Trend Toward Older Maternal Age Contributed To Growing Racial Inequity In Very-Low-Birthweight Infants In The US

Abstract

In 2016 the Centers for Disease Control and Prevention reported that for the first time, US women in their thirties were bearing more children than those in their twenties. Analyzing US vital statistics data from the period 1989–2019, we simulated the effect that the distributional shift to older maternal ages at first birth had on health inequity between Black and White infants. Net of maternal socioeconomic indicators, this shift increased the relative odds that White women gave birth to very-low-birthweight (VLBW) infants by 10 percent, versus 19 percent for Black women, largely accounting for the rise in VLBW and the increase in racial inequity seen in the years analyzed. Reductions in infant mortality over the period were dampened by the maternal age shift, especially among Black babies, exacerbating Black-White inequity. Policy implications for promoting reproductive justice include universal tertiary care access, increasing the supply and distribution of maternity care providers, addressing the holistic needs of mothers throughout pregnancy and postpartum, and expanding family support policies. Conceptually, we recommend centering the realities of pregnancy and parenting from the perspective of the populations at highest risk—centering on the margins—and taking into account their implications for maternal weathering (accelerated deterioration due to disparate impacts of structural racism).

Eliminating racial inequities in infant health has been a high-priority national public health objective in the US for more than thirty years, yet inequities remain entrenched.¹ The US ranks thirty-third of thirty-eight Organization for Economic Cooperation and Development (OECD) countries in rates of infant mortality.² Rates are far worse for US Black infants, who die at more than twice the rate of US White infants.³

In 2016 the Centers for Disease Control and Prevention (CDC) reported that for the first time, women in their thirties were bearing more children than those in their twenties.⁴ Direct empirical knowledge of whether this trend affects infant health on a population level and whether these impacts are equitable across Black and White mothers is lacking. On the one hand, ample evidence exists that a woman’s chances of conceiving and then carrying a healthy infant to term decrease during her thirties and forties because of developmental depletions of reproductive capacity.^5,6 On the other hand, the scientific literature suggests that the potential costs of postponing childbearing are small, increasingly remediable through sophisticated pregnancy screening and assisted reproductive technology, and outweighed by the socioeconomic advantages associated with delayed childbearing.⁷

There is, however, a theoretical basis for the possibility that delayed childbirth contributes to and exacerbates racial disparities in US birth outcomes. Specifically, “weathering theory” posits that the health of US Black women, on average, suffers accelerated deterioration compared withthatofUSWhitewomen.^8,9 This is not due to any essential biological differences among racialized groups, but to the disparate impacts of structural racism. Cumulative exposures to environmental toxins, chronic biopsychosocial stressors, and material hardships associated with structural racism and the chronic high-effort coping it entails weaken and dysregulate neuroendocrine, metabolic, cardiovascular, and immune systems, increasing allostatic load^10–12 and accelerating cellular damage and aging.^11,13 The literature confirms that the magnitude of inequities in Black-White stress-mediated morbidity increases over the reproductive ages.^12,14–19 This wear and tear across body systems increases the vulnerability of weathered women to pregnancy and birth complications with advancing maternal age.²⁰

Because of population inequities in the existence, pace, and severity of weathering, weathering could adversely affect fetal and infant health at younger maternal ages for US Black women compared with US White women.²⁰ In light of weathering and the well-established Black-White inequity in access to high-quality health services and technology that might remediate its adverse effects,^21,22 we hypothesize that the impact of increasing maternal age will be more adverse for Black compared with White infants and may have increased Black-White inequity in infant health.

Study Data And Methods

DATA

We analyzed very low birthweight (VLBW), neonatal mortality, and infant mortality rates among non-Hispanic Black and White (hereafter referred to as “Black” and “White”) mothers derived from US birth and death certificates provided by the CDC.²³ During the study period, the question asking mothers their race on birth registration forms changed, so the CDC created bridged race categories to compare samples across the period.²⁴ Therefore, we used bridged race categories to identify mothers’ race. Cohort files for neonatal and infant mortality were created using linked birth-death information provided by the CDC from period files.²⁵

We analyzed births during the period 1989–2019. The first year in which maternal race and Hispanic ethnicity were self-reported on the birth certificate was 1989. That year also began the significant historical period in which reducing social disparities in infant outcomes was placed on the front burner for redress.²⁶ The last year of births for which all infant deaths have currentlybeencapturedwas2019.Tocontrolfor parity, and to focus on the outcome of postponing the transition to parenthood, we limited our analyses to a mother’s first birth.

