New approaches in the haemostatic management of postpartum haemorrhage

Introduction Postpartum haemorrhage, or excessive bleeding after childbirth, is the leading cause of maternal mortality worldwide. Postpartum haemorrhage affects 1-10% of all deliveries and accounts for about 27% of all maternal deaths, especially in settings with limited resources. 1 The World Health Organization has estimated that 70 000 women globally die each year as a result of postpartum haemorrhage.2 Despite the high morbidity and mortality of postpartum haemorrhage, research to understand the haemostatic changes of postpartum haemorrhage and clinical trials to provide evidence based management are lacking. Consequently, management guidelines have extrapolated evidence from other clinical scenarios of blood loss, in particular traumatic bleeding. In contrast with postpartum haemorrhage, interest in traumatic bleeding in the past 20 years has increased. Multiple large trials assessing management have shown better clinical outcomes if fresh frozen plasma is used in a ratio of 1:1 or 1:2 initially, with red cell transfusion in patients with major bleeding.3 4 Usual advice for any patient with bleeding was essentially to give red blood cells to maintain adequate oxygenation and manage any coagulopathy later. Data for the initial use of fresh frozen plasma in trauma has been applied to the management of all major bleeding conditions, including postpartum haemorrhage. Physiological disparities exist, however, between the average trauma patient and a pregnant individual at term in a highly prothrombotic state within an expanded plasma volume. In this review, we report on the new understanding of the haemostatic changes of postpartum haemorrhage and how the coagulopathy of severe postpartum haemorrhage can be identified and managed. Individuals described in this review were pregnant women and other pregnant people.How postpartum haemorrhage is diagnosed The use of clear and standardised diagnostic criteria for postpartum haemorrhage is important for prompt identification and timely management. A lack of global consensus exists, however, about how best to diagnose postpartum haemorrhage in its varying severities, in terms of volume of blood loss or clinical parameters of haemodynamic instability, or both ( table 1).Table 1Definitions of postpartum haemorrhage in international guidelinesGuidelineDefinition of primary postpartum haemorrhageRoyal College of Obstetricians and Gynaecologists, UK44 Blood loss of ≥500 mL within 24 hours of birthMinor (500-1000 mL)Major (>1000 mL); subdivided into moderate (1001-2000 mL) and severe (>2000 mL). Additional recommendations:In women with lower body weight (eg, <60 kg), a lower level of blood loss may be clinically significantVisual estimation of peripartum blood loss is inaccurateClinical signs and symptoms should be included in the assessment of postpartum haemorrhage American College of Obstetrics and Gynaecology55 Cumulative blood loss of ≥1000 mL, or blood loss accompanied by signs or symptoms of hypovolaemia, within 24 hours of birthRecommended use of quantitative, rather than estimated, blood loss to improve the accuracy of postpartum haemorrhage diagnosis World Health Organization68 Blood loss of ≥500 mL within 24 hours of birth.Severe postpartum haemorrhage defined as blood loss >1000 mL International Federation of Gynaecology and Obstetrics (FIGO)49 Blood loss >500 mL in a vaginal birth and >1 L in a caesarean deliveryFor clinical purposes, any blood loss that has the potential to produce haemodynamic instability should be considered a postpartum haemorrhageClinical estimates of blood loss are often inaccurate. Postpartum haemorrhage has commonly been defined as blood loss ≥500 mL within 24 hours of delivery.2 A diagnosis based on volume of blood loss, however, is hampered by inherent difficulties in accurately determining blood loss at delivery. Visual estimation of blood loss is inaccurate, even by experienced clinicians, and often underestimates the volume of blood lost. As a result, the use of a definition of postpartum haemorrhage that relies on estimated volume of blood loss only is inadequate and likely to result in delays in treatment.5 Quantitative methods, such as weighing blood soaked items or using calibrated collection devices, have more precise measurements. The estimated blood volume of the individual needs to be factored in, however, because the same volume of blood loss will be more significant in a smaller individual. Measurement of haemodynamic instability in assessing the severity of blood loss is challenging in healthy pregnancies, because blood loss of >1 L can be tolerated with minimal changes in clinical observations, likely because of the increased circulating blood volume during pregnancy.6 7 A clear definition, combining clinical parameters with volume of blood loss, should be used in a multidisciplinary setting to identify women with postpartum haemorrhage in a timely manner and allow early monitoring of those that may develop more substantial bleeding.Haemostatic changes in normal pregnancy A gradual increase in the prothrombotic state occurs during pregnancy ( figure 1), so that at term, the risk of venous thromboembolism is 8-10 times that of the non-pregnant risk. Most coagulation factors increase during pregnancy. In particular, fibrinogen, von Willebrand factor, and factor VIII reach concentrations at least 2-3 times the normal level in a circulating plasma volume of 140% of the non-pregnant state.8 9 Factor XIII, in contrast, gradually decreases over the course of pregnancy, although levels usually remain within the normal reference range.10 Of all of these changes, the increase in fibrinogen is probably the most important because levels of fibrinogen are the highest of any coagulation factor in blood. Fibrinogen is important because it forms fibrin, the main constituent of a clot, and also is the ligand for platelet aggregation. A decrease in the level of physiological anticoagulant protein S is also seen, which further contributes to the prothrombotic state. These changes result in an increased ability to generate thrombin during pregnancy, as measured by tissue factor dependent thrombin generation assays.11 Figure 1Key haemostatic changes in pregnancy. PAI-1=plasminogen activator inhibitor 1; PAI-2=plasminogen activator inhibitor 2Complex changes are also seen in fibrinolysis, with increases in plasminogen activator inhibitor 1 and the production of plasminogen activator inhibitor 2 