Neonatal and Perinatal Epidemiology

Key Population Mortality Statistics

Maternal and child health in the population have traditionally been assessed by monitoring two key statistics—the maternal mortality ratio and the infant mortality rate. A maternal death is defined by the WHO as the death of a woman during pregnancy or within 42 days of pregnancy. Because maternal deaths are not part of the denominator of births, the resulting fraction is referred to as the maternal mortality ratio . When the cause of death is attributed to a pregnancy-related cause, it is described as direct . When pregnancy has aggravated an underlying health disorder present before pregnancy, the death is termed an indirect maternal death. The WHO recommends that both direct and total (direct plus indirect) maternal mortality rates be monitored. Typically, the ratio is indexed to 100,000 births.

Because pregnancy can contribute to deaths beyond 42 days, the term “late maternal death” has been used to describe the death of a woman from direct or indirect obstetric causes more than 42 days but less than 1 year after termination of pregnancy. These later deaths are not usually included in tabulations of maternal mortality in vital data, although they are included in “pregnancy-associated mortality” as defined by the Centers for Disease Control and Prevention (CDC).

Deaths unrelated to pregnancy, but occurring within 42 days of pregnancy, are termed incidental maternal deaths and are not included in maternal mortality. However, even incidental deaths may bear a relation to pregnancy; for example, homicide and suicide are more common in pregnancy and shortly thereafter and might not be entirely incidental to it.

In most geographic entities, infant mortality is defined as all deaths occurring from birth to 365 days of age. The infant mortality rate is the number of infant deaths in a calendar year divided by the number of births occurring in the same year. This approach makes for imprecision because some deaths in the examined year occurred to the previous year’s birth cohort, and some births in the examined year will die as infants in the following year. In recent years, birth-death linkage has permitted vital registration areas in the United States to provide infant mortality rates that avoid this imprecision. The standard infant mortality rate reported by the National Center for Health Statistics (NCHS) links deaths for the index year to all births to whom the death occurred, including births that took place the previous year. This form of infant mortality is termed period infant mortality. An alternative procedure is to take births for the index year and link them to infant deaths, including those taking place the following year. This is referred to as birth cohort infant mortality and is not used for regular annual comparisons because it cannot be completed in as timely a fashion as can period infant mortality. The denominator for all forms of infant mortality is 1000 live births.

Infant deaths are conventionally divided into deaths in the first 28 days of life ( neonatal deaths) and deaths later in the first year (postneonatal deaths). Neonatal deaths, which are largely related to preterm birth and birth defects, tend to reflect the circumstances of pregnancy and birth; postneonatal deaths, when high, are largely from infection, often in the setting of poor nutrition. In high-income countries (HICs), neonatal deaths have for many years made up a larger proportion of infant mortality than postneonatal deaths. This has been true of the United States since 1921, and in recent years the ratio of neonatal to postneonatal deaths in the United States has consistently been approximately 2 to 1. Until quite recently, postnatal deaths outnumbered neonatal deaths in low- and middle-income countries (LMICs), but in 2019, infant mortality was 28.2/1000 live births in LMICs while neonatal mortality was 17.9/1000 live births, indicating that infant mortality in LMICs is beginning to resemble patterns seen in HICs.

Perinatal mortality is a term used for a rate that combines stillbirths and neonatal deaths in some fashion. Stillbirth reporting prior to 28 weeks is probably incomplete, even in the United States, where such births are required to be reported in every state. Nonetheless, stillbirths continue to be reported at a level not much lower than that of neonatal deaths, and our understanding of the causes of stillbirth remains very uncertain.

Sources of Information on Mortality—Vital Data

All US mortality data depend upon the collection of information about all births and deaths. Routinely collected vital data are the nation’s key resource for monitoring progress in caring for mothers and children. Annual counts of births and deaths collected by the 52 vital registration areas of the United States (50 states, District of Columbia [DC], and NYC) are assembled into national data sets by the NCHS and described under the heading of National Vital Statistics Reports (NVSRs). Unlike data collected in hospitals or clinics, or even from nationally representative surveys, birth and death certificates are required by law to be completed for each birth and death. Birth and death registration have been virtually 100% complete for all parts of the United States since the 1950s. The universality of this process renders many findings from vital data analyses stable and generalizable, although formatting changes recommended in 2003, affecting both the birth and death certificates, have created some difficulties in interpretation because the NCHS can only recommend format revisions in vital data certificates; each state is free to adopt them or not.

