Racial Differences in Adverse Pregnancy Outcomes and Incident Hypertension: A Mediation Analysis

Abstract

Background

Adverse pregnancy outcomes (APOs) are associated with increased hypertension risk. Black women experience higher rates of APOs and hypertension compared with White women, but whether APOs contribute to Black‐White hypertension disparities remains unknown.

Methods

The nuMoM2b‐HHS (Nulliparous Pregnancy Outcomes Study: Monitoring Mothers‐To‐Be–Heart Health Study) prospectively recruited nulliparous pregnant people at 8 US centers (2010–2013) with follow‐up 2 to 7 years postpartum (2014–2017). Race, which represents a social construct, was self‐identified. The primary outcome was incident hypertension at follow‐up. APOs included hypertensive disorders of pregnancy, preterm birth, and small‐for‐gestational age birth. Potential confounders (psychological health, social factors, and health behaviors) were assessed via validated questionnaires. We used targeted maximum likelihood‐based estimation to estimate Black‐White hypertension differences that would remain if racial disparities in APOs were eliminated.

Results

Among 3335 participants (17% Black; 83% White; mean [SD] age, 27.6 [5.1] years), APO incidence was higher in Black versus White individuals (rate differences [RDs] per 1000 births: hypertensive disorders of pregnancy: 52.5 [95% CI, 20.0–85.1]; preterm birth: 45.4 [95% CI, 20.8–70.1]; small‐for‐gestational age: 102.4 [95% CI, 75.8–129.1]). Incident hypertension at median 3 years postpartum was higher in Black versus White individuals (RD, 107.1 [95% CI, 69.6–144.4]). APOs mediated the racial disparity in hypertension; eliminating Black‐White disparities in APOs would decrease the Black‐White disparity in hypertension by 18.2% for hypertensive disorders of pregnancy (RD, 87.6 [95% CI, 36.1–139.2]), 16.3% for preterm birth (RD, 89.6 [95% CI, 41.7–137.6]), and 13.9% for small‐for‐gestational age (RD, 92.2 [95% CI, 41.6–142.9]).

Conclusions

Black‐White differences in hypertension were partially mediated by APOs. APOs may represent a pregnancy‐specific pathway contributing to racial disparities in hypertension.

Nonstandard Abbreviations and Acronyms

APO

adverse pregnancy outcome

Black

non‐Hispanic Black

CDM

counterfactual disparity measure

HDP

hypertensive disorders of pregnancy

nuMoM2b

Nulliparous Pregnancy Outcomes Study: Monitoring Mothers‐To‐Be

nuMoM2b‐HHS

Nulliparous Pregnancy Outcomes Study: Monitoring Mothers‐To‐Be–Heart Health Study

PTB

preterm birth

RD

rate difference

SGA

small‐for‐gestational age

TMLE

targeted maximum likelihood estimation

White

non‐Hispanic White

Research Perspective.

What Is New?

• In women, the substantial racial differences in hypertension begin early in adulthood; the magnitude of these differences is similar for hypertension in pregnancy and after pregnancy.

• Prevention of new‐onset hypertension in pregnancy is a key target to mitigate disparities in hypertension after pregnancy.

What Question Should Be Addressed Next?

• Given increasing trends in hypertensive disorders of pregnancy and chronic hypertension after pregnancy in young women, research efforts should focus on early risk identification, prevention, and health policies that encourage monitoring of blood pressure beginning in young adulthood before an individual's first pregnancy.

Hypertension is one of the leading risk factors for preventable deaths worldwide, ¹ and, within the United States, hypertension‐related deaths have increased significantly in recent years, accounting for >500 000 deaths in 2019. ² , ³ An estimated 122 million US adults have hypertension, with annual health care expenditures attributable to hypertension exceeding $50 billion. ⁴ , ⁵ , ⁶ The burden of hypertension and hypertension‐related morbidity and mortality differs significantly by race and ethnicity, with 2‐ to 3‐fold higher rates among Black individuals compared with White individuals. ⁷ , ⁸ In addition to greater risk for hypertension among Black individuals, hypertension also develops 5 years earlier, on average, compared with White individuals. This underscores the importance of targeting prevention efforts during young adulthood when risk and disparities for hypertension may be emerging.