MEASURES

We identified infants weighing less than 1,500 grams as VLBW. VLBW reflects being born preterm or experiencing intrauterine growth retardation. Unlike gestational age, birthweight is highly reliably measured for all babies, regardless of health insurance, social class, region, or race and ethnicity. Weighing less than 1,500 grams is dangerous for any newborn, greatly increasing their chance of neonatal death or, if they survive, of facing life with disabilities.²⁷ Conceptually, whether or not a baby is born too early or too small directly reflects maternal health and well-being and, as such, was our primary indicator of any effects of maternal weathering.

We coded infants who were born a live but died within twenty-eight days as neonatal deaths and infants who died before their first birthday as infant deaths. Whether or not a live infant survives through its first month of life reflects its health at birth and, if unhealthy, the extent, timing, and quality of access to neonatal intensive care. Surviving the first year of life reflects these factors and other social determinants of child and family health.²⁸

We controlled for the indicators of socioeconomic position available in our data: mother’s education, age, and marital status and payment source for delivery. Between 1989–91 and 2017–19, the fraction of first births that were multiples(for example, twins)rosefrom1.3percent to 1.8 percent for White women and from 1.1 percent to 1.8 percent for Black women. The fraction of women using assisted reproductive technology rose to 2 percent for White first-time mothers and 1 percent for Black first-time mothers. To assess whether the increases in the fraction of births that involved multiples or assisted reproductive technology accounted for any change in the fraction of infants born VLBW, we reran our simulations including dichotomous variables for multiple births and assisted reproductive technology.

STATISTICAL ANALYSES

All analyses were performed separately for Black and White mothers. We analyzed secular trends in our birth outcomes by calculating the rate of each outcome, using three-year moving averages during 1989–2019. To standardize on maternal socioeconomic characteristics, we estimated logit models with birth outcome variables as dependent variables.²⁹

To approximate the effect that the secular change in the age distribution of births had on birth outcomes, we took estimated age effects for the2017–19samplesforeachpoorbirthoutcome and weighted them first by the shares of first births in each age group in 2017–19 and then by the 1989–91 shares. The difference between these two estimates simulated the effect that the change in the age distribution of first births, per se, had on infant outcomes in the later period. Detailed methods are in the online appendix.³⁰

LIMITATIONS

It is well established that in the US, maternal age at first birth is a function of characteristics and lived experiences that are, themselves, correlated with the risk for poor birth outcomes, such as socioeconomic resources. US mothers younger than age twenty are a highly select, structurally disadvantaged group. Some evidence points to their prior disadvantage as a more important cause of their poor birth outcomes than their young age, per se.^31–34 In contrast, US women postponing childbirth to the latest ages are disproportionally highly educated mothers with resources and greater access to the newest technologies in reproductive assistance, obstetrics, and neonatal intensive care. Standardizing on maternal socioeconomic indicators as we did is the customary approach for accounting for background differences between women who bear their children at different ages,^14,29 yet differences by age in lived experiences and health care access are not fully captured by theseindicators.^32–34 The precise magnitude of unobserved heterogeneity bias could not be gauged in this study. Prior evidence from natural experiments points to our conventionally adjusted models likely understating the effect that increasing maternal age has on adverse infant health outcomes.^32–34

The youngest mothers will not have greater than a high school education at the time they give birth, whatever their eventual educational attainment, which builds in a negative correlation between measured education and early maternal age at first birth. To confirm that our results were not being driven by this correlation, we replicated our analyses, restricted to mothers ages twenty-three and older. These results were virtually identical to those using the full sample; see appendix table 4.³⁰

Study Results

DESCRIPTIVE STATISTICS

In 1989–91, median age at first birth was twenty years for Black mothers and twenty-five years for White mothers. By 2017–19, it had increased to twenty-four years for Black mothers and twenty-eight years for White mothers. If one looks at the full age distributions (see exhibits 1 and 2), substantial decreases in the fraction of first births occurring to teenagers are apparent between 1989 and 2019. White mothers showed notable decreases in the proportion of first births through age twenty-five, a small increase in their late twenties, and more sizable increases in their thirties and forties. For Black mothers the fraction of first births increased beginning in their early twenties, with proportionally larger increases above age thirty.

EXHIBIT 1. Maternal age distribution of first births among non-Hispanic White mothers in the US, 1989–91 and 2017–19.

SOURCE Data derived from files provided by the Centers for Disease Control and Prevention^’s vital statistics data, specifically from the cohort birth files for 1989–2019.

