The systematic review question was formulated using the PECO framework, in which the components are defined as follows: P (Population: pregnant women), E (Exposure: maternal HBV infection), C (Comparison: pregnant women without HBV infection), and O (Outcome: PPH). Cohort studies and case–control studies evaluating the association between maternal HBV infection and PPH were eligible for inclusion. We excluded studies in which all pregnant women were diagnosed with HBV infection combined with other diseases (e.g., COVID-19 or ICP) or had received antiviral therapy. Reviews, comments, case reports, letters, and editorials were excluded. When data from the same study were reported in multiple publications, the report with the largest sample size was included.

We initially performed a search of eight databases from inception to 26 June 2023 with an update on 30 June 2024. These databases were five English sources (PubMed, Embase, Scopus, Cochrane Library, and Web of Science) and three Chinese databases (China National Knowledge Infrastructure, Wanfang databases, and Weipu databases). Searches in Chinese databases were limited to those journals included in the Peking University Core Periodical Catalog, which represents excellent Chinese journals (14). All searches were conducted without any language limitations. Both Medical Subject Headings (MeSH) and free-text terms—such as “Hepatitis B,” “HBV,” “Postpartum Hemorrhage,” “Pregnancy Outcome,” “Cohort Studies,” and “Case-Control Studies”—were used to develop the search strategy. The detailed search strategy is provided in Supplementary Appendix S1.

The retrieved studies were imported into the Endnote library. After removing duplicates, three researchers (JW, KZ, and MY) independently screened titles and abstracts and assessed full texts using the predefined criteria to select the final eligible studies. Any disagreements were resolved through discussion or, if necessary, mediation by a third party (YQ). Additional relevant studies were identified manually by searching the reference lists of the included studies.

Two researchers (JW and JG) independently extracted data from eligible studies using a standardized and predetermined form, which included the following information: study characteristics (e.g., first author, publication year, publication language, study location, study design, time of data collection, and sample size); included participant characteristics (e.g., maternal age); HBV infection characteristics (e.g., diagnostic criteria of HBV infection, the prevalence of HBV infection); PPH characteristics (e.g., diagnostic criteria of PPH, the incidence of PPH); exposed and unexposed characteristics (e.g., number of HBV infection pregnancies and total pregnancies in each group); and effect estimates (e.g., number of PPH cases in the exposed and control groups, adjusted data, and relevant confounders). Any disagreements were resolved through discussion.

In this study, maternal HBV infection was defined as HBsAg seropositivity during pregnancy, regardless of the hepatitis B e antigen (HBeAg) serostatus. In this study, PPH referred to primary PPH, defined as blood loss of more than 500 mL after vaginal delivery or more than 1,000 mL after cesarean delivery within 24 h of delivery (15). Since there has been a slight change in the definition of PPH in recent years, pregnant women who meet the American College of Obstetricians and Gynecologists criteria are also considered to have PPH; according to that criteria, PPH is defined as a cumulative blood loss of 1,000 mL or more or blood loss associated with signs or symptoms of hypovolemia within 24 h of delivery, regardless of the mode of delivery (16).

Two researchers (JW and PZ) independently assessed the risk of bias using the Newcastle–Ottawa scale (NOS), which consists of three parameters: (1) selection, (2) comparability, and (3) outcome (17). The NOS includes eight assessment items with a maximum score of 9, with comparability contributing up to 2 points. We categorized the studies as low, moderate, and high quality, with NOS scores of 1–3, 4–6, and 7–9, respectively. Disagreements were resolved through consensus with a third reviewer (YQ).

We pooled the crude relative risk (cRR), adjusted odds ratio (aOR), and corresponding 95% confidence intervals (CIs) between HBV infection and PPH using the random-effects model (DerSimonian–Laird). The heterogeneity of effect sizes among included studies was assessed using the I² statistic and Cochran's Q test (18), with I² > 50% indicating statistically significant heterogeneity between studies. We performed three subgroup analyses: (1) sample size (more than 1,000 vs. less than 1,000), (2) study region (Asia vs. non-Asia), and (3) prevalence of HBV infection (higher vs. lower). A study by Schweitzer et al., based on 1,800 HBsAg prevalence reports from 161 countries, defined a higher rate of prevalence of HBV infection as >3.61% (3). Furthermore, studies that did not explicitly report prevalence also referred to the data reported by Schweitzer et al. (3). We used the interaction test to estimate the differences between the subgroups (19). In addition, we performed seven sensitivity analyses: (1) omitting studies with NOS scores of less than 7; (2) omitting studies published in non-English languages; (3) omitting studies published as conference abstracts; (4) omitting studies with HBsAg serologic testing not performed during first trimester; (5) omitting studies including pregnancies that received antiviral therapy; (6) omitting studies without reported diagnostic criteria for PPH; and (7) performing a trim-and-fill analysis to evaluate the robustness of pooled effect estimates (20). Publication bias was assessed using funnel plots and Egger's tests (21). All statistical analyses were performed using Stata 18.0 (StataCorp LLC, College Station, TX, USA).

