, IL, area in 1930, 1940, and 1960. Although infant mortality among white infants generally decreased with increasing SES levels in each period, the difference in infant mortality between the lowest and highest SES groups diminished markedly between 1930 and 1960. Our national findings also differed from those of an ecological study of birth outcomes in 30 health districts of New York City, NY, which showed a significantly decreasing income gradient in infant mortality during 1988–2001; the relative rate ratio of infant mortality between the lowest and highest income quartiles decreased from 1.66 in 1988 to 1.25 in 2001.16 A recent Canadian study also showed a narrowing of the SES gap, both in terms of rate differences and rate ratios.28 Infant mortality in urban Canada in 1971 was 97% higher in the poorest quintile compared with the richest quintile. However, by 1996, the inequality between the richest and poorest quintiles had diminished to 61%.

Changing socioeconomic inequalities in infant mortality over time may reflect temporal inequalities in the material and social living conditions between SES groups, both in absolute and relative terms. Absolute differences between deprivation groups in income, wealth, assets (as measured by homeownership and median home value), poverty, unemployment, female-headed households, and health care personnel per capita widened between 1970 and 2000 and were more pronounced in the 1990s than in the 1970s and 1980s.8–10 Similarly, the gap in relative income disparity between deprivation groups increased markedly.8–10 Income differentials between educational strata also widened considerably. The mean earnings differential between those without a high school diploma and those with more than a bachelor's education increased from $10527 in 1975 to $54076 in 2001, and the ratio of mean earnings increased from 2.7 in 1975 to 3.9 in 2001.34

Temporal socioeconomic inequalities in infant mortality may also be related to geographical and SES inequalities in the other important social, behavioral, and health care factors known to be associated with infant mortality. The key risk factors, smoking during pregnancy, delayed or no prenatal care, and lack of health care coverage, all strongly associated with increased infant mortality risks, vary substantially by county deprivation levels and maternal education.2,35 Although rates of smoking during pregnancy have declined for all educational groups, the pace of decline has been faster for women in the higher educational strata.38 As a result, educational disparities in smoking during pregnancy increased during the last decade. In 1992, women with less than a high school education were 7.1 times more likely to smoke during pregnancy than those with a college degree.37 In 2002, this differential increased to 11.8 times.17 Although long-term trend data on prenatal smoking are lacking, they are expected to be similar to the trend of consistently increasing educational disparities in smoking prevalence in the general female population.36 SES disparities in prenatal care use have also persisted in the past 2 decades.17,38

Declining rates of teenage childbearing (births to mothers ≤20 years) and improved survival of LBW infants may also have contributed to the declines in overall US infant mortality. As shown here and based on other linked data, SES disparities have persisted despite substantial reductions in the incidence of teenage childbearing and in mortality of VLBW and MLBW infants over the long term.36 Socioeconomic differences in maternal health status, particularly such medical conditions as chronic and pregnancy-related hypertension, heart and lung disease, diabetes (including gestational diabetes), and anemia,22 may also have contributed to temporal disparities in infant mortality.

Other than reductions in smoking during pregnancy and improved medical care, the introduction of Medicaid in 1965, a federally funded public insurance program for poor and deprived children and socially disadvantaged groups, may have been responsible for declines in overall infant mortality and narrowing of the socioeconomic gap during 1969–1984.36 During the 1960s and 1970s, mortality from such leading causes of infant death as congenital anomalies, LBW, pneumonia and influenza, and infections all fell dramatically, possibly because of improved perinatal and neonatal medical care.12,36

What might explain increasing socioeconomic disparities in US infant mortality since 1985? Controlling for such risk factors as prenatal smoking, prenatal care, preterm birth, and LBW partly accounted for the observed SES differences in infant mortality. However, significant differences remained, particularly for the more recent cohorts, indicating that maternal education has become an increasingly important predictor of infant survival in both the neonatal and postneonatal periods.

Birth weight–specific analyses presented here may shed some light on the potential factors responsible for the increasing disparity in infant mortality. Substantial absolute declines in the mortality of VLBW and MLW infants between 1986 and 2001 across all of the education groups undoubtedly reflect the prominent role of improved access and use of obstetric and neonatal intensive care in reducing mortality.36,40 However, the relatively slower declines in mortality among LBW infants from the lower SES groups have resulted in a widening of the educational gradient in neonatal and postneonatal mortality over time, thus suggesting a continuing gap in access to high-quality neonatal and infant health care across various SES groups.

