The efficacy of tranexamic acid for the prevention of postpartum hemorrhage among women at high risk of postpartum hemorrhage undergoing cesarean section: a meta-analysis

Abstract

Objectives

This meta-analysis investigated the extent to which tranexamic acid reduces the risk of postpartum bleeding among pregnant women at high risk of postpartum hemorrhage undergoing cesarean section.

Methods

Scopus, the Cochrane Library, PubMed, and Web of Science were searched from their establishment to December 4, 2025, to identify randomized controlled trials comparing tranexamic acid with placebo for preventing postpartum hemorrhage in high-risk women undergoing cesarean section. Because of high heterogeneity, we employed a random-effects model (restricted maximum–likelihood) to pool data.

Results

Finally, 12 articles with 1240 high-risk pregnant women were included in this meta-analysis. Compared to placebo, tranexamic acid significantly reduced blood loss after cesarean delivery in high-risk pregnant women (WMD − 435.44 mL, 95% CI (-581.86, -289.01) I² = 94.71%).

Furthermore, tranexamic acid significantly prevented the drop in hemoglobin level (WMD 1.09 g/dL, 95% CI (0.20, 1.99) I2 = 97.63%), prevented postpartum hemorrhage (RR 0.45, 95% CI (0.30, 0.70) I2 = 49.34%) and more than 1000 mL blood loss (RR 0.23, 95% CI (0.14, 0.37) I2 = 0.00%), and reduced the need for transfusion (RR 0.28, 95% CI (0.20, 0.39) I2 = 0.00%) and uterogenic agents (WMD 0.28, 95% CI (0.18, 0.46) I2 = 85.63%). The beneficial effects of tranexamic acid on the length of hospital stay (WMD − 0.46 days, 95% CI (-0.93, 0.01) I2 = 73.44%) and the length of surgery (WMD − 8.20 min, 95% CI (-17.15, 0.74) I2 = 96.28%) did not reach the threshold of statistical significance. Compared to the placebo group, nausea/vomiting was significantly more frequent in the tranexamic acid group (RR 1.89, 95% CI (1.02, 3.52) I2 = 0.00%), while dizziness (RR 1.13, 95% CI (0.32, 3.92) I2 = 16.09%) and headache (RR 1.85, 95% CI (0.48, 7.19) I2 = 25.87%) showed no significant. There was no case of thromboembolic events in either group. Exploratory subgroup analysis showed that sample size, risk of bias, mother’s body mass index (BMI), tranexamic acid dose, the timing of administration, gestational age, and the reason for being high-risk significantly affected the beneficial effects of tranexamic acid on the volume of blood loss.

Conclusion

Our findings suggest that tranexamic acid may effectively reduce bleeding among women with a high risk of postpartum hemorrhage undergoing cesarean section.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12884-026-09594-0.

Introduction

Postpartum hemorrhage represents an obstetric emergency, which can increase the risk of morbidity and mortality after delivery [1]. Although there is no consensus, postpartum hemorrhages can be defined as blood loss ≥ 1000 ml or any blood loss leading to clinical and/or laboratory signs of shock/tissue hypoperfusion within 24 h after delivery [1]. It affects 1%-10% of pregnant women, but can be prevented in many cases [1]. A report from the United States indicated that from 2000 to 2019, of 76.7 million deliveries in hospitals, 2.3 million were affected by postpartum hemorrhage, and the rate of postpartum hemorrhage increased from 2.7% to 4.3% [2]. Adverse outcomes due to postpartum hemorrhages are often attributed to delayed recognition or inappropriate treatment, which necessitate more intensive care by gynecologists, midwives, and nurses, prolong hospitalization, and undermine the health of mothers and their neonates [1]. Identification of risk factors of postpartum hemorrhage, such as placenta previa, placenta accreta, anemia before delivery, pregnancy with uterine fibroids, uterine atony, history of postpartum hemorrhage, macrosomia, and placental abruption, can help identify high-risk patients and adopt preventive measures [3, 4]. For instance, doctor-nurse collaborative and hierarchical management combined with nursing risk management was found to significantly improve the nursing management of patients with postpartum hemorrhage and decrease bleeding volume [5].

Among the preventive measures, preoperative treatment with tranexamic acid has emerged as an effective measure to reduce the volume of bleeding after delivery [6, 7]. Tranexamic acid can be used shortly before delivery as a single-dose drug and provides many benefits as a preventive measure, particularly in resource-limited settings [6, 7]. Although previous meta-analyses have investigated the efficacy of tranexamic acid in the prevention of postpartum hemorrhage in the general population of pregnant women, its efficacy in high-risk patients undergoing cesarean section has not been assessed in meta-analyses so far because most original studies in this regard have been published in the last 3 years. Therefore, this meta-analysis investigated the extent to which tranexamic acid reduces the risk of postpartum bleeding and hemorrhage and improves clinical outcomes among high-risk women undergoing cesarean section.