Prevention efforts for hypertension focused on women, who are of particular interest given the higher relative risk for cardiovascular disease (CVD) and stroke associated with hypertension among women compared with men. ⁹ , ¹⁰ , ¹¹ Among women, pregnancy‐specific risk factors may represent a unique window of opportunity to target prevention of hypertension, which was recently highlighted in the 2025 American Heart Association/American College of Cardiology/Multi‐Society High Blood Pressure Guideline. ¹² Specifically, adverse pregnancy outcomes (APOs) have been associated with a 2‐fold increased risk for hypertension within 5 years after delivery, with the greatest magnitude of association observed for hypertensive disorders of pregnancy (HDP). ¹³ , ¹⁴ Emerging data also demonstrate a significant association for other APOs, such as preterm birth (PTB) and small‐for‐gestational age (SGA) infant, with incident hypertension. ¹³ , ¹⁴ This is especially relevant as these 3 APOs complicate ≈20% to 30% of pregnancies in the United States. ¹⁵ Furthermore, racial differences in rates of these APOs parallel those observed for chronic hypertension outside of pregnancy, with Black individuals at higher risk of HDP, PTB, and SGA birth compared with White individuals ¹⁶ , ¹⁷ As a result, growing emphasis has been placed on the prevention of APOs, not only to reduce the risk of short‐term perinatal complications, but also as a potential target to mitigate racial disparities in lifetime risk of hypertension and CVD. ¹³ , ¹⁸ , ¹⁹ , ²⁰ However, the extent to which prevention of APOs would mitigate racial differences in hypertension years after delivery remains unclear.

The goal of this study was to use statistical mediation methods to address this critical question with observational data. ²¹ , ²² In the present analysis, we leveraged data from a prospective, multicenter cohort of nulliparous individuals to examine the extent to which APOs mediate the Black‐White racial disparity in incident hypertension 2 to 7 years after delivery. First, we quantified Black‐White disparities in rates of APOs (ie, HDP, PTB, and SGA) and incident hypertension years after delivery. Second, we estimated a counterfactual disparity measure (CDM) for incident hypertension, which is an important public health metric that estimates the extent to which Black‐White disparities in hypertension would be reduced if APOs could be prevented.

METHODS

Data Source

The parent study, nuMoM2b (Nulliparous Pregnancy Outcomes Study: Monitoring Mothers‐To‐Be), enrolled 10 038 nulliparous pregnant individuals with singleton pregnancies between 2010 and 2013 at 8 clinical centers in the United States (Case Western Reserve University; Columbia University; Indiana University; University of Pittsburgh; Northwestern University; University of California at Irvine; University of Pennsylvania; and University of Utah). As part of the nuMoM2b‐HHS (nuMoM2b–Heart Health Study), a subcohort of 4508 individuals were recruited for an in‐person visit 2 to 7 years following delivery (2014–2017), at which time a rigorous assessment of cardiovascular risk factors was performed. All local Institutional Review Boards approved the study, and all participants provided written informed consent. Full details of nuMoM2b and nuMoM2b‐HHS protocols are available elsewhere. ²³ , ²⁴ Data from nuMoM2b and nuMoM2b‐HHS are available on BioData Catalyst (National Heart, Lung, and Blood Institute; National Institutes of Health; US Department of Health and Human Services). The present analysis was undertaken in accordance with the Strengthening the Reporting of OBservational studies in Epidemiology guidelines. ²⁵

Study Population

Analytic sample selection is depicted in Figure S1. The analytic sample included all non‐Hispanic Black and non‐Hispanic White adults aged ≥18 years at baseline who participated in the nuMoM2b‐HHS study. Given the focus on Black‐White disparities, we restricted our sample to those individuals who self‐reported race and ethnicity as non‐Hispanic Black (Black) or non‐Hispanic White (White). For this analysis, follow‐up began at study enrollment in the original nuMoM2b study, and continued until the nuMoM2b‐HHS visit, which occurred from 2 to 7 years after delivery. We excluded individuals with a pregnancy that ended before 20 weeks' gestation or those with a fetal demise. Given the outcome of incident hypertension after delivery, we additionally excluded those with prepregnancy hypertension (2.0% and 8.1% of White and Black individuals, respectively); this imbalance may be associated with conservative bias in effect estimates.