from the placenta, offset by smaller increases in tissue plasminogen activator, leading to overall reduced fibrinolysis. Also, levels of D-dimer gradually increase during pregnancy.8 9 Circulating platelet numbers increase in pregnancy, but this increase is less than the rise in plasma volume, hence causing a mild reduction in platelet count by volume. The decrease in platelet count tends to occur, or worsen, towards term, with 5-10% of women at term having a platelet count less than that of the non-pregnant adult normal range, but with a count of >100×109/L at delivery in most uncomplicated pregnancies.12 13 This so-called gestational thrombocytopenia does not result in a bleeding diathesis because the increase in fibrinogen is greater and more than compensates for the reduced number of platelets, and therefore the prothrombotic state remains.Causes of postpartum haemorrhage Several causes and risk factors for the development of postpartum haemorrhage are known ( table 2), highlighting areas where prevention is possible.Table 2Causes and risk factors for development of postpartum haemorrhage69–72 CausesRisk factorsUterine atonyMultiple pregnancyPrevious postpartum haemorrhageFetal macrosomiaFailure to progress in second stage of labourProlonged third stage of labourGeneral anaesthesiaTrauma related to deliveryInduction of labourInstrumental deliveryProlonged labourCaesarean sectionAbnormal placentationPlacenta praeviaPlacenta accretaPlacental abruptionChorioamnionitisHaematologicalInherited and acquired coagulopathy or haemostatic disordersThrombocytopeniaMaternal anaemiaOtherPre-eclampsia and HELLP syndromeAntepartum haemorrhagePolyhydramniosOlder maternal ageHELLP, haemolysis, elevated liver enzymes, and low platelets.Coagulopathy in the peripartum period can occur as a result of known coagulation disorders or from obstetric complications, such as HELLP (haemolysis, elevated liver enzymes, and low platelets) syndrome. These disorders can worsen the severity of postpartum haemorrhage and require prompt identification and treatment. Patients with known inherited bleeding disorders, such as von Willebrand disease or other deficiencies in coagulation factors, or inherited platelet disorders, should be reviewed by their specialist haemostasis centre during pregnancy to provide a clear plan about the precautions needed to reduce the risk of postpartum haemorrhage during delivery, and its management if bleeding occurs.14 A key determinant of normal haemostasis is a normal level of haemoglobin. Bleeding time (ie, time taken to start clot formation) is related to the level of anaemia: the more anaemic the patient, the longer it takes to form a thrombus or clot because of loss of axial flow. Axial flow means flow of red cells down the centre of an arteriole, with plasma proteins and platelets being thrown to the periphery, so that if a break occurs in the endothelium, platelet and coagulation proteins can quickly respond in forming a clot. With severe anaemia that relation is lost, with a mix of red cells, plasma proteins, and platelets close to the endothelium so that it takes longer for platelets and coagulation proteins to reach the breach in the vessel.15 Anaemia occurs in half a billion women of reproductive age worldwide and increases the risk of postpartum haemorrhage. This finding highlights an area where prevention of postpartum haemorrhage is possible by identifying and managing anaemia antenatally or ideally before conception.16 Haemostatic changes during postpartum haemorrhage Postpartum haemorrhage is exacerbated by any underlying haemostatic impairment, either pre-existing or developing over the course of the haemorrhage. Until recently, however, the characteristics of postpartum haemorrhage associated coagulopathy have been poorly defined.Reduced fibrinogen and factor XIII as predictors of postpartum haemorrhage Lower levels of fibrinogen have been shown to be predictive of ongoing bleeding in postpartum haemorrhage, with levels <2 g/L predictive of progression to massive postpartum haemorrhage. 17–19 The mechanisms of fibrinogen depletion are not clear, but are presumed to be consumption in coagulation, especially for uterine atony and bleeding into the uterine cavity, and also fibrinogenolysis caused by stimulated fibrinolysis. Marked hypofibrinogenaemia has been reported in association with amniotic fluid embolism20 and placental abruption,20 21 with some suggestion of disseminated intravascular coagulopathy.22 Disseminated intravascular coagulopathy is caused by widespread pathological activation of coagulation and fibrinolysis, leading to excessive or uncontrolled in vivo thrombin generation, microvascular thrombosis, and end organ damage, with consumption and depletion of coagulation factors and platelets, and raised fibrinogen degradation products. D-dimers are increased and prothrombin time and activated partial thromboplastin time are usually prolonged, together with hypofibrinogenaemia. In a review of 456 patients with postpartum haemorrhage of ≥1500 mL in a UK centre over three years, activated partial thromboplastin time and prothrombin time did not correlate with blood loss as strongly as fibrinogen levels, however, and prolonged prothrombin time and activated partial thromboplastin time were rare until blood loss was >3 L.19 Factor XIII seems to be another predictor of postpartum haemorrhage. The physiological role of factor XIII is to cross link fibrin strands, stabilising the blood clot. Lower levels of factor XIII before labour, although within the normal range of the population, have been shown in two observational studies23 24 (of 1300 and over 300 women, respectively) to be associated with an increased risk of postpartum haemorrhage.Acute obstetric coagulopathy The Obstetric Bleeding Study plus (OBS+) was a prospective exploratory study that examined haemostatic changes in 518 women recruited with postpartum haemorrhage of ≥1000 mL or placental abruption, amniotic fluid embolism, or concealed bleeding, from a cohort of >11 000 individuals. 