Method

Search strategy

We thoroughly searched PubMed, Scopus, the Cochrane Library, and Web of Science from their establishment to December 4, 2025, employing the following search keywords: (“tranexamic acid”) [title, abstract, keywords] AND (“cesarean section” OR " C-section” OR “postpartum” OR “cesarean delivery”) [title, abstract, keywords] AND (“clinical trial” OR “randomized controlled trial” OR “RCT”) [title, abstract, keywords] (Supplementary Material 1).

Additionally, the reference list of the identified trials and reviews was evaluated to detect more studies for inclusion. All articles were transferred to EndNote 9.0, and duplicate entries were eliminated. This systematic review adhered to The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline [8], and its protocol was registered with PROSPERO (CRD420251247018). PICOs were as follows: (P): pregnant women at high risk of postpartum hemorrhage undergoing cesarean section; (I): tranexamic acid; (C): placebo; (O): blood loss volume (main outcome), changes in hemoglobin level, length of hospital stay, length of surgery, the incidence of postpartum hemorrhage, more than 1000 mL blood loss, the need for transfusion, and the need for uterogenic agents.

Inclusion and exclusion criteria

Inclusion criteria: (I) study design was randomized controlled trial; (II) all patients were at high risk of postpartum hemorrhage; (III) all patients underwent cesarean section; (IV) including a placebo control group; and (V) reporting the outcomes of interest, including blood loss volume (main outcome), changes in hemoglobin level, length of hospital stay, length of surgery, the incidence of postpartum hemorrhage, more than 1000 mL blood loss, the need for transfusion, and the need for uterogenic agents.

Patients with a wide range of conditions that were previously proven to increase the risk of post-partum hemorrhage could be included in this article as high-risk patients.

Exclusion criteria: (I) observational studies, case series, case reports, and review articles; (II) absence of a placebo group; (III) patients underwent vaginal delivery; and (IV) patients were at low or moderate risk of postpartum hemorrhage.

Data extraction and outcome measures

Two authors systematically searched the literature and assessed the eligibility of the retrieved records. All discrepancies were addressed through consultation. First, the researchers evaluated the relevance of the retrieved records and assessed their compliance with the inclusion and exclusion criteria based on the titles and abstracts. Thereafter, the eligibility of the remaining articles was evaluated based on the full-text version.

The following data were collected: study characteristics (the first author’s name, year of publication, and country of study), population (sample size, age, baseline hemoglobin, baseline body mass index (BMI), reason for being high-risk, and gestational age), interventions (dose of tranexamic acid, route of administration, the timing of administration, type of placebo, and length of treatment), and the outcomes. In addition to efficacy outcomes, we collected safety outcomes, such as thromboembolic events, nausea/vomiting, dizziness, and headache.

Risk of bias assessment

Two authors independently assessed the risk of bias for all included records, and all discrepancies were addressed through consultation. The risk of bias was determined based on the revised Cochrane risk-of-bias tool for randomized trials (RoB2). The tool determines the risk of bias according to the randomization process, deviations from intended interventions, outcome measurement, selection of the reported results, and missing outcome data, providing an overall risk of bias. Each domain was assigned one of the following ranks: low risk, some concerns, and high risk of bias. Studies with high risk in at least one domain or some concerns in at least two domains were considered to have a high risk of bias [9].

Statistical analysis

For blood loss volume (main outcome), changes in hemoglobin level, length of hospital stay, and length of surgery, changes in mean and standard deviation (SD) were extracted from the original studies to calculate weighted mean difference (WMD) and confidence interval (CI). For the incidence of postpartum hemorrhage, more than 1000 mL blood loss, the need for transfusion, the need for uterogenic agents, and the incidence of adverse events, the number of events and non-events were extracted to calculate the risk ratio (RR) and 95% CI. We performed this meta-analysis using Stata version 17.0 (StataCorp, TX, USA). As an exploratory and hypothesis-generating step, we also repeated the meta-analysis for the main outcome after removing outlier studies.

Studies not reporting an outcome or having missing outcome data, were excluded from the analysis for that specific outcome. In addition, studies were inevitably excluded from subgroup analysis if they did not report a specific characteristic for which subgroup analysis was conducted.

The I² statistic was adopted to determine statistical heterogeneity between trials, and I² values of more than 50% was deemed high statistical heterogeneity. Because of significant methodological and statistical heterogeneity, a random-effects model (restricted maximum–likelihood) was employed to pool data. We used restricted maximum–likelihood because it provides less biased and more efficient estimates of between-study variance than several alternatives, especially DerSimonian–Laird (DL) and simple maximum–likelihood [10].

Other analyses

For the main outcome, subgroup analysis was conducted based on sample size (more than 100 vs. less than or equal to 100), dose of tranexamic acid (1 g vs. lower doses), the reason for being high-risk (only placenta previa or only placenta accreta vs. other reasons or the combination of different reasons in a study), mothers’ age (more than 30 years vs. less than or equal to 30 years), mother’s BMI (more than 27 kg/m² vs. equal to or less than 27 kg/m²), preoperative hemoglobin level (more than 11 g/dL vs. less than or equal to 11 g/dL), gestational age (more than 37 weeks vs. less than or equal to 37 weeks), the timing of administration (before skin incision vs. after clamping the cord), and risk of bias (high risk studies vs. other studies). We utilized the leave-one-out test for sensitivity analysis to assess the robustness of the results. Publication bias and small study effect were measured using the funnel plot and Egger’s and Begg’s tests. Finally, we adopted the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) tool to assess the certainty of efficacy outcomes in this meta-analysis. The online tool (https://gdt.gradepro.org/app/) was used in this regard.

