Background
In both the biomedical and public health literature, the risk for preterm birth has been linked to maternal racial/ethnic background, in particular African-American heritage. Despite this well-documented health disparity, the relationship of comorbid conditions, such as chronic hypertension, to maternal race/ethnicity and preterm birth has received relatively limited attention in the literature.
Objective
The objective of the study was to evaluate the interaction between chronic hypertension and maternal racial/ethnic background on preterm birth.
Study Design
This is a retrospective cohort study of singleton pregnancies among women who delivered between 2002 and 2015 at the University of California, San Francisco. The associations of chronic hypertension with both spontaneous and medically indicated preterm birth were examined by univariate and multivariate logistical regression, adjusting for confounders including for maternal age, history of preterm birth, maternal body mass index, insurance type (public vs private), smoking, substance abuse, history of pregestational diabetes mellitus, and use of assisted reproductive technologies. The interaction effect of chronic hypertension and racial/ethnicity was also evaluated. All values are reported as odds ratios, with 95% confidence intervals and significance set at P = .05.
Results
In this cohort of 23,425 singleton pregnancies, 8.8% had preterm deliveries (3% were medically indicated preterm birth, whereas 5.5% were spontaneous preterm births), and 3.8% of women carried the diagnosis of chronic hypertension. Chronic hypertension was significantly associated with preterm birth in general (adjusted odds ratio, 2.74, P < .001) and medically indicated preterm birth specifically (adjusted odds ratio, 5.25, P < .001). When evaluating the effect of chronic hypertension within racial/ethnic groups, there was an increased odds of a preterm birth among hypertensive, African-American women (adjusted odds ratio, 3.91, P < .001) and hypertensive, Asian-American/Pacific Islander women (adjusted odds ratio, 3.51, P < .001) when compared with their nonhypertensive counterparts within the same racial/ethnic group. These significant effects were also noted with regard to medically indicated preterm birth for hypertensive African-American women (adjusted odds ratio, 6.85, P < .001) and Asian-American/Pacific Islander women (adjusted odds ratio, 9.87, P < .001). There was no significant association of chronic hypertension with spontaneous preterm birth (adjusted odds ratio, 0.87, P = .4).
Conclusion
The effect of chronic hypertension on overall preterm birth and medically indicated preterm birth differs by racial/ethnic group. The larger effect of chronic hypertension among African-American and Asian/Pacific Islander women on medically indicated and total preterm birth rates raises the possibility of an independent variable that is not captured in the data analysis, although data regarding the indication for medically indicated preterm delivery was limited in this data set. Further investigation into both social-structural and biological predispositions to preterm birth should accompany research focusing on the effect of chronic hypertension on birth outcomes.
In both the biomedical and public health literature, the risk for preterm birth has been linked to maternal racial/ethnic background. African-American women have consistently had the highest rates, with almost 1 in 5 experiencing preterm delivery. These disparities carry over into the neonatal period, with approximately 54% of neonatal mortality among the non-Hispanic, African-American population attributable to preterm delivery.
Various etiologies for racial/ethnic disparities in preterm delivery have been hypothesized, from segregationist neighborhood policies to maternal stress and (epi)genetic factors. Despite well-documented health disparities, the relationship of comorbid conditions, such as chronic hypertension, to maternal race/ethnicity and preterm birth has received limited attention in the literature.
Multiple studies have shown the prevalence of chronic hypertension in pregnancy to be increasing, with current rates estimated to be between 1% and 10% in the United States. Multiple etiologies have contributed to the rise in the rate of chronic hypertension, with maternal obesity, increasing maternal age, and the growing prevalence of the metabolic syndrome likely to be important determinants. It has also been shown that chronic hypertension in pregnancy has impacts on not only the rates of pregnancy-related hypertensive diagnoses (eg, preeclampsia/eclampsia, hemolysis, elevated liver enzymes, and low platelet count [HELLP]) but also fetal/neonatal outcomes, such as intrauterine fetal demise, stillbirth, and preterm birth.