Educational inequalities in total infant mortality are driven largely by educational gradients in mortality among NBW infants (because they compose >90% of all births), which not only have increased over time but have become quite steep. This may reflect an increasingly important role of social and environmental influences on infant mortality risks in the United States.12,14 Cause-specific mortality analyses are needed to identify more precisely what factors might have been responsible for increasing or persistent socioeconomic inequalities in infant mortality. A recent US study showed substantial and persistent area SES disparities in infant mortality from such major causes of death as congenital anomalies, sudden infant death syndrome (SIDS), LBW, and injuries.41 Although birth defects mortality declined impressively during 1969–2000 for all groups, the deprivation gradient widened substantially since 1985, a temporal pattern consistent with that shown here in overall infant mortality. SIDS mortality showed much steeper deprivation gradients in the 1990s than in the 1980s. Infant mortality because of LBW rose significantly between 1986 and 1998 for more deprived groups but remained stable for less deprived groups, leading to increasing SES disparities in mortality from this cause.41

When first introduced in the general population, specific behavioral and public health interventions and advances in medical care may lead to an improved overall population health but may very well increase health disparities, particularly in the short run.29,42 Such is the case with SIDS mortality in the United States, which declined dramatically in the 1990s but saw social inequalities widen between 1989 and 1998 despite the introduction in June 1994 of the Back to Sleep campaign.43 Two recent Norwegian studies also found an increasing educational inequality in postneonatal and SIDS mortality in Norway between 1969 and 1995.44,45

Differentials in infant mortality between deprivation groups shown here are probably underestimated, because we used counties, rather than smaller and more homogeneous geographic areas, such as census tracts or neighborhoods, to define deprivation groups. Many US counties are large geographic areas or population units with substantial socioeconomic heterogeneity.8–11 Unfortunately, national vital records data do not identify geographic areas smaller than counties for confidentiality protection of individual information on birth and death certificates.8–10 Vital records for some states, such as those for California, Massachusetts, North Carolina, Virginia, and Kansas, may contain census tract or block group geocodes for the 1990 and 2000 censuses that could allow investigators to define deprivation levels at the neighborhood level, resulting in less misclassification of area-based SES. However, because the 1970 and 1980 censuses were not fully tracted, temporal analyses of census tract-based disparities in infant mortality, unlike our county-based analysis, may not extend as far back as 1969. The 1990 deprivation index has been shown to provide a stable socioeconomic classification of counties over time,8–11 and the use of the 1980 index produced infant mortality trends similar to those based on the 1990 index. The deprivation indices for the 1970, 1980, and 1990 censuses were highly correlated.8–11 The correlation of the 1990 index with the 1970 and 1980 indices was 0.90 and 0.94, respectively. Thus, the sole use of the 1990 index to construct deprivation quintiles for the entire study period is unlikely to have caused any substantial area misclassification, and the general trend of persistent inequalities in infant mortality holds regardless of which index is used.

Although this study focused on trends in overall socioeconomic disparities, it is important to note that both area- and individual-level SES gradients in infant mortality do vary by race/ethnicity. Our analysis revealed that the area deprivation gradients were more pronounced and consistent for whites than for blacks (data not shown). Disparities in infant mortality by maternal education were also greater for whites than for blacks, Hispanics, and Asian/Pacific Islanders.2 Changes in the extent of SES disparities in infant mortality over time for major racial/ethnic groups will be explored more fully in a subsequent article.

Behavioral and health policy interventions (eg, smoking reduction, tobacco regulation and advertising, improving access to and use of early and comprehensive prenatal care, and universal health care coverage) have the potential to reduce socioeconomic inequalities in infant mortality in both absolute and relative terms. Large socioeconomic disparities in infant mortality remain one of the primary reasons for the continued unfavorable international standing of the US infant mortality rate when compared with rates for the other industrialized countries; the US ranking dropped from being the 11th best in 1960 to only the 28th best in 2002.36 Continuing SES disparities in infant mortality may also prove to be a major obstacle in reaching the Healthy People 2010 goals. As shown here, the infant mortality rates for the 3 most deprived groups in 2000 (representing ~60% of the total population) and for those with high school education or less in 2001 (representing 53% of all mothers) were all greater than the year 2000 national target of 7 deaths per 1000 live births.3 It will be a formidable challenge to meet the 2010 target of 4.5 deaths per 1000 live births, particularly for infants in the most disadvantaged SES groups, of which the mortality rates would have to be reduced by 50% during this decade.3,4


   FOOTNOTES
 
Accepted Oct 9, 2006.

Address correspondence to Gopal K. Singh, PhD, Maternal and Child Health Bureau, Health Resources and Services Administration, 5600 Fishers La, Room 18-41, Rockville, MD 20857. E-mail: gsingh{at}hrsa.gov

The views expressed are the authors' and not necessarily those of the Health Resources and Services Administration or the US Department of Health and Human Services.

The authors have indicated they have no financial relationships relevant to this article to disclose.