Exposure, Mediators, and Outcome Measures

The primary exposure was self‐reported race and ethnicity, which included non‐Hispanic Black and non‐Hispanic White. Race and ethnicity, which are social constructs, were self‐identified at enrollment and used in this analysis to represent a complex array of differences in lived experiences, including racism and discrimination. The primary mediator was HDP, which was defined as a composite of preeclampsia (with or without severe features), eclampsia, and/or gestational hypertension (defined as new‐onset antepartum hypertension). ²⁶ Trained study personnel abstracted and verified HDP for all participants according to a priori definitions. ²³ , ²⁴ In secondary analyses, we investigated 2 additional potential mediators: PTB (defined as gestational age at birth of <37 weeks) and SGA (defined as birth weight of <10th percentile) based on Alexander birth weight curves. ²⁷

The primary outcome was incident hypertension at the nuMoM2b‐HHS follow‐up visit, defined by blood pressure (BP) ≥130/80 mm Hg or use of antihypertensive medications. Secondary outcomes included continuous measures of both systolic and diastolic BP (mm Hg), which were each averaged from direct measurements done in triplicate following a standardized research protocol (Omron HEM‐907XL). ²⁸

Covariates

Our proposed directed acyclic graph is visualized in the Figure1. We used 2 sets of confounders in these analyses. First, for estimation of the disparity in BP‐related outcomes, we adjusted for demographics (maternal age at first pregnancy) and socioeconomic markers (insurance, income, area deprivation index, and place of birth). ²⁹ , ³⁰ Then, to quantify mediation effects (eg, CDM), an assumption of no unmeasured confounding of the HDP‐outcome relationships was used. For these analyses, we additionally adjusted for early pregnancy health behaviors and factors (diet, physical activity, smoking, sleep, and body mass index), psychological health (anxiety, depression, and stress), social factors (experiences of racial discrimination, nativity, health literacy, resilience, and social support), and pregnancy‐specific factors (fetal sex). A detailed description of these confounders is included in Table S1.

Figure 1. A directed acyclic graph representing the hypothesized mechanism behind this health disparity: race and ethnicity as the social exposure of interest, HDP as a potential mediator, and cardiovascular health (and its components) 2 to 7 years postpartum as the outcome.

Nodes represent variables, and edges represent hypothesized 1‐directional relationships between variables (absence of edges implies independence between nodes). This diagram assumes that the exposure (race and ethnicity) is associated with confounders of the HDP‐outcome relationship, which we believe is plausible in this scenario. This diagram targets adverse pregnancy outcomes as the mediators and is meant to imply that this is 1 possible set of potentially causal pathways for disparities in cardiovascular health in the postpartum period. HDP indicates hypertensive disorders of pregnancy; PTB, preterm birth; and SGA, small‐for‐gestational age.

Statistical Analysis

First, we assessed the quantity of missing data. As study variables were missing in 0.2% to 27.1% of the sample (Table S2), we applied multiple imputation by chained equations to create 10 complete data sets. This procedure assumes missingness at random, here conditional on the confounders and the exposure, mediators, and outcomes. Data S1 contains more details about the multiple imputation process and presents a comparison of early pregnancy characteristics before and after imputation (Table S3). Wald‐style CIs were created for all estimated quantities using the Rubin rules. A 5% type‐I error rate was set a priori. We show these 95% CIs along with P values. ³¹

Next, we descriptively analyzed antenatal characteristics of the final sample stratified by race. We then calculated crude rates per 1000 births of APOs by race. Rate differences were reported to quantify the absolute Black‐White disparities in APOs and incident hypertension. Black‐White disparities in continuous BP outcomes (systolic BP and diastolic BP) were defined as marginal mean differences. Adjustment of the exposure‐outcome relationship is not required for identification of the CDM, so unadjusted differences are presented. To capture the proportion of the Black‐White disparity in BP outcomes that would remain if APOs were eliminated, we used the CDM framework described by Naimi et al. ²¹ This framework treats the mediator (ie, the APO) as the point of intervention, and captures the Black‐White disparity in the outcome that would remain if the mediator was set to a particular value. Like prior publications, we studied a hypothetical intervention where each APO was separately prevented from occurring (m=0). The results from this analysis can be viewed as a maximal upper bound for the effectiveness of an intervention targeted toward reducing the incidence of APOs and understood as conveying the degree to which disparities in incident hypertension and BP outcomes are mediated by APOs. The formal definition and key assumptions to this approach are outlined in Data S1.