The association of chronic hypertension with preterm birth, both medically indicated and spontaneous, has been reported in multiple studies. The role of endothelial dysfunction and abnormal placentation has been posited in the literature as a mechanism for spontaneous preterm birth as well as intrauterine fetal growth restriction and small-for-gestational-age neonates. Data also suggest that different maternal racial/ethnic backgrounds confer a different degree of risk for chronic hypertension, although many of these data are confounded by multiple socioeconomic factors, such as diet, living conditions, and access to medical care.
The interaction of maternal race/ethnicity, chronic hypertension, and preterm birth has the potential to provide important insight into the role of medical comorbidities and the complex social-biological factors influencing disparities in the rates of preterm birth. Few data have focused on this important subset of women. An abstract by Sabol et al noted an increased odds ratio of having preterm delivery in African-American women with chronic hypertension, as compared with hypertensive women of other racial/ethnic backgrounds. However, the data did not stratify for types of preterm delivery (spontaneous or medically indicated) or include other markers of disparity, such as insurance status. In another retrospective study, Samadi et al reported an increased risk of preterm delivery among African-American women with chronic hypertension, in particular those with pregnancy-aggravated hypertension.
We therefore sought to examine rates of medically indicated and spontaneous preterm birth by maternal race/ethnicity among a cohort of pregnant women with chronic hypertension delivering at a tertiary care institution. In particular, we chose to examine the interaction between maternal race/ethnicity and chronic hypertension, with a working hypothesis that African-American and Hispanic/Latina women would experience higher rates of preterm birth and, more specifically, spontaneous preterm birth. We further hypothesized that there would be a synergistic effect with chronic hypertension and maternal race/ethnicity, increasing the overall rate of preterm birth among high-risk racial/ethnic groups, such as African-American and Hispanic/Latina women.
Materials and Methods
Women who delivered at the University of California, San Francisco between Jan. 1, 2002. and Dec. 15, 2015, were included in the study. Our primary predictors were chronic hypertension and maternal race/ethnicity. The primary outcome was the rate of preterm birth (<37 weeks’ gestation); secondary outcomes were spontaneous (including preterm premature rupture of membranes) and medically indicated preterm birth.
Given the practice changes over the study time period, we chose to use the definition of chronic hypertension prior to the most recent American College of Obstetrics and Gynecology practice guidelines for hypertension in pregnancy in 2013: 2 blood pressure readings greater than 140 mm Hg systolic or 90 mm Hg diastolic recorded on at least 2 separate occasions greater than 6 hours apart prior to 20 weeks’ gestational age.
Exclusion criteria included multifetal gestation, known congenital anomalies in the fetus, delivery prior to 20 weeks’ gestational age, and women who received prenatal care elsewhere and were transported to University of California, San Francisco, for delivery.
The University of California, San Francisco, Perinatal Database includes birth and antepartum data on all women who have delivered at the institution since the mid-1970s. Prenatal data are collected by both nurses and physicians at the time of initial presentation to care and input into the database manually. Peripartum and postpartum data are input by physicians as well. Trained abstractors performing daily chart review and data abstraction to ensure accuracy. The database also undergoes monthly review by dedicated, trained physicians for quality assurance.
The demographic and clinical characteristics of the study sample were examined by descriptive statistics. Univariate and multivariable logistic regression modeling was used to assess the effects of maternal racial/ethnic background and chronic hypertension on rates of preterm birth. Multivariable regression modeling adjusted for maternal age, history of preterm birth, maternal body mass index, insurance type (public vs private), smoking, use of substances (ie, cocaine, methamphetamine, heroin/opiates, ecstasy, lysergic acid diethylamide, phencyclidine, and alcohol abuse), history of pregestational diabetes mellitus (type 1 or type 2), and use of assisted reproductive technologies.
The aforementioned variables were chosen for not only their previously documented effects on preterm birth but also their statistical significance on univariate analysis ( P < .05); many of the variables meeting statistical significance are well known in the literature to be correlated with preterm delivery. Maternal smoking was obtained via self-report at initiation of prenatal care or admission; substance use was also obtained via self-report or by positive urine toxicology at initiation of prenatal care or admission. We also examined interaction effects between maternal race/ethnicity and chronic hypertension in multivariable models to evaluate a potential differential impact of chronic hypertension on preterm birth by maternal race/ethnicity.