   REFERENCES
TOP
 ABSTRACT
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Hoyert DL, Heron MP, Murphy SL, Kung HC. Deaths: final data for 2003. Natl Vital Stat Rep. 2006;54(13) :1 –120
  2. Mathews TJ, MacDorman MF. Infant mortality statistics from the 2003 period linked birth/infant death data set. Natl Vital Stat Rep. 2006;54(16) :1 –32
  3. US Department of Health and Human Services. Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC: US Government Printing Office; 1990
  4. US Department of Health and Human Services. Tracking Healthy People 2010. Washington, DC: US Government Printing Office; 2000
  5. US Department of Health and Human Services. Healthy People 2010: Understanding and Improving Health. 2nd ed. Washington, DC: US Government Printing Office; 2000
  6. 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 :103 –109.[Abstract/Free Full Text]
  7. Feldman JJ, Makuc DM, Kleinman JC, Cornoni-Huntley J. National trends in educational differentials in mortality. Am J Epidemiol. 1989;129 :919 –933[Abstract/Free Full Text]
  8. Singh GK. Area deprivation and widening inequalities in US mortality, 1969–1998. Am J Public Health. 2004;93 :1137 –1143[ISI]
  9. Singh GK, Siahpush M. Widening socioeconomic inequalities in US life expectancy, 1980–2000. Int J Epidemiol. 2006;35 :969 –979[Abstract/Free Full Text]
  10. Singh GK, Siahpush M. Increasing inequalities in all-cause and cardiovascular mortality among US adults aged 25–64 years by area socioeconomic status, 1969–1998. Int J Epidemiol. 2002;31 :600 –613[Abstract/Free Full Text]
  11. Singh GK, Miller BA, Hankey BF, Feuer EJ, Pickle LW. Changing area socioeconomic patterns in U.S. cancer mortality, 1950–1998: part I — all cancers among men. J Natl Cancer Inst. 2002;94 :904 –915[Abstract/Free Full Text]
  12. Singh GK, Yu SM. Infant mortality in the United States: trends, differentials, and projections, 1950 through 2010. Am J Public Health. 1995;85 :957 –964[Abstract/Free Full Text]
  13. MacMahon B, Kovar MG, Feldman JJ. Infant mortality rates: socioeconomic factors, United States. Vital Health Stat. 1972;22(14) :1 –68
  14. Singh GK, Kposowa AJ. A comparative analysis of infant mortality in major Ohio cities: significance of socio-biological factors. Appl Behav Sci Rev. 1994;2 :77 –94[CrossRef][Medline]
  15. Singh GK, Wilkinson AV, Song FF, et al. Health and Social Factors in Kansas: A Data and Chartbook, 1997–98. Lawrence, KS: Allen Press; 1998
  16. Racine AD, Joyce TJ. The disappearing income gradient in New York City birth outcomes: thirteen years of convergence from 1988 to 2001. Pediatrics 2004;114 (1). Available at: www.pediatrics.org/cgi/content/full/114/1/e51
  17. Martin JA, Hamilton BE, Sutton PD, Ventura SJ, Menacker F, Munson ML. Births: final data for 2002. Natl Vital Stat Rep. 2003;52(10) :1 –113
  18. US Bureau of the Census. Census of Population and Housing, 1990. Summary Tape File 3A on CD-ROM. Washington, DC: US Bureau of the Census; 1992
  19. National Center for Health Statistics. Linked Birth/Infant Death Data Set, 1986 Birth Cohort. CD-ROM Series 20, No. 3. Hyattsville, MD: Public Health Service; 1995
  20. National Center for Health Statistics. Linked Birth/Infant Death Data Set, 1991 Birth Cohort. CD-ROM Series 20, No. 7. Hyattsville, MD: Public Health Service; 1996
  21. National Center for Health Statistics. Linked Birth/Infant Death Data Set, 1996 Birth Cohort. CD-ROM Series 20, No. 14a. Hyattsville, MD: Public Health Service; 1999
  22. National Center for Health Statistics. Linked Birth/Infant Death Data Set, 2001 Birth Cohort. CD-ROM Series 20, No. 19a. Hyattsville, MD: Public Health Service; 2005
  23. SAS Institute, Inc. SAS/STAT User's Guide, Version 9.1: The PHREG Procedure. Cary, NC: SAS Institute, Inc; 2003
  24. Allison, PD. Survival Analysis Using the SAS System: A Practical Guide. Cary, NC: SAS Institute, Inc; 1995
  25. SAS Institute, Inc. SAS/STAT User's Guide, Version 9.1: The GENMOD Procedure. Cary, NC: SAS Institute, Inc; 2003
  26. Percey JN, Keppel KG. A summary measure of health disparity. Public Health Rep. 2002;117 :273 –280[ISI][Medline]
  27. Harper S, Lynch J. Methods for measuring cancer disparities: using data relevant to Healthy People 2010 cancer-related objectives. In: NCI Cancer Surveillance Monograph Series, No. 6. Bethesda, MD: National Cancer Institute, 2005; NIH Pub. No. 05–5777
