Postpartum hemorrhage (PPH) is the leading cause of maternal mortality in low-income countries and represents 26.4% of all maternal deaths worldwide.1 Its prevalence varies across regions, ranging from 2.4% to 12.1%.2

The definition for PPH and severe PPH varies greatly worldwide. While some societies define PPH as any blood loss > 500 mL and severe PPH as blood loss > 1,000 mL, others define PPH as blood loss > 1,000 mL or even any blood loss that causes hemodynamic instability.3 Despite the introduction of novel approaches for early detection and intervention of PPH, parturients undergoing Cesarean delivery still face the risk of severe morbidity and mortality, or may require high-risk interventions such as hysterectomy that limit options for future fertility.4 Thus, prophylactic measures, including the administration of tranexamic acid (TXA), have been investigated in addition to uterotonics for the management of this clinical entity.

Tranexamic acid is a synthetic derivative of lysine that acts as an antifibrinolytic agent. It inhibits plasminogen activation, which stabilizes the preformed fibrin meshwork generated during secondary hemostasis. By preserving the integrity of the fibrin clot, TXA effectively reduces bleeding and promotes hemostasis.5 This pharmacologic property of TXA makes it an essential tool in managing excessive bleeding and preventing complications associated with fibrinolysis.

The prophylactic use of TXA involves its administration in situations before the onset of PPH. This approach has been investigated both prior to skin incision and after cord clamping, and across parturients with varying risk profiles. In fact, each parturient possesses their own individual risk for PPH. Parturients are at low risk for bleeding in the absence of prior uterine incisions, a singleton pregnancy, fewer than four vaginal deliveries, absence of bleeding disorders, and no history of postpartum bleeding. In contrast, a high risk of bleeding is conveyed by the presence of at least one of the following factors: 1) placenta previa or low-lying placenta, 2) suspected placenta accreta spectrum, 3) anemia, 4) active bleeding upon admission, and 5) known coagulopathy.

Prior meta-analyses have shown a significant decrease in the risk for PPH, the need of blood transfusion, and the requirement of additional uterotonics in parturients undergoing Cesarean delivery with prophylactic TXA.6,7,8 Nevertheless, the most recent and largest randomized controlled trial (RCT) on this topic to date reported no benefit of TXA in the prophylaxis of PPH, conflicting with the results from previous studies.9 In addition, there are limited data on the safety of TXA use in this patient population. Clinical practice guidelines have generally not recommended routine use of prophylactic TXA for the prevention of PPH.3

Herein, we sought to conduct an updated systematic review and meta-analysis of RCTs with a trial sequential analysis (TSA) evaluating the safety and efficacy of TXA in preventing PPH and related adverse events in parturients undergoing Cesarean delivery. We also aimed to explore the role of TXA in the prophylaxis of PPH according to strata of baseline bleeding risk.

Methods

This meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analyses Statement and recommendations from Cochrane Collaboration Handbook for Systematic Reviews of Interventions.10,11 We prospectively registered our research protocol (PROSPERO: CRD42023422188; first submitted 27 April 2023).

Eligibility criteria

We included in this meta-analysis studies that met all the following eligibility criteria: 1) RCTs, 2) trials comparing TXA with placebo or no treatment, and 3) trials in parturients undergoing Cesarean delivery. We excluded studies with 1) no control group; 2) TXA for treatment of PPH; 3) different outcomes of interest; or 4) overlapping populations, defined as trials recruiting from the same institution over an overlapping period; we also excluded 5) conference abstracts.

Search strategy and data extraction

Two authors (H. P. and M. B.) independently and systematically searched PubMed®, Embase, the Cochrane Library, and ClinicalTrials.gov from inception to 9 May 2023. The following terms were used without filters, publication date, or language restrictions: (“tranexamic acid” OR tranexamic OR TXA OR antifibrinolytic) AND (“cesarean section” OR “cesarean delivery” OR cesarean OR ceasarean) AND (random OR randomized OR randomized). The references from all included studies and previous systematic reviews and meta-analyses were also searched manually for any additional studies. Eventual conflicts were resolved by consensus among the authors. Two authors (P. C. and I. D.) independently extracted the following data from selected RCTs: 1) country, 2) number of patients, 3) timing of TXA, 4) control, 5) blinding, and 6) bleeding risk.

