Kogan MD, Alexander GR, Kotelchuck M, MacDorman MF, Buekens P, Martin JA, Papiernik E. Trends in Twin Birth Outcomes and Prenatal Care Utilization in the United States, 1981-1997. JAMA. 2000;284(3):335-341. doi:10.1001/jama.284.3.335
Author Affiliations: Maternal and Child Health Bureau, Health Resources and Services Administration, Rockville, Md (Dr Kogan); Department of Maternal and Child Health, University of Alabama at Birmingham (Drs Alexander and Papiernik); Department of Maternal and Child Health, University of North Carolina at Chapel Hill (Drs Kotelchuck and Buekens); Centers for Disease Control and Prevention, National Center for Health Statistics, Hyattsville, Md (Dr MacDorman and Ms Martin); and Department of Obstetrics and Gynecology, University Rene Descartes, Paris, France (Dr Papiernik).
Context Multiple births account for an increasing percentage of all low-birth-weight
infants, preterm births, and infant mortality in the United States. Since
1981, the percentage of women with multiple births who received intensive
prenatal care (defined as a high number of visits, exceeding the recommendation
of the American College of Obstetricians and Gynecologists by approximately
1 SD beyond the mean number of visits for women initiating care within each
trimester) has increased significantly.
Objectives To explore the hypothesis that more aggressive management of twin-birth
pregnancies may be associated with changes in birth outcomes in this population.
Design, Setting, and Subjects Cross-sectional and trend analysis of data from the National Center
for Health Statistics' birth and infant death records for all twin births
occurring in the United States between 1981 and 1997, excluding those with
missing or inconsistent data.
Main Outcome Measures Trends in preterm birth, low birth weight, preterm and term small-for-gestational-age
(SGA) births, and infant mortality, by level of prenatal care utilization.
Results The preterm birth rate for twins increased from 40.9% in 1981 to 55.0%
in 1997. The percentage of low-birth-weight infants increased from 51.0% to
54.0%. The preterm SGA rate also increased from 11.9% to 14.1%, while the
term SGA rate decreased from 30.7% to 20.5%. For women with intensive prenatal
care utilization, the preterm birth rate increased from 35.1% to 55.8%, compared
with an increase from 50.6% to 59.2% among women with only adequate use. Twin
preterm deliveries involving either induction or first cesarean delivery also
increased from 21.9% to 27.3% between 1989-1991 and 1995-1997. The twin infant
mortality rate for women with intensive prenatal care use declined between
1983 and 1996 and remained lower than the overall twin infant mortality rate.
Conclusions An apparent increase in medical interventions in the management of twins
may result in the seeming incongruity of more prenatal care and more preterm
births; however, these data suggest that women with intensive prenatal care
utilization also have a lower infant mortality rate.
National statistics indicate that the rates of both preterm birth and
low-birth-weight (LBW) infants have increased in the last 10 to 15 years.1 Since infants born with these conditions are at greatly
increased risk of infant mortality and subsequent developmental difficulties,2- 4 the rising incidence
of these adverse birth outcomes is a matter of importance.
During this same period, there has been a steady expansion in the percentage
of women with adequate prenatal care utilization,5
as defined by the standards recommended by the American College of Obstetricians
and Gynecologists,6 as well as intensive prenatal care utilization, defined as women receiving more
than the recommended number of visits.5,7,8
Early and adequate utilization of prenatal care has been associated with a
number of benefits, including more adequate immunizations and well-child care
for the infant,9 and possibly with a decrease
in LBW infants10- 12
and infant mortality.13 However, the purported
benefits of prenatal care have increasingly been questioned.14,15
The parallel rise in the preterm birth rate and the proportion of women
with intensive prenatal care utilization has led some researchers to suggest
that the benefits of prenatal care have been oversold.16
Other researchers have suggested that the increasing preterm birth rate may
be due to a number of factors, including an increase in multiple births, more
aggressive management of high-risk pregnancies, and an increased use of ultrasonography
to measure gestational age, which results in earlier gestational age estimates
than calculated by date of last menstrual period (LMP).17
Using intensive prenatal care utilization as a marker for more aggressive
management, research has indicated that there has been a dramatic increase
in intensive utilization among women with multiple births, a high-risk subgroup.5 In 1981, 8.5% of women with multiple births had intensive
prenatal care utilization compared with 22.8% by 1995 for this subgroup. Among
women with singleton births, 3.3% had intensive prenatal care utilization
in 1981 compared with 6.3% in 1995.
