An umbrella review of maternal pre-pregnancy characteristics and prior cesarean section as risk factors for stillbirth

Abstract

Background

Stillbirth is the death of a fetus at or after 20 weeks of gestation, making up 60% of perinatal deaths and having significant psychosocial and economic impacts. The present umbrella review assessed maternal risk factors before pregnancy and prior cesarean associated with stillbirth based on meta-analytic studies.

Materials and methods

We conducted a search of three major databases until January 2025, focusing on meta-analyses that evaluated environmental risk factors associated with stillbirth. We evaluated the strength of evidence for identified maternal characteristics associated with stillbirth using the validated Ioannidis classification system. We calculated summary effect estimates, 95% confidence intervals (CIs), heterogeneity (I²), 95% prediction intervals, small-study effects, excess significance biases, and sensitivity analyses. To assess the quality of the meta-analyses, we employed the AMSTAR 2 tool.

Result

This umbrella review included five studies, comprising eight meta-analyses that evaluated 365,158 cases of stillbirth across a total sample size of 70,124,598 participants. Race (black vs. white), chronic hypertension, and maternal age ≥ 35 years showed highly suggestive associations with stillbirth (Class II), while obesity and overweight showed suggestive associations (Class III). Associations for previous cesarean, and pre-existing diabetes were weak (Class IV). Given high heterogeneity and low AMSTAR-2 ratings, these findings should be interpreted as hypothesis-generating rather than causal.

Conclusion

These results synthesize meta-analytic evidence on maternal characteristics associated with stillbirth. The overall certainty is limited by considerable heterogeneity, and the low/critically low quality of included reviews; therefore, causal inferences are not warranted.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-025-08560-6.

Introduction

Stillbirth is the death of a fetus that occurs at or after 20 weeks of gestation, and it has significant psychosocial and economic impacts, and it accounts for 60% of all perinatal deaths [1]. The global health community considers the rate of stillbirth to be a significant indicator of the quality of care provided by health systems during pregnancy and childbirth [2]. This issue is addressed in a Every Newborn Action Plan (ENAP), which is led by UNICEF and WHO [3], as well as in the UN Global Strategy for Women’s, Children’s, and Adolescents’ Health for the years 2016–2030 [4].

Approximately two million stillbirths occur globally each year, with about half being potentially preventable [5]. However, some stillbirths cannot currently be prevented with evidence-based interventions, and others may never be [5]. Despite detailed clinical evaluations significant proportion of stillbirths remain unexplained [6]. Various studies have shown this percentage to vary between 25% and 60% [6]. As a result, researchers are actively exploring the various factors that influence the occurrence of stillbirth and its associated impacts. This variation arises from differences in geographic regions, socioeconomic conditions, reproductive and environmental health, and healthcare infrastructure [7–9].

Several risk factors for stillbirth have been documented, maternal factors, such as advanced maternal age, teenage pregnancies, poor nutritional status, a history of prior pregnancy losses, complicated pregnancies, and multiple pregnancies, are known to increase the risk of stillbirth [10]. Additionally, poor socio-economic conditions have been found to be associated with an increased likelihood of stillbirth [11]. Emerging evidence also suggests that certain environmental exposures, including extreme heat and air pollution, may be linked to stillbirths [12, 13].

To the best of our knowledge, no umbrella reviews have been conducted to evaluate the maternal risk factors before pregnancy and prior cesarean associated with stillbirth through meta-analyses and systematic reviews. Therefore, this umbrella review aims to assess the maternal risk factors before pregnancy and prior cesarean associated with stillbirth by examining findings from existing meta-analyses and systematic reviews.

Materials and methods

We followed state-of-the-art methods of umbrella reviews [14]. This umbrella review was conducted and reported in accordance with the PRISMA guidelines [15]. The study protocol was prospectively registered in the PROSPERO international prospective register of systematic reviews (Registration ID: CRD42025637355) from the earliest available records until 23 January 2025.

Identifying potential risk factors

To identify potential risk factors associated with stillbirth, a systematic search was performed across multiple online databases to compile all possible maternal risk factors. Following this, in consultation with a gynecologist and obstetrician, key risk factors were selected for further analysis. The next phase involved identifying relevant meta-analyses for these factors.