EXHIBIT 2. Maternal age distribution of first births among non-Hispanic Black mothers in the US, 1989–91 and 2017–19.

SOURCE Data derived from files provided by the Centers for Disease Control and Prevention’s vital statistics data, specifically from the cohort birth files for 1989–2019.

Appendix table 1 and appendix figure 1 display secular trends in VLBW and infant mortality rates by race.³⁰ In all cases, Black mothers had dramatically higher rates than White mothers. Although the trends were not completely smooth, between 1989 and 2019 the relative odds that first births were VLBW increased by roughly 16 percent (from 0.030 to 0.034) for Black mothers and 13 percent (from 0.010 to 0.012) for White mothers. Neonatal and infant mortality rates decreased between 1989 and 2019, with White rates falling about 10 percentage points more than Black rates.

Exhibit 3 shows maternal age patterns for VLBW infants separately by race for 2017–19; see appendix figure 2 for a similar analysis of infant mortality.³⁰ For all outcomes, the size of the Black-White gap increased with maternal age. White mothers experienced higher rates of adverse outcomes in their teens and after age thirty-two than in their twenties, yet Black mothers’ lowest-risk ages for VLBW were in their late teens, with risk increasing from age twenty through their thirties. Black infant mortality rates were lowest for mothers between ages eighteen and twenty-five, with rates continually increasing after age twenty-five.

EXHIBIT 3. Maternal age–specific rates of very low birthweight in the US, by maternal race, 2017–19.

SOURCE Data derived from files provided by the Centers for Disease Control and Prevention’s vital statistics data, specifically from the cohort birth files for 2017–19. NOTE The solid line represents the observed proportion, while the shaded region represents the 95% confidence interval around the observed rate.

LOGISTIC REGRESSIONS ADJUSTING FOR SOCIOECONOMIC INDICATORS

Appendix table 2 reports estimates, whereas appendix figure 3 visually displays the coefficients on age from logistic regressions predicting VLBW, using data from the period 2017–19.³⁰ By construction, the coefficient for the reference age group, mothers ages 23–25, is set to 0. Results are presented that did and did not adjust for socioeconomic indicators. Overall, the effect of socioeconomic indicators on the age coefficients was smaller for Black mothers than for White mothers.

Standardizing on socioeconomic indicators flattened the age profile for White women in their teens and early twenties, such that they accounted for all of the excess VLBW risk at maternal ages 16–17 through 20–22 compared with the reference age group. Adjusting for socioeconomic indicators, the association between maternal age and VLBW increased from ages 23–25 onward, and from ages 26–29 on it was as high as or higher than the VLBW odds for White mothers younger than age 16.

For Black mothers, adjusting for socioeconomic indicators steepened the maternal age profile of VLBW. After adjustment, mothers ages 16–17 had the lowest odds of VLBW, and even younger teenagers had lower VLBW odds than mothers ages 23–25 and older.

The logistic regression results suggest that the adjusted odds that a White mother would have a VLBW infant increased by 37.6 percent (exp^0.319 −1) if she postponed motherhood from her mid-twenties to her early thirties and 82.7 percent (exp^0.603 −1) if she postponed her first birth to her late thirties. For Black mothers, the increases were larger, at 64.4 percent and 106.3 percent, respectively.

Adjusting for socioeconomic indicators among White mothers reduced the odds of neonatal or infant mortality among teenagers older than age sixteen to roughly comparable or lower levels than those seen among mothers in their late twenties and thirties. For Black infants, only those with mothers younger than age sixteen (less than 1 percent of all Black births) experienced excess mortality relative to those with mothers in their early twenties and at levels comparable to or lower than those with mothers in their late twenties and thirties.

SIMULATIONS

In exhibit 4 we report simulated changes between 1989–91 and 2017–19 in the relative odds of mothers giving birth to VLBW infants or experiencing an infant death, given the change in the age distribution of first births. Results were based on the logistic regressions reported in the appendix that predicted each outcome, first controlling for maternal age and then also controlling for socioeconomic indicators.³⁰ A positive number indicates that the shift in the maternal age distribution of births increased the relative odds of poor birth outcomes by 2017–19; a negative number indicates the reverse.

EXHIBIT 4.