Results

Systematic search results

Initially, we identified 236 records in the Cochrane Library, 246 records in Web of Science, 68 records in PubMed, and 415 records in Scopus. Of 965 articles, 212 duplicate entries were eliminated. Thereafter, of the 753 remaining records, 694 were excluded after title and abstract screening. Next, the full-text version of 59 articles was assessed to determine their compliance with the inclusion and exclusion criteria. Finally, 12 articles with 1240 high-risk pregnant women undergoing cesarean delivery were included in this meta-analysis (Fig. 1 & Table 1). As mentioned in the method section, this meta-analysis did not tightly narrow down the definition of high-risk populations. Similarly, the existing clinical trials in this regard included several conditions, such as fetal macrosomia, placenta previa, prenatal anemia, multiple gestation, previous postpartum hemorrhage, multiple pregnancy, gestational hypertension, chronic hypertension, macrosomia, history of postpartum hemorrhage, polyhydramnios, and treatment with warfarin, as risk factors for postpartum bleeding. Although some of the trials only included one etiology, such as placenta previa, nearly half of the existing trials included multiple etiologies. Most studies used a single dose (1 gram) of tranexamic acid intravenously 10–20 min before skin incision and administered the same volume of sodium chloride 0.9% as the placebo. Generally, the included trials were similar in terms of participants’ age, BMI, and gestational age, route of administration, and dose of administration. However, the reason for being high risk widely varied across studies, and even there were different reasons for being high-risk in a single study; therefore, the included studies could not be categorized into appropriate groups for subgroup analysis. Egypt was the leading country in this field of research, followed by India, Nigeria, and Thailand (Table 1).

Fig. 1.

Systematic search flowchart

Table 1.

Study characteristics

First author’s name	Publication year	Age (year, mean)	Country of study	BMI (kg/m2)	Baseline hemoglobin level	Mean gestational age (weeks)	Timing of administration	Tranexamic acid dose/ route of administration	Number of participants in the tranexamic group	Number of participants in the placebo group	Type of placebo	Reference
Wongjariyakul et al.	2025	29.4	Thailand	29	12	Not reported	Before skin incision	1 gram, intravenous	25	25	0.9% sodium chloride	[11]
Ortuanya et al.	2024	31.9	Nigeria	29.7	11.2	Not reported	At least 10 min prior to commencement of the surgery	1 gram, intravenous	100	100	0.9% sodium chloride	[12]
Neumann et al.	2024	25.8	India	21.9	Not reported	Not reported	At least 10 min before skin incision	500 mg, intravenous	106	106	0.9% sodium chloride	[13]
Shojaei et al.	2024	31.3	Iran	29.8	Not reported	36.25	Immediately after the surgery	1 gram, intravenous	23	23	0.9% sodium chloride	[14]
Chaiyakarn et al.	2023	33.7	Thailand	26.9	Not reported	Not reported	Before skin incision	1 gram, intravenous	30	30	0.9% sodium chloride	[15]
Hany et al.	2023	31	Egypt	27.2	11	37.43	Applied during cesarean section	tranexamic acid-soaked gelatin sponge inserted between the anterior rectus sheath and the rectus abdominis muscle	21	21	Gelatin sponge not containing tranexamic acid	[16]
Bharati et al.	2022	29.8	India	Not reported	9.9	36.44	Just before skin incision	1 gram, intravenous	50	50	0.9% sodium chloride	[17]
Shalaby et al.	2022	28.7	Egypt	Not reported	Not reported	38	15 min before surgery	1 gram, intravenous	80	80	Glucose 5% solution	[18]
Abbas et al.	2019	30.6	Egypt	28.2	10.9	36.6	Just before skin incision	1 gram, intravenous	31	31	0.9% sodium chloride	[19]
Ifunanya et al.	2019	28.4	Nigeria	25	11.6	38	Prior to skin incision	1 gram, intravenous	84	84	0.9% sodium chloride	[20]
Shady et al.	2017	29.6	Egypt	Not reported	9.9	36.42	Just before skin incision	1 gram, intravenous	40	40	0.9% sodium chloride	[21]
Sujata et al.	2016	29.85	India	28.4	11.7	Not reported	Before skin incision	10 mg/kg, intravenous	31	29	0.9% sodium chloride	[22]

Fortunately, most outcomes, such as blood loss volume (main outcome), changes in hemoglobin level, length of hospital stay, and length of surgery, were reported in standard units by all studies (in mL, g/dL, days, and minutes, respectively). Similarly, there was no difference regarding the outcome of more than 1000 mL blood loss, contributing to the robustness of the results. In addition, we found no differences in the definition of adverse outcomes, particularly the incidence of nausea/vomiting, dizziness, headache, and thromboembolic events. Despite following the same practice guidelines, some outcomes, such as the need for transfusion, the need for uterogenic agents, and postpartum hemorrhage, partly relied on the sensitivity and clinical suspicion of physicians, which is unavoidable in clinical settings, and did not considerably affect the results. Although such minor differences in outcome assessment must be considered while interpreting the results, the extent of difference was not of high clinical importance, and this difference was only attributed to a small proportion of outcomes.