25 The findings have changed our understanding of postpartum haemorrhage by showing that most patients did not develop haemostatic impairment, but a subgroup of 12 women (1.06/1000 maternities) had a distinct coagulopathy, termed acute obstetric coagulopathy. These patients had poor outcomes, with associated fetal or neonatal death in 50% of patients and increased maternal morbidity. Acute obstetric coagulopathy is characterised by early, severe hypofibrinogenaemia, acute acquired dysfibrinogenaemia, hyperfibrinolysis, reduced levels of factor V, and decreased factor VIII to von Willebrand factor ratio, but with preservation of other procoagulant factors.25 The underlying cause is excessive generation of plasmin.26 These 12 patients with acute obstetric coagulopathy26 were first identified by assessing markers of fibrinolysis in a subset of 130 women in the OBS+ study who were at high risk of coagulopathy, with blood loss >2000 mL or presenting with placental abruption or amniotic fluid embolism, or having received blood products to manage postpartum haemorrhage. Fibrinolysis was assessed by D-dimer and plasmin α2 antiplasmin levels. Plasmin α2 antiplasmin is produced when plasmin is released and fibrinolysis is activated. The 12 women with acute obstetric coagulopathy had markedly increased levels of plasmin α2 antiplasmin (>40 000 ng/mL). In a subanalysis, these 12 women also had >25-fold higher levels of D-dimers with platelet counts marginally lower than those without acute obstetric coagulopathy. Levels of factors V and VIII were lower in the 12 patients with acute obstetric coagulopathy, whereas other coagulation factors were similar between the groups, suggesting specific rather than generalised depletion of coagulation factors.Clauss fibrinogen (a measure of fibrinogen functional activity) was lower in the acute obstetric coagulopathy group (median 2.1 g/L). Fibrinogen antigen levels were reduced to a lesser extent than Clauss fibrinogen. The Clauss fibrinogen to antigen ratio was lower in the group with acute obstetric coagulopathy, indicating an acquired fibrinogen dysfunction (ie, dysfibrinogenaemia), as well as reduced fibrinogen levels (ie, hypofibrinogenaemia). Plasminogen and factor XIII levels were also decreased in the acute obstetric coagulopathy group.The combination of changes seen in this newly described acute obstetric coagulopathy (figure 2) is typical of hyperfibrinolysis (ie, effects of excess plasmin). Plasmin is formed from circulating plasminogen by plasminogen activators, such as tissue plasminogen activator, and breaks down established fibrin clot generating fibrinogen degradation products. Fibrinogen degradation products, by coating fibrinogen, can interfere with the ability of fibrinogen to clot, and this action results in dysfibrinogenaemia. Fibrinogen is also a direct target for plasmin, especially if fibrin levels are low, causing hypofibrinogenaemia. Lastly, plasmin can inactivate factor V27 and factor VIII. Disproportionately low levels of factor V compared with other coagulation factors is also characteristic of fibrinolytic bleeding in situations such as trauma.28 29 Figure 2Haemostatic changes in acute obstetric coagulopathy. FDPs=fibrin and fibrinogen degradation productsAcute obstetric coagulopathy therefore seems to be caused by excessive and dysregulated generation of plasmin, although the underlying trigger for this effect remains to be elucidated. Acute obstetric coagulopathy was associated with placental abruption in five of the 12 patients, and one patient had amniotic fluid embolism, but acute obstetric coagulopathy was not restricted to these obstetric scenarios. A similar coagulopathy associated with excessive plasmin, fibrinogen depletion, and hyperfibrinolysis has also been described in a case-control study comparing women with amniotic fluid embolism (n=27) or placental abruption (n=12) to controls (n=23).30 The mechanisms underlying acute obstetric coagulopathy, and its relation to obstetric disseminated intravascular coagulation, and optimal management, need further investigation, especially given the poor fetomaternal outcomes.Haemostatic changes in patients with non-acute obstetric coagulopathy postpartum haemorrhage In the 506 patients recruited to the OBS+ study who did not have features of acute obstetric coagulopathy (median blood loss 1200 mL), coagulopathy was uncommon, with only 2.4% with a Clauss fibrinogen level of <2 g/L and 1.3% with an abnormal prothrombin time or activated partial thromboplastin time. 25 Median levels of factors II, V, VII, IX, X, and XI, and platelets decreased linearly and in similar proportions as bleed volume increased, but remained within the normal range of non-pregnant people, even at bleed volumes >3000 mL, except in two patients. Median levels of factor VIII and von Willebrand factor antigen did not decrease with increasing bleed volume; median levels of von Willebrand factor antigen increased with blood loss of 1000–1999 mL compared with non-bleeding pregnant controls. In women with postpartum haemorrhage, factor XIII decreased with blood loss >2000 mL compared with healthy term pregnant controls, and the median level (27 IU/dL) was below the non-pregnant normal range for blood loss >3000 mL.The decrease in coagulation factors with increasing bleed volume was not accompanied by a rise in D-dimers, suggesting that the cause of the decrease in levels of coagulation factors was consumption resulting from bleeding and clot formation, and dilution because of resuscitation, rather than disseminated intravascular coagulopathy. We can hypothesise that this finding would be particularly marked in those with uterine atony where a large uterus exists to fill with clot. When the combined effect of the procoagulant factors was assessed based on thrombin generation, parameters were above or at comparable levels with non-pregnant controls, even at high bleed volumes, suggesting adequate levels of coagulation factors to support thrombin generation. The importance of the decrease in factor XIII is uncertain and is currently being researched.Acute obstetric coagulopathy versus acute traumatic coagulopathy The pattern of fibrinolytic changes described in acute obstetric coagulopathy is similar to that seen in early traumatic coagulopathy. 