Search strategy and eligibility criteria

A systematic search was performed in PubMed, Scopus, and Web of Science to identify meta-analyses examining the association between maternal factors before pregnancy and prior cesarean associated with stillbirth. The search, conducted in 20 January 2025, had no restrictions on publication date, language, or geographic location. The search strategy incorporated MeSH terms and keywords (Table S1).

Study selection process

Two investigators (HA and EJ) independently screened the titles and abstracts of all identified studies and assessed the full texts of potentially eligible articles. Any discrepancies were resolved through discussion between two researchers.

Inclusion and exclusion criteria

This review included published meta-analyses of observational studies (cohort, case-control, and cross-sectional) that assessed pre-pregnancy and prior cesarean maternal factors for stillbirth, with no restrictions on publication date. We explicitly excluded conference abstracts, letters to the editor, original research articles, and studies focusing on genetic factors, maternal markers, or animal models. The primary inclusion criterion was the provision of essential data for cumulative analysis. In instances where multiple meta-analyses investigated the same specific factor, priority was given to the most comprehensive one, defined as the analysis encompassing the largest number of original studies. Incomplete data in the meta-analysis were addressed by contacting the respective corresponding authors.

Selection of meta-analyses

For each maternal factor, we selected the meta-analysis with the highest methodological quality, as assessed by the AMSTAR 2 tool. If multiple meta-analyses had the same quality score, we prioritized the most recent publication, and the largest sample size.

Data extraction

Two investigators (HA and EJ) independently extracted data from the included meta-analyses using a standardized form in Microsoft Excel. The following information was systematically collected: first author, publication year, number of participants, number of studies included in each meta-analysis, study designs of primary studies, measure of association (risk ratio, odds ratio, etc.) with 95% confidence intervals Random effects p-values and results of heterogeneity tests (I² statistic, Q-test p-value). All extracted data were cross-verified between reviewers to ensure accuracy. Discrepancies, if any, were resolved through discussion and consensus.

Quality assessment

The methodological quality of included meta-analyses was rigorously evaluated using the AMSTAR 2 tool [16], a validated critical appraisal instrument for systematic reviews. Two independent reviewers conducted the assessments, with any discrepancies resolved through discussion between two authors. The tool comprises 16 domains, including: 7 critical domains (fundamental methodological criteria) and 9 non-critical domains (supplementary quality indicators). Studies were classified into four quality tiers based on their fulfillment of these criteria: High Quality: No or only one non-critical weakness and all critical domains satisfied. Medium Quality: More than one non-critical weakness and all critical domains satisfied. Low Quality: One critical weakness (regardless of non-critical items): Critically Low Quality: Multiple critical weaknesses (regardless of non-critical items). This standardized approach ensured consistent and transparent quality evaluation across all included meta-analyses.

Statistical analysis

The meta-umbrella R-4.5.2 package (metaumbrella.org) was utilized to conduct umbrella reviews incorporating evidence stratification. All included studies reported odds ratios (ORs) and relative risks (RRs) along with 95% confidence intervals (CIs) for stillbirth risk factors. We applied a random-effects model to estimate the pooled ORs and RRs, and computed P-values for each risk factor. A significance threshold of P < 0.05 was used to determine statistical significance. Heterogeneity among the included studies was evaluated using Cochran’s Q test and the I² statistic [17]. The I² value represents the proportion of variation across studies that are due to heterogeneity rather than chance, with values over 50% indicating substantial heterogeneity. Additionally, we calculated the 95% prediction intervals (95% PIs) for each risk factor. The 95% PI provides a range within which the true effect size of a future study is expected to fall, based on the current data and predictor settings [18]. To detect potential publication bias, we employed Egger’s regression asymmetry test [19]. A P-value < 0.05 in Egger’s test was interpreted as evidence of such bias. We also applied the Ioannidis test for excess significance bias to evaluate the overall risk of bias across the meta-analyses [20]. This test helps detect various forms of bias, including selective reporting and publication bias. A P-value < 0.05 was taken as an indication of overall bias. The strength of evidence for small-study effects and publication bias was assessed using a pre-specified hierarchy. While both Egger’s regression test and the Ioannidis excess significance test were employed, any discordance between them was resolved by prioritizing the results of Egger’s test, as it directly tests for funnel plot asymmetry, the most common manifestation of publication bias. Furthermore, it is acknowledged that both statistical tests, particularly Egger’s test, can be underpowered and unreliable when fewer than ten studies are included in a meta-analysis. In such instances, the results of these tests were interpreted with extreme caution. For each risk factor, we reported Hedges’ g, a measure of effect size, which quantifies the magnitude of difference between groups—typically between an intervention and a control group [21]. According to Cohen’s benchmarks: Small effect = 0.2 (not visually apparent), medium effect = 0.5 and large effect = 0.8 (visibly detectable) [22].