  28. Wilkins R, Bethelot JM, Ng E. Trends in mortality by neighborhood income in urban Canada from 1971 to 1996. Health Rep. 2002;13(suppl) :45 –72
  29. Low A, Low A. Importance of relative measures in policy on health inequalities. BMJ. 2006;332 :967 –969[Free Full Text]
  30. Whitehead M. The Health Divide. London, England: Penguin Books; 1992
  31. Rowan S. Implications of changes in the United Kingdom social and occupational classifications in 2001 on infant mortality statistics. Health Stat Q. 2003;17 :33 –40
  32. Dummer TJB, Parker L. Changing socioeconomic inequality in infant mortality in Cumbria. Arc Dis Child. 2005;90 :157 –162[Abstract/Free Full Text]
  33. Kitagawa EM, Hauser PM. Differential Mortality in the United States: A Study in Socioeconomic Epidemiology. Cambridge, MA: Harvard University Press; 1973
  34. Walt CD, Proctor BD, Mills RJ. Income, Poverty, and Health Insurance Coverage in the United States: 2003. Current Population Reports, P60-226. Washington, DC: US Census Bureau; 2004
  35. Health Resources and Services Administration. Area Resource File, February 2003 Release. Rockville, MD; 2004
  36. National Center for Health Statistics. Health, United States, 2005 With Chartbook on Trends in the Health of Americans. Hyattsville, MD: US Department of Health and Human Services; 2005
  37. Ventura SJ, Martin JA, Taffel SM, Mathews, Clarke SC. Advance report of final natality statistics, 1992. Monthly Vital Stat Rep. 2005;43 (5 suppl):1–88
  38. National Center for Health Statistics. Health, United States, 1998 With Socioeconomic and Health Chartbook. Hyattsville, MD: US Department of Health and Human Services; 1998
  39. Singh GK, Yu SM. US childhood mortality, 1950 through 1993: trends and socioeconomic differentials. Am J Public Health. 1996;86 :505 –512[Abstract/Free Full Text]
  40. Kramer MS, Barros FC, Demissie K, et al. Does reducing infant mortality depend on preventing low birthweight? An analysis of temporal trends in the Americas. Pediatr Perinat Epidemiol. 2005;19 :445 –451[CrossRef][ISI][Medline]
  41. Singh GK. Trends in area- and individual-level socioeconomic inequalities in US infant mortality, all-cause and cause-specific mortality, 1970–2000. Presented to the Special Joint Session on Population Health and SES Disparities in the US and Canada at the Annual Meeting of the Population Association of America; April 2005; Philadelphia, PA
  42. Mechanic D. Disadvantage, inequality, and social policy. Health Aff (Millwood). 2002;21 :48 –59[Abstract/Free Full Text]
  43. Pickett KE, Luo Y, Lauderdale DS. Widening social inequalities in risk for sudden infant death syndrome. Am J Public Health. 2005;95 :1976 –1981[Abstract/Free Full Text]
  44. Arntzen A, Samuelsen SO, Bakketeig LS, Stoltenberg C. Socioeconomic status and risk of infant death: a population-based study of trends in Norway, 1967–1998. Int J Epidemiol. 2004;33 :279 –288[Abstract/Free Full Text]
  45. Arntzen A, Samuelsen SO, Daltveit AK, Stoltenberg C. Post-neonatal mortality in Norway, 1969–1995: a cause-specific analysis. Int J Epidemiol. 2006;35 :1083 –1089[Abstract/Free Full Text]
PEDIATRICS (ISSN 1098-4275). ©2007 by the American Academy of Pediatrics



This article has been cited by other articles:


Home page
 Int J EpidemiolHome page
C. V Ananth, S. Liu, K S Joseph, M. S Kramer, and for the Fetal and Infant Health Study Group of the
A comparison of foetal and infant mortality in the United States and Canada
Int. J. Epidemiol., September 20, 2008; (2008) dyn194v1.
[Abstract] [Full Text] [PDF]

Home page
 JWatch Women's HealthHome page
Socioeconomic Gaps in U.S. Infant Mortality
Journal Watch Women's Health, May 17, 2007; 2007(517): 5 - 5.
[Full Text]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow P3Rs: Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when P3Rs are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow E-mail this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My File Cabinet
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via CrossRef
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Singh, G. K.
Right arrow Articles by Kogan, M. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Singh, G. K.
Right arrow Articles by Kogan, M. D.
Related Collections
Right arrow Office Practice