Prophylactic plasma transfusion for patients without inherited bleeding disorders or anticoagulant use undergoing non‐cardiac surgery or invasive procedures

Summary of findings for the main comparison. Prophylactic plasma transfusion before surgery/invasive procedures compared to no prophylactic plasma transfusion before surgery/invasive procedures (colloid, crystalloid, placebo, or no treatment) for patients undergoing non‐cardiac surgery or invasive procedures

Prophylactic plasma transfusion before surgery/invasive procedures compared to no prophylactic plasma transfusion before surgery/invasive procedures (colloid, crystalloid, placebo, or no treatment) for patients undergoing non‐cardiac surgery or invasive procedures
Patient or population: patients undergoing non‐cardiac surgery or invasive procedures Setting: individuals in ICU undergoing invasive procedures. Studies conducted in The Netherlands Intervention: prophylactic plasma transfusion before surgery/invasive procedures Comparison: no prophylactic plasma transfusion before surgery/invasive procedures (colloid, crystalloid, placebo, or no treatment)
Outcomes	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**			Certainty of the evidence (GRADE)
		Without prophylactic plasma transfusion before surgery/invasive procedures	With prophylactic plasma transfusion before surgery/invasive procedures	Difference
All‐cause mortality up to 30 days № of participants: 72 (1 RCT)	RR 0.38 (0.13 to 1.10)	Study population			⊕⊝⊝⊝ Very low^a,b
		297 per 1000	113 per 1000 (39 to 327)	184 fewer per 1000 (259 fewer to 30 more)
Major bleeding within 24 hours № of participants: 148 (2 RCTs)	Two studies; reported separately^c FFP group events 0/38; no FFP group events 1/38; RR 0.33, 95% CI 0.01 to 7.93 (Müller 2015) FFP or platelets or both group events 3/35; no transfusion group events 2/37; RR 1.59, 95% CI 0.28 to 8.93 (Veelo 2012)				⊕⊝⊝⊝ Very low^b,d,e
Number of transfusions per participant within 7 days № of participants: 76 (1 RCT)	Study authors reported no difference ("P = 0.91 (RBC), P = 0.06 (FFP), P = 0.43 (PLT)") between the 2 groups (76 participants; Müller 2015). See Table 1				⊕⊝⊝⊝ Very low^d,f,g
Number of individuals requiring a transfusion within 7 days № of participants: 72 (1 RCT)	Study authors reported "no... transfusion of packed red cells for blood loss during or after [the procedure]" (72 participants; Veelo 2012)				⊕⊝⊝⊝ Very low^d,g,h
Serious adverse events measured by plasma transfusion‐related complications within 24 hours № of participants: 76 (1 RCT)	Study authors reported that a difference between the 2 groups "did not reach statistical significance" (Müller 2015). See Table 1				⊕⊝⊝⊝ Very low^d,f,i
Serious adverse events measured by surgery or procedure‐related complications within 30 days ‐ not reported	‐	‐	‐		‐
Quality of life ‐ not reported	‐	‐	‐		‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; FFP: fresh frozen plasma; ICU: intensive care unit; PLT: platelet; RBC: red blood cell; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded two points due to indirectness (single trial, ICU setting, single bedside procedure; Veelo 2012). ^bDowngraded one point due to imprecision (wide confidence intervals, which cross the line of no difference). ^cGiven only two studies, small numbers of participants, low event count, and very poor‐quality evidence by GRADE assessment with heterogenous groups and wide confidence intervals that cross the line of no difference, we have not pooled the results, as doing so would be neither statistically nor clinically meaningful. ^dDowngraded one point due to high risk of bias over multiple domains. ^eDowngraded one point due to indirectness (2 trials, ICU setting, participants undergoing bedside procedures). ^fDowngraded one point due to indirectness (ICU setting, participants undergoing bedside procedures; Müller 2015). ^gDowngraded one point due to imprecision (low or absent event incidence). ^hWould be downgraded two points due to very serious indirectness. However downgraded only one point as already downgraded two points in other domains (maximum 3 downgrade points allowed by GRADE method). ⁱDowngraded one point for imprecision (single study, 76 participants, terminated early, likely under‐powered to demonstrate an effect; Müller 2015).