Evidence strength assessment

We evaluated the strength of evidence for identified risk factors using the validated Ioannidis classification system. This standardized approach categorizes associations into five hierarchical evidence classes based on stringent methodological criteria: Convincing evidence (Class I), highly suggestive evidence (Class II), suggestive evidence (Class III), weak evidence (Class IV) and Non-significant (ns). The specific quantitative criteria for each classification level are detailed in Table S2.

Results

This umbrella review included four studies [23–26], comprising seven factors that evaluated 365,158 cases of stillbirth across a total sample size of 70,124,598 participants. The process of selecting eligible meta-analyses for inclusion in this review is outlined in Fig. 1. The reasons for exclusion studies whose full text was reviewed are reported in Table S3. In this umbrella review, a total of 60 primary studies were analyzed, comprising 55 cohort studies, 4 case-control studies, and 1 cross-sectional study. The included studies investigated seven Pre-pregnancy maternal factors potentially associated with the risk of stillbirth. These factors included chronic hypertension, race (black vs. white), maternal age, obesity, overweight, and prior caesarean, as summarized in Table 1.

Fig. 1.

The flowchart of the meta-analyses included in the umbrella review

Table 1.

The pre-pregnancy and prior cesarean maternal factors for stillbirth in this umbrella review

Risk factors	Source (year)	Number of population	Number of included studies	Study design	Effect metrics	Random effect summary estimate	AMSTAR2 quality	Credibility of evidence
Prior cesarean	Moraitis (2015)	798,631	3	Cohort	Odds ratio	1.39 (1.11, 1.74)	Critically low	Weak
Race (black vs. white)	Arechvo (2022)	62,098,323	21	Cohort	Relative risk	2.03 (1.85, 2.22)	Critically low	Class II
Chronic hypertension	Al Khalaf (2021)	1,287,412	18	Case-control/Cohort/cross-sectional	Odds ratio	3.0 (2.70, 3.32)	low	Class II
Pre-existing diabetes	Flenady (2011)	41,604	5	Case-control/Cohort	Odds ratio	2.88 (2.04, 4.07)	Critically low	Weak
Maternal age (≥ 35 years)	Flenady (2011)	3,735,837	4	Case-control/Cohort	Odds ratio	1.64 (1.56, 1.72)		Class II
Obesity	Flenady (2011)	723,733	4	Cohort	Odds ratio	1.63 (1.33, 1.99)		Class III
Overweight	Flenady (2011)	1,439,058	5	Cohort	Odds ratio	1.23 (1.09, 1.38)		Class III

Class II: Highly suggestive; Class III: Suggestive; class IV: Weak

The following three risk factors for stillbirth were identified as highly suggestive evidence (Class II): chronic hypertension (OR 3.0, 95% CI: 2.70, 3.32), race (black vs. white) (OR 2.03, 95% CI: 1.85, 2.22), and maternal age (≥ 35 years) (OR 1.64, 95% CI: 1.56, 1.72). Obesity (OR 1.63, 95% CI: 1.33, 1.99) and overweight (OR 1.23, 95% CI: 1.09, 1.38) were classified as risk factors with suggestive evidence (Class III). In contrast, pre-existing diabetes (OR 2.88, 95% CI: 2.04, 4.07), and prior cesarean (OR 1.39, 95% CI: 1.11, 1.74) were classified as weak evidence (Class IV) (Table 1).

Seven of the seven associations evaluated demonstrated statistically significant results under the random-effects model. Importantly, six of the included meta-analyses investigated over 1,000 cases of stillbirth. Furthermore, one study reported low heterogeneity (I² < 50%). As presented in Table 2, there was no evidence of excess significance bias or small-study effects. The quality of the three available meta-analyses was rated as critically low, and one study were evaluated as low quality, based on the AMSTAR2 criteria (Table 1 and Table S4).