While constituting only 3% of births in the United States in 1997, multiple
births are a high-risk group, which accounted for 21% of all LBW births, 14%
of preterm births, and 13% of infant deaths in 1997.1,18
Moreover, the incidence of multiple births has increased notably. From 1980-1997,
the number of twin births increased 52%, while the number of higher-order
multiple births increased 404%.19
Recent research based on Canadian national data indicated that the preterm
birth rate has risen much faster among multiple births compared with singleton
births.20 It was suggested that obstetric interventions
were changing the upper tail of gestational age distributions among multiple
births, although this relationship has not been explored more directly.
Our goals in this article are to (1) present trends in birth outcomes
for twins in the United States; and (2) explore the hypothesis that more aggressive
management of multiple pregnancies, as suggested by intensive prenatal care
utilization and induction of labor and first cesarean deliveries, may be associated
with changes in birth outcomes in this high-risk subgroup.
Two different data sources were used. The National Center for Health
Statistics natality files from 1981-1997 were used for all analyses, except
for infant mortality, which used the National Center for Health Statistics–linked
birth-infant death cohorts for 1983-1984, 1989-1990, and 1995-1996. The analysis
is limited to twin births, as they constitute the large majority of multiple
births. There were 1,479,862 twin births over this 17-year period.
Records with inconsistent or missing values for the month prenatal care
began, the number of prenatal visits, or the length of gestation were excluded
from the analysis. Records with 1 or more missing values for gestational age,
the month care began, or number of visits ranged from 5.0% to 6.8% for each
Gestational age in completed weeks is computed from the interval between
the first day of the LMP and the date of birth. Records missing the date of
the LMP are imputed when there is a valid month and year. From 1989-1997,
clinical estimate of gestation was used in the computation of gestational
age in cases where the date of the LMP was not reported or where the LMP date
was inconsistent with the birth weight. Inconsistency and imputation procedures
have been described in detail elsewhere.1,21
Approximately 4% to 5% of the gestational ages during the period were based
on clinical estimate of gestation.
While recent research indicated that intensive prenatal care utilization
increased notably among multiple births according to both the R-GINDEX and
the Adequacy of Prenatal Care Utilization Index,5
we chose to use the R-GINDEX based on the recommendation that the R-GINDEX
was useful for research focusing on birth outcomes and to simplify the article
by limiting the number of indices revealing similar patterns.7
The R-GINDEX is based on calculations of when a woman began care and the number
of visits she received, and is adjusted for the length of gestation at delivery.
It is derived from the American College of Obstetricians and Gynecologists
recommendations for prenatal care. A woman's utilization of prenatal care
can be classified as "intensive," "adequate," "intermediate," "inadequate,"
"no care," or "missing." The intensive utilization category of the R-GINDEX
reflects women who had an excessively large number of prenatal care visits
(approximately 1 SD beyond the mean number of visits) given their gestational
age at delivery and the trimester that prenatal care began. For example, a
woman who starts care in the first trimester, delivers at 39 weeks' gestation,
and receives 17 or more prenatal care visits would be defined as exhibiting
intensive prenatal care utilization. The American College of Obstetricians
and Gynecologists recommendation for adequate prenatal care utilization for
this case would be 12 visits. Detailed descriptions of the R-GINDEX and the
algorithm used to calculate the index are available elsewhere.7,22
We examined 5 outcome measures: trends in preterm birth, LBW, preterm
small for gestational age (SGA), term SGA, and infant mortality. Preterm birth was defined as delivery from 20 to 36 weeks of gestation.
Although twins have a different fetal growth pattern than singletons,23 we chose to use the same definition for preterm as
have others examining preterm birth among twins for purposes of comparability.2,24Low birth weight was defined as infants weighing less than 2500 g. Infants were classified
as SGA using the 10th percentile of birth weight values of a previously reported
1991 US birth weight for gestational age reference curve25
and were categorized as preterm SGA or term (≥37 weeks' gestation) SGA. Infant mortality was defined as a death before the first
completed year of life. We chose to use infant rather than neonatal mortality
since an increasing proportion of neonatal deaths are being postponed until
the postneonatal period with advances in neonatal intensive care, and because
a significant proportion of postneonatal deaths are related to perinatal causes.