The definition of the high-risk population for postpartum hemorrhage after cesarean section

Since there is no unified definition of the high-risk” population, and many conditions are deemed to increase the risk of postpartum bleeding in clinical practice, patients with a wide range of conditions were eligible for inclusion. Herein, Wongjariyakul et al. included placenta previa or low-lying placenta, prenatal anemia (hemoglobin levels less than 9.9 g/dL), previous cesarean section, transverse presentation, fetal macrosomia defined as fetal weight more than 4.0 kg in ultrasound, multiple gestation, previous history of postpartum hemorrhage, and polyhydramnios as high-risk conditions [11]. Ortuanya et al. included co-existing uterine fibroid, multiple pregnancy, a previous history of postpartum hemorrhage, previous cesarean section, fetal macrosomia, severe preeclampsia, eclampsia, placenta previa, multiple gestation, and polyhydramnios as high-risk conditions [12]. Neumann et al. include chronic hypertension, gestational hypertension, obesity, preeclampsia, eclampsia, multiple pregnancy, previous cesarean section, anemia, use of oxytocin augmentation for more than 4 h, parity more than 4, abnormally implanted placenta, placenta previa, abruption, uterine leiomyomas, chorioamnionitis, polyhydramnios, fetal macrosomia, history of postpartum hemorrhage, and cholestasis of pregnancy as high-risk conditions [13]. Shojaei et al. included only pregnant women with placenta accreta syndrome as high-risk patients [14]. Chaiyakarn et al. Bharati et al. Abbas et al. and Shady et al. included only pregnant women with placenta previa as high-risk patients [15, 17, 19, 21]. Hany et al. included only pregnant women receiving warfarin [16]. Shalaby et al. included placenta previa, uterus overdistension (due to multiple gestation, fetus macrosomia > 4500 g or polyhydramnios with an amniotic fluid index > 24), anemia, and intraoperative blood transfusion during or before cesarean section as the reasons for being high-risk [18]. Ifunanya et al. included hypertensive disorders during pregnancy, multiple pregnancy, previous cesarean section, labor augmentation, failed induction of labor, chorioamnionitis, placenta previa, placental abruption, polyhydramnios, multiparity more than > 4, previous postpartum hemorrhage, fetal macrosomia, obstructed labor, and coexisting fibroids as high-risk conditions [20]. Sujata et al. included placenta previa, pregnancy-induced hypertension, oxytocin augmentation for at least 4 h, multiple pregnancy, cholestasis, more than two previous cesarean deliveries, chorioamnionitis (defined as oral temperature more than 38.5 °C with a high leukocyte count, after ruling out other sources of infection), general anesthesia, genital-tract injury, macrosomia (estimated birth weight more than 4 kg), polyhydramnios (amniotic fluid index more than the 95th percentile for the length of pregnancy on prenatal ultrasonography), fibroids, and multiparity (parity more than 4) as the reasons for being high-risk [22]. Table 2 summarizes the risk factors included in each study.

Table 2.

The reasons for being high-risk in each study

First author’s name	The reason for being high-risk	Reference
Wongjariyakul et al.	Placenta previa or low-lying placenta, prenatal anemia (hemoglobin levels less than 9.9 g/dL), previous cesarean section, transverse presentation, fetal macrosomia defined as fetal weight more than 4.0 kg in ultrasound, multiple gestation, previous history of postpartum hemorrhage, and polyhydramnios	[11]
Ortuanya et al.	Co-existing uterine fibroid, multiple pregnancy, a previous history of postpartum hemorrhage, previous cesarean section, fetal macrosomia, severe preeclampsia, eclampsia, placenta previa, multiple gestation, and polyhydramnios	[12]
Neumann et al.	Chronic hypertension, gestational hypertension, obesity, preeclampsia, eclampsia, multiple pregnancy, previous cesarean section, anemia, use of oxytocin augmentation for more than 4 h, parity more than 4, abnormally implanted placenta, placenta previa, abruption, uterine leiomyomas, chorioamnionitis, polyhydramnios, fetal macrosomia, history of postpartum hemorrhage, and cholestasis of pregnancy	[13]
Shojaei et al.	Placenta accreta syndrome	[14]
Chaiyakarn et al.	Placenta previa	[15]
Hany et al.	Using warfarin	[16]
Bharati et al.	Placenta previa	[17]
Shalaby et al.	Placenta previa, uterus overdistension (due to multiple gestation, fetus macrosomia > 4500 g or polyhydramnios with an amniotic fluid index > 24), anemia, and intraoperative blood transfusion during or before cesarean section	[18]
Abbas et al.	Placenta previa	[19]
Ifunanya et al.	Hypertensive disorders during pregnancy, multiple pregnancy, previous cesarean section, labor augmentation, failed induction of labor, chorioamnionitis, placenta previa, placental abruption, polyhydramnios, multiparity (> 4), previous postpartum hemorrhage, fetal macrosomia, obstructed labor, and coexisting fibroids	[20]
Shady et al.	Placenta previa	[21]
Sujata et al.	Placenta previa, pregnancy-induced hypertension, oxytocin augmentation for at least 4 h, multiple pregnancy, cholestasis, more than two previous cesarean deliveries, chorioamnionitis (defined as oral temperature more than 38.5 °C with a high leukocyte count, after ruling out other sources of infection), general anesthesia, genital-tract injury, macrosomia (estimated birth weight more than 4 kg), polyhydramnios (amniotic fluid index more than the 95th percentile for the length of pregnancy on prenatal ultrasonography), fibroids, and multiparity (parity more than 4)	[22]

