outcomes fetal

advertisement
At the Starting Gate: Pregnancy as a Mediator of Black-White Inequality
Scott Winship, Harvard University
draft of 4/24/2002
Introduction
Americans' views on poverty and inequality are perhaps best defined by their
views on the extent to which opportunity is equally distributed and the extent to which
individuals are responsible for their own fates. Along these dimensions, conservatives
generally believe that equal opportunity exists, and so it follows that individuals have no
one to blame for their outcomes but themselves. Liberals, on the other hand, tend to
believe that various classes of people face barriers in pursuing their desired goals and
even that one's goals themselves are limited by one's social environment. Justice, then,
dictates that the state intervene to equalize opportunities and social environments.
The most cursory review of social scientific evidence, newspaper headlines, or
social life in general leaves the strongest version of the conservative position untenable.
Everyone can recognize, for instance, many (but not all) advantages and disadvantages
they have faced in their own lives. Nowhere is the evidence for disadvantage more
obvious -- nor, in some quarters, more denied -- than in the case of inequality between
Afro-Americans and Euro-Americans. Liberal egalitarian theories of black-white
inequality recognize the important role of social environments in producing ethnoracial
inequality, focusing in particular on the role of institutional actors such as schools and
labor markets. Such emphasis is not misguided, but explanations that focus on these
institutions are undercut by evidence of early inequalities between black and white
children and even infants.
1
In fact, black-white inequality within a given cohort has its roots in prenatal
inequality. This paper reviews the evidence that prenatal (intrauterine) environments are
unequal between black and white fetuses, discusses trends in perinatal inequality, and
makes the case that perinatal inequality within a cohort has consequences for
socioeconomic inequality in adulthood. It then considers the (still mysterious) causes of
perinatal inequality. Relying mainly on ethnoracial differences in low birth weight rates,
I explore the extent to which the disadvantages that Afro-American women suffer -- be
they related to class, status as a "racial" minority, or some other combination of factors -are passed on to their fetuses during pregnancy. In particular, I consider two potential
sources of ethnoracial inequality that have been neglected in empirical work in this area:
1) the extent to which the lower status position afforded blacks in the U.S. produces less
favorable birth outcomes via its effect on maternal psychological well-being, and 2) the
extent to which past black-white socioeconomic inequality has lingering biological
effects on birth outcomes in subsequent generations. Finally, I conduct my own original
analysis using a nationally representative dataset to explore the causes of black-white
inequality in low birthweight rates in greater detail than previous studies have considered.
Ethnoracial Inequality in Perinatal Outcomes
Figure 1 provides a summary of current ethnoracial inequality in perinatal
outcomes. The indicators shown represent outcomes related to fetal and infant health
throughout the perinatal period. All figures are from the National Center for Health
Statistics and based on birth and death certificates. See the Appendix for detailed
information on official vital statistics measures. Fetal mortality rates are from 1998 and
2
represent the number of fetal losses of twenty weeks or more gestation as a proportion of
births and fetal losses combined.1 Induced abortions are not incorporated into the
measure. Data on fetal loss rates are unavailable for groups other than blacks and whites,
and unlike the other measures, Hispanics are included within these two groups. Preterm
birth rates indicate the percentage of infants born premature (births of less than thirtyseven completed weeks of gestation). Figures are from 2000. Preterm births account for
a large share of low birthweight births, or births of less than 2,500 grams (five pounds,
eight ounces). Epidemiologists further distinguish births of very low birthweight (less
than 1,500 grams, or three pounds, four ounces). The birthweight figures are also from
2000. Finally, Figure 1 shows neonatal and infant mortality rates from 1999. Neonatal
mortality refers to death within 28 days of birth, while infant mortality to death within
one year of birth.
The obvious feature of Figure 1 is the extent to which black outcomes diverge
from those of other groups. The prevalence of most poor perinatal outcomes among
blacks is twice that among whites. Other minority groups' outcomes more closely
resemble those of whites rather than blacks. American Indian outcomes are noticeably
worse than those of other groups on some indicators but are never more like black
outcomes than white outcomes. I discuss the possible causes for these differences below.
For now, I simply hint at the complexity of the question by presenting Figure 2, which
shows ethnoracial differences in low birthweight rates stratified by education level.
To provide some additional context within which to interpret these figures, it is
useful to consider trends in perinatal outcomes. Figures 3 through 7 provide these trends
1
The vast majority of fetal losses occur prior to the twentieth week of gestation. Under federal reporting
requirements, states need not collect data on these earlier losses. Indeed, some of these losses involve
3
for blacks and whites, this time including Hispanics in each group. Data availability for
other ethnoracial groups is limited prior to 1990. Each figure also shows how the
absolute gap between blacks and whites -- the white rate subtracted from the black rate -changes over time.
On many of the outcomes, disparities between blacks and whites have been
reduced to a striking degree. Consider first fetal health. In 1950, the fetal mortality rate
among blacks was 32.1 per 1,000 potential births compared to just 16.6 per 1,000 for
whites (see Figure 3). By 1980, however, black fetal mortality stood at just 14.7 per
1,000. Fetal mortality declined among whites as well, from 16.6 to 8.1 per 1,000.
Though fetal loss among black mothers in 1980 was nearly as high as it was among white
mothers thirty years earlier, over half of the 1950 black-white gap had been erased.
Improvement was slower for both groups after 1980, and the black-white gap remained
constant at about 6.6 per 1,000.
While I emphasize the closing of the absolute black-white gap (the difference in
mortality rates), others prefer instead to focus on the relative gap between blacks and
whites -- the ratio of their rates. For instance, the relative gap between black and white
fetal mortality has increased slightly over the past fifty years, so that the black fetal
mortality rate was roughly twice that of whites in both 1950 and 1998. Such figures have
led some observers to argue that no progress has been made in reducing ethnoracial
disparities (e.g., Williams, 1990 and 1992). I have chosen to emphasize the narrowing of
the absolute gap for two reasons. First, the sheer magnitude of improvement in the
absolute gap is much more striking than the increase in the relative gap (all of which
occurred after 1980). Consider Figure 3 again and ask yourself whether your impression
pregnancies of which the mother is never aware.
4
is one of decreasing inequality or increasing inequality. Second, a narrowing of the
absolute gap implies that the absolute benefits accruing to blacks from declining fetal
mortality have been greater than the benefits accruing to whites, even if we were more
successful at cutting the white rate.
Another factor complicating the interpretation of trends in fetal mortality is the
increase in the number of induced abortions during this period. If women who obtained
abortions would have been more likely to experience a non-induced fetal loss than other
women, then the ethnoracial gap in fetal mortality would not have narrowed as much as it
did in the presence of rising abortion rates after 1973. This is true because the abortion
rate among black women has been quite a bit higher than among white women (U.S.
Bureau of the Census, 2002).