The findings of the study are essentially the same if neonatal mortality is
We present trends in preterm birth, LBW, preterm SGA, and term SGA for
US twin births from 1981-1997. The percentage change for each prenatal care
utilization level for women delivering twins from 1981-1997 is then discussed.
To examine whether more prenatal care may have altered birth outcomes,
we explore the trends from 1981-1997 for preterm birth, preterm SGA, and term
SGA comparing women with intensive prenatal care utilization with adequate
and less than adequate (intermediate, inadequate, and no care) utilizers.
Low birth weight was not included in further analyses because it may be reflective
of either preterm birth or intrauterine growth retardation.26
Beginning in 1989, obstetric procedures were added to the natality files.
We explore the trends in preterm birth among twins who were delivered with
induction of labor or first cesarean deliveries by utilization level from
We then compare the odds of delivering preterm in 1997 and 1990 with
1981, as a marker of more aggressive prenatal care, using multiple logistic
regression within 3 models of prenatal care utilization. The first model was
limited to women delivering twins and who had intensive utilization in either
1981, 1990, and 1997; the second was limited to women with adequate utilization
in those years; and the third was limited to women with less than adequate
utilization. The parameters in the logistic model were estimated by the maximum
likelihood method. Adjusted odds ratios (ORs) and 95% confidence intervals
(CIs) were calculated from the logistic analyses.
Last, we explore the changes in infant mortality rates (IMRs) among
twin births in 1983-1984, 1989-1990, and 1995-1996 using the linked birth-infant
death files stratified by level of prenatal care utilization, within the gestational
age categories of less than 32, 32 to 36, and 37 or more weeks. Confidence
intervals for IMRs were calculated based on a Poisson probability distribution,
and z tests were used to compare the differences
between the total IMR and each level of prenatal care utilization during each
Table 1 indicates that there
were notably different trends for preterm birth and LBW among twin births
in the United States from 1981 to 1997. The preterm rate for twins increased
markedly from 40.9% to 55.0%, while the LBW rate displayed lesser change;
51.0% in 1981 to 54.0% in 1997. Preterm SGA births increased from 11.9% to
14.1%, while the proportion of term SGA births decreased from 30.7% to 20.5%.
The twin IMR also declined 44.4% from 1983 to 1996. Moreover, Table 1 indicates that twins are accounting for an ever-increasing
percentage of all preterm and LBW births in the United States.
During the same period, the percentage of women who delivered twins
and who had intensive utilization increased dramatically from 8.3% in 1981
to 22.7% in 1997. The percentage of women who delivered twins and who had
adequate utilization increased from 47.8% to 55.7%. The percentage of women
who delivered twins and who had less than adequate utilization declined steeply
from 43.9% in 1981 to 21.6% (data available on request).
Figure 1 shows that the increase
in the twin preterm birth rate per 100 twin births was conspicuously sharper
among women who had intensive prenatal care utilization: from 35.1 in 1981
to 55.8 in 1997, an increase of 59%, although the rate for adequate utilizers
remained higher. The preterm SGA rate also rose most sharply among intensive
utilizers: from 8.7 in 1981 to 14.0 in 1997. The rate rose only from 13.4
to 14.6 among adequate utilizers, and from 10.9 to 12.4 among less than adequate
utilizers. Term SGA exhibited a very different pattern, decreasing from a
rate of 28.9 in 1981 to 19.1 in 1997 among women with intensive utilization;
from 22.0 to 17.0 among adequate utilizers; and from 40.5 to 31.9 among less
than adequate utilizers (data available on request).
Exploring the change in preterm birth among intensive utilizers further,
we found that the increase was primarily for births between 32 and 36 weeks'
gestation. The rate rose from 26.8 in 1981 to 46.2 in 1997, or an increase
of 72.4%. The rate for preterm birth less than 32 weeks' gestation rose 16%:
from 8.3 per 100 in 1981 to 9.6 per 100 in 1997.
We then explored the association between other obstetric procedures
and twin preterm birth. Combining 3 years of data for statistical stability,
we found that both the number and percentage of twin births delivered preterm
with induction of labor or first cesarean delivery increased from 21.9 in
1989-1991 to 27.3 in 1995-1997 (Table 2), while the percentage of twin births delivered preterm without
these procedures essentially remained the same at 25%. Among intensive utilizers,
the percentage of twin births delivered preterm with induction or first cesarean
delivery increased from 23.2% to 29.7% during this period, while the percentage
of twin births delivered term or postterm declined from 54.0% to 46.2%.