Given the differences in the type of risk factors among the included studies, the results of this meta-analysis may not be directly applicable to clinical practice; instead, this study provides a research guide for evaluating the effects of tranexamic acid in specific risk groups. Furthermore, this wide variety in the type of etiologies for being high-risk in individual studies and among different studies can substantially contribute to heterogeneity; therefore, we conducted a feasible subgroup analysis based on the reason for being high-risk (only placenta previa or only placenta accreta vs. other reasons or the combination of different reasons in a study) to assess its effect on the outcome and heterogeneity. Of course, the etiology for being high-risk must be considered while interpreting the results and assessing their generalizability.

Risk of bias assessment

The RoB2 tool indicated a high risk of bias for the articles reported by Wongjariyakul et al. Shojaei et al. Hany et al. Abbas et al. and Sujata et al. but other studies were found to have a low risk of bias (Supplementary material 2). The main problem leading to the high risk of bias in the aforementioned studies was the lack of researcher blinding. Due to the nature of the intervention, this was inevitable in some studies.

Meta-analysis

Regarding the primary outcome, pooling data from 12 studies indicated that compared to placebo, tranexamic acid significantly reduced blood loss after cesarean delivery in high-risk pregnant women (WMD − 435.44 mL, 95% CI (-581.86, -289.01) I² = 94.71%) (Fig. 2); however, the heterogeneity remained high. Based on the primary result, we then repeated the analysis after eliminating two studies conducted by Shojaei et al. and Neumann et al. whose results were at the two ends of the confidence interval. After removing the two studies, the heterogeneity markedly decreased while the results still remained significant, confirming the accuracy of our findings (WMD − 366.11 mL, 95% CI (-395.08, -337.14) I² = 0.00%) (Fig. 3).

Fig. 2.

The effect of tranexamic acid on the volume of blood loss

Fig. 3.

The effect of tranexamic acid on the volume of blood loss after removing the studies conducted by Shojaei et al. and Neumann et al

Regarding secondary outcomes, the results indicated that compared to placebo, tranexamic acid significantly prevented the drop in hemoglobin level (WMD 1.09 g/dL, 95% CI (0.20, 1.99) I² = 97.63%) (Fig. 4), prevented postpartum hemorrhage (RR 0.45, 95% CI (0.30, 0.70) I² = 49.34%) (Fig. 5) and more than 1000 mL blood loss (RR 0.23, 95% CI (0.14, 0.37) I² = 0.00%) (Fig. 6), and reduced the need for transfusion (RR 0.28, 95% CI (0.20, 0.39) I² = 0.00%) (Fig. 7) and uterogenic agents (WMD 0.28, 95% CI (0.18, 0.46) I² = 85.63%) (Fig. 8).

Fig. 4.

The effect of tranexamic acid on changes in hemoglobin level

Fig. 5.

The effect of tranexamic acid on postpartum hemorrhage

Fig. 6.

The effect of tranexamic acid on the incidence of postpartum blood loss of more than 1000 mL

Fig. 7.

The effect of tranexamic acid on the need for transfusion

Fig. 8.

The effect of tranexamic acid on the need for uterogenic agents

Although compared to placebo, tranexamic acid reduced the length of hospital stay (WMD − 0.46 days, 95% CI (-0.93, 0.01) I² = 73.44%) (Fig. 9) and the length of surgery (WMD − 8.20 min, 95% CI (-17.15, 0.74) I² = 96.28%) (Fig. 10), its effect did not reach the threshold of statistical significance.

Fig. 9.

The effect of tranexamic acid on the length of hospital stay

Fig. 10.

The effect of tranexamic acid on the length of operation

All adverse events were reported to be tolerable, and no thromboembolic event was observed in studies reporting adverse events. Compared to the placebo group, nausea/vomiting was significantly more frequent in the tranexamic acid group (RR 1.89, 95% CI (1.02, 3.52) I² = 0.00%) (Fig. 11), while dizziness (RR 1.13, 95% CI (0.32, 3.92) I² = 16.09%) (Fig. 12) and headache (RR 1.85, 95% CI (0.48, 7.19) I² = 25.87%) (Fig. 13) showed no significant difference between the two groups.