It is not clear a priori whether women obtaining abortions would be more or less
likely to experience a fetal loss. Such women are disproportionately young and
unmarried, which are risk factors for poor perinatal outcomes. On the other hand, many
women seek an abortion because the opportunity cost of becoming a mother is relatively
high. These women may have access to resources that would make a fetal loss less
likely. I was able to crudely model post-1970 trends in fetal mortality under a
counterfactual in which no abortions occurred and in which the within-group fetal
mortality rate for these additional hypothetical fetuses was twice as high as it was among
actual fetuses. The model indicates that the black-white gap in fetal mortality would
have narrowed by 1.8 per 1,000 during the 1970's rather than 4.3 per 1,000. Trends in the
5
absolute gap after 1980 would not have differed much from the actual trends that
occurred.2
It is not clear that this is an appropriate counterfactual, however, especially when
one considers that the availability of safe and legal abortion may have altered pregnancy
rates differentially among women of varying backgrounds. Had abortion remained
illegal, the number of pregnancies might have increased less than it actually did, possibly
resulting in fewer pregnancies that were at high risk for fetal loss. At any rate, if we
assume that pregnant women maximize their utility in making decisions about whether to
carry their pregnancy to term, then abortion trends alter the picture of improvement in
perinatal outcomes I have described only if rising abortion rates in the 1970's are
indicative of declines in welfare among women due to economic or other constraints.
That is, the magnitude of improvement in perinatal outcomes becomes questionable only
if abortion trends imply that pregnant women increasingly felt compelled to obtain an
abortion against their preferences. Reduced ethnoracial inequality becomes questionable
only if this occurred to a greater extent among black than white women. I doubt that data
are available to answer these questions.
Turning to trends in other perinatal outcomes, Figure 4 shows that the incidence
of premature births was roughly flat for Afro-Americans after 1981. Because births to
white women were increasingly likely to be premature, however, the gap between blacks
and whites narrowed during the 1990's. Between 1990 and 2000, one-third of the blackwhite gap at the start of the decade was eliminated. Unfortunately, data are unavailable
prior to 1981, so it is impossible to know whether there were large improvements in
preterm birth rates prior to 1980 as in the case of fetal mortality rates. The post-1980
2
A spreadsheet containing the model and its data are available from the author upon request.
6
improvement in the ethnoracial gap in preterm birth rates, occurring during a period in
which ethnoracial inequality in fetal mortality rates was unchanged, also points to the
difficulties in relying on a single measure of perinatal outcomes to gauge progress.
This point becomes critical in considering trends in low and very low birthweight
rates (Figures 5 and 6). There appears to be much less in these charts to inspire
optimism: the incidence of low birthweight is more or less static over the past forty years,
and the incidence of very low birthweight has increased among blacks since at least 1970.
Ethnoracial disparities have been increasing on both measures, particularly in the case of
very low birthweight.
Considering trends in infant mortality rates makes the low birthweight trends even
more curious (see Figure 7). The black-white gap in infant mortality -- deaths in the first
year of life -- has narrowed by seventy percent since 1940 and by over sixty percent since
1960. The black infant mortality rate declined from 72.9 deaths per 1,000 births in 1940
to 14.6 in 1999. Preliminary vital statistics data indicates that infant mortality continued
its decline through the middle of 2001 (National Center for Health Statistics, 2002).3
How to explain the paradox of strong improvements in most perinatal outcomes
and narrowing ethnoracial inequality on the one hand and stagnation -- and even
regression -- in terms of low birthweight rates? An especially important but neglected
part of the answer lies in the way that these perinatal outcomes are interrelated. For
example, it is entirely possible that declines in fetal mortality have propped up low
birthweight rates, keeping them higher than a secular improvement in perinatal health
would predict. As a consequence of advances in medical technology, public health
3
Data (not shown) also indicate dramatic narrowing of the black-white gap in neonatal mortality after
1960.
7
efforts, wider health care coverage, and rising education levels and health awareness,
more pregnancies are now successfully carried to term. But this undeniably welcome
progress means that relatively deprived newborns who in the past would not have
survived to birth increasingly live to take their first breath and enter the world. In other
words, yesterday's "fetal losses" may have become today's low birthweight newborns.
To examine the possible effect of this interrelationship on low birthweight trends,
I constructed a measure of "poor birth outcome", which is the number of fetal losses and
low birthweight newborns expressed as a proportion of all potential births (again,
excluding induced abortion from consideration). The resulting trends (not shown)
indicate that poor birth outcomes among black women declined from 14.2 percent in
1970 to 12.2 percent by 1998. This implies that declining fetal mortality propped up low
birthweight rates among blacks by a small amount -- the low birthweight rate only
declined from 13.9 to 13.1 percent over this period -- but that the effect did not much
alter the low birthweight trend. The same is true among whites. Furthermore, the
ethnoracial gap in low birthweight is also not much affected: the absolute gap in low
birthweight rates was cut by 7 percent over the period, compared to 14 percent for the
gap in the rate of poor birth outcomes.
On the other hand, declines in fetal mortality may have had a bigger impact on
very low birthweight rates. When "poor birth outcomes" are defined as fetal losses plus
very low birthweight births, the black rate declines from 4.7 percent to 3.9 percent during
the 1970's (when the very low birthweight rate increased from 2.4 to 2.5 percent). After
1980, the rate of poor birth outcomes increases to 4.3 percent but is still lower in 1998
than in 1970. In contrast, the rate of very low birthweight rose to 3.1 percent, a rate that
8
exceeded the 1970 rate. The ethnoracial gap in poor birth outcomes remained roughly
constant over the period (at about 2.5 percentage points), compared to the small increase
in the very low birthweight gap.
But while declining fetal mortality appears to account for the apparent increase in
black very low birthweight rates after 1970, it does not lead us to the conclusion that the
incidence of low or very low birthweight births actually showed much improvement,
certainly not of the magnitude we see on the other perinatal outcomes. Again, however, a
consideration of the interrelationships among these outcomes suggests an improved
picture. Specifically, declining infant mortality rates mean that more at-risk girls survive
infancy than in the past. This necessarily increases the proportion of at-risk pregnancies
in the cohort when these girls reach their child-bearing years. Potentially, then, a secular
improvement in perinatal health -- or simply advances in medical technology that allow
more endangered newborns to survive -- might paradoxically prop up fetal mortality
rates, rates of preterm birth, and rates of low birthweight.4
If one is willing to make certain assumptions, it is possible to estimate, albeit
crudely, the effect of declining infant mortality on other perinatal outcomes. I modeled
trends between 1970 and 2000 in low birthweight rates under a minimal set of such
assumptions:

The infant mortality rate remained at its 1940 level in all years. This
allowed me to model what would have happened had declining infant
mortality not generated increasingly at-risk maternal cohorts. I
determine how much smaller a maternal cohort would have been had
infant mortality claimed the same proportion of the cohort as it did in
1940. The additional women -- the beneficiaries of declining infant
4
The argument holds for "at-risk" boys too -- to the extent that infant health problems are correlated with
poor outcomes later in life, declining infant mortality means that an increasing proportion of future
pregnancies will be compromised by family stress or environmental factors (e.g., cigarette smoking) related
to the father.
9




mortality -- are then assumed to have had relatively high rates of fetal
loss and low birthweight. These assumptions alter both the numerator
and the denominator of the low birthweight rate, producing a lower
counterfactual rate.
All of the infants who survive past age one are assumed to survive to
childbearing age (nineteen years later). Thus, the size of a maternal
cohort is the same as the birth cohort twenty years earlier (minus infant
deaths).
