TRAAP2: Tranexamic Acid for the Prevention of Blood Loss after Cesarean Delivery

Sentilhes L. New Eng J Med 2021; 384:1623-1634. doi:0.1056/NEJMoa2028788

Clinical Question

  • In patients undergoing cesarean section (CS) does the administration of tranexamic acid (TXA) when compared to a placebo result in a lower incidence of calculated estimated blood loss or red blood cell transfusion by day 2?


  • The use of TXA has been widely studied in post partum haemorrhage (PPH)
  • The WOMAN trial showed that TXA may be beneficial in reducing the risk of death due to PPH
    • This has been lauded as a landmark trial but there are some controversies that are excellently summarised here
  • The evidence surrounding the use of TXA post CS is limited and consists mainly low quality evidence and there is no strong evidence to support its use in a prophylactic manner


  • Multi-centre, randomised, placebo controlled, double blind trial
  • Randomised to receive uterotonic agent plus either TXA or placebo
  • Randomised in 1:1 ratio
    • Computer based, in blocks of 4
    • Stratified by site and whether CS done pre or during labour
  • TXA and placebo in identically labelled packaging and vials
  • Appropriate consent process – written informed consent obtained by treating teams if CS considered likely
  • Blood loss estimated using following equation:
    • Estimated blood volume x [(pre-op haematocrit-post-op haematocrit) / post-op haematocrit]
    • Pre-op haematocrit was most recent in 8 days prior to caesarean
    • Post-op was closest to day 2 post delivery
  • Physician responsible for the delivery prospectively recorded the procedures used during the third stage of labor and clinical outcomes identified in the immediate postpartum period
  • Adverse events were assessed until hospital discharge and a telephone interview was conducted at 3 months post delivery
  • 4525 patients required
    • 4072 patients estimated (80% power to detect a 20% relative difference in incidence of primary outcome, with 5% type 1 error rate)
    • Allowed for up to 10% to be lost to follow, deliver vaginally or not have blood tests taken on day 2
    • Predominantly based on data from ECSSIT trial
  • Performed with modified intention to treat population
    • Those who withdrew consent, delivered vaginally, or deemed ineligible after randomisation were excluded
  • There were two pre-specified pre-protocol populations: those who received TXA within 3 minutes and those who received TXA within 10 minutes
  • Registered with NCT3431805


  • 27 centres in France
  • March 2018 – January 2020


  • Inclusion:
    • > 18 yo
    • >/= 34 weeks gestation
    • expected to undergo Caesarean before or during labour
  • Exclusion:
    • Known or suspected increased risk of VTE / arterial embolism
    • Increased risk of bleeding (including administration of LMWH or anti-platelet in week prior)
    • History of seizures or epilepsy
    • Eclampsia
    • Autoimmune disease
    • Failed operative vaginal delivery
    • Sensitivity to trial drugs
    • Hb < 9g/dL in the week prior
    • Poor comprehension of spoken French
  • 4551 randomised
    • 2276 – TXA
    • 2275 – placebo
  • Baseline demographics similar
    • Age: 33.3 (TXA) vs 33.5 (Placebo)
    • BMI: 26.3 vs. 26.1
    • Primip: 37.2% vs. 36.6%
    • Prior CS: 51.8% vs. 52.4%
    • History of PPH: 5.1% vs. 4.5%
    • Multiple pregnancies: 7.2% vs. 7.2%
    • Timing of CS – during labour: 28.9% vs. 29.2%
    • GA: 3.0% vs. 3.8%
    • Prophylactic uterotonic: 98.9% vs. 99.0%
    • Anticoagulant prophylaxis post delivery: 58.8% vs. 59.1%
    • Median duration of interval from birth to TXA: 2 mins vs. 2 mins


  • 1g TXA
    • 10ml vial


  • Normal saline
    • Identically labeled vial

Management common to both groups

  • Either TXA or placebo given over 30-60 seconds in the first three minutes after birth
  • This was given after bolus prophylactic uterotonics given (5-10u oxytocin or or 100 μg of carbetocin)
    • Subsequent oxytocin infusion at discretion of treating unit
  • All spent at least 2 hours in PACU / Recovery until bleeding diminished to expected amount