To account for multiple pregnancies by the same individual in the cohort, maximum likelihood estimates were generated that fit a generalized linear model. This procedure used generalized estimating equations to deal with correlated data arising from repeated measurements. Maternal medical record number was used to identify repeated measures in the models. Maternal white race/ethnicity was used as reference, in line with previous studies focusing on racial/ethnic disparities in obstetric and neonatal outcomes. All statistical analyses were performed with SAS 9.4 (SAS Institute Inc, Cary, NC). Results are reported as odds ratios and 95% confidence intervals, with significance set at P = .05.
Results
There were 23,425 singleton deliveries over the time period of the study. A total of 86 births were excluded from the analyses on the specific type of preterm birth because of limited coding data regarding the type of preterm delivery. Table 1 shows maternal demographic and clinical characteristics across racial/ethnic categories. Nonwhite women constituted more than half of the sample: 46% (n = 10843) were white/Caucasian, 9.7% (n = 2272) were black/African-American, 13.6% (n = 3193) were Latina/Hispanic, and 18.9% (n = 4431) were Asians/Pacific Islanders.
| Total c | White (n = 10,843) | African-American (n = 2272) | Latina/Hispanic (n = 3193) | Asian/Pacific Islander (n = 4431) | Other (n = 2686) a | |
|---|---|---|---|---|---|---|
| Mean maternal age (SD) | 31.7 (5.9) | 33.2 (5.0) | 26.6 (6.5) | 29.5 (6.5) | 32.5 (5.2) | 30.7 (5.8) |
| Nulliparous | 12964 (55.3%) | 6602 (60.9%) | 974 (42.9%) | 1499 (46.9%) | 2416 (54.5%) | 1473 (54.8%) |
| Mean gestational age at delivery (SD) | 38.8 (±2.1) | 39.0 (±2.0) | 38.5 (±2.4) | 38.6 (±2.3) | 38.7 (±1.9) | 38.7 (±2.3) |
| History of prior preterm birth | 843 (3.6%) | 282 (2.6%) | 155 (6.9%) | 151 (4.8%) | 153 (3.5%) | 102 (3.8%) |
| Public insurance | 5876 (25.1%) | 1618 (14.9%) | 1386 (61.0%) | 1220 (38.2%) | 824 (18.6%) | 828 (30.8%) |
| BMI >30 kg/m 2 | 2838 (13.5%) | 908 (9.3%) | 643 (33.0%) | 675 (23.6%) | 344 (8.4%) | 268 (11.8%) |
| In vitro fertilization | 739 (3.2%) | 429 (4.0%) | 14 (0.6%) | 51 (1.6%) | 162 (3.7%) | 83 (3.1%) |
| Chronic hypertension | 913 (3.9%) | 363 (3.3%) | 166 (7.3%) | 126 (3.9%) | 171 (3.9%) | 87 (3.2%) |
| Pregestational diabetes mellitus | 588 (2.5%) | 202 (1.9%) | 60 (2.6%) | 105 (3.3%) | 144 (3.2%) | 77 (2.9%) |
| Cigarette smoking | 562 (2.4%) | 205 (1.9%) | 153 (6.7%) | 71 (2.2%) | 55 (1.2%) | 78 (2.9%) |
| Substance use b | 851 (3.6%) | 302 (2.8%) | 301 (13.2%) | 90 (2.8%) | 55 (1.2%) | 103 (3.8%) |
a Native American, mixed race, other, or unknown
b Defined as cocaine, methamphetamine, heroin/opiates, ecstasy, lysergic acid diethylamide, phencyclidine, and alcohol abuse
c Eighty-four preterm births were unknown for medically indicated or spontaneous preterm birth.
