Maternal Hepatitis B Infection and Pregnancy Outcomes in the United States: A Population-Based Cohort Study

Abstract

Hepatitis B virus (HBV) is a leading cause of death globally, with chronic infection commonly causing cirrhosis and hepatocellular carcinoma [1]. Though the United States is considered a low-burden country, with routine childhood vaccination recommended since 1991 [2], a substantial number of people are HBV-infected, with estimates around 847 000 [3]. Mother-to-child transmission (MTCT) of HBV is an important mode of viral propagation. To prevent transmission, HBV screening is part of routine prenatal care in the United States, and infants born to hepatitis B surface antigen (HBsAg)–positive mothers are given HBV immune globulin and vaccination [4].

In addition to known risk of MTCT, emerging research suggests that maternal hepatitis B carrier status may contribute to other important adverse pregnancy outcomes, such as increased risk of gestational diabetes mellitus (GDM) [5–8], miscarriage [9], preterm delivery (PTD) [7, 10–12], and antepartum or postpartum hemorrhage [5, 8], but results are inconsistent across studies [13–15]. The biological basis for poor pregnancy outcomes in HBV-infected women is not fully understood, though HBV infection is recognized as a chronic inflammatory state that can precipitate acute hepatitis flares and further complications [16–18]. Limitations of published studies that could explain conflicting results include small sample sizes, heterogeneous patient populations, differences in study designs, and incomplete adjustment for important confounders.

Few data from large population samples have been published in the United States and other low-burden, developed country settings. The epidemiology of HBV infection differs between the United States and highly endemic regions like East Asia. In Asia, mother-to-child and bloodborne HBV transmission predominate [19]. In contrast, in the United States, transmission occurs most commonly through sexual exposure or intravenous drug use in young adults [20]. Due to variance in chronicity of infection related to mode of transmission, comorbid conditions, and clinical care models, we hypothesize that pregnancy-related outcomes may differ between these settings [20]. The primary purpose of our study was to determine whether there is an association between maternal HBV infection and risk of adverse pregnancy outcomes, including GDM, pre-eclampsia, eclampsia, placenta previa, PTD, low birth weight (LBW), small for gestational age (SGA), and large for gestational age (LGA) in the United States, using a large birth cohort from Washington State.

METHODS

Study Population

We conducted a population-based retrospective cohort study using Washington State singleton birth certificates from 1992–2014 linked to the Washington Comprehensive Hospital Abstract Reporting System (CHARS), which contains maternal and infant hospital discharge data. For women who had multiple births during this period, only 1 randomly selected birth was included in the analysis.

The exposed group consisted of mothers with HBV infection ascertained from birth certificates (denoted as “hepatitis B” or “hepatitis B surface antigen positive”) or relevant ICD-9 codes in CHARS (070.20–23, 070.30–33, or V02.61). The unexposed cohort was selected from the full 1992–2014 birth cohort and consisted of women without HBV by birth certificate or ICD-9 codes, frequency matched by birth year in a 5:1 ratio to HBV-infected women.

Maternal-Neonatal Outcomes and Covariates

We ascertained maternal and neonatal outcomes from birth certificates including maternal GDM, pre-eclampsia, eclampsia, placenta previa, PTD (live birth before 37 weeks’ gestation determined by prenatal obstetric estimate), delivery of an LBW neonate (weighing less than 2500 grams), SGA, and LGA. Neonates were classified as SGA if they were in the lowest 10% or LGA if they were in the highest 10% of Washington State birth weight range for a given gestational age for their gender, using previously described methods [21]. We compared LBW infants with a reference category of normal birth weight infants (2500–4500 grams), SGA and LGA newborns with normal for gestational age newborns (10%–90% of the Washington State birth weight range), and PTD infants with those born after 37 weeks in our analysis. Women with preexisting diabetes were excluded from the analysis of maternal GDM.

Maternal demographics and other medical and obstetric history data were obtained from birth certificates, including data on maternal age (stratified as ≤25, 26–35, >35 years), parity (nulliparous or multiparous), level of education (<12 or ≥12 years of education), maternal race (white, black, Hispanic, Asian, Pacific Islander, or other), maternal country of birth (US or foreign born), marital status (married or not married), any tobacco use during pregnancy (no/yes), any alcohol use during pregnancy, trimester of first prenatal visit, number of prenatal visits (stratified as <6, 6–14, >14 visits), and maternal sexually transmitted infections (syphilis, gonorrhea, chlamydia, hepatitis C, and herpes simplex virus). We determined prepregnancy body mass index (BMI) combining maternal height and weight from birth certificates, available from 2003–2014. Mothers were classified as underweight (BMI < 18.5 kg/m²), normal weight (BMI, 18.5–24.9 kg/m²), overweight (BMI, 25.0–29.9 kg/m²), or obese (BMI > 30.0 kg/m²). Excess weight gain during pregnancy was defined as gaining ≥40 pounds for underweight women, ≥35 pounds for normal weight women, and ≥25 pounds for obese women [22].