Fig. 11.

The effect of tranexamic acid on the risk of nausea/vomiting

Fig. 12.

The effect of tranexamic acid on the risk of dizziness

Fig. 13.

The effect of tranexamic acid on the risk of headache

Despite the beneficial effects of tranexamic acid, it should be noted that heterogeneity was high for most efficacy outcomes, which may undermine the robustness of our findings. Although we conducted additional analyses, such as subgroup and sensitivity analyses, to identify the source of heterogeneity, the results still remain hypothesis-generating rather than definitive.

Subgroup analysis

We conducted subgroup analysis to identify the source of heterogeneity and determine in which subgroups tranexamic acid was more effective; however, it should be considered that the results of subgroup analysis are exploratory and subgroup analyses were not prespecified. Regarding sample size, tranexamic acid was significantly more effective in studies with a sample size of less than or equal to 100 (WMD − 439.51 mL, 95% CI (-516.16, -362.85), I² = 83.34%) compared to studies with a sample size of more than 100 (WMD − 341.24 mL, 95% CI (-370.11, -312.36), I² = 81.35%). In addition, tranexamic acid was significantly more effective in studies using 1 g of tranexamic acid (WMD − 377.64 mL, 95% CI (-407.15, -348.13), I² = 82.18%) than studies using lower doses of tranexamic acid (WMD − 227.90 mL, 95% CI (-295.12, -160.68), I² = 27.10%). We also found that using tranexamic acid after clamping the cord (WMD − 520.25 mL, 95% CI (-657.54, -382.96), I² = 97.19%) was significantly more effective than using tranexamic acid before skin incision (WMD − 346.72 mL, 95% CI (-374.28, -319.17), I² = 59.30%), which may be due to the pharmacokinetics and mechanism of action of tranexamic; however, it only two studies were included in one of the subgroups and the results remain exploratory. Additionally, tranexamic acid was significantly more effective in studies including only placenta previa or only placenta accreta (WMD − 510.66 mL, 95% CI (-614.33, -406.99), I² = 89.20%) than in studies including other reasons or the combination of different risk factors (WMD − 341.99 mL, 95% CI (-369.97, -314.00), I² = 64.94%) (Supplementary material 3).

Mothers’ age (more than 30 years (WMD − 382.23 mL, 95% CI (-436.80, -327.65), I² = 90.86%) vs. less than or equal to 30 years (WMD − 344.10 mL, 95% CI (-375.20, -313.00), I² = 67.49%)) and preoperative hemoglobin levels (more than 11 g/dL (WMD − 368.76 mL, 95% CI (-400.10, -337.42), I² = 0.00%) vs. less than or equal to 11 g/dL (WMD − 399.59 mL, 95% CI (-504.31, -294.88), I² = 21.16%)) did not significantly affect the beneficial effects of tranexamic acid, while mother’s BMI more than 27 kg/m² was associated with superior, benefits of tranexamic acid (WMD − 396.29 mL, 95% CI (-450.60, -341.97), I² = 88.13%) compared to mother’s BMI equal to or less than 27 kg/m² (WMD − 335.84 mL, 95% CI (-368.47, -303.20), I² = 87.66%) (Supplementary material 3).

We also found that tranexamic acid provided significantly greater benefit for women with gestational ages of less than or equal to 37 weeks (WMD − 638.38 mL, 95% CI (-766.59, -510.16), I² = 88.47%) compared to for women with gestational age more than 37 weeks (WMD − 363.19 mL, 95% CI (-397.86, -328.52), I² = 0.00%). Moreover, the beneficial effect of tranexamic acid was greater in studies with a high risk of bias (WMD − 491.81 mL, 95% CI (-593.66, -389.95), I² = 89.31%) compared to other studies (WMD − 342.97 mL, 95% CI (-370.99, -314.95), I² = 67.74%). Although tranexamic acid showed greater efficacy in some subgroups, its beneficial effects remained significant in all subgroups (Supplementary material 3).

Sensitivity analysis

The leave-one-out test for the primary outcome indicated that removing individual studies did not change the overall results of this study regarding the impact of tranexamic acid on blood loss after high-risk cesarean delivery, and the confidence interval ranged from − 615.50 mL to -267.98 mL after removing individual studies (Supplementary material 4).

Publication bias

A funnel plot was drawn for blood loss, which showed asymmetry in visual inspection; therefore, Begg’s and Egger’s tests were conducted. Unlike Begg’s test (p = 0.147), Egger’s test showed a significant risk of publication bias (p = 0.014); therefore, we conducted the fill-and-trim test. The fill-and-trim test showed that after imputing the missing data, the results still remained significant (WMD − 353.44, 95% CI (-380.44, -326.44) (Supplementary material 5).

Certainty of evidence

Using the GRADE system for assessing certainty of evidence, we found that the quality of evidence was moderate for the volume of blood loss, while low for other efficacy outcomes, including hemoglobin level, postpartum hemorrhage, the need for uterogenic agents, the need for transfusion, and more than 1000 mL of blood loss (Supplementary material 6).