Simulated percent change in the relative odds of poor birth outcomes in 2017–19 due to the shift in the age distribution of first births since 1989–91, by race

	Change in relative odds
Poor birth outcomes	Unadjusted change	95% CI	Adjusted change	95% CI
Very low birthweight
White	−0.9%	(−1.5, −0.3)	9.5%	(8.7, 10.4)
Black	11.9	(9.4, 13.4)	18.9	(16.8, 20.9)
Black-White inequity	12.9	(11.1, 14.7)	8.5	(6.4, 10.6)
Neonatal mortality rate
White	−7.8	(−8.8, −6.7)	2.9	(1.3, 4.5)
Black	4.7	(1.9, 7.6)	9.9	(6.3, 13.5)
Black-White inequity	13.5	(10.2, 16.9)	6.7	(2.9, 10.7)
Infant mortality rate
White	−10.2	(−11.1, −9.4)	1.7	(0.4, 3.0)
Black	0.8	(−1.4, 3.1)	6.7	(3.8, 9.6)
Black-White inequity	12.3	(9.6, 15.1)	4.9	(1.7, 8.1)

SOURCE Data derived from files provided by the Centers for Disease Control and Prevention's vital statistics data, specifically from the cohort birth and death files for 1989–2019. NOTES A positive number indicates that the simulation suggests that the shift to an older maternal age distribution of first births between 1989–91 and 2017–19 increased the relative odds of the poor birth outcome by that percentage. A negative number indicates the reverse. Black-White inequity is the percent change in the relative odds (Black/White) of the birth outcome in question. Adjusted models controlled for maternal education and marital status and payment source for delivery.

The adjusted simulations suggest that the shift in the age distribution of first births for Black mothers would have predicted an 18.9 percent rise in their odds of VLBW (exhibit 4), more than fully accounting for the 15.9 percent actual rise in their relative odds of VLBW between 1989 and 2019 (appendix table 1).³⁰ For White infants, adjusted simulations predict that the shift in the age distribution increased the relative odds of VLBW by 9.5 percent (exhibit 4), accounting for most of the observed rise of roughly 13.1 percent (appendix table 1).³⁰

Net of socioeconomic indicators, infant mortality odds increased with advancing maternal age distribution. The simulations suggest that the shift in the age distribution of births had a substantially larger impact on Black mothers than on White mothers, exacerbating the already large infant mortality inequities.

MULTIPLE BIRTHS AND ASSISTED REPRODUCTIVE TECHNOLOGY

Reestimating our simulations by adding dummies for multiple birth and assisted reproductive technology suggested that together, the increase in multiples and assisted reproductive technology accounted for 25 percent of our estimates of the effect that the change in the maternal age distribution of White mothers had on the odds that they gave birth to a VLBW infant, and for 6 percent of our estimates for Black mothers (see appendix tables 3 and 5).³⁰

Discussion

We found that the historic shift toward a later distribution of maternal age at first birth in the US had adverse implications for infant health, accounting for the full increase in the odds that Black infants were VLBW and most of that increase in White infants, net of socioeconomic indicators. The rise in the maternal age distribution at first birth also accounted for the growing Black-White inequity in infant mortality, despite technological advances that lowered infant mortality rates overall. Temporally, this continued and growing inequity occurred despite the explicit targeting of racial disparities in infant health to be eliminated.²⁶ Preventing early childbearing was expected to contribute importantly to this effort. However, the adverse health impacts of weathering appear to be detectable as early as the beginning of a woman’s twenties. That the decreases in early childbearing did not translate into improved or equitable infant health outcomes underscores the fact that Black and socioeconomically disadvantaged mothers begin their race against the adverse reproductive health impacts of weathering sooner than they begin their race against the developmental biological clock.

The shift in the age distribution of births between 1989–91 and 2017–19 accounted for a significant fraction of the increase in women having multiple births and using assisted reproductive technology. Conceptually, these factors are likely mediators of the adverse effect of the older maternal age distribution in response to the deterioration in oocyte number or quality and the general depletion of reproductive capacity that women experience in their thirties and forties.^5,6

Policy Implications

Findings that VLBW rates had increased by 2019 compared with 1989 yet that neonatal mortality rates and infant mortality rates were reduced are consistent with technological advances in averting death in high-risk newborns during the study period, and in particular the introduction in the 1990s of surfactant therapy to treat respiratory distress syndrome.³⁵ (Before surfactant therapy, respiratory distress syndrome was the second-leading cause of infant death.)³⁶ However, new technologies have not been applied equitably across racial and ethnic or socioeconomic groups.^37,38 Our findings provide augmented impetus toward health care equity, including in access to the tertiary care technologies that can avert deaths in infants born too early or too small. This is a stop-gap measure, as the goal of preventing VLBW itself will be more beneficial toward reducing the inequitable human costs and the unsustainable economic costs associated with relying on advanced technology. The health care costs in just the first six months of life for a single preterm or low-birthweight baby are estimated to range from $76,143 to $603,778.³⁹

If US society wants to facilitate childbearing at a later age and promote health equity, attention must be devoted to reducing the weathering that makes it unsafe. The fact that other OECD countries have substantially lower VLBW and infant mortality rates,⁴⁰ despite their trend toward older maternal age at first birth, implies that improvement is possible. Important differences in maternity and postpartum care and in the generosity of family policies between the US and other OECD countries may contribute to this disparity. Emulating the approaches in these other countries would change the context in which births to US mothers take place and theoretically reduce the severity of weathering.