Statistical Analysis

We used the chi-square test to compare baseline demographic and clinical characteristics between mothers with and without HBV. For each maternal and neonatal outcome, we calculated a crude odds ratio, adjusted odds ratio, and 95% confidence intervals (CIs) associated with maternal HBV using multivariate logistic regression. As these outcomes were rare, the odds ratios estimated crude (RR) and adjusted relative risks (aRR). Potential confounders for adjusted models were selected based on review of the literature [6, 7, 10, 23]. We considered maternal age, race, country of origin, marital status, education level, parity, gravidity prepregnancy BMI, and tobacco use as potential confounders for all outcomes. Additionally, we considered confounding by hypertension for pre-eclampsia and eclampsia outcomes and sexually transmitted diseases for PTD, LBW, SGA, and LGA outcomes. We included potential confounders in final models if they altered the effect estimate by 10% or more. Subgroup analysis included examination of the association between HBV and GDM among women with excess weight gain. We also considered country of origin as a proxy for chronicity of HBV infection and examined the potential for country of origin to modify the association between HBV status and GDM using an interaction term. Analysis was performed using STATA, version 14 (StataCorp, College Station, TX). This study was exempt from review by the Washington State Department of Health Institutional Review Board as we did not have access to personal identifiable information.

RESULTS

Compared with women without HBV, women with HBV were more likely to be Asian (61% vs 8%, P < .001), foreign born (76% vs 23%, P < .001), and older in age (77% vs 64% ≥26 years, P < .001). They were less likely to be overweight or obese (33% vs 50%, P < .001). Women born in Vietnam and China comprised half of all Asian mothers with HBV. Other baseline demographic and clinical characteristics were generally similar between exposure groups (Table 1). Baseline hypertension, alcohol use in pregnancy, and hepatitis C infection were uncommon (≤2% each).

HBV-infected women had a similar risk of GDM (GDM prevalence, 9%) compared with HBV-negative women (GDM prevalence, 5%) after adjustment for mother’s age, race, country of origin, and prepregnancy BMI (adjusted relative risk [aRR], 1.11; 95% confidence interval [CI], 0.92–1.34) (Table 2). In the subgroup of women with excess weight gain during pregnancy, HBV was not significantly associated with GDM (aRR, 1.12; 95% CI, 0.80–1.57), and there was no interaction between HBV and country of origin with respect to GDM.

Women with HBV had a lower risk of delivering SGA infants (13% among HBV-positive vs 10% among HBV-negative; aRR, 0.79; 95% CI, 0.67–0.93). No statistically significant differences were observed in pre-eclampsia (5% among HBV-positive vs 4% among HBV-negative; aRR, 1.06; 95% CI, 0.82–1.35), eclampsia (0.5% among HBV-positive vs 0.3% among HBV-negative; aRR, 2.31; 95% CI, 0.90–5.91), placenta previa (0.4% among HBV-positive vs 0.2% among HBV-negative; aRR, 1.16; 95% CI, 0.35–3.84), PTD (7% among HBV-positive vs 7% among HBV-negative: aRR, 1.15; 95% CI, 0.98–1.34), LBW (5% among HBV-positive vs 5% among HBV-negative; aRR, 1.08; 95% CI, 0.90–1.29), and LGA (7% among HBV-positive vs 10% among HBV-negative; aRR, 1.01; 95% CI, 0.82–1.24) between HBV-infected and -uninfected women.

DISCUSSION

Routine prenatal screening for hepatitis B infection has been recommended by the US Centers for Disease Control and Prevention (CDC) since 1988 and is important for the prevention of MTCT through targeted HBV immunoglobulin and vaccination postdelivery [24]. Our retrospective cohort study, the largest to date conducted in a low-HBV-endemic setting, provides important insights into maternal HBV and risk of adverse maternal and newborn outcomes. We found that infants born to HBV-infected pregnant women did not experience a higher risk of adverse outcomes.