Women who would not have survived to childbearing age had infant
mortality not declined were assumed to have the same probability of
becoming pregnant as other women.5
Women who would not have survived to childbearing age were
assumed to be twice as likely as other women within their ethnoracial
group to experience a fetal loss. The results of the model are quite
insensitive to this assumption, due to comparatively low fetal loss
rates.
I assume that three out of every four live births to the mothers who
would not have survived to childbearing age were born low
birthweight.
Figure 8 provides the results of the modeling. Considering that it
necessarily relies on assumptions of uncertain validity and makes no attempt to
incorporate more complex factors such as endogeneity of pregnancy rates with
respect to expected infant mortality rates, the model results should only be taken
as suggestive. The chart redisplays the actual trends in low birthweight rates for
blacks and whites from Figure 5. In addition, it displays the same trends under
the counterfactual of static infant mortality. In contrast to the relatively flat
observed trends and invariant ethnoracial gap, the counterfactual scenario implies
that had improvements in infant mortality not produced populations that were
increasingly at risk for poor birth outcomes, black and white low birthweight rates
would have declined substantially. Furthermore, some of the absolute blackwhite gap would have closed between 1980 and 2000. By 2000, the black low
10
birthweight rate would have been just two percentage points higher than the level
that the white rate actually took that year.6
Of course, even a mistaken acceptance of the model estimates as highly
precise would imply that an obvious gap between blacks and whites would still
have existed -- the counterfactual model indicates that over four-fifths of the
black-white gap would have remained in 2000 even under demographic
conditions in which infant mortality would have barred the most disadvantaged
girls from reaching their childbearing years. The greater survival rate of infants
over the course of the Twentieth Century is unquestionably a wonderful trend and
an incredible medical and public health achievement. It is not a "cause" of low
birthweight in any useful sense. Nor is it a "cause" of the black-white gap in low
birthweight, as the continued large gap in infant mortality itself attests. If
anything, the foregoing discussion should lead us to question how we measure
success in combating inequality and improving material well-being. Great
improvements in health may paradoxically lead to greater challenges on other
5
Abortion trends are not incorporated into the model. That is, no attempt is made to model a
counterfactual in which abortion rates differed from their actual trends, and abortion rates do not enter into
the model at any point.
6
I constructed my model to allow for alternative assumptions. Perhaps the strongest of my assumptions is
that among women who themselves would not have survived infancy under the counterfactual scenario,
three-fourths of their births were low birthweight. When I drop this parameter to fifty percent, the black
low birthweight rate declines from 12.8 percent in 1970 to 10.3 percent in 2000. That is, the decline in the
black rate is smaller, though the counterfactual rate is still 20 percent lower than the actual rate in 2000.
The black-white gap still shrinks, though only by one percentage point, and only after 1990. If this
parameter is set much below fifty percent, the counterfactual trends do not differ greatly from the actual
trends. On the other hand, it may not be unreasonable to suppose that all of the births to the women who
benefited from declining infant mortality were low birthweight. After all, these were women whose health
as infants was such that they would not have survived beyond their first year under 1940 conditions. It is
not so farfetched to argue that they carried considerable health problems with them into their childbearing
years, problems that almost guaranteed poor perinatal outcomes. When I set my parameter so that one
hundred percent of the births to these women were low birthweight, the black-white gap narrows by onefourth, and the 2000 black counterfactual rate equals the rate that whites actually had that year. A
spreadsheet containing the model and its data is available from the author upon request.
11
fronts. If we measure "progress" too narrowly -- i.e., using cross-sectional
measures of socioeconomic outcomes -- we may convince ourselves that
improvements have been less significant than they actually have been. Because
low birth weight is associated with various problems later in life, it is possible that
secular improvements in fetal, neonatal, and infant mortality have actually made it
tougher for us to narrow black-white socioeconomic inequality.
Consequences of Low Birthweight in Childhood and Adulthood
[UNDER CONSTRUCTION. Medical literature has established strong
associations between low birthweight and infant and child development. Less
work has been devoted to the effect of LBW on adult outcomes. Problems of
causal direction, multicollinearity, and omitted variable bias, however, make even
the developmental findings less reliable than we might think. It is also unclear
whether it is LBW itself that matters or whether it is LBW for gestational age. In
other words, some argue that preterm LBW babies "catch up" developmentally
and suffer no long-term consequences, while babies who are small for their
gestational age have worse outcomes. Regarding methods, family fixed effect
models, in which one uses variation in an independent variable among siblings to
estimate the effect on the dependent variable, are ideally suited for the question of
LBW-outcome relationships. Such models control for all genes and environment
that siblings share. Conley and Bennett (2000) use such a model to show that
LBW has a big effect on the probability of high school graduation. There is still
12
the possibility that omitted non-shared genes or environmental factors (e.g., peer
groups) produce biased estimates of LBW's effect.]
Risk Factors for Low Birthweight
Though a vast epidemiological literature has developed around perinatal
outcomes, surprisingly little is known about the causes of these outcomes generally. The
most work has been devoted to exploring low birthweight. Low birthweight is more
reliably measured than preterm birth rates. The determination of whether a newborn is
preterm or not generally relies on the mother's recall of when the pregnancy began.
Because preterm birth is a binary outcome based on a somewhat arbitrary criterion, small
errors in recall can mean the difference between accurate and inaccurate data. Birth
weight, on the other hand, is measured by medical professionals at the time of birth.
Neonatal and infant mortality, of course, are also easily measured, but researchers appear
to have devoted more energy to studying low birthweight. Most likely, this reflects the
fact that low birthweight infants have high mortality rates. For example, about onefourth of very low birthweight infants die in their first year, a much higher rate than for
moderately low birthweight infants (two percent) or normal birthweight infants (less than
0.5 percent) (Mathews et. al., 2001).
There are, proximally, two "causes" of low birthweight: preterm delivery and
intrauterine growth retardation. That is, an infant may be small because she was born
prematurely or because she experienced health problems in utero that slowed the rate of
her growth. These two conditions often occur together however. About forty percent of
low birthweight infants are small for their gestational age (Lieberman et. al., 1987).
13
Specific biological processes are not yet well understood, however, so most of the
epidemiological research has undertaken to identify "risk factors", or correlates of low
birthweight.
Many of the risk factors that researchers have identified are related directly to
maternal physical health. The actual causal effects of even some of these correlates,
however, are questionable. For instance, many researchers have identified primiparity, or
first birth, as a risk factor (Helsel et. al., 1992; Miller, 1994). Recently, however, Conley
and Bennett (2000) found that controlling for infant's sex, maternal age, education,
income, and race, being a first-born child had no statistically significant effect on the
probability of low birthweight. And in a study that distinguished between preterm low
birthweight and low birthweight due to intrauterine growth retardation, Kallan (1992)
found that the odds of the former showed no relationship to parity (cumulative number of
births), but that as parity increased, the odds of the latter increased. Similarly, other
reviews have proposed high parity as a risk factor (i.e., having a fourth or higher birth).
One confounding risk factor in this case is maternal weight. Women whose
weight is low for their height are more likely to have low birthweight births. Research
indicates that low weight gain during pregnancy also increases the odds of a low
birthweight birth (e.g., Cramer, 1987). Shortness of stature (height) is associated with
greater risk of low birthweight birth as well.