Endpoints and subgroup analyses

Hemorrhagic endpoints were analyzed as a binary endpoint of PPH, defined as blood loss equal or greater than 1,000 mL within 24 to 48 hr after birth, as well as a continuous outcome of total estimated blood loss. Secondary endpoints included 1) any blood transfusion, 2) additional uterotonics (administration of a higher dosage of oxytocin or use of other uterotonics than the standard protocol of oxytocin outlined in the study centres), 3) hysterectomy, 4) side effects (nausea, vomiting, dizziness, photopsia, diarrhea, and myalgia), 5) serious adverse events (thromboembolic events, ischemic stroke, myocardial infarction, seizure, and maternal death), and 6) thromboembolic events.

We performed subgroup analyses of data restricted to 1) high bleeding risk, 2) low bleeding risk, 3) TXA administered before incision, 4) TXA administered after delivery, and 5) RCTs with low risk of bias.

Quality assessment

We evaluated the risk of bias using version 2 of the Cochrane Risk of Bias Assessment Tool (RoB-2) for RCTs, wherein each study is scored as high, moderate, or low risk of bias. The assessment was performed by two independent authors (L. F. and M. S. B.) and disagreements were resolved through consensus after discussing reasons for discrepancy. We performed sensitivity analyses using leave-one-out, Baujat and L’abbé analyses. Publication bias was assessed for the outcome of PPH through the generation of a funnel plot. Additionally, an exploratory analysis was conducted excluding outliers.

To assess the certainty of evidence, we used the Grading Recommendations, Assessment, Development, and Evaluation (GRADE) tool. Using the GRADEpro Guideline Development Tool, four independent authors rated the strength of recommendations and another author resolved disagreements.12

Statistical analysis

We computed risk ratios (RRs) using the Mantel–Haenszel test for dichotomous outcomes and used 95% confidence intervals (CIs) as a measure of effect size. We considered P values of less than 0.05 to be statistically significant. We used mean differences (MD) as the effect measure for continuous outcomes, also with 95% CI.

To assess heterogeneity, we used Cochran’s Q test and I2 statistics. We classified I2 values of < 25%, 25–75%, and > 75% as representing low, moderate, and high heterogeneity, respectively. To account for potential disparities in both clinical and methodological aspects across trials, we applied the restricted maximum-likelihood estimator and random effects models for outcomes. We also performed a funnel plot and Egger’s regression test as needed to investigate heterogeneity between study-specific estimates. Our meta-analysis was conducted using the meta package for RStudio version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria).

Trial sequential analysis

We conducted a TSA to determine whether the cumulative evidence was appropriately powered. An intervention effect of a 20% RR reduction was established for the assessed outcomes. Specifically, we focused our analysis only on subgroups of studies with a low risk of bias. The outcomes examined included PPH, the requirement for blood transfusion, the use of additional uterotonics, and the occurrence of thromboembolic events. We conducted two-sided testing with a type I error of 5% and aimed to achieve a type II error of 20% (power of 80%). To compare the intervention and control groups, we constructed both conventional boundaries (with an α of 5%) and trial sequential monitoring boundaries. In the TSA, we applied a variance-based heterogeneity correction and used the random effects model. To evaluate the strength of the evidence, we constructed a cumulative sequential z-score curve.13

Additionally, we calculated the diversity-adjusted required information size (RIS), which represents the number of participants needed in a meta-analysis to detect or reject a specific intervention effect, using the aforementioned modelling (TSA version 9.5.10, Copenhagen, Denmark).14