Table 2.

The assessment of the evidence credibility for maternal factors associated with stillbirth

Risk factors	Number of cases	Summary associations (p-value) per random-effects calculations	Small-study effects (p-value for Egger)	Excess of significance bias (p-value)	PI	Largest study (95% CI)	I2%
Prior caesarean	1672	0.003	0.074	0.558	0.33, 5.91	0.94–1.8	7.92
Race (black vs. white)	263,621	< 0.000001	0.493	0.459	1.35, 3.05	2.01–2.06	95.76
Chronic hypertension	55,699	< 0.000001	0.473	0.648	2.09, 4.29	1.96, 3.73	79.78
Pre-existing diabetes	358	< 0.000001	0.824	0.1	1.03, 8.07	2.46–4.55	50.17
Maternal age (≥ 35 years)	7202	< 0.000001	0.275	0.783	1.36, 1.96	1.51–1.68	52.18
Obesity	3580	< 0.001	0.221	0.167	0.66, 4.02	1.69, 2.39	83.97
Overweight	33,026	< 0.001	0.516	0.612	0.86, 1.76	1.08, 1.38	60.64

PI Prediction intervals, I² Heterogeneity

Discussion

This initial umbrella review offers a comprehensive synthesis of systematic reviews and meta-analyses focused on pre-pregnancy maternal risk factors associated with stillbirth. Increased clinical and community awareness of the maternal risk factors and implementation of best practice guidelines might improve management and lower the associated stillbirth rates [5]. Overall, three risk factors received strong support from highly suggestive evidence (class II), including race (black vs. white), maternal age (≥ 35 years), and chronic hypertension; also, two risk factors (Obesity and overweight) received strong support from suggestive evidence (class III). In contrast, evidence for factors such as prior cesarean, and pre-existing diabetes was weak evidence (class IV). The present umbrella review’s assessment of race as a risk factor (comparing black to white individuals) is constrained by the scope of the included meta-analyses. Thus, the findings are specific to this comparison and should not be extrapolated to other ethnicities, such as South Asian, East Asian, or Hispanic populations. The failure to achieve a high level of evidence may be attributed to factors such as heterogeneity and potential biases [27].

Some associations between maternal risk factors before pregnancy and stillbirth were relatively consistent in both strength and direction. This consistency may be attributed to the strong interrelationships and plausible causal links among many of these risk factors. Obesity is one of the leading factors contributing to the overall burden of disease worldwide [28]. Individuals who are obese or overweight are more likely to develop conditions such as hypertension [29]. Additionally, the prevalence of obesity is higher among black individuals [30]. Therefore, it can be inferred that the complex interactions among these risk factors significantly contribute to the causal pathway of stillbirth. Although this risk factor is suggestive, however, there was high heterogeneity among studies (I² = 83.97%). The reason for this heterogeneity was due to different definitions for BMI categories in the study.

This heightened risk may be attributed to socioeconomic inequalities, lower levels of education, and a higher prevalence of specific pregnancy complications such as preeclampsia, eclampsia, and premature rupture of membranes within this population [31, 32]. Additionally, Davis (2019) introduced the term “obstetric racism” to describe the intersection of obstetric violence and medical racism, which leads to the mistreatment and abuse of black women by healthcare providers during pregnancy, labor, birth, and the postpartum period [33]. This obstetric racism is evident in research involving black women, who frequently report experiencing racist treatment from healthcare providers [32]. Although this risk factor is highly suggestive, however, the results may have been influenced by several confounders and high heterogeneity. The heterogeneity between studies was due to the lack of adjustment for confounders in most of the studies (I² = 95.76%).

The risk of stillbirth increases for all women of advanced childbearing age, with a more significant rise in those ≥ 35 years old. While advanced maternal age is linked to a higher risk of various other factors, such as diabetes, and infertility, it is also an important independent highly suggestive risk factor for stillbirth, although a moderate heterogeneity [34]. Therefore, Improved community awareness of the associated risks might lower the proportion of women becoming pregnant at older ages [23].