Another confounding risk factor is age. Many studies have found that teenage
women are more likely to have low birthweight births than somewhat older women (e.g.,
Ketterlinus et. al., 1990; Kallan, 1992). Younger women are more likely to be
primiparous however. At any rate, the association between teenage pregnancy and
14
greater low birthweight risk has itself been reversed by Rosenzweig and Wolpin (1995)
and Geronimus and Korenman (1991), who found using sibling fixed effects models that
young maternal age was actually advantageous. The authors argue that omitted risk
factors associated with early pregnancy are really to blame. Ketterlinus et. al. (1990) and
Kallan (1992) found that women over the age of thirty-five had worse birth outcomes
than somewhat younger women. This finding, too, has been challenged. Older women
are more likely to have multiple births (twins, triplets), and infants involved in multiple
births are more likely to be low birthweight. When low birthweight rates for singleton
births are examined, the effect of older age is weakened substantially (Martin et. al.,
2002).
Other risk factors closely linked to maternal physical health include short
interpregnancy intervals (Kallan, 1992; Klebanoff, 1988), various medical conditions,
and having had a prior problematic pregnancy. Anemia, infection, and hypertension are
among the medical conditions that increase the odds of low birthweight birth. Women
who have already experienced a fetal loss or given birth to a low birthweight infant are at
greater risk for delivering another low birthweight infant (Institute of Medicine, 1985).
Finally, some researchers have proposed genetic factors, generally citing associations
between maternal and paternal low birthweight on the one hand and child low
birthweight on the other. Conley and Bennett (2000) found that the probability of being
born low birthweight was four times higher if an infant's mother had herself been born
low birthweight, controlling for income, maternal education and age, parity, and race.
Infants with fathers who were low birthweight were six times more likely than other
15
infants to be born low birthweight (see also Coutinho et. al., 1997; Wang et. al., 1995).
These intergenerational associations, however, need not imply genetic inheritance.
Various behaviors that affect maternal health have also been identified as risk
factors. Researchers have found associations between low birthweight on the one hand
and maternal alcohol consumption, smoking, and drug use during pregnancy on the other
(Overspect and Moss, 1991; Bakketeig et. al., 1993; Cornelius et. al., 1995; Abma and
Mott, 1990). Nutrition levels are also correlated with birth outcomes (Institute of
Medicine, 1985). In addition, the quantity, quality, and timing of prenatal care have
been linked to birth outcomes, though Williams (1990) notes that variation in medical
care explains little of the variation in health status by socioeconomic status in general.
Marsiglio and Mott (1988) found that the extent to which a pregnancy was wanted has a
weak effect on low birthweight by reducing the amount of prenatal care a mother obtains.
It is reasonable to postulate that psychological well-being influences perinatal
outcomes as well. Stress causes the adrenal glands to secrete hormones that can cause
constriction of blood vessels in the fetus, reducing fetal oxygen supply. Relatively little
research has been done in this area due to the difficulty of measuring stress levels.
Homer et. al. (1990), using National Longitudinal Survey of Youth data, found that
working women who did not wish to work and whose job was stressful were more likely
to give birth to low birthweight infants.
Socioeconomic status may affect fetal health in other ways. At least one study
has found that women who work are more likely to have low birthweight births (PeoplesSheps et. al., 1991). Bakketeig et. al. (1993) found that exposure to second-hand smoke
on the job worsened birth outcomes. Many researchers have considered the importance
16
of income and poverty to birth outcomes, but this research is plagued by omitted variable
bias, with findings often varying in direction depending on the set of controls used and
the subgroups considered (see Aber et. al., 1997; Gortmaker, 1979; Starfield et. al., 1991;
Duncan and Laren, 1990; Korenman and Miller, 1997). In particular, Conley and
Bennett's (2000) finding that the effect of income on the probability of a birth being low
birthweight disappears when parental low birthweight status is controlled calls into
question previously asserted causal associations. Maternal education has also been
considered in much of the research, but it too is particularly prone to omitted variable
bias. Duncan and Laren (1990) found that increased education lowers the risk of low
birthweight birth, but Conley and Bennett (2000) found no association once income,
maternal age, parity, and race were controlled.
Finally, it is worth considering two other interesting findings from the literature
before proceeding to consider the causes of black-white inequality in low birthweight
rates. Abma and Mott (1990) considered whether a mother had smoked or drank during
pregnancy and whether she delayed seeking prenatal care. They found that the total
number of these three risk factors present in a mother had an effect on her birth outcome,
regardless of the particular risk factors in which she engaged. This finding suggests that
the cumulation of risk factors may be more important than individual risk factors.
Diamond (1999) used a gruesome natural experiment to consider the effect of
poor maternal health on birth outcomes over multiple generations. Near the end of World
War II, the Nazi siege of the Netherlands subjected part of the Dutch population to
starvation. Diamond's research followed the daughters of women who were pregnant
during this period to examine the effects of the dramatically worsened health status of the
17
mothers on the children that they bore. Not surprisingly, birth outcomes were markedly
worse than those for Dutch women not subjected to famine. Some of the affected
daughters, though, did not experience noticeable health disadvantages despite the famine.
Unexpectedly, those daughters' children (the grandchildren of the women who were
pregnant during the famine) had higher low birthweight rates than other children, even
though their mothers were outwardly as healthy as their peers. In other words, the
grandmother's health status somehow skipped a generation to affect her grandchild's
perinatal health. Though this effect is not yet well understood, it would seem to have
implications for ethnoracial inequality in perinatal outcomes. I now turn to the literature
on the sources of such inequality.
The Causes of Ethnoracial Disparities in Perinatal Outcomes
If the causes of low birthweight in general are uncertain, it is no less the case that
the causes of the black-white gap in low birthweight rates are also poorly understood.
Most studies begin by comparing the observed black and white rates in a sample or in a
population (using vital statistics data). Research consistently reveals that AfroAmericans are twice as likely as Euro-Americans to be born low birthweight. This
finding is consistent with national figures that are based on one hundred percent of birth
certificates. In 2000, for instance, the low birthweight rate for non-Hispanic blacks was
13.1 percent, compared to 6.6 percent for non-Hispanic whites (Martin et. al., 2002).
Epidemiological and sociological studies then estimate black and white rates
conditional on various risk factors. Typically, among women who do or do not possess
any one risk factor, the black low birthweight rate remains close to twice that of whites.
18
In fact, many studies find that as risk status improves -- e.g., as one moves from high
school dropouts to college graduates -- the relative risk for blacks actually increases (as in
Figure 2).
So, for instance, Kleinman and Kessel (1987) found that the black low
birthweight rate in 1983 was 2.4 times the white rate. Among high school dropouts, the
black rate was 1.6 times the white rate, while among college graduates, it was 2.3 times
as high.7 The relative gap among unmarried women was 1.8 versus 2.0 for married
women. Among high-parity births -- "high-parity" being defined as having had a large
number of births for one's age -- the relative gap was 1.7, while for low-parity births it
was 1.8. All of these gaps control for the other covariates listed. Kallan (1992) estimated
that the relative risk (the odds ratio) of preterm low birthweight for blacks versus whites
was 2.6, controlling for age at pregnancy, pregnancy order, interpregnancy interval,
education, maternal smoking, and whether the mother had previously had a fetal loss or
low birthweight birth. The relative risk of intrauterine growth retardation low
birthweight was 4.4. (See also Gould and LeRoy, 1988; Wilcox and Russell, 1990;
Frisbie et. al. 1997.)