29 Both disorders cause marked activation of fibrinolysis. The major difference is that pregnant women have an expanded plasma volume and platelet number with marked prothrombotic changes and so can withstand the early losses of coagulation factors and platelets. Therefore, theoretically, initial use of fresh frozen plasma to provide additional coagulation factors is not as critical as for traumatic bleeding. In both traumatic bleeding and postpartum haemorrhage, levels of factor VIII and von Willebrand factor are preserved. In traumatic coagulopathy, increased levels of factor VIII and von Willebrand factor are attributed to adrenaline (epinephrine), cytokine, and thrombin activation, leading to endothelial cell activation and discharge of Weibel-Palade bodies releasing von Willebrand factor. The presumption is that a similar pathophysiological mechanism occurs in postpartum haemorrhage.Identifying coagulopathy in postpartum haemorrhage Coagulopathy has traditionally been detected in various clinical settings by coagulation screens (comprising prothrombin time, from which an international normalised ratio can be derived, activated partial thromboplastin time, and fibrinogen) and platelet count. Prothrombin time is a measure of the extrinsic (factor VII) and common (factors II, V, and X) coagulation pathways whereas activated partial thromboplastin time measures intrinsic (factors VIII, IX, XI, and XII) and common pathways. The turnaround time for a coagulation screen and full blood count will depend on sample time to the laboratory and the laboratory workflow, but even for the most urgent samples is generally about 30-60 min. Neither prothrombin time nor activated partial thromboplastin time are sensitive to deficiencies in factor XIII or abnormalities of fibrinolysis.In contrast, viscoelastic haemostatic assays, including thromboelastography (TEG, Haemonetics, Braintree, MA, USA) and rotational thromboelastometry (ROTEM, Tem, Munich, Germany) provide rapid and comprehensive information on whole blood coagulation dynamics at the point of care and are increasingly being used to identify and monitor haemostatic changes in postpartum haemorrhage. The clinical use of viscoelastic haemostatic assays is now well established in cardiac surgery, liver transplantation, and trauma, but not currently in obstetric settings.31 32 TEG and ROTEM both measure the mechanical properties of clot formation, as well as clot lysis, in whole blood.31 32 The basic principle is to place whole blood in a cup, add activators as required, and then place a pin or probe in the middle of the blood. The cup moves relative to the pin or probe, or vice versa. As clot formation and then lysis occur, the resistance to movement changes. The changes are translated by various methods into a graphical trace, which appears in real time over 20-30 min. From the trace, several key parameters are reported (figure 3 and table 3). Although TEG and ROTEM traces look identical, the parameters are not directly interchangeable and should not be regarded as equivalent. Also, recorded parameters depend on the device used, and normal or target ranges will depend on the clinical setting, with different normal ranges for many parameters seen in pregnancy.33 Figure 3Example thromboelastography (TEG) and rotational thromboelastometry (ROTEM) trace with key measured parameters. R=reaction time; CT=clotting time; K=kinetics time; α=α angle; CFT=clot formation time; MA=maximum amplitude; MCF=mean clot firmness; Ly=lysis; CL=clot lysisTable 3Comparison of thromboelastography (TEG) and rotational thromboelastometry (ROTEM) parameters and their significanceCoagulation phaseTEG parameterROTEM parameterPhysiological significanceStart of coagulationR (reaction time)CT (clotting time)Time to initial fibrin formation; reflects clotting factors.Amplification of coagulation; formation of stable clotα angleK (kinetics time)α angleCFT (clot formation time)Speed of clot propagation; reflects thrombin generation, and fibrin-platelet interactionsMaximum strengthMA(maximum amplitude)MCF(mean clot firmness)Peak clot strength; reflects platelets and fibrinogenLysisLy(lysis)CL (clot lysis)ML (maximal lysis)Rate of fibrinolysisThe latest devices (eg, TEG 6S and ROTEM sigma) are compact and use new cartridge based techniques with lyophilised reagents designed to improve ease of use and precision, facilitating their use as point-of-care devices by trained healthcare professionals. The cartridges have separate test channels with different reagents to allow different aspects of coagulation to be assessed, with separate traces (and corresponding parameters) for each test. Table 4 provides a list of tests and reagents for TEG and ROTEM.Table 4Thromboelastography (TEG) and rotational thromboelastometry (ROTEM) tests and reagents31 TestReagents usedWhat the trace assessesTEG-6S  CKKaolin, calciumStandard clot formation by intrinsic pathway activation  CKHKaolin, calcium, heparinaseWhen compared with CK trace, a shorter reaction time suggests the presence of heparin  CRTKaolin, calcium, tissue factorStandard clot formation by extrinsic pathway activation  CFFTissue factor, Reopro (glycoprotein IIb/IIIa inhibitor, inhibiting platelet activity), calciumClot formation with platelet contribution removed; assesses fibrinogen contribution. If compared with standard kaolin trace, platelet contribution can be assessedROTEM sigma  INTEMEllagic acidStandard clot formation activating the intrinsic pathway  EXTEMTissue factorStandard clot formation activating the extrinsic pathway  HEPTEMEllagic acid and heparinaseWhen compared with INTEM trace, a shorter clotting time suggests the presence of heparin  APTEMTissue factor and aprotininWhen compared with EXTEM trace, differences in maximum amplitude or mean clot firmness suggest contribution of fibrinolysis  FIBTEMTissue factor and cytochalasin CPlatelets inhibited, contribution of fibrinogen to clot remains. If results compared with EXTEM trace, platelet contribution can be assessedCompared with laboratory coagulation testing, the viscoelastic haemostatic assays offer the benefits of reduced turnaround times and provide a more global assessment of haemostasis. The disadvantages include potentially less stringent quality control processes and greater inter-user variability. In addition, there are cost implications of purchasing the relevant machine and cartridges and reagents, which currently precludes use in many resource limited healthcare settings.Viscoelastic haemostatic assays Studies have shown moderate to good correlation between measures of clot strength with the FIBTEM test (ROTEM), measures of clot strength with the CFF test (TEG), and laboratory Clauss fibrinogen levels during postpartum haemorrhage, and acceptable sensitivity and specificity in detecting hypofibrinogenaemia. 