The 2003 revision of the birth certificate emphasized recording of data from medical records rather than maternal interview and recommended the reformatting of some elements, such as date of first prenatal visit, in ways that produced differences in findings compared to an earlier revision made in 1989. To complicate matters further, states adopted the 2003 revision at different times, and for much of the next decade, both versions of the birth certificate—1989 and 2003 revisions—were in use, leading to the NCHS deciding not to issue national data for several years for the prevalence of gestational diabetes, gestational hypertension, and gestational age at initiation of prenatal care. This problem has now been resolved because, as of 2016, all 50 states, the DC, Puerto Rico, Guam, Commonwealth of the Northern Marianas, and US Virgin Islands reported data based on the 2003 US Certificate of Live Birth. American Samoa continues to report based on the earlier 1989 birth certificate revision.

In 2003, the NCHS also recommended revisions to the US Standard Certificate of Death, including a special checkbox for identifying whether the decedent, if female, was pregnant or had been pregnant in the previous 42 days. As with the birth certificate, this revision was variably followed by states, and it has been found that the number of deaths recognized as maternal in states that adopted the checkbox is higher than in those that did not ( Fig. 1.1 ).

The limitations of vital data are well known. Causes of death are subject to certifier variability and, perhaps more importantly, to professional trends in diagnostic categorization. The accuracy of recording of conditions and measures on birth certificates is often uncertain and variable from state to state and hospital to hospital. Yet the frequencies of births and deaths in subgroups defined objectively and recorded consistently, such as birthweight and mode of delivery, are likely to be valid.

Time Trends in Mortality Rates of the Perinatal Period in the United States

Maternal mortality and infant mortality declined steadily through the 20th century; by 2000, neonatal mortality had declined by 90% from its 1915 value, postneonatal mortality by 93%, and maternal mortality by 98%. These extraordinary and unprecedented changes are the product of a variety of complex social factors including improvements in income, housing, birth spacing, and nutrition, as well as ecological-level public health interventions that produced cleaner food and water. Public health action at the individual level, including targeted maternal and infant nutrition programs and immunization programs have made a lesser, but still notable contribution. Medical care per se was, until recently, less critically involved in these declines, with the exception of the decline in maternal mortality, which was very sensitive to the developments in blood banking and antibiotics that began in the 1930s. Maternal mortality remains a major public health problem in much of the world, and such manageable complications as hemorrhage and infection continue to account for a large fraction of the world’s maternal deaths.

A notable feature of the last half of the 20th century was the sharp decline in all three mortality rates beginning in the 1960s, following a period of stagnation in the 1950s. The decline began with maternal mortality, followed by postneonatal and then neonatal. The contribution of medical care of the neonate was most clearly seen in national statistics in the 1970s, a decade that witnessed a larger proportional decline in neonatal mortality than in any previous decade of the century. All of the change in neonatal mortality between 1950 and 1975 was in mortality for a given birthweight; no improvement was seen in the birthweight distribution. This finding suggested the effectiveness of newborn intensive care, whose impact on mortality in very small babies has been striking. In 1960, shortly before the development of newborn intensive care, survival of an infant with birthweight of 1000 g was no more than 5%. Forty years later, survival at that birthweight was 95%. In retrospect, several factors seem to have played critical roles in the rapid development of the newborn intensive care programs that largely accounted for this rapid decline in birthweight-specific neonatal mortality. Perhaps the most important was the provision of more than nursing care to marginal populations such as the premature infant. Although the death of the mildly premature son of President Kennedy in 1963 provided a stimulus to the development of newborn intensive care, it should be noted that the decline in infant mortality that began in the 1970s was paralleled by a similar decline in mortality for the extremely old, perhaps an indicator that the availability of federal funding through Medicare and Medicaid enabled previously underserved populations at the extremes of age to receive greater medical attention than before. The Medicaid program, adopted in 1965, may have made it feasible for the first time to pay for the intensive care of premature newborns, among whom the medically indigent are overrepresented. While financial support for newborn intensive care may have been a necessary ingredient in its development, finances would not have been sufficient to improve neonatal mortality had not new medical technologies, especially those supporting ventilation of the immature newborn lung, been developed at about the same time.

Advances in newborn care have ameliorated the impact of premature birth and birth defects on mortality. Progress has come from improved medical care of the high-risk pregnancy and the sick infant, rather than through understanding and prevention of the disorders themselves. Unfortunately, the frequencies of underlying disorders that drive perinatal mortality have shown less improvement.

With the very important exception of neural tube defects, the prevalence of which has declined with folate fortification of flour in the United States and programs to encourage intake of folate in women of child-bearing age, prevalence rates of the major causes of death—preterm birth and birth defects—have not declined. The incidence of cerebral palsy, the major neurodevelopmental disorder that can be of perinatal origin, was remarkably stable for decades, notwithstanding advances in obstetric and neonatal care. However, there are now suggestions from some parts of the world that the birth prevalence of this disorder is on the decline.