To estimate the CDM, we used targeted maximum likelihood estimation (TMLE), which is a statistical estimation technique that is preferred for its doubly robust properties in observational data. ³² It begins by calculating initial counterfactual outcome predictions conditional on the exposure‐outcome and mediator‐outcome confounders ³³ ; in our setting, these are the risks of incident hypertension at the postpartum visit in White and Black individuals who do not experience a particular APO (eg, HDP). However, as traditional methods are targeted to estimates of the conditional outcomes, the resulting estimate of the CDM may be biased if the model used to capture the confounding structure is misspecified. The next step in TMLE performs a bias correction by including an inverse probability weight, which is labeled as “clever covariate” as an independent variable in a regression model with the logit of the predicted counterfactual outcomes from step 1, allowing the counterfactual outcome predictions to “fluctuate” in another direction that still targets the same outcome prediction. This second step makes TMLE a doubly robust estimator: TMLE is a consistent (statistically unbiased) estimator of the CDM if: (1) the model for the mediator is correctly specified, or (2) the model for the outcome and the model that incorporates the fluctuation are correctly specified. Additionally, the TMLE approach includes exposure‐mediator interactions. Before implementation of TMLE for each analysis with a nondichotomous outcome, the outcome was scaled to the interval [0,1]; results were translated back to the original scale for reporting. The TMLE “clever covariate” was truncated at 0.05, with a maximal weight of 20, for stability. More details about the TMLE approach are available in Data S1.

Results are displayed as the residual or remaining mean Black‐White differences in BP outcomes if no pregnant individual experienced an APO (ie, a case in which there was parity in APO frequency). Using the estimated marginal mean differences in BP outcomes, we derived the proportion of the disparity reduction attributable to prevention of HDP as follows: [marginal mean difference—CDM (when mediator=0)]/marginal mean difference × 100. To calculate Wald‐style 95% CIs for the estimates of CDM, we used estimates of the asymptotic SE calculated from the efficient influence function. ³⁴ , ³⁵

We used Stata 17 for descriptive data analysis, and R v4.2.1 for implementation of TMLE. P<0.05 was considered statistically significant.

RESULTS

Baseline Characteristics

Of the 3335 individuals included in the current analysis (Figure S1), 2753 (82.5%) were White and 582 (17.5%) were Black (Table 1). White individuals were older than Black individuals at enrollment in the first trimester of pregnancy (mean, 28.3 [SD, 5.1] versus 24.0 [SD, 5.2] years). Most (84.7%) of White individuals had private insurance, whereas most (58.6%) of Black individuals had public insurance. White individuals had higher income (451% [SD, 294%] versus 150% [SD, 245%] of the 2013 poverty line) and resided in less disadvantaged areas (39.0 [SD, 25.4] versus 74.3 [SD, 25.8] on the 2015 Area Deprivation Index) than Black individuals.

Table 1.

Baseline Characteristics of the Final Study Cohort, Stratified by Race of the Birthing Individual (n=3335)