The research that has been conducted on black-white differences in preterm birth
and infant mortality rates finds similar results. Lieberman et. al. (1987) found that the
7
The reader may have noticed that I have shifted from using the absolute gap as a measure of inequality to
the relative gap. There is no inconsistency in doing so -- I am considering a different question here than in
the trend analysis earlier in the paper. In considering the causes of the black-white gap at a given point in
time, we ask whether being black affects the probability of being born low birthweight net of other factors.
The "effect" of being black is to increase the probability of low birthweight relative to being white. In the
trend analysis, we are asking how the probability of low birthweight for blacks has changed over time
compared to the probability of low birthweight for whites. It is then up to the observer to determine
whether the change is best assessed by considering the absolute or relative gap. Another way to think about
the different measures is to consider that improvements in educational attainment among blacks probably
helped to narrow the absolute gap between them and whites in a time series, producing real absolute gains
19
relative risk of preterm birth for black versus white women was 1.94 among a large
sample of Boston women with health insurance. The relative risk declined to just 1.88
when the authors controlled for a dozen medical risk factors, including anemia, cigarette
smoking, urinary tract infection, diabetes, and weight-for-height. Schoendorf et. al.
(1992) found that the relative risk of infant mortality among black college graduates as
compared to white college graduates was 1.9. The relative risk only declined to 1.8 after
controlling for age, parity, timing of prenatal care, and marital status.
Some studies have had more success driving down the relative risk of poor birth
outcomes for blacks versus whites. Starfield et. al. (1991) estimated that while a large
black-white low birthweight gap remained after controlling for a number of factors
(including income), among poor women the gap was statistically insignificant. Abma
and Mott (1990) indicate that the black-white gap disappeared when they controlled for
whether a mother drank or smoke during pregnancy, the timing of prenatal care, and
maternal scores on a test of cognitive skills.
Another study (Lieberman et. al., 1987) reported a relative risk of 1.22 conditional
on the number of risk factors a mother possessed (out of four -- age less than twenty,
unmarried, high school dropout, or welfare receipt). Controlling for a continuous
measure of red blood cell count (hematocrit level) and nothing else dropped the relative
risk to 1.38. Controlling for both hematocrit level and number of risk factors resulted in a
relative risk of 1.03. Unfortunately, however, only one in ten white mothers in the
sample had even one risk factor, so the findings related to risk factors may be sensitive to
the composition of the sample. And the hematocrit level finding is difficult to interpret.
for blacks that were greater than for whites. On the other hand, education differences may fail to "explain"
any of the relative gap in the cross section.
20
Low hematocrit level is an indication of anemia, which is itself a symptom of more
chronic health problems. Anemia is associated, for instance, with deficiency of iron,
vitamin B12, and folic acid, as well as with inflammatory diseases, autoimmune disorders,
chronic infection, cancer, exposure to lead or chemicals, and alcoholism. Some forms of
anemia are genetically inherited (Blinder, 1998). Thus, the Lieberman results imply that
differences in maternal health between blacks and whites lead to differences in birth
outcomes, which almost certainly must be true. The finding does not tell us anything
about the nature or source of the maternal health differences however.
A related line of research has explored the effect of nativity on ethnoracial
inequality in birth outcomes. The literature has consistently found that American women
born outside the U.S. or in Puerto Rico have better birth outcomes than their co-ethnics
born in the fifty states. This has been found for blacks and for women of Mexican,
Puerto Rican, Cuban, Central/South American, Chinese, Japanese, and Filipino descent
(Singh and Yu, 1996. See also Ventura and Taffel, 1985 and Guendelman et. al., 1990 on
women of Mexican descent, Alexander et. al. on women of Japanese descent. But see
Alexander et. al. and Mor et. al. who failed to find such an effect for Filipino and Korean
women, and Collins and Shay, 1994 who found that the effect for Mexican women
declined as neighborhood income increased.).
Several studies have found that sub-Saharan-African- and Caribbean-born black
women in the U.S. have more favorable birth outcomes than U.S.-born black women.
David and Collins (1997) estimate that the low birthweight rates of black women born in
sub-Saharan Africa are half those of U.S. born black women. Pallotto et. al. (2000) show
that moderately-low birthweight rates (i.e., 1,500-2,499 grams) of Caribbean-born black
21
women are closer to those of white American-born women than to black American-born
women. This result was found even among college educated women between the ages of
twenty and thirty-four who were married to college-educated men, received prenatal care
in the first trimester, and were of low parity.
The apparent effect of nativity on birth outcomes among women of American
minority groups -- and the apparently negligible effect of nativity on white women
(Kleinman et. al., 1991) -- has led some researchers to surmise that minority status itself
affects birth outcomes in some way. This interpretation has been subject to criticism.
The studies generally rely on cross-sectional data and so do not compare the outcomes of
foreign-born women to those of their daughters. The findings could be entirely explained
by a combination of favorable self-selection of immigrants to the U.S. (i.e., those women
who emigrate from their country may tend to be healthier or otherwise more advantaged)
and intergenerational regression to the mean. Regression to the mean refers to the
tendency for children of advantaged or disadvantaged individuals to have outcomes
closer to the average -- it is simply more likely that a person whose parents are far from
the mean will migrate toward the mean rather than even further away. Nevertheless,
Singh and Yu (1996) found that the effect of nativity held for all of the ethnoracial groups
previously mentioned except for Japanese and Filipino women when they controlled for
maternal age, marital status, education, birth order, timing of prenatal care, metropolitan
residence, and whether the delivery involved a multiple birth. Nativity effects for nonHispanic whites were not statistically significant.
Other evidence that is suggestive of the importance of racial ascription in
explaining ethnoracial inequality in birth outcomes may be found in the extent to which
22
Puerto Rican outcomes deviate from those of other Hispanic groups (see Figure 9).
Phenotypically, Puerto Ricans resemble Afro-Americans more closely than do other
Hispanic groups, and they are much more likely to self-identify as being "black"
compared to other Hispanics. Collins and Shay (1994) show that the relative risk of low
birthweight for Puerto Ricans versus other Hispanics is close to 2.0, even controlling for
maternal age, education, marital status, prenatal care adequacy, and neighborhood
income. Furthermore, black and white women have different age-risk profiles. That is,
among Euro-American women, low birthweight rates decline as maternal age increases
from less than fifteen years old to 30-34 years old. Low birthweight rates then increase
with age. In contrast, low birthweight rates among Afro-American women are lowest
among 25-to-29-year-olds and then increase steadily thereafter. This pattern has led
some researchers to hypothesize that the health of black women is worsened by minority
status, a "weathering" effect that cumulates as these women age (Geronimus, 1992).
Little theoretical or empirical work has been done, however, to determine the
mechanisms by which minority status might come to affect maternal health and thus birth
outcomes. Black-white inequality in perinatal outcomes is just one form that ethnoracial
inequality in health status takes. There are also large disparities between blacks and
whites on a number of other indicators of morbidity and mortality (Williams, 2001).