Results

Study selection and characteristics

The initial search yielded 493 results. After removing duplicate studies, 227 records were identified through database searching and their summaries were screened for eligibility. Of these, 95 remained and were fully reviewed based on predefined eligibility criteria (Fig. 1). Thereafter, 38 RCTs were included comprising 22,940 parturients, 11,595 (50.3%) of whom were in the TXA group; 30 RCTs compared TXA with placebo, and nine with no treatment. Table 1 summarizes the individual trials’ characteristics.9,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51

Fig. 1
figure 1

PRISMA flow diagram of study screening and selection

Full size image
Table 1 Baseline characteristics of included trials
Full size table

Postpartum hemorrhage and total blood loss

In a pooled analysis of 16 RCTs, TXA was associated with a 49% relative reduction in the risk of PPH when compared with the control group (RR, 0.51; 95% CI, 0.38 to 0.69; P < 0.001; I2 = 72%; 16 RCTs; 17,795 parturients; Fig. 2A), representing 32 fewer parturients with PPH per 1,000 parturients when TXA is used. Postpartum hemorrhage was reported within 24 hr after skin incision in eight studies, and within 48 hr in three RCTs.

Fig. 2
figure 2

Forest plots of the risks of (A) postpartum hemorrhage and (B) transfusion needs

CI = confidence interval; RR = risk ratio

Full size image

We also found a significant decrease in total blood loss in the TXA group (MD, −197.7 mL; 95% CI, −237.2 to −158.3; P < 0.001; I2 = 96%; 25 RCTs, 9,882 parturients; Electronic Supplementary Material [ESM] eFig. 1). Total blood loss encompassed blood lost both during surgery and two to six hours after surgery.

Sensitivity analysis with removal of each individual study did not change the overall conclusion in these outcomes (ESM Fig. 2). The L’abbé test and Baujat plot showed that two studies primarily elevated heterogeneity (ESM eFigs 3A and 3B).17,27 Funnel plot analyses showed an asymmetric distribution of studies of different weights relative to their standard error, suggestive of nonreporting (publication) bias (ESM eFig. 4).

Transfusion needs

Twenty-two RCTs reported the need for blood transfusion and the pooled analysis showed a 57% relative reduction in the risk in the TXA group compared with the control group (RR, 0.43; 95% CI, 0.30 to 0.61; P < 0.001; I2 = 57%; 22 RCTs; 20,393 parturients; Fig. 2B), consisting of one event prevented for every 77 parturients treated.

Additional use of uterotonics and need for hysterectomy

The TXA group had a 48% relative reduction in the risk of needing additional uterotonics when compared with the control group (RR, 0.52; 95% CI, 0.40 to 0.68; P < 0.001; I2 = 84%; 17 RCTs; 19,072 parturients; Fig. 3). Nevertheless, no significant effect of prophylactic TXA was noted for the risk of hysterectomy (RR, 0.92; 95% CI, 0.51 to 1.67; P = 0.79; I2 = 0%; four RCTs; 7,962 parturients; ESM eFig. 5).

Fig. 3
figure 3

Forest plot of the need for additional uterotonics

CI = confidence interval; RR = risk ratio

Full size image

Adverse events and thromboembolic risk

Eleven RCTs reported side effects and four also reported severe adverse events. There were no significant differences between groups for either outcome (RR, 1.22; 95% CI, 1.00 to 1.48; P = 0.05; I2 = 77%; 11 RCTs; 16,637 parturients; ESM eFig. S6A and RR, 1.27; 95% CI, 0.72 to 2.22; P < 0.41; I2 = 0%; four RCTs; 15,568 parturients; ESM eFig. 6B; respectively). There were no significant differences between groups in the incidence of thromboembolic events (RR, 1.20; 95% CI, 0.37 to 3.95; P = 0.76; I2 = 54%; three RCTs; 14,344 parturients; Fig. 4).

Fig. 4
figure 4

Forest plot of the incidence of thromboembolic events

CI = confidence interval

Full size image

Subgroup analyses

All subgroup analyses showed the effectiveness of TXA when compared with control. There was a significant interaction (Fig. 5) for the outcome of PPH (P < 0.01) between TXA after cord clamping (RR, 0.86; 95% CI, 0.80 to 0.93; P < 0.01; I2 = 0%) and before incision (RR, 0.53; 95% CI, 0.40 to 0.72; P < 0.01; I2 = 67%).