  • Primary outcome:
    • Postpartum hemorrhage (defined as a calculated estimated blood loss greater than 1000 ml or red-cell transfusion by day 2): – significantly reduced in tranexamic acid group
    • 556/2086 (26.7%) TXA vs. 653/2067 (31.6%) Placebo
      • Adjusted RR 0.84 (95% CI – 0.75 – 0.93, p = 0.003)
      • No difference between pre-specified subgroups (PPH risk and timing of CS)
  • Secondary outcomes:
    • No significant differences in:
      • Gravimetrically estimated blood loss: 689ml vs. 719ml
        • Estimated by suction volume and swab weight
      • Clinically significant blood loss: 13.6% (TXA) vs. 14.8% (Placebo)
      • Additional uterotonic agents for excessive bleeding: 5.9% vs. 7.2%
      • Blood transfusion: 1.9% vs. 1.8%
      • Number of units transfused: 3.1 vs. 3.2
    • Significant difference in pre and post-operative Hb:
      • Peripartum change -1.2 g/L vs. -1.4g/L (p < 0.001)
  • Adverse Events:
    • In PACU or OR:
      • Higher rates of nausea and vomiting
        • 43.0% vs. 36.3% (p < 0.001)
    • D2 after delivery:
      • No significant differences between groups for urea, creatinine, AST, ALT
    • Up to 3 months after delivery:
      • No significant difference for:
        • Seizure: <0.1% vs. <0.1% (1 in each arm)
        • DVT/PE: 0.4% vs. 0.1% (8 cases in TXA arm, 2 in placebo)

Authors’ Conclusions

  • The use of TXA in this setting reduced the incidence of PPH (as defined by calculated estimated blood loss > 1000ml or red cell transfusion by day 2).
  • However there was no reduction in clinical secondary outcomes


  • Large, multi-centre trial
  • Well balanced balanced baseline characteristics
    • High proportions of risk factors for PPH (previous caesarean, prior PPH, obesity, multiple pregnancy, emergency caesarean) present in baseline characteristics
    • This is important to ensure that two low risk cohorts not being analysed
    • Sensible pre-specified subgroup looking at the use of TXA in those at high risk of PPH vs. low risk
      • Defined as presence of a one or more risk factor with an OR > 3
  • Good randomisation and blinding strategy results in strong internal validity
  • Pragmatic statistical analysis plan
    • Although large numbers of women were randomized but not assessed for the primary outcome in the modified intention to treat analysis, the statistical analysis plan allowed for up to 10% drop out following randomisation
  • Sensible use of a standardised method of estimating blood loss given clinicians’ estimations are often incorrect
    • The values of estimated blood loss obtained by the gravimetrical method (suction volume, and weighing swabs) and the standardised equation similar
  • Minimal cross-over / protocol violations
    • Only 1.4% in each group received no treatment
    • Only 0.1% in each group received the opposite treatment
  • Good follow up at three months – important for reporting adverse events
    • 93.5% (TXA) vs. 94.4% (Placebo) completed follow up interviews at three months


  • The primary outcome is not patient orientated
    • Although the trial is not adequately powered for clinically relevant secondary outcomes there is no suggestion that TXA reduces the rate of these
      • The authors rightly acknowledge this in the conclusion
  • As discussed, the statistical plan allowed for up to 10% drop out from those who are randomised to receive a treatment, however:
    • The rates were slightly higher than 10% in each arm:
      • TXA – 10.5%
      • Placebo – 11.8%
        • This may introduce an attrition bias
    • 210 (4.6%) patients excluded for unknown reasons or meeting exclusion criteria following randomisation
      • This may introduce a selection bias
    • 278 (6.1%) had missing data for the primary outcome
      • The appendix has a comparison of where only complete cases are analysed (the data presented above) against the imputation of missing values as failures (i.e. they were assumed to have PPH)
        • If the missing values are treated as failures then the p value > 0.005
          • TXA 31.1% vs Placebo 36.0% p = 0.006
  • PPH is a heterogenous process
    • The 4Ts is a common acronym for PPH reasons: uterine tone, retained tissue, trauma, thrombin
      • TXA does not have a plausible biological mechanism of action in the treatment of PPH for anything other than trauma
      • No documentation of suspected reasons for PPH is recorded
      • The protocol allowed for the use of a oxytocin infusion as per local hospital policy following the uterotonic bolus
        • No mention is made of how many had an oxytocin infusion after
          • This could be an important confounder
  • The study excluded those who are increased risk of bleeding – would these patients potentially benefit more?
  • The BNF suggests that TXA should be given as a slow IV injection (at a rate not exceeding 100mg/minute)
    •  The protocol states it should be given over 30 to 60 seconds (20x as fast)
      • This may be a reason for the increased rate of vomiting and nausea
  • Approximately 25% did not receive TXA within three minutes in each arm
    • This is understandable as it is an already frenetic time ensuring that uterotonics are given, and that the baby and parents are well!
    • This delay did not seem to have an effect though as in the protocol analysis which included those who received TXA/placebo within 10 minutes still showed a significant reduction in the primary outcome

The Bottom Line

  • The use of prophylactic TXA for caesarean delivery reduces the rates of PPH as defined by an estimated blood loss of > 1000ml or red cell transfusion by day 2
  • I will not use TXA in a routine prophylactic manner given this did not seem to show any trend to a reduction in meaningful clinical outcomes, and an increased risk of nausea and vomiting

External Links


Summary author: George Walker
Summary date: 9th May 2021
Peer-review editor: David Slessor

Image by: iStock


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.