MATERNITY AND POSTPARTUM CARE

Other high-income countries have more than double the number of providers per birth than the US (and some have four or even five times as many).⁴¹ Many high-poverty US rural counties currently have no maternity care providers.⁴² Increasing the supply of midwives and doulas⁴³ in the US would increase the supply of maternity care providers and lower health care costs without sacrificing quality of care. Randomized controlled trials show that midwife- or doula-assisted births are associated with significantly lower cesarean section rates, less use of anesthesia, and more positive experiences of childbirth overall.⁴⁴

To address the possible role of clinician bias in birth outcome inequity, best-practice protocols that override clinicians’ discretion in life-threatening situations can be institutionalized. The Alliance for Innovation on Maternal Health⁴⁵ provides instructions for standardizing clinical response to the most common causes of delivery complications, ensuring that necessary equipment is in place and actions taken. The alliance and the American College of Obstetricians and Gynecologists are working with state leaders to increase the number of states that have contracted to institutionalize and evaluate these protocols. Legislation could be used to mandate these best-practice protocols in all states.

US culture does not acknowledge the holistic needs of mothers and babies at any stage of the pregnancy experience—least of all postpartum.^46,47 Standard care in the US is one postpartum visit at six weeks. Other high-income countries mandate and subsidize multiple postpartum visits starting in the first week, most withintwenty-fourhoursofhospitaldischarge.⁴¹ Home visitors—midwives or nurses—are mandated to spend several hours per visit to support mothers recovering from childbirth. Their efforts include identifying postpartum complications in mothers and newborns before they become life-threatening, helping mothers establish breast-feeding to promote infant health, reassuring mothers and assisting them with domestic tasks to reduce maternal stressors, and identifying home health hazards and challenges that can be remediated by linking families to appropriate services. Canadian Healthy Child programs have successfully implemented a targeteduniversalapproach⁴⁸ by contacting all families with children to provide services from pregnancy to school entry. Families found to be facing greater challenges are matched with trained home visitors, who provide multifaceted support for up to three years, contingent on need.⁴⁹

FAMILY POLICIES

We celebrate the increased educational and professional opportunities that Black and White women have realized during the past thirty years. However, we and others^50–52 found that the benefits to infant health of higher maternal socioeconomic position are smaller for Black mothers than for White mothers and that White mothers experienced higher VLBW rates in 2017–19 even with increasing income during the study period. This suggests that societal structures have not kept pace with providing the support that working families need. US policies are ranked the least family friendly of high-income countries, and the social safety net is the weakest.^53,54 The stressors associated with parental obligations are multiplied for working-class women, who are unlikely to have any paid personal time or maternity leave or to be able to afford to hire help and are likely to have stressful jobs and long commutes; live in overcrowded or dilapidated housing in disinvested urban and rural areas; and struggle with balancing complex, multigenerational caregiving duties. They also face stagnant wages, weakened unions, soaring health care and housing costs, and job in security, which undermine parents’ capacity to make ends meet.^55,56 Policies such as paid parental leave, subsidized child care, and child allowances are commonplace in peer countries. Consideration should be given to reinstating the expanded Child Tax Credit, which provided $3,600 per child per year during the pandemic. It was estimated to have rescued about one-third of the nine million food-insecure households from food insecurity and ameliorated the stressors of daily life for parents, with positive impacts for infant health.^57–59

Conclusion

To achieve reproductive justice in the US, researchers, practitioners, and policy makers would benefit from employing structural humility and “centering on the margins”⁶⁰—namely centering, not erasing, the ground-truth realities of pregnancy and parenting from the perspectives of the populations that experience the highest risks. Research would benefit from employing theoretical frames that encompass social and epidemiological influences on maternal and fetal health, such as weathering, that are population specific in contour, pace, and severity, owing to inequitable patterning of the social determinants of health, instead of considering maternal age through the lens of universal reproductive development alone. ▪