The pace of decline in infant, neonatal, and postneonatal mortality in the United States began to slow in 1995 and changed little in the following decade. However, a decline of nearly 20% in both neonatal and postneonatal mortality has been seen since 2005 ( Table 1.1 ). For infants weighing 501 to 1500 g at birth, data from the Vermont Oxford Neonatal Network encompassing more than a quarter of a million newborns from hundreds of largely North American neonatal units showed a decline in mortality of 12.2% in the final decade of the 20th century and a further decline of 13.3% from 2000 to 2009. For infants at the threshold of viability (born at 22 to 24 weeks), the large multicenter National Institute of Child Health and Human Development (NICHD) neonatal network has reported that mortality declined by 12.6% between 2000 and 2011.

These declines are more modest than in the early days of newborn intensive care. From 1960 to 1985, a greater than 50% decline in mortality for infants weighing 501 to 1500 g at birth was recorded in national data, even though much of the first decade of that interval preceded the use of newborn intensive care technology in all but a few pioneering centers. The pace of advances in newborn medicine and the expansion of newborn intensive care to populations previously underserved, factors that have exerted a constant downward pressure on infant mortality since the 1960 s, have lessened in the past two decades or so.

Reported maternal mortality has actually climbed substantially in recent years, but this almost certainly reflects the effect of improved reporting. The CDC has a special unit dedicated to the problem of maternal mortality, the Pregnancy Mortality Surveillance System (PMSS). Established in 1987, its counts of “pregnancy-related” deaths, based on more in-depth exploration than is possible from a vital registration system alone, have provided consistently higher estimates of maternal mortality than data reported by the HCHS, as shown in Fig. 1.2 , in part because the CDC count includes deaths occurring up to 1 year after delivery.

The major reason for the increase in reported maternal mortality was the recommendation by the NCHS in 2003 that all death certificates to females include a checkbox indicating whether the decedent had been pregnant in the prior year. This recommendation was initially adopted by some states and not others, producing considerable variability across states’ reported maternal mortality ratios. The inconsistency led the NCHS to not report on maternal mortality ratios in the United States from 2008 to 2017, as seen in Fig. 1.2 .

Inasmuch as use of the checkbox has now been adopted by all states, the NCHS resumed reporting maternal mortality ratios in 2018 and has provided a detailed overview of issues in defining this important health parameter in vital data. The checkbox on the death certificate has proven to be a mixed blessing. While it uncovers many otherwise unknown maternal deaths, it also produces a small number of false positives. For example, in 2013, seven births were reported to women in their 60s, yet 53 death certificates for women of that age had indicated a recent pregnancy. The careful assessments by NCHS of the procedures for recording maternal deaths may account for a welcome convergence of estimates of maternal mortality from the two systems. PMSS estimated the maternal mortality ratio at 17.3/100,000 in 2017, and the NCHS estimated it at 17.4 in 2018. However, the NCHS reported an increase in the maternal mortality ratio to 20.1/100,000 in 2019, although with continued cautions about data quality.

The risk of preterm birth (<37 weeks’ gestation) increased steadily in the first years of the present century, peaked in 2007, and has declined since. These changes occurred largely in moderately preterm babies and likely reflected a period in which obstetricians became more comfortable with earlier inductions of labor as well as the increased prevalence of twins and triplets, who are generally born preterm, resulting from in vitro fertilization. The newer data suggest some modifications of these earlier practices.

The recording of diabetes in pregnancy on birth certificates has increased substantially in the 21st century, with the 2019 prevalence 2.5 times higher than in 2000 ( Table 1.2 ). It is possible that some of this increase might reflect more complete reporting on the 2003 birth certificate revision, but the obesity epidemic has likely contributed to increased prevalence of both prenatal and gestational diabetes.

The cesarean section rate continues its long-term increase, from 5% in 1970 to 23% in 1990, peaking at nearly 33% in 2010, with a faint decline to 31.7% in 2019. The reasons for this long-term increase are multifactorial and include pressures from patients, physicians, and the medical malpractice system. The steady reduction in smoking in pregnancy, which is now reported to be just 6%, is encouraging and parallels the decline in smoking in the general population but may unfortunately also reflect the uncertainty of medical records from which this information is obtained since the 2003 birth certificate revision. The proportion of mothers unmarried when they gave birth peaked at more than 50% in 2007 to 2008 but declined to 40% in 2019.

The long-term decline in the US fertility rate was accentuated in 2020, the year of the COVID-19 pandemic, when it was lower than at any time since the Depression. The number of US births declined by 4% in 2020. The preterm birth rate declined as well, by slightly more than 1%, with a suggestion that the decline in births before 34 weeks was slightly greater.