Fig. 5
figure 5

Forest plot of the risk of postpartum hemorrhage regardless of the timing of tranexamic acid administration (before surgical incision and after cord clamping)

CI = confidence interval; PPH = postpartum hemorrhage; TXA = tranexamic acid

Full size image

There was no statistically significant interaction in the risk for PPH (P = 0.06) between parturients with high risk for bleeding (RR, 0.30; 95% CI, 0.21 to 0.43; P < 0.01; I2 = 0%) and low risk for bleeding (RR, 0.51; 95% CI, 0.34 to 0.75; P < 0.01; I2 = 54%; ESM eFig. 7).

Upon examining the varying economic settings of the RCTs and categorizing them into high, upper-middle, and lower-middle income countries, we found that the outcomes pertaining to our primary outcome continued to exhibit statistical significance, favouring the preventive administration of TXA. On the other hand, there was variation in the decrease in relative risk across the different groups. High-income countries had a comparatively smaller fall in relative risk, amounting to 15% (RR, 0.85; 95% CI, 0.79 to 0.92; P < 0.001; I2 = 26%; ESM eFig. 11), while lower-middle income countries observed the most substantial reduction in relative risk, reaching 61% (RR, 0.39; 95% CI, 0.24 to 0.63; P < 0.001; I2 = 77%).

Finally, we performed a subgroup analysis for the incidence of PPH by considering only RCTs with low risk of bias and this analysis showed similar findings, with a significant reduction in the incidence of PPH in the TXA group (RR, 0.63; 95% CI, 0.45 to 0.90; P < 0.01; I2 = 68%; eFig. 9).

Quality assessment

Electronic Supplementary eFig. 10 summarizes the individual evaluation of each RCT included in the meta-analysis using the RoB-2 quality assessment tool. Fourteen included studies were rated as having a low risk of bias, 24 as having some concerns of bias, and just one as having high risk of bias. According to the GRADE tool, the overall certainty of the evidence for the outcomes assessed was high and was downgraded according to the judgement of the risk of bias effect, inconsistency, imprecision, and publication bias (Table 2).

Table 2 Evidence profile: prophylactic tranexamic acid compared with control for postpartum hemorrhage in Cesarean delivery
Full size table

Trial sequential analysis

The TSA for PPH and the need for blood transfusion provided firm evidence of a relative risk reduction for the TXA group. Moreover, for the primary outcome, the cumulative sequential z-score curve not only reached but also surpassed the RIS line (Fig. 6). Concerning the use of additional uterotonics, the cumulative sequential z-curve did not cross the RIS line (ESM eFig. 11A). Regarding the need for blood transfusion, the cumulative sequential z-curve crossed the conventional boundary for benefit, suggesting a potential beneficial effect. Nevertheless, it did not cross the trial sequential monitoring boundary for benefit by a narrow margin (ESM eFig. 11B). In our TSA examining the occurrence of thromboembolic events following the use of TXA, we encountered a unique situation. These events are rare, necessitating a substantial number of cases to achieve an appropriate RIS for plotting the cumulative sequential z-score graph. Because these events are scarce, is it impractical to generate a graph that accurately represents the trend. Consequently, no specific results can be obtained for this particular outcome.

Fig. 6
figure 6

Trial sequential analysis for postpartum hemorrhage incidence with the use of prophylactic tranxamic acid. Low risk-of-bias trials showed benefit of tranxamic acid use with a sufficiently powered sample.

