Reports on Selected Racial/Ethnic Groups
Special Focus: Maternal and Child Health
Infant Mortality Among Racial/Ethnic Minority Groups, 1983-1984
Joel C. Kleinman, Ph.D.
Division of Analysis
National Center for Health Statistics
Summary
Infant mortality varies considerably among racial/ethnic groups
in the United States. For groups other than whites and blacks,
previously published rates based on the vital statistics system
have been underestimated because of inconsistencies in the
classification of race and Hispanic status on birth and death
certificates. For this report, infant mortality rates (IMRs) are
based on the 1983 and 1984 linked birth and infant-death files,
and mother's race and Hispanic origin are reported in accordance
with information shown on the birth certificates.
Overall, Asians have somewhat lower infant mortality rates than
whites, but the rates vary from 6.0/1,000 among Japanese mothers
to 9.0/1,000 among "other Asian" mothers. Hispanic mothers show
even wider variation: from 7.8/1,000 among Cubans to 12.9/1,000
among Puerto Ricans. Blacks have an IMR twice as high as that for
whites, and the rate for American Indians is nearly 60% above the
rate for whites.
Mexicans are the third largest minority group in the United
States, accounting for one-quarter million births per year.
Despite a high rate of poverty and low use of prenatal care,
Mexicans have approximately the same IMR (9.0/1,000) as
non-Hispanic whites. Further study of this group could assist in
the development of prevention strategies.
INTRODUCTION
Although infant mortality has been decreasing since the
mid-1960s, progress in reducing infant mortality slowed in the
1980s. Approximately 20 countries have infant mortality rates
(IMRs) lower than the IMR in the United States. A major concern
related to infant mortality in the United States is the high rate
among black infants. In 1987, the IMR among blacks was twice that
among whites. Furthermore, the black-white ratio of IMRs has not
improved since 1950 (1).
Although the difference between black and white infant mortality
has received much attention, limited information is available on
infant mortality among other minority groups in the United
States. For such information to be provided on a regular basis, a
national surveillance system for monitoring trends would be
required. However, because of inconsistencies in the recording of
race on birth and death certificates, the use of routine vital
statistics for monitoring trends can be misleading for minority
groups other than blacks. A better method of assessing infant
mortality is to link infant death certificates with the
corresponding birth certificates. These linked files provide IMRs
for minorities other than blacks that differ considerably from
IMRs obtained from routine vital statistics. Furthermore, because
birth certificates in the United States contain more information
about the mother (e.g., education, marital status, prenatal care)
and the infant (e.g., birth weight, period of gestation) than
death certificates, more detailed analysis is possible to
identify high-risk populations and to develop and evaluate
interventions for reducing mortality among high-risk groups. The
National Center for Health Statistics, CDC, has implemented a
program to produce such linked files on an annual basis,
beginning with the 1983 birth cohort. Data from the 1983 and 1984
cohorts are now available and are used in this report to examine
IMRs among minority groups.
Infant mortality rates are usually calculated by dividing the
number of infant deaths in a given year (obtained from death
certificates) by the number of live births in the same year
(obtained from birth certificates). Race-specific IMRs calculated
in this way are valid only when the coding of race on both birth
and death certificates is comparable. However, results from the
1983 and 1984 National Linked Birth-Death Files show that the
coding for races other than white or black is not comparable. In
studies based on these files, the race of the child on the birth
certificate (used as the denominator of the usual IMR) is
compared with race of the child on the death certificate (used as
the numerator of the usual IMR) for all infant deaths. For whites
and blacks, the race coding on the birth certificate differed
from that on the corresponding death certificate in less than 2%
of the linked files; however, 25%-40% of infant deaths among
births coded as American Indian/Alaskan Native or Asian on the
birth certificate were coded to a different race on the death
certificate. For this reason, the IMRs for minorities other than
blacks need to be tracked by using the National Linked
Birth-Death Files.
METHODS
Infant mortality rates from the linked birth-death files were
calculated by using the race and Hispanic origin* of the mother
recorded on the birth certificate. The race of the mother was
used instead of the race of the child primarily for two reasons.