After using multiple logistic regression to control for other factors, Table 3 indicates that the ORs for preterm
twin birth increased for all women who delivered twins in 1997 compared with
1981. However, the largest increase occurred among women with intensive prenatal
care utilization in 1997 compared with women who had the same level of prenatal
utilization in 1981 (OR = 2.29; 95% CI = 2.13-2.44). Women who had less than
intensive utilization were only about 45% more likely to deliver preterm in
1997 compared with 1981.
An analysis of IMRs among twin births by level of prenatal care utilization
and gestational age in 1983-1984, 1989-1990, and 1995-1996 indicated significant
declines in twin infant mortality among intensive utilizers who delivered
preterm (Table 4), paralleling
those across other utilization levels. During each period examined, both the
preterm and overall twin IMR among intensive utilizers was significantly lower
than the twin IMR among all twins. Further, the twin IMR for preterm births
less than 32 weeks for intensive utilizers was significantly lower compared
with other levels of utilization. For adequate utilizers, the term IMR was
significantly lower than the overall twin IMR.
This study documents a notable increase in preterm births and preterm
SGA among twin births in the United States from 1981 to 1997. This increase
has not been distributed equally across levels of prenatal care utilization.
Indeed, the rise in the twin preterm birth and preterm SGA rates has been
much steeper among women who had intensive utilization. The preterm birth
rate increase among the intensive utilizers occurred almost entirely among
deliveries of 32 to 36 weeks' gestation. At the same time, we observed a noteworthy
decrease in term SGA births and in the twin IMR.
Collectively, these observed trends in preterm births, preterm SGA,
and term SGA indicate that twin births are increasingly less likely to be
delivered at term. We believe that changes in obstetric practice and interventions
may be partly responsible for this change. This interpretation is supported
by the greater increases in preterm birth rates among women with intensive
utilization. Previous research has suggested that the maturation process of
plural births is accelerated in comparison with that of singleton births,
although opposing findings exist.28- 31
As the optimal intrauterine growth and lowest morbidity may occur earlier
for twins than for singletons, the traditional wisdom of maintaining a pregnancy
to "term," ie, until 37 weeks' gestation, may not be equally applicable to
twin births due to the increasing morbid and mortal risks associated with
postmaturity that may occur much earlier in twins.28,29,32- 34
More intensive utilization of prenatal care services may allow for the
earlier detection of problems requiring early intervention, such as the recognition
of fetal growth retardation, which could account for the decline in term SGA
and the concomitant increase in preterm SGA, since obstetric interventions
may be motivated by the finding that 1 or both fetuses are failing to thrive.
It may explain why the percentage of preterm twins delivered with induced
labor or by first cesarean delivery rose between 1989 and 1997, while the
percentage of preterm twins delivered without these procedures remained the
same. It may also explain why women with intensive utilization in 1997 were
over twice as likely to deliver preterm compared with women with intensive
utilization in 1981. Similar large increases were not evident for other levels
of prenatal care utilization. Yet, despite this trend toward earlier deliveries,
twin infant mortality was lowest among women who had intensive utilization
during this period.
Other explanations are possible. The association between intensive utilization
and birth outcomes for this high-risk subgroup also suggests that these women
and infants probably have access to more and better neonatal care. Intensive
utilization for high-risk groups may act as a marker for access to tertiary
care facilities with neonatal intensive care units equipped to care for preterm
births. Recent evidence suggests that among high-risk pregnancies, women who
began prenatal care in the first 2 trimesters were significantly more likely
to deliver in a high-technology hospital.35
Changes in methods of gestational age dating over time might also have
affected the results, wherein the entire gestational age distribution shifted
slightly to the left. If true, it might account for the increasing twin preterm
rate, the decreasing term SGA rate, and the smaller increase in LBW rate.
However, there are a number of factors that argue against this theory. The
rise in the singleton preterm birth rate has not been nearly as steep as for
twins.36 There was not a uniform shift in the
gestational age distribution. Almost the entire change occurred among twins
of 32 to 36 weeks' gestation, with the largest change occurring at 35 to 36
weeks, when the median birth weights for twins are 2353 and 2500 g, respectively.23 Moreover, the increase in preterm birth occurred
only among twins delivered with the assistance of obstetric procedures such
as induction of labor or first cesarean delivery. Finally, the vast majority
(about 95%-96%) of gestational age calculations are based on LMP throughout
the study period. The data from 1990 to 1997 were examined excluding clinical
estimate of gestation, and the results were essentially the same (data available
In addition to potential innovations in obstetric practice, a number
of other factors may also be affecting both the incidence and outcomes of
twin births. Changes in sociodemographic characteristics of the population,
eg, the increasing percentage of births to older mothers, may play a role.