PPH = postpartum hemorrhage; RCTs = randomized controlled trials; TXA = tranexamic acid

Full size image

Exploratory analysis

We performed an exploratory (sensitivity) analysis removing the studies identified as contributing the most to overall heterogeneity.17,27 The results were consistent with the overall analysis, favouring the use of TXA for the reduction of PPH (RR, 0.62; 95% CI, 0.48 to 0.80; P < 0.001; I2 = 51%; ESM eFig. 13). We also performed a sensitivity analysis restricted to studies with a low risk of bias, which also favoured the TXA group (RR, 0.63; 95% CI, 0.45 to 0.90; P < 0.001; I2 = 68%). Finally, to minimize small-study effects and publication bias, we performed a sensitivity analysis removing the studies in the lower quartile of study weights (RR, 0.55; 95% CI, 0.41 to 0.73; P < 0.001; I2 = 75%; ESM eFig. 14) and the studies below the median of study weights (RR, 0.62; 95% CI, 0.44 to 0.86; P < 0.001; I2 = 78%; ESM eFig. 15), both of which consistently showed a benefit of TXA in the prevention of PPH.

Discussion

In this systematic review and meta-analysis, we identified 38 RCTs with a total of 22,940 parturients that compared the use of prophylactic TXA with placebo or no treatment in parturients undergoing Cesarean delivery. Our main findings were: 1) there was a significantly reduced risk of PPH, total blood loss, need for blood transfusion, and use of additional uterotonics with TXA use; 2) there were no statistical differences in side effects, serious adverse events, or thromboembolic events between TXA and control groups; and 3) results were consistently in favour of TXA use across subgroups, TSA, and exploratory analysis.

Our results provide compelling data supporting the efficacy of prophylactic TXA in reducing the risk of PPH. We observed a significant 49% reduction in the risk of PPH with the use of TXA compared with the control group. Additionally, TXA administration was associated with a decrease in total blood loss. Importantly, the mean difference in blood loss between groups was 197 mL, favouring the TXA group. Although this difference may not appear clinically relevant, it may be particularly important in parturients with pre-existing anemia or cardiovascular comorbidities. Moreover, a cost-effectiveness analysis showed that implementing routine prophylaxis with TXA is likely to yield significant cost savings and a decrease in adverse maternal outcomes within the context of PPH.52

Postpartum hemorrhage represents a substantial contributor to maternal morbidity and mortality on a global scale.1 Inadequate recognition and management of PPH can lead to severe maternal complications, including hypovolemic shock, organ failure, and mortality. For this reason, implementing effective strategies for the prevention and appropriate management of PPH is of paramount importance in reducing maternal mortality rates and improving overall maternal health outcomes.10

Previous meta-analyses on the role of TXA for preventing PPH focused only on efficacy outcomes and found highly heterogeneous results without performing sensitivity analyses.6,7,8 Tranexamic acid is generally considered safe. A meta-analysis of 125,500 patients in nonpregnant patients found no increased risk of thromboembolic events with TXA use.53 Nonetheless, there remains a potential concern regarding the association of TXA with thromboembolic events in this specific population, considering the hypercoagulable state during pregnancy and the postpartum period.

Our updated meta-analysis brings new insights regarding the safety of TXA. For this purpose and to verify its potential benefits for parturients with different bleeding risks, we incorporated new data from 12,281 parturients. Our report also included safety outcomes, while exploring relevant subgroups for clinical practice, conducting sensitivity analyses for heterogeneity, and performing TSA to determine whether additional studies are needed or if there is conclusive evidence to support prophylactic use of TXA.

The robustness of our findings was confirmed through sensitivity analysis, which consistently supported the overall conclusions regarding PPH risk reduction and total blood loss. The leave-one-out analysis further emphasized the clear benefit of TXA. Our analysis of the L’abbé and Baujat plots provided that only two studies contributed to most of the observed heterogeneity. Furthermore, the results were consistent among subgroups and various exploratory (sensitivity) analyses, strengthening the evidence supporting the use of TXA in the prevention of PPH.