First, the race of the child is assigned in an arbitrary manner,
depending upon the race of the mother and father. For example, if
one parent is white and the other is not, the race of the child
is coded as the race of the parent who is not white. Second, the
race of the father is unknown for nearly 20% of the birth records
compared with less than 1% for which the race of the mother is
unknown (the race of the child is coded to the known race if one
parent's race is unknown). Similarly, in approximately 4% of
birth records, the mother's Hispanic origin is not recorded
compared with 14% for the father's.
Infant mortality rates are presented by mother's race for the
entire United States and by mother's Hispanic origin for the 23
states in which data on Hispanic origin are collected and
recorded on the birth certificate. Data are aggregated for 1983
and 1984 in order to provide more stable estimates of rates based
on the small numbers of deaths that occur in some minority
groups.
Postneonatal mortality among normal birth-weight infants (greater
than or equal to 2,500 g) has been used as an indicator of
preventable mortality (2). Excluding deaths from congenital
anomalies further refines this indicator. This measure is not a
perfect indicator of preventable mortality, because some
congenital anomalies are not lethal when optimal medical care is
given, and other deaths among these infants could not be
prevented even with the best of medical care. Nevertheless, as a
practical indicator, this measure has the advantage of being
easily implemented. This method is also least subject to
reporting differences among population subgroups. Thus, for this
report, preventable mortality among minority groups was estimated
by using postneonatal mortality among normal birth-weight infants
from all causes except congenital anomalies.
RESULTS
Table 1 shows infant, neonatal, and postneonatal mortality rates
by racial/ethnic category. Japanese mothers had substantially
lower IMRs than any other group. This advantage was evident in
both the neonatal and postneonatal period. Cuban mothers,
however, had even lower postneonatal rates than Japanese mothers.
Black mothers had the highest rates, followed by American Indians
and Puerto Ricans. American Indians had relatively low neonatal
mortality but the highest postneonatal mortality rates of all the
groups. They are unique in that more than half of their infant
deaths occurred in the postneonatal period compared with
approximately one-third for the other groups.
The five leading causes of infant death are shown in Table 2.
Congenital anomalies were the leading cause for all groups except
blacks and American Indians; in these groups, sudden infant death
syndrome (SIDS) ranked first. Of all the causes of death,
congenital anomalies showed the least variation among ethnic
groups: from 198/100,000 live births among Cubans to 255/100,000
among Puerto Ricans. SIDS mortality, on the other hand, showed
the greatest variation. SIDS rates among blacks were twice as
high as those among whites, and SIDS rates among American Indians
were nearly three times higher than those among whites. Asians,
Cubans, Mexicans, and (especially) Central and South Americans
had unusually low SIDS rates. Mortality rates from respiratory
distress syndrome (RDS) and other disorders related to short
gestation and low birth weight were much higher among blacks and
Puerto Ricans, with the differential especially pronounced for
the latter cause among blacks. Complications of pregnancy were
twice as high among blacks and 35% lower among Asians than among
whites.
Birth weight is strongly associated with infant mortality;
therefore, data were analyzed to determine whether differences in
IMRs resulted from differences in birth-weight distribution or
from differences in birth-weight-specific mortality. Following
methods used in other analyses (3), Table 3 disaggregates low
birth weight (LBW) into two components: very low birth weight
(VLBW) ( less than 1,500 g) and moderately low birth weight
(MLBW) (1,500-2,499 g). Blacks had nearly three times and Puerto
Ricans 1.6 times the incidence of VLBW compared with non-Hispanic
whites. Other minority groups showed little excess VLBW
incidence. However, Chinese and Japanese mothers had 20% lower
VLBW rates than whites. Blacks and Puerto Ricans had high MLBW
rates, but some of the other groups--Filipinos, other Asians, and
other and "unknown Hispanics" (i.e., Hispanic mothers whose
Hispanic origin is not known)--also had MLBW rates 30%-40% above
the rate for whites. Although Japanese mothers had low VLBW
rates, they had somewhat higher MLBW rates than whites. The high
rates of both LBW components among black and Puerto Rican mothers
is reflected in their high IMRs, especially from RDS and
disorders related to short gestation and LBW.
Table 4 shows birth-weight-specific IMRs. Little variation among
minority groups was found in mortality for VLBW infants, although
Japanese and Filipino mothers had particularly low rates.