34–37 For example, in a prospective observational cohort study of 521 women with moderate to severe obstetric haemorrhage (>1000 mL of blood loss), the sensitivity and specificity of TEG 6S CFF A10 ≤17 mm to detect fibrinogen ≤2 g/L were 0.74 and 0.97, respectively.37 In the same cohort, sensitivity and specificity of ROTEM sigma FIBTEM A5 ≤11 mm to detect fibrinogen ≤2 g/L were 0.76 and 0.96, respectively.34 Furthermore, in a prospective observational study of a consecutive cohort of 356 women with postpartum haemorrhage (blood loss 1000-1500 mL), FIBTEM and fibrinogen were measured and subsequent transfusions, invasive procedures, and bleed volume recorded. On multivariate analysis, FIBTEM was an independent predictor for progression to blood loss >2500 mL.38 ROTEM and TEG cannot identify specific deficiencies in coagulation factors, other than fibrinogen. Normal test results for ROTEM sigma EXTEM clotting time or TEG 6S CK R (kaolin and calcium reaction time), however, have been shown to have a high negative predictive value (≥0.95) in identifying abnormalities in prothrombin time or activated partial thromboplastin time, giving a potential use in identifying when fresh frozen plasma is not needed.34 37 Haematological management of postpartum haemorrhage Tranexamic acid Tranexamic acid, an antifibrinolytic agent, is an analogue of lysine and binds to the lysine binding sites on plasminogen. Essentially, tranexamic acid is a competitive inhibitor of plasminogen. WOMAN (World Maternal Antifibrinolytic Trial) compared the use of tranexamic acid with placebo in 20 000 women and is currently the largest randomised controlled trial of postpartum haemorrhage. 39 The trial was conducted in low, middle, and high income countries where management of traumatic bleeding and postpartum haemorrhage varied widely. The trial showed a reduction in bleeding deaths, with maximal benefit if tranexamic acid was given as early as possible after giving birth; no benefit was found if tranexamic acid was given three hours after birth. Tranexamic acid was safe, with no adverse effects for mothers or babies, and in particular no increased rates of venous thromboembolism. 39In WOMAN, tranexamic acid was established as an antifibrinolytic agent to reduce perioperative bleeding, with little knowledge about the state of fibrinolysis in postpartum haemorrhage when the trial started. Collins et al have now shown that acute obstetric coagulopathy was driven by hyperfibrinolysis.26 A large multicentre randomised controlled trial was conducted in >4000 women to investigate if tranexamic acid could prevent bleeding in caesarean deliveries.40 Women underwent caesarean delivery and received prophylactic uterotonic agents. Treatment with tranexamic acid resulted in a significantly lower incidence of estimated blood loss >1000 mL or red cell transfusion by day 2 than placebo (P=0.003), but a lower incidence of haemorrhage related secondary clinical outcomes was not seen. The results suggest little fibrinolytic activation during caesarean section.WOMAN-2 (World Maternal Antifibrinolytic Trial 2), a randomised, double blind, placebo controlled trial, enrolled >16 000 participants with moderate and severe anaemia, who were given tranexamic acid within 15 min of the umbilical cord being clamped. Tranexamic acid did not reduce the risk of clinically diagnosed postpartum haemorrhage, showing that the use of tranexamic acid could not compensate for anaemia.41 Tranexamic acid is a cost effective strategy and has been incorporated into all national and international guidelines for the management of established postpartum haemorrhage (table 5). Currently, however, many low income countries do not have access to tranexamic acid, and multiple agencies are working together to improve the manufacturing and supply of the drug in sub-Saharan Africa where high mortality from postpartum haemorrhage exists.42 Table 5Haematological management of postpartum haemorrhage according to international guidelinesInstitutionYearTestingManagementLaboratoryPoint of careTranexamic acidRed blood cells, fresh frozen plasma, or plateletsFibrinogenInternational Federation of Gynaecology and Obstetrics (FIGO)49 2022No specific recommendationsNo specific recommendationsTranexamic acid 1 g intravenously after delivery no more than 3 hours after onset of postpartum haemorrhage. Second dose if bleeding continues after 30 minRed blood cells, fresh frozen plasma, and platelets are applied in a 1:1:1 ratioAim for target fibrinogen levels of ≥150-200 mg/dLBritish Society for Haematology45 2022Serial monitoring of coagulation tests is recommendedNo specific recommendationsTranexamic acid 1 g intravenously within 3 hours of postpartum haemorrhage onset. Second dose if bleeding continues after 30 minOptimum red blood cells to fresh frozen plasma ratio unknown.If major ongoing bleeding and no coagulation results, use 1:2 fresh frozen plasma to red blood cells. Use fresh frozen plasma to maintain prothrombin time ratio <1.5×mean normal. Keep platelets >50×109/LReplace fibrinogen if levels are <2.0 g/L in the context of ongoing bleedingRoyal Australian and New Zealand College of Obstetrics and Gynaecologists48 2021Monitor full blood count, coagulation screen, calcium, and arterial blood gas every 30-60 minNo specific recommendations although noted that thromboelastography has an increasing role in the assessment of postpartum haemorrhageTranexamic acid 1 g intravenously followed by an infusion of 1 g over 8 hoursFresh frozen plasma transfusion at 15 mL/kg if international normalised ratio is >1.5Replacement with cryoprecipitate or fibrinogen concentrate is recommended where fibrinogen is <1.5-2g/LNetwork for the Advancement of Patient Blood Management, Haemostasis, and Thrombosis46 2019Full blood count and coagulation screen should be done in severe postpartum haemorrhage (>1000 mL blood loss)Viscoelastic testing can be used to guide product replacement. No algorithm recommended for testingTranexamic acid 1 g intravenously as soon as possible and within 3 hours of postpartum haemorrhage onsetInitial 4 units of red blood cells, followed by red blood cells to fresh frozen plasma in 2:1 ratio. Platelet transfusion if <75×109/L or as guided by point-of-care testingMaintain fibrinogen level >2 g/L or FIBTEM A5 >12 mmAmerican College of Obstetrics and Gynaecology55 2017No recommendationsNo recommendationsFor consideration in management of postpartum haemorrhage when initial medical treatment failsWhen massive transfusion is required (>10 units of red blood cells in 24 hours or 4 units of red blood cells in 1 hour) recommend a ratio of 1:1:1 red blood cells to fresh frozen plasma to plateletsEarly use of cryoprecipitate is recommended in disseminated intravascular coagulopathy, placental abruption, or amniotic fluid embolismRoyal College of Obstetrics and Gynaecology44 2016Full blood count and coagulation screen should