In contrast to our meta-analysis, a recent large RCT by Pacheco et al. found no significant reduction in the incidence of PPH or the need for blood transfusion with prophylactic TXA in parturients who underwent Cesarean deliveries.9 The potential mechanisms for this discrepancy may be multifactorial. First, Pacheco et al. conducted a multicentre trial in a high-income country, whereas the majority of other RCTs were conducted in low-income countries. In these low-resource settings, the shortage of blood products or uterotonics may increase the severity of PPH, potentially increasing the benefit of TXA. Second, the prevalence of baseline anemia and other comorbidities was also substantially higher in other studies compared with the recent study by Pacheco et al., which had strict exclusion criteria related to comorbidities. And, finally, the RCT by Pacheco et al. administered TXA after cord clamping, whereas most other RCTs administered TXA before surgical incision. The latter approach may maximize the benefit of TXA. Pharmacokinetic data indicates that TXA has an onset of action of approximately three to five minutes. Therefore, it may be preferred to initiate prophylactic use prior to incision. This is corroborated by our finding showing a higher magnitude of benefit in studies with TXA administration prior to surgical incision vs after cord clamping, with a significant test for subgroup differences (P < 0.01).

While our findings indicate a difference in TXA administration timing, favouring the pre-incision period, the results of Seifert et al. suggested a decrease in therapeutic serum TXA concentration after one hour.54 In light of this, further investigations should be conducted to compare different administration times or to consider the potential benefits of continuous infusion of TXA.

The TSA results align with the findings of the meta-analysis and provide strong support for most of the examined outcomes. For our primary outcome, the analysis showed that the information size was sufficient to yield robust evidence of a 20% RR reduction in PPH. Moreover, the RIS line was significantly surpassed, indicating that our sample size was adequately powered to confidently assert that the prophylactic use of TXA reduces the incidence of PPH. Consequently, based on these results, it is unlikely that further studies in this topic would be of any added value.

The present study has limitations. First, there were methodological differences between the individual studies, some of which included only parturients with low or high risk of bleeding, or different timing of TXA administration. For example, patients with previous anemia would be at a higher risk of being transfused; however, most studies do not report incidence of anemia among their patients and the incidence of anemia may vary across different countries and socioeconomic conditions. To address these limitations, we conducted additional analyses, including leave-one-out, Baujat, and L’abbé tests, as well as subgroup analyses based on parturients' bleeding risk, timing of TXA administration, and risk of bias. Second, the asymmetry in the funnel plots for PPH suggest publication bias. To further explore this and to minimize the small-study effects, we conducted an exploratory analysis by removing the studies with lower weight (below median and lower quartile). The results were consistent with the overall analysis, suggesting that publication bias did not affect our results.

Despite these limitations, the present study also has important strengths. First, it represents the largest and most updated review of TXA for the prevention of PPH in parturients undergoing Cesarean delivery. Second, the analysis was restricted to RCTs, which minimizes the risk of any confounding factors. Third, we conducted a comprehensive search for eligible RCTs, encompassing studies published in any language, thus minimizing potential language bias. And lastly, our meta-analysis is the first to assess the safety of prophylactic TXA in terms of serious adverse effects and thromboembolic events in this population. These strengths contribute to the robustness and novelty of our findings, enhancing the overall value and significance of the study.

The findings of our meta-analysis carry important implications for clinical practice. The favourable safety profile of TXA, coupled with its shown effectiveness in reducing the risk of PPH, the need for blood transfusion, and the additional use of uterotonics, provides strong justification for considering its widespread prophylactic use in the context of Cesarean delivery. These results suggest that incorporating TXA as a preventive measure may contribute to improved parturient outcomes and potentially reduce the burden on health care resources associated with PPH management.

Conclusion

Our meta-analysis of 38 RCTs supports the prophylactic use of TXA in Cesarean deliveries to reduce the risk of PPH, except in cases of contraindication, and preferably before surgical incision and in the presence of maternal comorbidities. Tranexamic acid showed significant benefits in terms of reduced PPH incidence, total blood loss, and the need for blood transfusion and additional use of uterotonics. Safety analysis indicated no significant differences in serious adverse events or thromboembolic events between TXA and control groups. These findings highlight the potential of TXA to improve parturient outcomes and justify its consideration as a preventive measure in Cesarean deliveries.