American Indian mothers had nearly 50% greater IMRs for MLBW
infants than did white mothers, whereas Japanese, Filipino, and
Cuban mothers had rates approximately 40% lower. The IMRs among
normal birth-weight infants were particularly high for black,
American Indian, and Puerto Rican mothers.
Tables 3 and 4 imply that the high IMRs among blacks and Puerto
Ricans were due to adverse outcomes for both birth weight and
birth-weight-specific survival. The high IMR among American
Indians, on the other hand, was due almost entirely to poorer
birth-weight-specific survival.
Postneonatal mortality rates among normal birth-weight infants
(excluding congenital anomalies) provide an indicator of
preventable mortality. American Indians had the highest rate,
almost three times the rate for whites (Table 5). Blacks had
twice the rate of whites. Puerto Ricans were the only other group
with an elevated risk (29% above the rates for whites). If these
groups could achieve the same rate of postneonatal mortality
(excluding congenital anomalies) among normal birth-weight
infants as the white non-Hispanic group, the overall IMR would
decline by almost 25% among American Indians, 10% among blacks,
and 5% among Puerto Ricans. Asians, Mexicans, and Central and
South Americans had similar rates; other and unknown Hispanics
had a somewhat higher risk (16% above whites).
DISCUSSION
Infant mortality among racial/ethnic minority groups in the
United States varies widely. For groups other than whites and
blacks, previously published rates based on the usual vital
statistics system have been underreported because of
inconsistencies in the classification of race and Hispanic status
on birth and death certificates. Now that a national system of
linked birth and death records is available, surveillance of
these groups will be possible.
Infant mortality rates based on linked files differ in other
respects from the those based on routine vital statistics. In
particular, they are measures of risk for the birth cohort, i.e.,
they measure the probability of death among the cohort of live
births, whereas the usual IMR is a ratio of deaths to births
occurring in the same calendar year. However, some deaths are not
accounted for because of the lack of a match on birth files. In
1983, 1.6%--and in 1984, 2.2%--of the infant deaths could not be
matched to a birth certificate.
Overall, Asians had somewhat lower IMRs than whites, but the
rates varied from 6.0/1,000 among Japanese mothers to 9.0/1,000
among "other Asian" mothers. Hispanic mothers showed even wider
variation: from 7.8/1,000 among Cubans to 12.9/1,000 among Puerto
Ricans, with Mexicans (the largest group) having about the same
rate (9.0/1,000) as non-Hispanic whites. Blacks had an IMR twice
as high as that of whites, and the rate for American Indians was
nearly 60% above that of whites. Other studies have shown
considerable variation in infant mortality among different
American Indian communities (4).
Congenital anomalies were the leading cause for all groups except
blacks and American Indians (for whom SIDS ranked first). SIDS
mortality varied the most, with rates being twice as high among
blacks and nearly three times as high among American Indians as
they were among whites. Cubans, Mexicans, and (especially)
Central and South Americans had unusually low SIDS rates.
Mortality rates from RDS and other disorders related to short
gestation and LBW were much higher among blacks and Puerto
Ricans.
These results are consistent with data on birth-weight
distribution among live births. Blacks and Puerto Ricans had a
higher incidence of VLBW and MLBW than non-Hispanic whites. The
other minority groups had little excess VLBW.
Infant mortality varied little among the minority groups for VLBW
infants, and only American Indian mothers had substantially
higher IMRs than white mothers for MLBW infants. However, this
lack of variation should be interpreted with caution for two
reasons. First, finer birth-weight intervals need to be used in
these ranges because of the very steep gradient in IMR with
increasing birth weight and the differences in birth-weight
distribution among minority groups. Second, some investigators
argue that comparison of birth-weight-specific mortality should
be made with explicit reference to the birth-weight distribution
(5). That is, birth-weight-specific mortality rates in two
populations with different mean birth weights should be compared
on the basis of how far a particular birth weight is from the
mean, rather than using the absolute value of the birth weight.
When additional years of data become available, such detailed
analysis will be possible.
The high IMRs among normal-birth-weight infants for black,
American Indian, and Puerto Rican mothers identifies a problem
that would be evident even with more detailed
birth-weight-specific analysis. A further refining of this
comparison by using postneonatal mortality among normal
birth-weight infants (excluding congenital anomalies) as an
indicator of preventable deaths shows that American Indians had
three times and blacks had twice the risk of preventable deaths
that whites had. If these groups could achieve the rate among
whites, their overall IMRs would decline by nearly 25% among
American Indians and 10% among blacks.