Characteristic	Black	White
No. of birthing individuals	582	2753
Demographics and socioeconomic markers
Age in early pregnancy, mean±SD, y	24.0±5.2	28.3±5.1
Health insurance, %
Private	36.8	84.7
Public	58.6	12.4
Other	4.6	2.9
Income as % of the 2013 poverty level, mean±SD	150±245	451±294
Nativity (US born), %	94.5	96
Health literacy, %*	64.8	92.6
Area deprivation index (2015), mean±SD	74.3±25.8	39.0±25.4
Psychosocial factors
Anxiety, mean±SD†	15.3±9.0	13.3±8.6
Depression, mean±SD‡	6.4±4.7	5.4±4.0
Stress, mean±SD§	15.0±7.3	11.8±6.2
Experienced racial discrimination, %||	18.9	2.0
Resilience, mean±SD¶	79.1±13.5	79.7±11.0
Social support, mean±SD#	57.8±17.2	63.6±13.4
Early pregnancy** health factors, mean±SD
Body mass index, kg/m2	30.2±8.5	25.9±5.9
Systolic blood pressure, mm Hg	112.4±11.9	109.6±10.4
Diastolic blood pressure, mm Hg	68.2±8.9	67.5±8.3
Mean arterial pressure, mm Hg	83.0±8.8	81.6±8.0
Glucose, mg/dL	86.9±29.3	87.6±28.1
Non‐HDL cholesterol, mg/dL	104.3±28.0	117.0±31.5
Early pregnancy** health behaviors
Smoking, %	7.5	4.8
Physical activity, mean±SD, min/wk	138±178	190±177
Dietary qualityh, mean±SD	53.5±11.4	64.9±12.2
Sleep, mean±SD, average hours per night	7.9±1.6	7.9±1.1
Pregnancy factors
Female fetal sex, %	50.1	52.1

Black indicates non‐Hispanic Black; HDL, high‐density lipoprotein; and White, non‐Hispanic White.

^*Health literacy assessed via the Rapid Estimate of Adult Literacy in Medicine short form (range, 0–7).

^† Anxiety measured as the State–Trait Anxiety Inventory (range, 20–80).

^‡ Depression measured via the Edinburgh Postnatal Depression Scale (range, 10–40).

^§ Stress measured via the perceived stress scale‐10 (range, 10–50).

^|| Experiences with racism assessed via the Krieger Scale (score of at least 2).

^¶ Resilience measured via the Connor‐Davidson Resilience Scale (range, 25–125).

^# Social support measured via the Multidimensional Scale of Perceived Social Support (range, 12–84).

^**Measurements taken at first study visit, occurring between 6 and 13 weeks of gestation.

Black and White individuals reported differences in psychosocial stressors during pregnancy. White individuals reported more perceived social support compared with Black individuals (63.6 [SD, 13.4] versus 57.8 [SD, 17.2] points). In contrast, Black individuals reported more stress (15.0 [SD, 7.3] versus 11.8 [SD, 6.2] points), depression (6.4 [SD, 4.7] versus 5.4 [SD, 4.0] points), and anxiety (15.3 [SD, 9.0] versus 13.3 [SD, 8.6] points) than White individuals (Table 1). In addition, 18.9% of Black individuals reported experiencing racial discrimination, versus 2.0% of White individuals.

In early pregnancy, White individuals compared with Black individuals had similar or better levels of measured health factors: more physical activity (190 [SD, 177] versus 138 [SD, 178] minutes per week), better diet quality (64.9 [SD, 12.2] versus 53.5 [SD, 11.4] points), and comparable sleep duration (7.9 [SD, 1.1] versus 7.9 [SD, 1.6] hours per night). Black individuals had higher mean body mass index (30.2 [SD, 8.5] versus 25.9 [SD, 5.9] kg/m²) and blood pressure in early pregnancy (systolic BP: 112.4 [SD, 11.9] versus 109.6 [SD, 10.4] mm Hg; and diastolic BP: 68.2 [SD, 8.9] versus 67.5 [SD, 8.3] mm Hg) and were more likely to smoke (7.5% versus 4.8%).

Racial Disparities in APOs

Table 2 presents unadjusted rates of APOs, including HDP, PTB, and SGA, in Black and White birthing individuals per 1000 births. We observed higher rates of HDP in Black individuals (200.2 versus 147.6 per 1000 births; rate difference [RD], 52.5 [95% CI, 20.0–85.1] per 1000 births; P<0.01). Rates of both PTB and SGA were also higher in Black individuals (RD, 45.4 [95% CI, 20.8–70.1] and 102.4 [95% CI, 75.8–129.1] per 1000 births, respectively; P<0.01).

Table 2.