1. Additional data: outcomes reported for FFP versus no FFP with median and IQR

Outcome	Prophylactic transfusion			No prophylactic transfusion			P value
Outcome	Median	IQR	Total number of participants in group	Median	IQR	Total number of participants in group	P value
Number of transfusions after intervention, per participant (within 24 hours)
Müller 2015
RBC	1	0 to 2	38	1	0 to 3	38	0.91
FFP	0	0 to 1	38	2	0 to 2	38	0.06
PLT	1	0 to 2	38	0	0 to 1	38	0.43

Volume of blood loss per participant (within 7 days)
Veelo 2012	3.0	1.0 to 6.0	35	3.0	2.0 to 6.0	37	0.96

Lung injury score (within 24 hours after randomisation)
Müller 2015	2	0.8 to 2.5	38	1.25	0.4 to 2.4	38	0.28

ICU length of stay (within 30 days)
Müller 2015	12	6 to 19	38	7	3 to 17	38	0.13
Veelo 2012	15	8 to 29	35	21	14 to 26	37	0.21

Coagulation test abnormalities (within 24 hours)
Müller 2015
INR before FFP intervention	1.8	1.5 to 2.5	38	‐	‐	‐
INR after FFP intervention	1.4	1.3 to 1.63	38	‐	‐	‐	< 0.001
FFP: fresh frozen plasma; ICU: intensive care unit; INR: international normalised ratio; IQR: interquartile range; PLT: platelets; RBC: red blood cells.

Summary of findings 2. Prophylactic plasma transfusion before surgery/invasive procedures compared to alternative haemostatic agents for patients undergoing non‐cardiac surgery/invasive procedures

Prophylactic plasma transfusion before surgery compared to alternative haemostatic agents for patients undergoing non‐cardiac surgery or invasive procedures
Patient or population: patients undergoing non‐cardiac surgery/invasive procedures Setting: individuals with chronic liver disease undergoing liver needle biopsy. Study conducted in Italy in 1976 Intervention: prophylactic plasma transfusion before surgery/invasive procedures Comparison: alternative haemostatic agents
Outcomes	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**			Certainty of the evidence (GRADE)
Outcomes	Relative effect (95% CI)	Without prophylactic plasma transfusion before surgery/invasive procedures	With prophylactic plasma transfusion before surgery/invasive procedures	Difference	Certainty of the evidence (GRADE)
All‐cause mortality up to 30 days ‐ not reported	‐	‐	‐		‐
Major bleeding within 24 hours № of participants: 21 (1 RCT)	Study authors reported, "in none of the patients was there any clinical or laboratory evidence of bleeding" (21 participants; Mannucci 1976)				⊕⊝⊝⊝ Very low^a,b
Number of transfusions per participant within 7 days ‐ not reported	‐	‐	‐		‐
Number of individuals requiring a transfusion within 7 days ‐ not reported	‐	‐	‐		‐
Serious adverse events measured by plasma transfusion‐related complications within 24 hours № of participants: 21 (1 RCT)	RR 9.82 (0.59 to 162.24)	Study population			⊕⊝⊝⊝ Very low^d,e
		Low
		10 per 1000^c	98 per 1000 (6 to 1000)	88 more per 1000 (4 fewer to 1612 more)
		High
		100 per 1000^c	982 per 1000 (59 to 1000)	882 more per 1000 (41 fewer to 16124 more)
Serious adverse events measured by surgery or procedure‐related complications within 30 days ‐ not reported	‐	‐	‐		‐
Quality of life ‐ not reported	‐	‐	‐		‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded two points due to indirectness (single study, 21 participants, single bedside procedure; Mannucci 1976). ^bDowngraded one point due to imprecision (low or absent event incidence, small study). ^cAssumed risks taken from available drug product information by Baxter AG, Austria, for the use of Prothromplex TOTAL (HPRA 2018). Prothromplex TOTAL is a 4 factor concentrate (II, VII, IX, and X) with added protein C. Prothromplex, Immuno Vienna as used in Mannucci 1976 was a three‐factor concentrate (II, VII, IX) after which factor VII, Immuno, Vienna was administered. No published safety data found for Prothromplex, Immuno, Austria, as used in Mannucci 1976. ^dWould have been downgraded two points due to indirectness (differences in plasma preparation in modern practice limit applicability of results from a study conducted in 1976, and use of prophylactic steroids before plasma transfusion in the study is not recommended by modern guidelines; Mannucci 1976). However downgraded only one point as already downgraded two points in another domain (maximum three downgrade points allowed by GRADE method). ^eDowngraded two points due to imprecision (very wide confidence intervals, crossing the line of no difference, which could include both significant harm and benefit, and clinicians instituted a practice change in a single arm during the study, by giving participants prophylactic steroids before FFP transfusion, following four febrile or allergic reactions in the group).