These, too, however, are poorly understood. Health disparities -- and perinatal disparities
in particular -- could arise from differences in health status among people with equivalent
access to quality health care or could be the result of unequal access to or quality of care
among people with the same initial health status. Differences in health status could arise
from differences in health practices and behavior or differences in exposure to unhealthy
23
environments, be they physical or social. All of these "explanations" could, theoretically,
themselves be the result of black-white differences in income or wealth, education,
neighborhood environments, or genes.8 Income and education seem to matter much less
than might be expected.
One way in which minority status could affect birth outcomes is by affecting
psychological well-being. Afro-Americans might experience more psychological stress
than Euro-Americans simply due to differences in socioeconomic status. Alternatively, it
could be that Afro-American stress levels might be affected by the regular experience of
prejudice and discrimination, the necessity of bracing oneself against the possibility of
such exposure, or repeated questioning of others' motives and beliefs. I have come across
no research that has explored the extent to which ethnoracial differences in stress levels
accounts for inequality in perinatal outcomes.
Ethnoracial inequality in perinatal outcomes could also result from the
combination of even greater ethnoracial inequalities in previous generations and
intergenerational transmission of health status. A girl born low birthweight might
overcome any disadvantage this imposes and have quite successful adult outcomes. Yet
she might still have poor birth outcomes in adulthood if the biological disadvantages with
which she entered the world remain with her. These disadvantages may be such that they
are not even apparent to her or others. The Dutch war famine research cited earlier
8
Genetic explanations of ethnoracial inequalities should, of course, be advanced with the utmost care, if at
all. Two points are important to consider here. The first is that even if health status were highly heritable
between parents and offspring, it would not follow that ethnoracial disparities in health status must be even
partially genetic in origin. Two groups of plants grown from genetically identical seeds could attain highly
unequal "outcomes" if one group of seeds is planted in fertile soil while the other is planted on arid parched
land. The second is that it would appear to be extraordinarily difficult to distinguish any theory in which
genetic inheritance of health status causes ethnoracial disparities from an alternative theory in which what
is inherited are the phenotypic characteristics that serve as "racial" markers, markers that then affect the
treatment that one receives from others.
24
provides evidence that such a scenario is indeed plausible. If the current black-white
disparity in perinatal outcomes reflects not just socioeconomic inequality within the
maternal generation but inequality within the previous generation, the large disparity
becomes less perplexing. The United States has made great -- if obviously incomplete -strides in equalizing the average environments faced by Afro-Americans and EuroAmericans (Smith and Welch, 1986). But intergenerational transmission of health status
-- through non-genetic biological pathways, if we are to assume the most responsible
position -- would imply that perinatal inequality will be more persistent than other
ethnoracial inequalities and will in fact be a barrier to reducing them.
To summarize, previous studies have shown that black-white inequality in
perinatal outcomes have little to do with many of the risk factors that we might expect
would be important. There is some evidence that the gap may be related to the
cumulative number of risk factors faced by Afro-American and Euro-American women
and to health behaviors (including the timing of prenatal care). There is also evidence
that perinatal inequality is related to inequality in maternal performance on tests of
cognitive skills. Finally, there is reason to believe that factors related to "racial" minority
status -- including effects linked to earlier generations -- may be especially consequential.
25
References
Aber, J. Lawrence, Neil G. Bennett, Dalton C. Conley, and Jiali Li. 1997. "The Effects
of Poverty on Child Health and Development." Annual Review of Public Health
18: 463-483.
Abma, Joyce C. and Frank L. Mott. 1990. "Is There a 'Bad Mother' Syndrome?
Evidence of Overlapping High Risk Behavior During Pregnancy Among a
National Cross-section of Young Mothers." Paper presented at the Annual
Meetings of the Population Association of America, Toronto.
Alexander, G.R., G. Baruffi, J. Mor, and E. Kieffer. 1992. "Maternal Nativity Status and
Pregnancy Outcome Among United-States-Born Filipinos." Social Biology 39(34): 278-284.
Alexander G.R., J.M. Mor, M. D. Kogan, N. A. Leland, and E. C. Kieffer. 1996.
"Pregnancy outcomes of U.S. and foreign born Japanese-Americans." American
Journal of Public Health 86(6): 820-825.
Bakketeig, L. S., G. Jacobsen, H.J. Hoffman, G. Lindmark, P. Bergsjo, K. Molne, and J.
Rodsten. 1993. "Pre-Pregnancy Risk Factors of Small-for-Gestational Age
Births among Parous Women in Scandinavia." Acta Obstetrica and Gynecologica
Scandinavia 72: 273-279.
Blinder, Morey A. 1998. "Anemia and Transfusion Therapy," in Charles F. Carey, Hans
H. Lee, and Keith F. Woeltje, eds., The Washington Manual of Medical
Therapeutics (Philadelphia and New York: Lippincott-Raven Publishers).
Collins, J. W. and D. K. Shay. 1994. "Prevalence of Low Birthweight among Hispanic
Infants with United States-Born and Foreign-Born Mothers: The Effect of Urban
Poverty." American Journal of Epidemiology 139: 184-192.
Conley, Dalton and Neil G. Bennett. 2000. "Is Biology Destiny? Birth Weight and Life
Chances." American Sociological Review 65: 458-467.
Cornelius, M.D., P.M. Taylor, D. Geva, and N.L. Day. 1995. "Prenatal Tobacco and
Marijuana Use among Adolescents: Effects on Offspring Gestational Age,
Growth, and Morphology." Pediatrics 95: 738-743.
Coutinho, R., R. J. David, and J. W. Collins, Jr. 1997. "Relation of parental birth
weights to infant birth weight among African Americans and whites in Illinois: a
transgenerational study." American Journal of Epidemiology 146: 804-809.
Cramer, James C. 1987. "Family Structure and Infant Health." Paper presented at the
Annual Meetings of the Population Association of America, Chicago.
26
David, R. J. and J. W. Collins, Jr. 1997. "Differing Birth Weight Among Infants of U.S.Born Blacks, African-Born Blacks, and U.S.-Born Whites." New England
Journal of Medicine 337: 1232-1233.
Deaton, Angus and Darren Lubotsky. 2002. "Mortality, inequality and race in American
cities and states." Working Paper, Center for Health and Wellbeing, Princeton
University.
Diamond, Jared. 1999. "War Babies." in Stephen J. Ceci and Wendy M. Williams eds.,
The Nature-Nurture Debate: The Essential Readings (Malden, MA: Blackwell
Publishers).
Duncan, Greg J. and Deborah Laren. 1990. "Neighborhood and Family Correlates of
Low Birthweight: Preliminary Results on Births to Black Women from the PSID
Geocode File." Mimeo, Survey Research Center, Ann Arbor, Michigan.
Frisbie, W. Parker, Monique Biegler, and Peter de Turk. 1997. "Racial and Ethnic
Differences in Determinants of Intrauterine Growth Retardation and Other
Compromised Birth Outcomes." American Journal of Public Health 87: 19771983.
Geronimus, Arline T. 1992. "The Weathering Hypothesis and the Health of AfricanAmerican Women and Infants: Evidence and Speculations." Ethnicity and
Disease 2: 207-221.