be done at onset of postpartum haemorrhage and repeated at >1000 mL or if shock presentPoint of care testing can be used according to locally agreed algorithm as long as quality control checks in placeFor consideration in management of postpartum haemorrhageAfter 4 units of red blood cells transfused then fresh frozen plasma should be infused at 12–15 mL/kg. Platelets should be transfused when <75×109 /L based on laboratory monitoringPlasma fibrinogen level >2 g/L should be maintained during ongoing postpartum haemorrhage and cryoprecipitate should be used for fibrinogen replacementInternational Society of Thrombosis and Haemostasis consensus47 2016Serial laboratory testing is useful and should always include Clauss fibrinogenPoint of care testing can be used according to locally agreed algorithm as long as quality control checks are in placeTranexamic acid 1 g intravenously should be considered in ongoing postpartum haemorrhageAfter 4 units of red blood cells transfused, transfusion should be red blood cells to fresh frozen plasma in 1:1 ratio until haemostatic test results available. Transfuse fresh frozen plasma if prothrombin time or activated partial thromboplastin time is prolonged to limit progression to 1.5×normal.Platelets should be transfused at 75×109/ L to maintain a level >50 × 109 /L during ongoing postpartum haemorrhageSuggest that fibrinogen level ≥2 g/L should be maintained during ongoing obstetric bleeding. Either cryoprecipitate or fibrinogen concentrate may be usedTransfusion strategies The longstanding lack of randomised clinical trials of the use of blood product in postpartum haemorrhage and lack of evidence of the pathophysiology of postpartum haemorrhage resulted in extrapolation of management approaches from traumatic haemorrhage. Initial use of fresh frozen plasma to avoid the development of coagulopathy, or to treat it early, compared with waiting for formal laboratory tests to be available, was first found to be useful in trauma 3 and was then applied to guidelines on postpartum haemorrhage. The large dataset from Collins et al,25 however, showed that in most patients with postpartum haemorrhage, little derangement of coagulopathy exists, unless blood loss is >3 L, and that acute obstetric coagulopathy is rare, in contrast with traumatic bleeding where haemostatic changes are common early on.29 These findings strongly suggest that early empiric plasma to increase levels of coagulation factors and platelet transfusion is not necessary, leading to unwarranted use of blood products, with an associated risk of adverse events such as transfusion associated circulatory overload and transfusion associated acute lung injury. Also, for the subset of patients with postpartum haemorrhage and hypofibrinogenaemia, if fresh frozen plasma is given, the low fibrinogen concentration of fresh frozen plasma may worsen hypofibrinogenaemia because of the dilutional effect.43 Although the approach to haemostatic management in postpartum haemorrhage is being re-evaluated, some international guidelines still advise the use of early fresh frozen plasma. Other guidelines advise against the use of fresh frozen plasma, cryoprecipitate, or platelets unless coagulopathy or thrombocytopenia is present, or in the event of massive postpartum haemorrhage with ongoing blood loss where four or more units of red blood cells have already been transfused (table 5).Recommended triggers for requesting laboratory coagulation screens or full blood counts vary, as well as the frequency of repeat screens. Some guidelines acknowledge a role for viscoelastic haemostatic assays as point-of-care tests to guide replacement of blood products. All guidelines emphasise the need for specific fibrinogen monitoring, however, ideally Clauss fibrinogen, reflecting the substantial evidence that low fibrinogen levels is a predictor of severe postpartum haemorrhage; fibrinogen levels may be low despite normal prothrombin time and activated partial thromboplastin time in postpartum haemorrhage.17 18 Transfusion targets Evidence to guide transfusion thresholds or targets in postpartum haemorrhage is scarce, with variation between international guidelines in defined transfusion triggers. Broad consensus states that platelets should be maintained >50-75×10 9/L,44–47 and prothrombin time or activated partial thromboplastin time, or both, <1.5 times the upper limit of non-pregnant individuals.45 47 48 Variability exists in terms of target fibrinogen levels (>1.5-2 g/l,48 49 or >2 g/L44–47).Cryoprecipitate versus fibrinogen concentrate When fibrinogen levels are low with ongoing bleeding, correcting fibrinogen deficiency is essential to control bleeding. Both cryoprecipitate and fibrinogen concentrate can be used. 46 Not all fibrinogen concentrates are licenced for treatment of acquired fibrinogen deficiency (eg, in the UK), and cryoprecipitate is currently standard of care in the UK.44 Cryoprecipitate is not universally available worldwide. The volume and content of cryoprecipitate varies internationally but a commonly used adult dose is two pools (10 single units) of cryoprecipitate, which should increase plasma fibrinogen levels by about 1 g/L, depending on the clinical setting. Increasing the level of fibrinogen by 1 g/L requires about 60 mg/kg of fibrinogen concentrate.50 Observational data indicate that fibrinogen concentrate is as efficacious as cryoprecipitate in postpartum haemorrhage, with a suggestion of reduced transfusion reactions with fibrinogen concentrate.51 52 Correction of fibrinogen deficiency can generally be achieved more quickly with fibrinogen concentrate than with cryoprecipitate because cryoprecipitate must be defrosted in blood banks and sent to the labour ward, whereas fibrinogen concentrate can be kept in the clinical area, and quickly reconstituted and administered. Factor XIII content in fibrinogen concentrate and cryoprecipitate is variable, with more von Willebrand factor, factor VIII, and fibronectin in cryoprecipitate than in fibrinogen concentrate.53 In vitro data suggest that the additional factors in cryoprecipitate may allow for stronger fibrin clot formation.54 Empiric fibrinogen replacement is recommended by both the International Society of Thrombosis and Haemostasis and the Royal College of Obstetrics and Gynaecology if bleeding is ongoing after eight units of red blood cells with fresh frozen plasma have been given, in the absence of formal Clauss fibrinogen levels or viscoelastic testing.44 47 Fresh frozen plasma For ongoing bleeding