Rates of APOs and Racial Differences in Black and White Birthing Individuals (n=3335)

APO	Black (n=582)		White (n=2753)		Absolute disparity
	No. of cases	Rate per 1000 live births	No. of cases	Rate per 1000 live births	Rate difference (95% CI)	P value*
Hypertensive disorders of pregnancy†	117	200.2	406	147.6	52.5 (20.0–85.1)	<0.01
Preterm birth‡	70	120.3	206	74.8	45.4 (20.8–70.1)	<0.01
Small‐for‐gestational age§	107	183.5	223	81.1	102.4 (75.8–129.1)	<0.01

APO indicates adverse pregnancy outcome; Black, non‐Hispanic Black; and White, non‐Hispanic White.

^* P value from an unadjusted logistic regression model with race as the independent variable and APO as the dependent variable.

^†Hypertensive disorders of pregnancy include gestational hypertension, preeclampsia, and eclampsia.

^‡ Preterm birth defined as before 37 weeks' gestation.

^§ Small‐for‐gestational age defined as <10th percentile using Alexander et al (1996). ²⁷

Racial Disparity Analysis: Hypertension Years After Delivery

The mean (SD) follow‐up time from the delivery visit to the nuMoM2b‐HHS follow‐up visit was 3.2 (1.0) years in Black and 3.2 (0.9) years in White birthing individuals. As shown in Table 3, we observed higher unadjusted BP levels in Black individuals than in White individuals (systolic BP: 114.1 versus 111.2 mm Hg; mean difference, 2.9 [95% CI, 1.9–3.9] mm Hg; P<0.01; diastolic BP: 73.9 versus 71.8 mm Hg; mean difference, 2.1 [95% CI, 1.2–3.0] mm Hg; P<0.01). Black individuals had significantly higher rates of incident hypertension (312.5 versus 205.4 per 1000 births; RD, 107.1 [95% CI, 69.6–144.4]; P<0.01).

Table 3.

Rates of Incident Hypertension and Continuous BP Levels and Absolute Differences Years After Delivery in Black and White Birthing Individuals (n=3335)

Primary outcome	Black (n=582)	White (n=2753)	Absolute disparity
	Rate per 1000 live births	Rate per 1000 live births	Rate difference (95% CI)	P value*
Incident hypertension†	312.5	205.4	107.1 (69.6–144.4)	<0.01

Secondary outcomes	Mean value	Mean value	Mean differences (95% CI)	P value*
Systolic blood pressure, mm Hg	114.1	111.2	2.9 (1.9–3.9)	<0.01
Diastolic blood pressure, mm Hg	73.9	71.8	2.1 (1.2–3.0)	<0.01

Black indicates non‐Hispanic Black; BP, blood pressure; and White, non‐Hispanic White.

^* P values from an unadjusted linear regression model with race as the independent variable and BP as the dependent variable.

^† Incident hypertension defined as BP ≥130/80 mm Hg or on antihypertensive treatment at study visit 2 to 7 years postpartum.

As illustrated in Table 4, which displays CDM estimates, the mean differences in BP outcomes at 2 to 7 years after delivery were significantly mediated by differences in HDP, PTB, and SGA. Specifically, the Black‐White disparity in incident hypertension (the marginal CDM) would be reduced by 18.2% (ie, to an RD of 87.6 [95% CI, 36.1–139.2] per 1000 births) if HDP was hypothetically eliminated. The proportion mediated was slightly lower for PTB (16.3%) and SGA (13.9%). For the secondary outcomes of systolic BP and diastolic BP, HDP, PTB, and SGA also significantly mediated racial differences. Specifically, for these 3 APOs, the Black‐White disparities in systolic BP and diastolic BP would be attenuated by an expected 1.9 to 2.2 mm Hg (percentage mediated ranging from 23.0% to 36.0%) and by 1.4 to 1.7 mm Hg (percentage mediated from 22.6% to 35.6%).

Table 4.