Discussion

This meta-analysis of 12 clinical trials indicated that compared to placebo, tranexamic acid can significantly decrease blood loss, prevent the decline in hemoglobin level, reduce the risk of postpartum hemorrhage and more than 1000 mL blood loss, and decrease the need for transfusion and uterogenic agents. Tranexamic acid also reduced the length of hospital stay and the duration of surgery, although the effect was not statistically significant. Despite these benefits, the meta-analysis indicated high heterogeneity and a low level of evidence for most efficacy outcomes, which considerably undermines the robustness and generalizability of the findings. Although high statistical heterogeneity existed regarding the primary outcome, we found the two studies leading to this heterogeneity, whose removal significantly reduced heterogeneity. Unfortunately, there was no consistency among the included studies regarding the definition of the high-risk group for postpartum hemorrhage, and a range of medical conditions, such as placenta previa/accreta spectrum, maternal anemia, hypertensive disorders, multiple gestation, and anticoagulation use, could lead to the inclusion of pregnant women in this group. Although this reflects real-world practice, it significantly limits the internal consistency of the pooled analyses, and the results may not be equally applicable to all high-risk subgroups. The high heterogeneity detected in this meta-analysis may suggest that future clinical trials should consider narrowing down the spectrum of conditions regarded as high-risk. Thus, meta-analyses conducted on a specific risk category may provide more homogeneous and reliable results. Furthermore, sensitivity analysis confirmed the robustness of our findings. Notably, the extent of the effect observed in this meta-analysis suggests that prophylactic tranexamic acid, as a simple and non-invasive preventive measure, not only offers a statistically significant benefit but also clinically provides benefits for high-risk women undergoing cesarean section. In this meta-analysis, there was no case of thromboembolic events in either group, and we observed only a modest increase in the risk of tolerable nausea/vomiting after administering tranexamic acid. It should be noted that the included studies were mostly from low- and middle-income countries, and data from developed countries in Europe and America with different ethnic backgrounds and access to medical resources are needed to validate these findings.

For sensitivity analysis, we repeated the main analysis after removing two outlier studies; however, this step was exploratory and hypothesis-generating rather than confirmatory. Thus, the main pooled estimate should still be interpreted in light of substantial between-study variability. To identify the sources of heterogeneity, we also conducted subgroup analysis, which showed that a sample size of equal to or less than 100, high risk of bias, mother’s BMI more than 27 kg/m², tranexamic acid dose of 1 g, administering tranexamic acid after clamping the cord, gestational age equal to or less than 37 weeks, and only inclusion of those with placenta previa or placenta accreta as the reason for being high-risk were associated with a significantly greater decrease in the volume of blood loss.

Guinness et al. conducted a meta-analysis of 61 clinical trials comprising pregnant women, regardless of their risk level and method of delivery, and reported that although tranexamic effectively improved bleeding outcomes in the general population of pregnant women, individuals at high risk of postpartum hemorrhage, those from developing countries, and those undergoing cesarean section are more likely to benefit from prophylactic tranexamic acid [7]. Consistent with our results, this finding suggests that tranexamic acid represents a feasible and simple strategy for preventing postpartum hemorrhage in high-risk pregnant women undergoing cesarean delivery, particularly in low-income and resource-limited settings [7]. Unlike our promising findings in high-risk pregnant women undergoing cesarean delivery, a meta-analysis reported that among the general population of pregnant women undergoing vaginal delivery, tranexamic acid offers limited or no benefit in terms of preventing bleeding and postpartum hemorrhage and reducing the need for transfusion [23]. This finding suggests that physicians, nurses, and midwives pay more attention to risk factors and case selection before administering tranexamic acid prior to delivery [23, 24]. Consistently, most reports suggesting the low efficacy of tranexamic acid for preventing postpartum bleeding and hemorrhage focused on the general population of pregnant women, particularly those undergoing vaginal delivery [24]. Interestingly, data from 2245 Chinese pregnant women with risk factors for postpartum hemorrhage undergoing vaginal delivery indicated that compared to placebo, tranexamic acid did not reduce the incidence of postpartum hemorrhage, but significantly prevented severe postpartum hemorrhage and decreased the use of additional uterotonic agents [25]. This finding suggests that tranexamic acid is more effective in preventing bleeding after cesarean section, which may be due to the nature of cesarean section as a surgical intervention, which is associated with increased blood loss.

Although studies included in this meta-analysis mostly focused only efficacy outcome, data from other meta-analyses that included clinical trials of the general population of pregnant women suggest that using tranexamic acid for the prevention of postpartum hemorrhage does not increase the risk of thromboembolic events [26], but may increase the risk of nausea and vomiting [27], indicating the safety of tranexamic acid as a prophylactic treatment. However, it must be noted that due to the limited sample size and short follow-up length, the included studies could not reliably detect uncommon but clinically relevant adverse events, and caution is needed in this regard. Herein, the absence of reported thromboembolic events should not be interpreted as evidence of absence.