with prolonged prothrombin time or activated partial thromboplastin time, fresh frozen plasma is recommended at standard doses of 12-15 ml/kg 44 45 47 48 if prothrombin time and activated partial thromboplastin time are more than 1.5 times the upper limit of normal. Many guidelines also recommend empiric fresh frozen plasma for massive postpartum haemorrhage where four units of red blood cells have already been transfused44 46 47 55 because coagulopathy is more likely at this stage.No evidence exists for the use of prothrombin complex concentrate in postpartum haemorrhage. Prothrombin complex concentrate is not currently recommended in the management of postpartum haemorrhage outside of clinical trials because deficiencies in factors II, VII, IX, and X are rare in postpartum haemorrhage25 and because of the theoretical associated thrombotic risk.Alternative approaches to standard laboratory testing for coagulopathy in postpartum haemorrhage The turnaround time for urgent formal laboratory coagulation screens, Clauss fibrinogen, and full blood counts is a minimum of 30-40 min, which limits their clinical use in the management of bleeding. In major ongoing bleeds, the results will inevitably reflect past rather than current haemostatic status. Relying on formal laboratory testing will not allow early detection and correction of coagulopathy, and rapid control of bleeding will be less likely. Alternative approaches have therefore been investigated, such as empiric initial administration of fibrinogen or fresh frozen plasma to prevent coagulopathy, or using viscoelastic haemostatic assays to detect coagulopathy.Empirical early fibrinogen replacement or initial fresh frozen plasma to prevent coagulopathy A role for empiric early fibrinogen replacement has been investigated. Two randomised controlled trials (n=249, n=437, respectively) of initial fibrinogen concentrate (at doses of 2 g 56 or 3 g57) versus placebo at diagnosis of postpartum haemorrhage showed no benefit in terms of transfusion requirements or blood loss. The mean fibrinogen level at the time of randomisation, however, was about 4 g/L in both studies. No thromboembolic events were seen. A pilot cluster randomised controlled trial (n=180) of early cryoprecipitate (two pools) versus placebo in postpartum haemorrhage requiring at least one unit of red blood cells showed a trend to lower requirement for red blood cells, surgical intervention, and admission to the intensive care unit in the cryoprecipitate group.58 Only 32% of women in the intervention group received the planned cryoprecipitate within 90 min in accordance with the protocol, however, highlighting the fact that giving cryoprecipitate in a timely manner is challenging. In summary, initial empiric fibrinogen replacement has not been shown to be of clinical benefit and is not recommended.Transfusion of red cells and fresh frozen plasma in a fixed ratio of 3:2 for blood loss >1500 mL as part of a structured comprehensive postpartum haemorrhage protocol was introduced in a US centre.59 Comparison of outcomes in a cohort study before and after implementation for >5800 deliveries showed increased use of fresh frozen plasma from before to after implementation, as well as a shift to resolution of bleeding at earlier stages, reduced use of blood products overall, and a reduction in disseminated intravascular coagulopathy rates. The extent to which the empirical plasma transfusion alone versus the other components of the protocol contributed to these outcomes, however, is unknown, whereas rates of transfusion reactions, such as transfusion associated circulatory overload, were not reported. In a single centre retrospective study in France, 142 patients with severe postpartum haemorrhage over a four year period were analysed.60 The fresh frozen plasma to red blood cells ratio increased over the time period from 1:1.8 to 1:1. Propensity scoring showed that a high fresh frozen plasma to red blood cells ratio was associated with a lower odds of arterial embolisation or surgical procedures, perhaps suggesting earlier control of bleeding, although the authors concluded that randomised controlled trials are needed.Role of viscoelastic haemostatic assays in early identification of coagulopathy Evidence is growing for the use of viscoelastic haemostatic assays as point-of-care tests in postpartum haemorrhage. Viscoelastic haemostatic assays allow early identification and correction of hypofibrinogenaemia and other coagulopathy, reduce the use of blood products, and improve outcomes ( table 6).Table 6Evidence for use of viscoelastic haemostatic assays in postpartum haemorrhageStudy designCountryYearPatientsTest usedOutcomeReferenceSingle centre prospective studyUK2011-16893 patients, postpartum haemorrhage >1500 mLROTEMSignificant reduction in use of blood products after implementation of ROTEM guided algorithm, including a decrease in median use of fresh frozen plasma from 4 units to 0 units, with an associated reduction in transfusion associated circulatory overloadMcNamara,73 Mallaiah65 Single centre retrospective cohort studyUSA2011-1586 patients with severe postpartum haemorrhageROTEMComparison of 28 patients managed with ROTEM v 58 patients managed with the empirical transfusion protocol showed reduced transfusions, lower postpartum haemorrhage volumes, and reduced admissions to intensive care unit and emergency caesarean section rates with the ROTEM guided protocolSnegovskikh74 Single centre retrospective studyAustralia2016900 patients with postpartum haemorrhageROTEMImplementation of ROTEM guided transfusion protocol in 2016 did not show a significant change in red cell transfusions, although ROTEM was only used in 32% of eligible patients. In more severe cases of postpartum haemorrhage (>1500 mL), ROTEM was used more often with reported reduction in use of fresh frozen plasma and plateletsTsang75 Multicentre randomised controlled trial(OBS2)*UK2013-15663 patients with postpartum haemorrhage enrolled; 57 randomisedROTEMROTEM guided infusion of fibrinogen did not improve outcomes in postpartum haemorrhage. Fresh frozen plasma can be safely excluded in postpartum haemorrhage based on ROTEM parameters (see text)Collins61 62 Multicentre prospective study(OBS Cymru)†UK2017-186024 patients with postpartum haemorrhageROTEMDemonstrated feasibility of using ROTEM to rapidly identify the minority of patients with coagulopathy to guide targeted treatment (see text) Bell63 *Obstetrics Bleeding Study 2.