The median maternal age was 32 years, 55.3% of the cohort was nulliparous, and 3.6% of the women carried a history of prior preterm birth. In the overall cohort, 913 (3.9%) carried the diagnosis of chronic hypertension. On univariate analysis, nonwhite women tended to be younger (except Asian/Pacific Islander), parous, smokers (except Asian/Pacific Islander women), on public insurance, to have had a prior history of preterm birth, to have had pregestational diabetes mellitus, and to deliver 0.3–0.5 weeks earlier than white women; they were less likely to have used assisted reproductive technologies to conceive. African-American women had higher rates of chronic hypertension, smoking, and substance use than all other race/ethnicities.
A notable finding was the near equivalence of odds for total preterm delivery on multivariate analysis for pregestational diabetics (adjusted odds ratio, 2.75, 95% confidence interval, 2.18–3.47) and women with chronic hypertension (adjusted odds ratio, 2.74, 95% confidence interval, 2.28–3.29), listed in Table 2 .
| n (column %) | No preterm birth | Preterm birth | Univariate | Multivariate a | |||
|---|---|---|---|---|---|---|---|
| n (row%) | n (row%) | OR (95% CI) | P value | aOR (95% CI) | P value | ||
| Chronic hypertension | |||||||
| No | 22512 (96.1%) | 20653 (91.7%) | 1859 (8.3%) | Reference | Reference | ||
| Yes | 913 (3.9%) | 703 (77.0%) | 210 (23.0%) | 3.20 (2.71–3.79) | <.001 | 2.74 (2.28–3.29) | <.001 |
| Race/ethnicity | |||||||
| White | 10843 (46.3%) | 9965 (91.9%) | 878 (8.1%) | Reference | Reference | ||
| African-American | 2272 (9.7%) | 1996 (87.9%) | 276 (12.1%) | 1.54 (1.33–1.79) | <.001 | 1.08 (0.89–1.30) | .45 |
| Latina/Hispanic | 3193 (13.6%) | 2871 (89.9%) | 322 (10.1%) | 1.27 (1.10–1.45) | <.001 | 1.04 (0.88–1.22) | .66 |
| Asian/Pacific Islander | 4431 (18.9%) | 4087 (92.2%) | 344 (7.8%) | 0.95 (0.83–1.08) | .43 | 0.91 (0.79–1.05) | .20 |
| Other b | 2686 (11.5%) | 2437 (90.7%) | 249 (9.3%) | 1.13 (0.97–1.32) | .11 | 0.92 (0.77–1.10) | .37 |
| Maternal age, y c | |||||||
| <25 | 3110 (13.3%) | 2767 (89.0%) | 343 (11.0%) | 1.45 (1.26–1.66) | <.001 | 1.29 (1.08–1.54) | .005 |
| 25–29 | 4300 (18.4%) | 3919 (91.1%) | 381 (8.9%) | 1.15 (1.00–1.31) | .04 | 1.05 (0.90–1.22) | .55 |
| 30–34 | 8167 (34.9%) | 7536 (92.3%) | 631 (7.7%) | Reference | Reference | ||
| 35–39 | 6070 (25.9%) | 5548 (91.4%) | 522 (8.6%) | 1.10 (0.98–1.24) | .11 | 1.08 (0.95–1.24) | .24 |
| ≥40 | 1746 (7.5%) | 1559 (89.3%) | 187 (10.7%) | 1.42 (1.20–1.69) | <.001 | 1.27 (1.05–1.55) | .02 |
| History of preterm delivery d | |||||||
| No | 22357 (96.4%) | 20531 (91.8%) | 1826 (8.2%) | Reference | Reference | ||
| Yes | 843 (3.6%) | 624 (74.0%) | 219 (26.0%) | 3.18 (2.63–3.85) | <.001 | 2.72 (2.19–3.37) | <.001 |
| BMI, kg/m 2 e | |||||||
| Normal (18–24.9) | 13297 (63.3%) | 12301 (92.5%) | 996 (7.5%) | Reference | Reference | ||
| Overweight (25–29.9) | 4868 (23.2%) | 4448 (91.4%) | 420 (8.6%) | 1.17 (1.04–1.32) | 0.01 | 1.03 (0.91–1.17) | .63 |
| Obese (>30) | 2838 (13.5%) | 2499 (88.1%) | 339 (11.9%) | 1.66 (1.45–1.90) | <.001 | 1.21 (1.04–1.40) | .01 |
| Use of assisted reproductive technology | |||||||