Geronimus Arline T. and Sanders Korenman. 1991. "Maternal Youth or Family
Background? Preliminary Findings on the Health Disadvantages of Infants with
Teenage Mothers." Research Report No. 91-204 (March). Ann Arbor:
Population Studies Center, The University of Michigan.
Gortmaker, Stephen L. 1979. "Poverty and Infant Mortality in the United States."
American Sociological Review 44: 280-297.
Gould, J. B. and S. LeRoy. 1988. "Socioeconomic Status and Low Birth Weight: A
Racial Comparison." Pediatrics 82: 896-904.
Guendelman S., J. B. Gould, M. Hudes, and B. Eskenazi. 1990. "Generational differences
in prenatal health among the Mexican American population. Findings from
HHANES 1982-1984." American Journal of Public Health 80(suppl): 61-65.
Helsel, D., D.B. Petitti, and P. Kunstadter. 1992. "Pregnancy among the Hmong:
Birthweight, Age, and Parity." American Journal of Public Health 82: 13611364.
27
Homer, Charles J., Shirley A. Berresford, Sherman A. James, Earl Siegel, et. al. 1990.
"Work-related Physical Exertion and Risk of Preterm, Low Birthweight
Delivery." Paediatric and Perinatal Epidemiology 4: 161-174.
Hoyert D. L. 1996. "Fetal mortality by maternal education and prenatal care, 1990."
Vital Health Statistics 20(30). National Center for Health Statistics.
Institute of Medicine. 1985. Preventing Low Birthweight. Washington, D.C.: National
Academy Press.
Kallan, Jeffrey E. 1992. "Effects of Interpregnancy Intervals on Preterm Birth,
Intrauterine Growth Retardation, and Fetal Loss." Social Biology 39(3-4): 231245.
Ketterlinus, Robert D., Sandra Henderson, and Michael E. Lamb. 1990. "Maternal Age,
Sociodemographics, Prenatal Health, and Behavior: Influences on Neonatal Risk
Status." Journal of Adolescent Health Care 11(5): 423-431.
Klebanoff, Mark A. "Short Interpregnancy Interval and the Risk of Low Birthweight."
American Journal of Public Health 78: 667-670.
Kleinman, J.C., L.A. Fingerhut, and K. Prager. 1991. "Differences in Infant Mortality
by Race, Nativity Status, and Other Maternal Characteristics." American Journal
of Diseases of Children 145(2): 194-199.
Kleinman, J. C. and S. S. Kessel. 1987. "Racial Differences in Low Birth Weight:
Trends and Risk Factors." New England Journal of Medicine 317: 749-753.
Korenman, Sanders and Jane E. Miller. 1997. "Effects of Long-Term Poverty on
Physical Health of Children in the National Longitudinal Survey of Youth." in
Greg Duncan and Jeanne Brooks-Gunn, eds., Consequences of Growing Up Poor
(New York: Russell Sage Foundation).
Lieberman, Ellice, Kenneth J. Ryan, Richard R. Monson, and Stephen C. Schoenbaum.
1987. "Risk Factors Accounting for Racial Differences in the Rate of Premature
Birth." New England Journal of Medicine 317(12): 743-748.
Marsiglio, William and Frank L. Mott. 1988. "Does Wanting to Become Pregnant with a
First Child Affect Subsequent Maternal Behaviors and Infant Birth Weight?"
Journal of Marriage and the Family 50: 1023-1036.
Martin, Joyce A., Brady E. Hamilton, Stephanie J. Ventura, Fay Menacker, and Melissa
M. Park. 2002. "Births: Final Data for 2000." National Vital Statistics Reports
50(5). National Center for Health Statistics.
28
Mathews T.J., M.F. MacDorman, F. Menacker. 2001. "Infant mortality statistics from
the 1999 period linked birth/infant death data set." National Vital Statistics
Reports 50(4). National Center for Health Statistics.
Miller, Jane. 1994. "Birth Order, Interpregnancy Interval and Birth Outcomes among
Filipino Infants." Journal of Biosocial Science 26: 243-259.
Mor, J.M., G. R. Alexander, E. C. Kieffer, and G. Baruffi. 1993. "Birth Outcomes of
Korean Women in Hawaii." Public Health Reports 108(4): 500-505.
National Center for Health Statistics. 2002. "Births, Marriages, Divorces, and Deaths:
Provisional Data for August 2001." National Vital Statistics Reports 50(7).
http://www.cdc.gov/nchs/data/nvsr/nvsr50/nvsr50_07.pdf.
National Center for Health Statistics. 2000. "Vital Statistics of the United States, 1998
Natality." Technical Appendix (Hyattsville, MD).
National Center for Health Statistics. 1999. "Vital Statistics of the United States, 1998
Mortality." Technical Appendix (Hyattsville, MD).
Overspect, M. D. and A. J. Moss. 1991. "Children's Exposure to Environmental
Cigarette Smoke Before and After Birth." Advanced Data No. 202. Vital and
Health Statistics, National Center for Health Statistics.
Pallotto, Eugenia K., James W. Collins, Jr., and Richard J. David. 2000. "Enigma of
Maternal Race and Infant Birth Weight: A Population-Based Study of U.S.-Born
Black and Caribbean-Born Black Women." American Journal of Epidemiology
151(11): 1080-1085.
Peoples-Sheps, Mary D., E. Siegel, and C. M. Suchindran. 1991. "Characteristics of
Maternal Employment During Pregnancy: Effects on Low Birthweight."
American Journal of Public Health 81: 1007-1012.
Phillips, Meredith, Jeanne Brooks-Gunn, Greg J. Duncan, Pamela Klebanov, and
Jonathan Crane. 1998. "Family Background, Parenting Practices, and the BlackWhite Test Score Gap." in Christopher Jencks and Meredith Phillips, eds., The
Black-White Test Score Gap (Washington, D.C.: The Brookings Institution
Press).
Rosenzweig, Mark R. and Kenneth I. Wolpin. 1995. "Sisters, Siblings, and Mothers:
The Effect of Teen-Age Childbearing on Birth Outcomes in a Dynamic Family
Context." Econometrica 63: 303-326.
Schoendorf, K.C., C. J. R. Hogue, J. C. Kleinman, and D. Rowley. 1992. "Mortality
among Infants of Black as Compared with White College-Educated Parents."
New England Journal of Medicine 326: 1522-1526.
29
Singh, Gopal K. and Stella M. Yu. 1996. "Adverse Pregnancy Outcomes: Differences
Between US- and Foreign-Born Women in Major US Racial and Ethnic Groups."
American Journal of Public Health 86: 837-843.
Smith, James P. and Finis R. Welch. 1986. Closing the Gap: Forty Years of Economic
Progress for Blacks (Santa Monica, CA: The Rand Corporation).
Starfield, Barbara, S. Shapiro, J. Weiss, K. Liang, K. Ra, D. Paige, and X. Wang. 1991.
"Race, Family Income, and Low Birth Weight." American Journal of
Epidemiology 134: 1167-1174.
U.S. Bureau of the Census. 2002. Statistical Abstract of the United States: 2001. Table
92. http://www.census.gov/prod/2002pubs/01statab/stat-ab01.html.