†Obstetric Bleeding Strategy for Wales.ROTEM, rotational thromboelastometry.In 2017, Collins and colleagues published the results of a large scale, multicentre prospective study of women with postpartum haemorrhage of 1000-1500 mL in the UK (Obstetrics Bleeding Study 2, OBS2).61 62 Those with FIBTEM A5 ≤15 mm (corresponding to a fibrinogen level of about <3 g/L) were randomly assigned to receive either fibrinogen concentrate or placebo based on double blind methodology. Only 55 of 663 enrolled women met this criterion. Early infusion of fibrinogen concentrate in these patients did not significantly reduce transfusion requirements or blood loss. Prespecified subgroup analyses suggested that a FIBTEM A5 >12 mm or fibrinogen >2 g/L were adequate for haemostasis and therefore fibrinogen replacement was not necessary in these patients. An effect of fibrinogen infusion on the use of blood products and bleed volume at a FIBTEM A5 ≤12 mm or Clauss fibrinogen <2.5 g/L could not be excluded and was thought to warrant further investigation.62 In the same cohort, fresh frozen plasma was withheld in those with FIBTEM >15 mm or in whom bleeding stopped, and none of these women developed haemostatic impairment. The authors concluded that restrictive use of fresh frozen plasma guided by clinical assessment and ROTEM is feasible.61 The same group developed the Obstetric Bleeding Strategy for Wales (OBS Cymru) care bundle for the management of women with postpartum haemorrhage in Wales. One element was ROTEM guided early fibrinogen replacement (when FIBTEM A5 <12 mm) and withholding of fresh frozen plasma unless evidence of coagulopathy was found (from EXTEM clotting time, or prothrombin time or activated partial thromboplastin time). Other elements included universal risk assessment of postpartum haemorrhage based on measured rather than estimated blood loss, and a structured escalation strategy at defined volumes of blood loss. Implementation of the bundle in all Welsh maternity units (>30 000 maternities/year) as a quality improvement project reduced (by 29%) the proportion of moderate postpartum haemorrhage progressing to severe over a two year period. A reduction in red cell transfusion of >20% was found, with a 42% fall in fresh frozen plasma transfusion, without an increase in women with ongoing haemostatic impairment.63 Although the exact contribution of the viscoelastic testing component to the overall outcomes is unknown, the project illustrated that using ROTEM to rapidly identify the minority of patients with coagulopathy to guide targeted treatment is feasible, thus avoiding unnecessary fresh frozen plasma in the remaining patients.ROTEM was also incorporated into a structured multicomponent algorithm (the D-A-CH algorithm) for managing postpartum haemorrhage >1000 mL, developed in German speaking nations. A retrospective review of patients managed before (n=141) and after (n=176) implementation of the algorithm in a Swiss centre did not show a significant reduction in blood loss, but medical interventions occurred in a more timely manner without increased associated costs.64 Two single centre prospective randomised controlled trials investigated the role of ROTEM in guiding transfusion management of postpartum haemorrhage. In one US centre, patients with postpartum haemorrhage after vaginal or caesarean delivery were randomised to receive standard care or ROTEM guided transfusion according to the algorithm previously validated by Mallaiah and colleagues.65 No difference was seen in the primary endpoint of blood components transfused in the first 48 hours, although recruitment was slow and the study ended early after recruitment of 49 patients (the target was 100).66 A similar Finnish randomised controlled trial of ROTEM guided transfusion management versus standard care in postpartum haemorrhage >1500 mL (with 54 patients in the final analysis) did not show a difference in the use of red blood cells between the groups but the ROTEM group showed reduced use of plasma.67 In summary, the evidence is growing for the use of viscoelastic haemostatic assays to detect coagulopathy in women with postpartum haemorrhage, and to guide blood components, reducing unnecessary administration of blood components. Further trials are needed, such as the Obstetric Bleeding Study UK (OBS UK, www.obsuk.org), a stepped wedge cluster randomised controlled trial currently open in 36 centres in the UK. This study aims to test the impact of the OBS Cymru postpartum haemorrhage care bundle and will provide evidence about the effect of implementing a postpartum haemorrhage care bundle that incorporates point-of-care coagulation testing (with ROTEM or TEG) and targeted replacement of blood products on maternal outcomes.Conclusions The challenge of improving care of severe postpartum haemorrhage is being dealt with, and we now have a greater understanding of what happens to haemostasis in postpartum haemorrhage. The use of viscoelastic haemostatic assays, such as ROTEM and TEG, to identify those women with major haemostatic changes may be revolutionary, allowing blood transfusion to be used in those who need it while preventing unnecessary use of blood products and side effects, such as transfusion associated circulatory overload, in those who do not. Further trials are needed to determine best transfusion practice in postpartum haemorrhage. Continuing international collaboration is also needed to prevent the ongoing high mortality caused by postpartum haemorrhage in low and middle income countries.Questions for future research What are the underlying precipitants or mechanisms of acute obstetric coagulopathy?What is the relation of acute obstetric coagulopathy to obstetric disseminated intravascular coagulopathy?How should acute obstetric coagulopathy be managed?What is the impact of low factor XIII during postpartum haemorrhage?What is the role of viscoelastic haemostatic assays as point-of-care tests in the management of postpartum haemorrhage?What is the optimal ratio of packed red cells to fresh frozen plasma in the management of obstetric haemorrhage?Compared with cryoprecipitate, what is the role of fibrinogen concentrate in the management of postpartum haemorrhage?Compared with fresh frozen plasma, what is the role of prothrombin complex concentrate in the management of postpartum haemorrhage?Should the role of factor XIII replacement, with cryoprecipitate or factor XIII concentrate, during severe postpartum haemorrhage be investigated?Patient involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.