Risk profiles for these minority groups varied widely. Blacks,
American Indians, Puerto Ricans, and Mexicans had the lowest
proportion with prenatal care beginning in the first trimester
(approximately 60% in 1984) compared with nearly 80% for
non-Hispanic whites, Cubans, and Asians (1). The proportion of
births to unmarried mothers was highest among blacks (59% in
1984), Puerto Ricans (51%), and American Indians (40%). Mexicans
had a lower proportion (24%), but it was still above the
proportions for Cubans (16%), non-Hispanic whites (11%), and
Asians (10%). The proportion of births to teenagers was higher
among blacks (24%), Puerto Ricans (21%), American Indians (20%),
and Mexicans (18%) than among non-Hispanic whites (11%), Cubans
(9%), and Asians (6%).
The low IMR among Mexicans is of particular interest because
Mexican mothers have relatively high maternal risk profiles
compared with non-Hispanic whites. Another finding is that
American Indians have risk profiles that are nearly as high as
those of blacks, yet their incidence of LBW and their neonatal
mortality rates are only 10%-15% above those of whites.
Some of these anomalous results could be due to underreporting of
infant deaths and live births of VLBW. For example, illegal
immigrants have incentives to obtain U.S. birth certificates for
infants who survive but to avoid the reporting of infants who die
shortly after birth (6). Specific studies based on field work
outside the hospital and vital statistics system could help
clarify this issue.
Continued surveillance of infant mortality among minority groups
in the United States is important for several reasons. Black
mothers are at particularly high risk of virtually all adverse
pregnancy outcomes. Puerto Rican mothers have an intermediate
risk, between the risks for blacks and whites. American Indians
are at especially high risk for postneonatal mortality. Each of
these minority groups will have population targets set up in the
year 2000 objectives. Mexicans are the third largest minority
group in the United States, accounting for one-quarter million
births per year. They appear to have relatively good pregnancy
outcomes, despite a high rate of poverty and low use of prenatal
care. Further study of this group might provide helpful
information for prevention strategies.
Further specificity of minority-group categories would also be
helpful. The group "other Asian or Pacific Islanders" is now the
largest Asian group and the one with the highest IMR. Specific
groups within this category may warrant special attention.
Similarly, the group "other and unknown Hispanic" is the second
largest category of Hispanic births. Coding of birth certificates
does not allow this group to be further subdivided; therefore,
whether the unknown Hispanic is the dominant category within this
group cannot be determined. The further consideration of coding
guidelines and the development of studies in localities with
relatively large concentrations of minority groups should be
encouraged.
References
National Center for Health Statistics. Health, United States,
1989. Hyattsville, Maryland: Public Health Service, 1990.
WHO Collaborating Center in Perinatal Care. Unintended
pregnancy and infant mortality/morbidity. In RW Amler, HB Dull
(eds). Closing the gap: the burden of unnecessary illness. New
York: Oxford University Press, 1987.
Kleinman JC, Kessel SS. Racial differences in low birth
weight: trends and risk factors. N Engl J Med 1987;317:749-53.
Honigfeld L, Kaplan D. Native American postneonatal mortality.
Pediatrics 1987;80:575-8.
Wilcox A, Russell I. Why small black infants have a lower
mortality rate than small white infants: the case for
population-specific standards for birth weight. J Pediatr
1990;116:7-10.
Kleinman JC. Underreporting of infant deaths: then and now. Am
J Public Health 1986;76:365-6.
*For reporting purposes, "Hispanic" is defined as "a person of
Mexican, Puerto Rican, Cuban, Central or South American or other
Spanish culture or origin, regardless of race."
All MMWR HTML documents published before January 1993 are electronic conversions from ASCII text into HTML. This conversion may have resulted in character translation or format errors in the HTML version. Users should not rely on this HTML document, but are referred to the original MMWR paper copy for the official text, figures, and tables. An original paper copy of this issue can be obtained from the Superintendent of Documents, U.S. Government Printing Office (GPO), Washington, DC 20402-9371; telephone: (202) 512-1800. Contact GPO for current prices.
**Questions or messages regarding errors in formatting should be addressed to mmwrq@cdc.gov.