Interestingly, a cost-effectiveness study reported that early administration of tranexamic acid to a theoretical cohort of 100,000 women in the United States can prevent 16 maternal deaths due to hemorrhage, 9 laparotomies, and 155 reoperations, saving 438 quality-adjusted life years and an annual cost of $23.15 million [28]. Based on our findings and the results of previous studies, future guidelines may recommend the prophylactic use of tranexamic acid for high-risk pregnant women undergoing cesarean section to reduce the risk of postpartum massive bleeding and maternal morbidity and mortality. This meta-analysis showed that among high-risk pregnant women, the use of tranexamic acid can lower blood loss volume after cesarean section by -435.44 mL, which is equal to nearly 2 units of packed red blood cells (RBC) and can be life-saving for many patients. Our results also showed that the use of tranexamic acid was associated with decreased risk of postpartum hemorrhage (330 fewer cases per 1000 high-risk patients), more than 1000 mL blood loss (250 cases fewer per 1000 high-risk patients), using uterogenic agents (445 fewer cases per 1000 high-risk patients), and transfusion (208 fewer cases per 1000 high-risk patients) (Supplementary material 6). These findings indicate that the use of tranexamic acid not only led to a statistically significant change in blood loss but also exhibited a clinically meaningful effect. Consistently, the International Federation of Gynecological and Obstetrics (FIGO) strongly recommends the administration of tranexamic acid for treating postpartum hemorrhage as a component of the first response bundle when bleeding is believed to be due or partly due to trauma [29].

This meta-analysis found that tranexamic acid was more effective in studies with smaller sample sizes and a higher risk of bias, reflecting the need for larger studies with higher quality to address any inconsistencies. The main reason for the high risk of bias among the included studies was the lack of blinding, which may explain the greater effect size observed in this subgroup and result in low certainty of evidence. Consistently, the GRADE system indicated a moderate certainty for the volume of blood loss but low certainty for other efficacy outcomes.

Placenta previa or placenta accreta as the sole etiology and gestational age less than 37 weeks were associated with greater benefits of tranexamic acid. This may be due to the greater risk of bleeding in this population, where tranexamic acid can provide many more benefits. Given the safety profile of tranexamic acid, this drug may be considered for the management of these particular subgroups of high-risk pregnant women. This study also showed that a higher risk of tranexamic acid (1 g) was more effective. Since no thromboembolic event was observed and tolerable nausea/vomiting remained as the main side effect, the administration of this dose over lower doses seems rational. We also found that mothers’ BMI of more than 27 kg/m² was associated with a significantly lower blood loss after receiving tranexamic acid, which may be due to the effect of body composition on drug metabolism, bleeding tendency, and coagulation.

Limitations

This meta-analysis faced several limitations. First, some of the included studies were not conducted in a blinded manner. Second, there were methodological differences among trials, particularly in terms of the reason for being high-risk. Third, the included clinical trials enrolled a few participants and were from a single country, undermining the generalizability of their results. In addition, more studies from China, Europe, and North America are needed to cover regions with different ethnic backgrounds and different levels of access to medical resources. Multi-center, high-quality, large-scale clinical trials from different countries are necessary to validate our findings. Fourth, there was significant statistical heterogeneity for some outcomes; thus, we adopted a random-effects model. In addition, we found that two studies were causing the high heterogeneity. Fifth, we conducted subgroup analysis to better address the source of heterogeneity. Regarding the outcome definition, all outcomes were uniformly reported using standard units or objectively based on their incidence. However, Ortuanya et al. defined postpartum hemorrhage as blood loss of 1000 mL or more only due to uterine atony within the first 24 h after the delivery of the baby, but Shady et al. and Bharati et al. did not mention the definition for this outcome in their studies.

Conclusion and clinical implications

This meta-analysis indicated that among high-risk pregnant women undergoing cesarean section, tranexamic acid can significantly decrease blood loss, prevent the decline in hemoglobin level, lower the risk of postpartum hemorrhage and more than 1000 mL blood loss, and decrease the need for transfusion and uterogenic agents; however, heterogeneity remained considerable and the GRADE system exhibited low to moderate certainty of evidence, necessitating caution while interpreting the results. It should be noted that the observed effect represents an average effect across highly heterogeneous risk profiles, and extrapolation to specific high-risk subgroups should be made with caution. While no increase was observed in terms of serious adverse effects, larger datasets are required to confidently exclude small but clinically relevant risks. In addition, despite several subgroup and sensitivity analyses, statistical heterogeneity remains high, and the results should be considered hypothesis-generating rather than definitive. Therefore, large, high-quality, multicenter trials with more standardized definitions of risk and outcomes are needed to confirm the findings of this meta-analysis and assess the effects of tranexamic acid on predefined high-risk subgroups.

Authors’ contributions

Qian Gao and Pan Li (literature search, data extraction, and writing the draft), Haining Wang (conceptualization, literature search, data extraction, data analysis, and writing the draft), and Cui Li (conceptualization, writing, and editing the article). All authors reviewed and edited the manuscript and approved the final version of the manuscript.

Funding

None.

Data availability

Data will be available by the corresponding author on a reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.