Estimated CDM for Black‐White Disparities in BP Outcomes Years After Delivery (n=3335)

Variable	Hypertensive disorders of pregnancy		Preterm birth		Small‐for‐gestational age
	CDM (95% CI)*	% M	CDM (95% CI)*	% M	CDM (95% CI)*	% M
Incident hypertension	87.6 (36.1–139.2)	18.2 (14.5–22.0)	89.6 (41.7–137.6)	16.3 (13.3–19.4)	92.2 (41.6–142.9)	13.9 (11.7–16.0)
Systolic BP, mm Hg	2.2 (0.7–3.8)	23.0 (17.8–28.3)	1.9 (0.4–3.3)	36.0 (32.6–39.4)	2.2 (0.8–3.6)	24.1 (21.3–27.0)
Diastolic BP, mm Hg	1.6 (0.3–2.9)	26.3 (21.5–31.0)	1.4 (0.2–2.6)	35.7 (31.7–39.7)	1.7 (0.4–2.9)	22.6 (19.1–26.1)

% M indicates percentage mediated; Black, non‐Hispanic Black; BP, blood pressure; CDM, counterfactual disparity measure; and White, non‐Hispanic White.

^*CDMs per 1000 live births were calculated via targeted maximum likelihood estimation. Associated Wald‐style 95% CIs were calculated using empirical estimates of asymptotic influence function‐based variance. Analyses were adjusted for variables listed in Table S3.

DISCUSSION

In this multicenter longitudinal cohort study, rates of APOs (HDP, PTB, and SGA) and incident hypertension 2 to 7 years after delivery were significantly higher among Black individuals compared with White individuals. The racial difference in incident hypertension was significantly mediated by each of the 3 APOs studied here. As such, APOs may represent a pathway and target for prevention that contribute to approximately one‐fifth of the racial disparities in chronic hypertension. Conversely, this suggests that much of the Black‐White disparity observed in incident hypertension in young adulthood remained after hypothetical elimination of APOs. This signifies that the broader context of racial disparities in APOs, hypertension, and CVD, and future work, should examine additional pathways that represent structural and systemic barriers (eg, racism) experienced before the pregnancy. In the present sample, modest differences in BP levels were present between Black and White individuals even in early pregnancy; however, even exposure to modestly higher BP in adults has been shown to have a meaningful impact on CVD risk. Accordingly, opportunities to fully mitigate racial disparities in hypertension should focus on targeting risk factors upstream of APOs in addition to after pregnancy.

The current analysis extends and expands up the available epidemiologic evidence describing significant racial disparities in incident hypertension and newly considers the role of APOs with a life‐course approach. It is well established in nonpregnant populations that BP levels are higher among Black individuals compared with White individuals as early as late adolescence. ³⁶ The present analysis highlights pregnancy‐specific pathways via APOs as a partial contributor to racial disparities in incident hypertension even among young adults. A similar degree of mediation was observed for each of these interrelated APOs, confirming their role as a window into future cardiovascular health. In addition to higher rates of APOs among Black individuals compared with White individuals, retrospective data additionally suggest that postpartum BP trajectories after HDP may manifest a less rapid decline among Black individuals that may translate into higher rates of chronic hypertension after HDP. ³⁷

Nearly one‐fifth of the racial disparity in incident hypertension was mediated by APOs. Therefore, prevention of APOs may be an important target for interventions beginning before an individual's first pregnancy and to promote health equity. Improving prepregnancy health care access coupled with support for health behavior interventions to optimize cardiovascular health among pregnancy‐capable individuals is imperative to achieve health equity both for the pregnant person as well as for the offspring. Individuals are increasingly entering pregnancy with worse levels of cardiovascular health factors, including body mass index, physical activity, and dietary quality. ³⁸ This loss of ideal cardiovascular health by late adolescence is associated with more frequent pregnancy complications and higher risk of CVD. ³⁹ Worsening prepregnancy cardiovascular health may also have broader public health implications, as there is evidence of intergenerational transmission of cardiovascular risk associated with maternal cardiovascular health, regardless of whether an APO occurs. ⁴⁰ Greater emphasis on the preservation and optimization of cardiovascular health is particularly important during the reproductive years. A joint statement by the American College of Obstetricians and Gynecologists and the American Heart Association highlighted recommendations by the US Preventive Services Task Force and the American College of Obstetricians and Gynecologists for clinicians to provide pregnancy‐capable persons with referrals to behavioral interventions or other resources to support healthful behaviors before conception. ¹³ , ⁴⁰ , ⁴¹ , ⁴² In addition to individual‐level interventions, broader policy‐level efforts are needed to enhance access to and continuity of health care for individuals before, during, and after pregnancy to equitably achieve optimal cardiovascular health. Moreover, once HDPs are identified, postpartum interventions to prioritize screening and awareness for the risk of chronic hypertension is needed given racial disparities in chronic hypertension that persist across the life course.