Ventura S.J., W.D. Mosher, S.C. Curtin, J.C. Abma, and S. Henshaw. 2000. "Trends in
pregnancies and pregnancy rates by outcome: Estimates for the United States,
1976–96." Vital Health Statistics 21(56). National Center for Health Statistics.
Ventura S.J. and S.M.Taffel. 1985. "Childbearing characteristics of U.S.- and foreignborn Hispanic mothers." Public Health Reports 100: 647--52.
Wang, X., B. Zuckerman, G.A. Coffman, and M. J. Corwin. 1995. "Familial
Aggregation of Low Birth Weight Among Whites and Blacks in the United
States." New England Journal of Medicine 333(26): 1744-1749.
Wilcox, A. and I. Russell. 1990. "Why Small Black Infants Have a Lower Mortality
Rate Than Small White Infants: The Case for Population-Specific Standards for
Birth Weight." Journal of Pediatrics 116: 7-10.
Williams, David R. 1990. "Socioeconomic Differentials in Health: A Review and
Redirection." Social Psychology Quarterly 53: 81-99.
Williams, David R. 1992. "US Socioeconomic and Racial Differences in Health: Patterns
and Explanations." American Review of Sociology 21: 349-386.
Williams, David R. 2001. "Race and Health: Trends and Policy Implications." in James
A. Auerbach and Barbara K. Krimgold, eds., Income, Socioeconomic Status, and
Health: Exploring the Relationships (Washington, D.C.: National Policy
Association).
30
Appendix: Notes on Vital Statistics Data
Federal law requires the collection of vital statistics data, which the National
Center for Health Statistics coordinates through the National Vital Statistics System.
Birth data is based on 100 percent of birth certificates in 1940, 1950, 1960, 1985 and
from 1990 on. In 1970, data was derived from a fifty-percent sample of birth certificates.
In 1980 (and 1981), some states' data were derived from fifty-percent samples while most
are based on 100 percent of birth certificates. Prior to 1960, birth data were corrected for
underregistration; by 1960, registration was over 99 percent complete. Births are
categorized by race based on the child's race (as reported by the mother) from 1940 to
1970 but by mother's race from 1980 to the present. Mother's race is imputed for birth
certificates with missing information (less than one percent of all birth certificates).
Hispanic origin items that are missing are coded as "non-Hispanic" (1.2 percent of births
in 1998). In general, other birth certificate data are missing for one percent of certificates
or fewer. Gestation age -- and thus whether a birth is preterm or not -- is based on the
mother's recall of the first day of her last menstrual period or -- for a small fraction (five
percent of births in 2000) -- a clinical estimate. This measure is the least reliable of the
perinatal outcomes listed due to recall error. Since 1981, gestation age is imputed for
birth certificates for which the month of the last menstrual period is reported but the day
is not. The definitions for preterm births and low birth weight are based on International
Classification of Diseases, Ninth Revision standards. Mortality data are based on 100
percent of death certificates. Infant deaths with missing race data (0.1 percent in 1995)
receive an imputed race. Race and Hispanic origin data is supplied by a funeral director,
31
who determines the information from observation or from an informant. Evidence
suggests that the mortality rates of non-black minority groups are biased downward by
misclassification. This bias is worst for Japanese infants but is also large for Filipino and
Chinese infants (understated by twenty to one hundred percent). Bias is less substantial
for American Indian, and Puerto Rican infants (ten to fifteen percent understatement) and
unimportant for other groups. Mortality rates are technically the number of deaths in a
calendar year divided by the number of births in the calendar year. Thus, they are only
estimates of the actual rate at which infants born in a calendar year die within 28 days or
within one year. Fetal mortality rates include only fetal losses occurring at a gestation of
twenty weeks or more and so are subject to maternal error in recalling last menstrual
period. A fetal loss is a pregnancy outcome other than a live birth or an induced abortion.
(Prior to 1970, induced abortions were not separated from fetal loss data.) The vast
majority of fetal losses occur prior to the twentieth week, and the reliability of reported
fetal losses of twenty weeks gestation or more is not as strong as the reliability of birth
data. Fetal losses are categorized by race based on the race of the child before 1980 but
the race of the mother from 1980 to the present. (See National Center for Health
Statistics, 2000; Martin et. al., 2002; National Center for Health Statistics, 1999; Hoyert,
1996; Ventura et. al., 2000.)
32
Figure 1. Ethnoracial Inequality in Perinatal Outcomes at the
Close of the Twentieth Century
White
Black
Hispanic
Asian
18
American Indian
16
14
Rate
12
10
8
6
4
2
0
Fetal Mortality (p.
1,000)
Preterm Births (p.
100)
Very Low Birthweight Low Birthweight (p. Neonatal Mortality (p.
(p. 100)
100)
1,000)
Figure 2. Low-Birthweight Births by Ethnoracial Group and
Education, 1999
16.0
14.0
Infant Mortality (p.
1,000)
Non-Hisp. White
Non-Hisp. Black
Hispanic
Asian
Amer. Indian
12.0
10.0
8.0
6.0
4.0
2.0
0.0
<12 years
12 years
13+ years
Years of Schooling
33
Figure 3. Fetal Mortality Rates
(Fetal Deaths of 20+ Weeks Gestation per 1,000 Live Births Plus Fetal Deaths)
40
16
White
Black
12
20
8
10
4
0
0
1950
1960
1970
1980
1990
Gap
Mortality Rate
Absolute Gap
30
1998
20
15
16
12
12
9
8
6
Gap
Percent
Figure 4. Percent of Live Births Preterm
White
Black
4
3
Absolute Gap
0
0
1981
1985
1990
1995
2000
34
Figure 5. Low-Birthweight Births as Percent of Live Births
14
7
12
6
10
5
8
4
6
3
Percent
8
White
4
Gap
16
2
Black
2
1
Absolute Gap
0
0
1960
1970
1980
1990
2000
Figure 6. Very Low-Birthweight Births as Percent of Live Births
4
2.0
3
1.5
Black
2
Absolute Gap
1
1.0
Gap
Percent
White
0.5
0
0.0
1970
1980
1990
1999
35
Figure 7. Infant Mortality Rates (Deaths per 1,000 Live Births)
80
40
White
70
60
30
Absolute Gap
50
25
40
20
30
15
20
10
10
5
0
Gap
Mortality Rate
35
Black
0
1940
1950
1960
1970
1980
Figure 8. Low-Birthweight Births as a Percent of
Live Births
1990
1999
White
Black
Counterfact. White
Counterfact. Black
16
Absolute Gap
8
14
7
12
6
10
5
8
4
6
3
4
2
2
1
0
Gap
Percent
Counterfact. Absolute Gap
0
1970
1980
1990
2000
36
Figure 9. Low-Birthweight Births -- Hispanics
All Births
10.0
All Hispanics
9.0
Puerto Rican
Mexican
Cuban
Central/South Amer.
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0
1980
1990
1999
Low-Birthweight Births -- Asians
All Births
16.0
All Asian
Chinese
Japanese
14.0
Filipino
Hawaiian
12.0
10.0
8.0
6.0
4.0
2.0
0.0
1970
1980
1990
1999
37
Download