The characteristics of the included studies are summarized in Table 1. A total of 21 studies enrolling 379,782 participants, with 31,069 HBV-infected pregnant women, were included. Sample sizes ranged from 163 to 87,889, with 12 studies recruiting more than 1,000 pregnant women (7, 8, 10–12, 25–30, 37). These studies were conducted in China (n = 15) (7, 8, 10–12, 22–28, 30, 32, 36), Thailand (n = 2) (34, 35), Korea (n = 1) (37), Iran (n = 1) (33), Australia (n = 1) (29), and Romania (n = 1) (31). Seventeen studies were published in English (7, 8, 10–12, 29–31, 33, 35, 37), three in Chinese (28, 32, 36), and one in Thai (34). Three studies included pregnancies that received antiviral therapy. Of these, two studies specified the number: Lok et al. reported 44 individuals (0.5% of the total HBV-infected pregnancies) and Tu et al. reported 69 individuals (17.4%) (22, 27). However, Chen et al. did not specify the number of individuals who received antiviral therapy (24). Among the included studies, 14 did not specify the diagnostic criteria for PPH (7, 10, 12, 24, 28–37). Of the remaining seven, five provided detailed definitions (11, 22, 23, 25, 27), while two defined the outcome based on ICD criteria (8, 26).

Of these 21 studies, 14 reported the prevalence of HBV infection. Among them, 10 reported a prevalence rate higher than 3.61% (7, 10–12, 25–28, 30, 37), while four reported a prevalence rate lower than 3.61% (8, 29, 34, 35). Eleven studies adjusted for the effect of confounders when examining the association between HBV infection and PPH. Five studies were assessed using a multivariate analysis (10, 22, 23, 29, 30) and the remaining six using matching (24, 31, 33–36). All 21 studies scored in the 5–8 range, with 12 studies scoring 7 and above, indicating a relatively low risk of bias (Supplementary Table S1).

Twenty-one studies evaluated the association between maternal HBV infection and PPH, of which 11 provided adjusted results (10, 28, 29, 31, 33–36). The pooled results of the unadjusted data showed that maternal HBV infection was associated with an increased risk of PPH (1,426/31,069 vs. 22,556/348,713, cRR = 1.18, 95% CI: 1.06–1.31, I² = 56.1%) (Figure 2). Furthermore, the pooled results of the adjusted data indicated a similar risk of PPH (295/3,813 vs. 8,058/52,674, aOR = 1.50, 95% CI:1.29–1.73, I² = 0.0%) (Figure 3).

According to the prespecified subgroup analyses, the majority of results indicated that maternal HBV infection was associated with an increased risk of PPH, which was consistent with the overall pooled results (Table 2). No statistically significant differences were found between sample size, study region, and prevalence of HBV infection in subgroup analyses in both unadjusted and adjusted pooled analyses (all P-values for interaction were greater than 0.05, Table 2).

Sensitivity analyses results were mainly consistent with the primary results. After excluding studies with NOS scores less than 7, it was found that the pooled cRR was 1.16 (95% CI: 1.03–1.31) and the aOR was 1.48 (95% CI: 1.21–1.83) (Supplementary Table S2). After excluding studies published in non-English language, it was found that the pooled cRR was 1.16 (95% CI: 1.03–1.31) and the aOR was 1.49 (95% CI: 1.27–1.75). Upon excluding studies published as conference abstracts, it was found that the pooled cRR was 1.16 (95% CI: 1.04–1.30) and the aOR was1.49 (95% CI: 1.23–1.81). In addition, when only studies with HBsAg seropositivity during the first trimester were included, the pooled cRR was 1.10 (95% CI: 0.98–1.23) and the aOR was 1.40 (95% CI: 1.13–1.75). When only studies without antiviral therapy were included, the pooled cRR was 1.20 (95% CI: 1.06–1.36) and the aOR was 1.48 (95% CI: 1.27–1.72). Among the studies that specified the diagnostic criteria of PPH, the pooled cRR was 0.99 (95% CI: 0.89–1.11) and the aOR was 1.90 (95% CI: 0.94–3.86). In addition, trim-and-fill analyses showed that the pooled cRR was 1.16 (95% CI: 1.04–1.29) and the aOR was 1.50 (1.29–1.74) (Supplementary Table S2).

All funnel plots appeared symmetric, and Egger's test did not reveal significant publication bias either in the overall unadjusted pooled analyses (P = 0.12) or adjusted pooled analyses (P = 0.99) (Supplementary Figures S1 and S2).