Fertility therapies have been indicated as markedly contributing to the rise
in multiple births and in preterm and LBW births in the United States and
elsewhere.37,38 In the early 1990s,
approximately 2.4% of twin infants were conceived through assisted-reproductive
technologies.37 No national estimates exist
on the contribution of ovulation-stimulation drugs without the use of assisted-reproductive
technologies to twin births, although data from the East Flanders, Belgium
Prospective Twin Survey indicated that 14% of twin births were associated
with use of fertility-enhancing drugs.39 There
are conflicting reports regarding whether twin births conceived through assisted-reproductive
technologies are more fragile and at greater risk of perinatal mortality or
There are potential limitations to this study. Intensive utilization
refers only to the number of visits based on the initiation of care; it does
not refer to the content of prenatal care. The accuracy and completeness of
reporting on vital records are also possible limitations. It was unknown whether
the first cesarean delivery was an elective obstetric procedure. There was
also incomplete information on potential confounders, such as pregnancy complications
like sexually transmitted diseases, which could lead to both an increase in
utilization and independently to preterm birth. Research has indicated that
reporting of gestational age on the birth certificate may be less accurate
for preterm births, which can introduce measurement error and affect the accurate
classification of preterm birth and prenatal care use.42,43
However, recent research has indicated that improbable gestational ages for
the recorded birth weights were considerably less for preterm twin births.44 Also, this was not a matched set of twins, so disparities
in twin pairs could not be examined.
We did not have accurate data on twin fetal mortality. An examination
of twin fetal mortality ratios for 20 or more weeks' gestation over the study
period indicated that they declined from 31.2 to 20.7 twin fetal deaths per
1000 live twin births (data available on request). However, a more detailed
analysis of fetal mortality data by prenatal care utilization was not possible
because a significant portion were missing information on prenatal care utilization.
Nonetheless, the reduction in the fetal mortality ratio is important if the
hypothesis is to be accepted that the increase in the preterm delivery rate
may result from obstetric interventions. It also suggests that a shift from
infant to fetal mortality does not account for the observed changes. Further
research is needed on this topic.
Changes in reporting areas could have affected the results,27 but our analyses did not reveal this. Two states
did not collect information on the number of prenatal visits during the earlier
part of the study period and were excluded from the analysis for those years.
For the years 1989-1997, we examined the data both including and excluding
these states and found negligible differences. Therefore, we included all
states for 1989-1997 (data available on request).
Studies that have examined the relationship between prenatal care utilization
and preterm birth or LBW have commonly begun with the hypothesis that receiving
the recommended levels of prenatal care or more would reduce these adverse
birth outcomes.15 This study suggests that
for the population of multiple births, increasing levels of prenatal care
use may occur during periods of increasing preterm and LBW rates, while simultaneously
exhibiting a decrease in IMRs. This research indicates that the relationship
between prenatal care and birth outcomes may be far more complex, particularly
when infant mortality is examined as well.
The declines in infant mortality among twin births in the United States
may largely reflect the same technological and medical advances in high-risk
obstetric and neonatal intensive care that have dramatically reduced birth
weight and gestational age–specific neonatal mortality in the United
States since the 1970s, including the increased use of corticosteroids.45- 47 However, the continuing
increase in multiple and preterm births and their better survival rates has
potential adverse consequences. While most of these preterm twins will be
without marked problems, a small but growing number may have both short- and
long-term disabilities and developmental delays.48- 50
The increasing number of surviving infants with disabilities that extend into
their childhood and adult years has cost implications related to the ongoing
provision of a wide array of health care, educational, and support services,
as well as significant human costs in meeting the special needs of these children.
Twin births are accounting for a greater percentage of all LBW and preterm
births in the United States, during a period when the overall rates for these
outcomes and prenatal care have been increasing. This study may help us understand
some of the seeming incongruity. More aggressive and more successful management
of twin pregnancies may be accounting for part of the elevation in the preterm
rate. To the extent that earlier deliveries of twins may result in reducing
the risk of a fetal or infant death, the importance of early and ongoing utilization
of prenatal care continues to be emphasized.