Pre-gestational chronic hypertension can negatively impact the remodeling of the maternal spiral arteries that supply blood to the placenta [35]. This impairment can reduce blood flow to both the uterus and the placenta, leading to fetal hypoxia, prolonged hypoxia may result in fetal distress and even stillbirth [36]. Additionally, women with chronic hypertension face a higher risk of complications such as preeclampsia and placental abruption, both of which can further increase the risk of stillbirth [37]. Although this risk factor is highly suggestive, however, the results have considerable heterogeneity (I² = 79.78%).

Pre-gestational diabetes is linked to various mechanisms that elevate the risk of stillbirth. These include hyperglycemia, which can lead to infections, reduced fetal movement, metabolic dysfunction, and placental problems [38]. Such factors may result in fetal distress, hypoxia, growth abnormalities, and other complications that threaten fetal health [39]. Additionally, women with pre-gestational diabetes face a higher risk of developing preeclampsia [40]. In the present meta-analysis, stratification between Type 1 and Type 2 diabetes was not available in the primary data.

Substantial heterogeneity was prevalent among the meta-analyses of stillbirth risk factors. While most factors (excluding prior cesarean) had I² statistics greater than 50%, the highest levels of heterogeneity (I² >80%) were identified for two risk factors with suggestive or highly suggestive evidence: race (black vs. white) and obesity.

In our study, all meta-analyses, except for the one on race (black vs. white), included fewer than 10 studies. It’s important to recognize that making judgments based solely on statistical tests can be misleading. Research suggests that tests for heterogeneity can be underpowered or inconclusive when a meta-analysis includes fewer than 10 studies. Therefore, caution should be taken when concluding solely from statistical analyses in these cases [20, 41].

This umbrella review has several interpretive and methodological limitations. Methodologically, the included systematic reviews often received low AMSTAR-2 ratings, indicating potential weaknesses in their conduct. Furthermore, there was an absence of quality assessment for the primary studies within these reviews. The synthesis was constrained by a limited number of meta-analyses per risk factor (fewer than 10) and small-study, increasing uncertainty. The potential for publication bias was a concern, exacerbated by the restriction to published literature. Consequently, the conclusions of this review are inherently dependent on the completeness and quality of the underlying systematic reviews. Finally, the generalizability of the findings is limited by the underrepresentation of certain populations in the available data. Interpretively, the application of the evidence-grading software was limited to meta-analyses containing three or more studies. Additionally, several important domains, namely psychological, autoimmune, and nutritional factors, could not be assessed, as no relevant meta-analyses met the inclusion criteria.

Conclusion

We believe this study could provide valuable insights into the relationship between pre-pregnancy maternal risk factors and stillbirth. Our findings indicate that being black race, having chronic hypertension, and being the age of ≥ 35 are highly suggestive maternal characteristics associated with stillbirth. Additionally, obesity and being overweight also show suggestive evidence of association. In contrast, factors such as having a caesarean section, and pre-existing diabetes appear to be weakly related to the maternal characteristics associated with stillbirth. Given the low and critically low quality of the primary studies, as assessed by the AMSTAR 2 tool, the conclusions derived from this umbrella review should be considered tentative and viewed as generating hypotheses rather than providing definitive evidence.

Abbreviations

CI

Confidence Interval

OR

Odds Ratio

RR

Relative Risks

Ns

Non-significant

PROSPERO

Prospective Register of Systematic Reviews

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Authors’ contributions

Conceptualization: EJ, AS. Data curation: EJ, HA. Methodology: EJ. Software: HA. Writing—original draft: AS. Writing review and editing: All authors. The authors read and approved the final manuscript.

Funding

Hamadan University of Medical Sciences supported the protocol of this study (140407226685).

Data availability

Access to data is possible with permission from the Corresponding authors (EJ OR AMS).

Declarations

Ethics approval and consent to participate

The Ethics Committee of the Hamadan University of Medical Sciences approved the protocol of this study (IR.UMSHA.REC.1404.633). This umbrella review analyzed data extracted exclusively from previously published research. No new human participants were recruited, and no identifiable or sensitive personal data were collected. Consequently, informed consent was not applicable, and the requirement for informed consent was waived as it was unnecessary for secondary data analyses. The conduct of this review adhered to the ethical principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.