Though the issue remains controversial, several studies in high-burden settings in Asia have suggested an association between maternal HBV infection and other non-transmission-related adverse pregnancy outcomes [5, 6, 8]. Findings across the literature remain inconsistent, with small sample sizes limiting statistical power. A recent large cohort study in China by Liu et al. found maternal HBV carrier status (positive HBsAg) to be modestly associated with increased risk of preterm delivery (odds ratio [OR], 1.18; 95% CI, 1.04–1.34) [11]. However, though statistical significance was achieved in the setting of large sample sizes, the overall effect size was small. As pregnant HBV-infected women may differ with respect to chronicity of infection, degree of liver damage, comorbid conditions, and access to health care, findings from a highly endemic setting may not be relevant for women in the United States where chronic HBV infection is less common [19].

Our large cohort study conducted in a low endemic US setting does not demonstrate statistically significant associations between HBV and adverse pregnancy outcomes. The magnitude of the risk estimate for PTD was, however, similar to findings presented by Liu et al. [11]. Moreover, we found that HBV was associated with a 21% reduction in the risk of SGA. Though the clinical significance of this finding is uncertain, a smaller retrospective cohort study of 1458 HBV-infected women in Florida also found that women with hepatitis B infection were less likely to have SGA infants, with a similar magnitude of association (OR, 0.79; 95% CI, 0.66–0.95) [14]. Another US-based retrospective cohort study encompassing 37 states and 814 HBV-infected women did not find an association between maternal HBV and GDM, intrauterine growth restriction, or pre-eclampsia, though the risk of PTD was significantly higher in HBV-infected women (adjusted OR, 1.65; 95% CI, 1.34–2.02) [10]. A direct comparison of adverse pregnancy outcomes is limited by notable differences in these study populations as compared with ours. For one, fewer HBV-infected women were Asian (14% [10] and 16% [14] vs 61% in our cohort). Furthermore, substance use, which was not routinely ascertained in our study, was more common in infected women than controls in both studies. Finally, the association between HBV infection and pre-eclampsia remains unclear, with some studies hypothesizing that impaired maternal immune response leads to immunotolerance of the fetus and reduced incidence of pre-eclampsia [25]. We did not identify such an association in our cohort. Given the lack of consensus regarding HBV-associated maternal and neonatal outcomes and geographic heterogeneity of demographic patterns throughout the United States, our study contributes important information about HBV-related pregnancy outcomes among infected women of Asian descent.

Our study has several limitations. Most importantly, we were not able to confirm maternal HBV status by serologic testing. In the United States, routine screening of all women for HBsAg during each pregnancy is recommended to prevent neonatal transmission. Specifically, HBV envelope antigen positivity and high HBV DNA levels are important risk factors for vertical transmission [26–28]. Furthermore, liver dysfunction has been independently associated with adverse neonatal outcomes [18]. We were not able to ascertain HBV envelope antigen status, quantify HBV viremia, assess the administration of preventive antiviral therapy, or measure abnormalities in liver function tests, factors that influence the likelihood of viral transmission as well as pregnancy and neonatal outcomes.

Additional limitations include lack of complete information on adequate prenatal care in our data sources. Notably, prenatal care data were missing more frequently among foreign-born and Asian populations, for whom the prevalence of chronic HBV is expected to be the highest. Thus, HBV exposure is likely underascertained in our cohort. Precision of risk estimates for eclampsia and placenta previa was limited as data for these outcomes were only available from 1992–2002. Finally, missing data limited complete evaluation of additional potential confounders, including BMI, HIV status, alcohol consumption, socioeconomic status, and hepatitis C.

CONCLUSIONS

In our Washington State study population, HBV was not associated with clinically significant adverse pregnancy outcomes. Women with HBV were more likely to be Asian, foreign born, and older in age. Given the overall heterogeneity of findings in the literature, it remains important to identify and counsel women with HBV during routine prenatal screening.

Acknowledgments

We thank Seth Rowley for his assistance with Washington State birth certificate and CHARS data management, Alyson Littman for methodologic guidance, and the Washington State Department of Health for data access. Seth Rowley and Alyson Littman are employed by the University of Washington and have no funding sources to disclose.

Financial support. This work was supported by the Host Defense Training in Allergy and Infectious Diseases Grant (grant numer T32 AI007044; KLB and MWT) and the STD and AIDS Research Training Grant (grant number T32 AI007140; HCSK).

Potential conflicts of interest. All authors: no reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References