| No | 22686 (96.8%) | 20712 (91.3%) | 1974 (8.7%) | Reference | Reference | ||
| Yes | 739 (3.2%) | 644 (87.1%) | 95 (12.9%) | 1.51 (1.20–1.89) | <.001 | 1.50 (1.16–1.94) | .002 |
| Insurance status | |||||||
| Private | 16817 (71.8%) | 15449 (91.9%) | 1368 (8.1%) | Reference | Reference | ||
| Public | 5876 (25.1%) | 5249 (89.3%) | 627 (10.7%) | 1.34 (1.22–1.49) | <.001 | 1.06 (0.93–1.21) | .36 |
| Unknown | 732 (3.1%) | 658 (89.9%) | 74 (10.1%) | 1.25 (0.98–1.61) | .08 | 1.10 (0.81–1.50) | .54 |
| Cigarette smoking | |||||||
| No | 22863 (97.6%) | 20883 (91.3%) | 1980 (8.7%) | Reference | Reference | ||
| Yes | 562 (2.4%) | 473 (84.2%) | 89 (15.8%) | 1.88 (1.47–2.39) | <.001 | 1.34 (0.99–1.80) | .06 |
| Pregestational diabetes mellitus | |||||||
| No | 22837 (97.5%) | 20911 (91.6%) | 1926 (8.4%) | Reference | Reference | ||
| Yes | 588 (2.5%) | 445 (75.7%) | 143 (24.3%) | 3.26 (2.63–4.03) | <.001 | 2.75 (2.18–3.47) | <.001 |
| Substance use | |||||||
| No | 22574 (96.4%) | 20646 (91.5%) | 1928 (8.5%) | Reference | Reference | ||
| Yes | 851 (3.6%) | 710 (83.4%) | 141 (16.6%) | 2.10 (1.74–2.53) | <.001 | 1.45 (1.12–1.88) | .005 |
a Adjusted for maternal age, history of preterm birth, maternal body mass index >30 kg/m 2 , public insurance status, smoking, use of substances (ie, cocaine, methamphetamine, heroin/opiates, ecstasy, lysergic acid diethylamide, phencyclidine, and alcohol), history of pregestational diabetes mellitus, and use of assisted reproductive technology
b Native American, mixed race, other, or unknown
c A total of 32 entries were excluded because of missing data
d A total of 225 entries were excluded because of missing data
e A total of 2422 entries were excluded because of missing data.
Total preterm birth
Table 2 illustrates the model used for multivariate analysis of preterm birth. A total of 2069 women (8.83%) had preterm birth; 694 (2.97% of all births) had medically indicated preterm births, and 1289 (5.52% of all births) had spontaneous preterm births. The rate of preterm birth was significantly higher in women with chronic hypertension as compared with women without this diagnosis (23% and 8.3%, respectively).
Other notable variables that increased the odds for preterm birth on multivariate analysis include body mass index >30 kg/m 2 , smoking, pregestational diabetes mellitus, substance use, use of assisted reproductive technologies, and maternal age less than 25 years or older than 40 years. Importantly, there was no significant increase in the odds ratio for a preterm birth based on race/ethnicity on multivariate analysis, although there was a modest increase on univariate analysis among African-American (odds ratio, 1.54, 95% confidence interval, 1.33–1.79) and Latina/Hispanic women (odds ratio, 1.27, 95% confidence interval, 1.10–1.45).
Table 3 illustrates the different relationships between chronic hypertension and type of preterm birth. Chronic hypertension was associated with an odds ratio of 3.32 (95% confidence interval, 2.82–3.90) for preterm birth on univariate analysis and an adjusted odds ratio of 2.74 (95% confidence interval, 2.28–3.29) for preterm birth when controlling for confounders.