Limitations

This study should be interpreted with the known limitations of all observational analyses, including potential for unmeasured confounding. We tried to mitigate unmeasured confounding as much as possible by adjusting for many available measured confounders in nuMoM2b‐HHS. However, data on family medical history and alcohol intake were not available, but it is unlikely that these variables would significantly alter the findings. Prepregnancy BP was not available, which is an important confounder of the mediator‐outcome association. However, prepregnancy BP and early pregnancy BP are highly correlated.

This study has a few additional limitations to note. First, we focused only on non‐Hispanic Black and White individuals in the present analysis given our focus on the health implications of anti‐Black racism and the disproportionate burden of BP‐related outcomes experienced by Black individuals in pregnancy and throughout the life course. These results may not generalize to other minoritized populations, and futures studies in disaggregated Hispanic and Asian populations are needed. Second, although completely eradicating APOs is not plausible, this approach allows the modeling of the portion of incident hypertension that is mediated by APOs; the estimates of CDMs represent the maximum possible benefit that could be achieved from an intervention targeting APOs, which is an important public health metric. Third, we were unable to examine individual components of HDP as well as pathways involving multiple APOs in this analysis. Fourth, the nuMoM2b cohort is still young at follow‐up (mean age of 31 years at the HHS visit), so we were unable to examine differences in clinical CVD events (eg, coronary heart disease) in this sample. Therefore, we focused on hypertension as a causal and key CVD risk factor for the study outcome. Fourth, formal model assessment was not conducted for each of the underlying models contributing to estimation of the CDM; model specification was dictated by the causal model. Last, the estimand used here standardizes the CDM estimates to the overall population instead of the Black subpopulation, which may not reflect their covariate distribution. ²¹

CONCLUSIONS

In this multicenter, longitudinal cohort, we studied the pathways between self‐identified race, APOs, and incident hypertension years after delivery using state‐of‐the‐art causal inference techniques. We demonstrate that racial disparities in incident hypertension after delivery were significantly mediated, in part, via APOs. However, most of the racial disparity remained even following hypothetical elimination of APOs. These data suggest that interventions upstream of the first pregnancy before an APO occurs are needed to complement postpartum preventive strategies to mitigate racial inequities in birth outcomes and incident hypertension after delivery.

Sources of Funding

This study was supported by grants from the National Heart, Lung, and Blood Institute (R01 HL161514) to S.S.K. Grant funding was obtained from the Eunice Kennedy Shriver National Institute of Child Health and Human Development: U10 HD063036; U10 HD063072; U10 HD063047; U10 HD063037; U10 HD063041; U10 HD063020; U10 HD063046; U10 HD063048; and U10 HD063053. In addition, support was provided by Clinical and Translational Science Institutes: UL1TR001108 and UL1TR000153. Cooperative agreement funding was obtained from the National Heart, Lung, and Blood Institute and the Eunice Kennedy Shriver National Institute of Child Health and Human Development: U10‐HL119991; U10‐HL119989; U10‐HL120034; U10‐HL119990; U10‐HL120006; U10‐HL119992; U10‐HL120019; U10‐HL119993; U10‐HL120018, and U01HL145358; and the National Center for Advancing Translational Sciences through UL‐1‐TR000124, UL‐1‐TR000153, UL‐1‐TR000439, and UL‐1‐TR001108; and R01HL124649, U54 AG065141, the Barbra Streisand Women's Cardiovascular Research and Education Program, Cedars‐Sinai Medical Center, Los Angeles, the Linda Joy Pollin Women's Heart Health Program, the Erika Glazer Women's Heart Health Project, and the Adelson Family Foundation, Cedars‐Sinai Medical Center (Los Angeles, CA); and supplemental support to U10‐HL119991 from the Office of Research on Women's Health, Office of Behavioral and Social Sciences Research, and the Office of Disease Prevention.

Disclosures

Dr Petito reports receiving funding for unrelated work from Omron Healthcare Co, Ltd.

Supporting information

Figure S1

Tables S1–S3

Data S1. Supplemental Methods

References 43–45

STROBE Checklist