See Characteristics of included studies, Characteristics of excluded studies,Characteristics of studies awaiting classification, and Characteristics of ongoing studies.

Summary of findings 3. Thromboelastography threshold compared to standard of care (laboratory parameters) for patients undergoing non‐cardiac surgery or invasive procedures

Thromboelastography threshold compared to standard of care (laboratory parameters) for patients undergoing non‐cardiac surgery or invasive procedures
Patient or population: patients undergoing non‐cardiac surgery/invasive procedures Setting: individuals with cirrhosis undergoing invasive procedures. Study conducted in Italy Intervention: thromboelastography threshold Comparison: standard of care (laboratory parameters)
Outcomes	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**			Certainty of the evidence (GRADE)
		Without thromboelastography threshold	With thromboelastography threshold	Difference
All‐cause mortality up to 30 days ‐ not reported	‐	‐	‐		‐
Major bleeding within 24 hours № of participants: 60 (1 RCT)	RR 0.33 (0.01 to 7.87)	Study population			⊕⊝⊝⊝ Very low^a,b,c
		33 per 1000	11 per 1000 (0 to 262)	22 fewer per 1000 (33 fewer to 229 more)
Number of transfusions per participant within 7 days ‐ not reported	‐	‐	‐		‐
Number of individuals requiring a transfusion within 7 days № of participants: 60 (1 RCT)	RR 0.18 (0.08 to 0.39)	Study population			⊕⊕⊝⊝ Low^a,b
		1000 per 1000	180 per 1000 (80 to 390)	820 fewer per 1000 (920 fewer to 610 fewer)
Serious adverse events measured by plasma transfusion‐related complications within 24 hours № of participants: 60 (1 RCT)	RR 0.33 (0.01 to 7.87)	Study population			⊕⊝⊝⊝ Very low^a,b,c
		33 per 1000	11 per 1000 (0 to 262)	22 fewer per 1000 (33 fewer to 229 more)
Serious adverse events measured by surgery or procedure‐related complications within 30 days ‐ not reported	‐	‐	‐		‐
Quality of life ‐ not reported	‐	‐	‐		‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one point due to high risk of bias across multiple domains (De Pietri 2016). ^bDowngraded one point due to indirectness (single trial, 60 participants with cirrhosis; De Pietri 2016). ^cWould be downgraded two points due to very serious imprecision (low event incidence, very wide confidence intervals including both serious harm and benefit). However downgraded only one point as already downgraded two points in other domains (maximum three downgrade points allowed by GRADE method).

Background

Description of the condition

A coagulopathy has been defined as a condition leading to impairment of the blood's clotting ability (Hunt 2014). People undergoing surgical procedures may have a coagulopathy for a myriad of reasons including co‐existing medical conditions; nutritional or absorptive abnormalities leading to vitamin K deficiency (which results in a reduction in vitamin K‐dependent clotting factors); abnormal physiological states such as hypothermia or acidosis; coagulant factor dilution due to intravenous fluids or red cell transfusion; use of antiplatelet or anticoagulant medication; or clotting factor consumption due to bleeding (McGilvray 2001).

There are also people requiring surgery in whom both prothrombotic (procoagulant or hyper‐coagulant) and coagulopathic (anticoagulant) states may coexist, such as those with liver disease, disseminated intravascular coagulation, renal failure, and systemic inflammatory response syndrome or sepsis (Martlew 2000). People with perioperative critical illness and sepsis are associated with a net procoagulant state, despite laboratory measurements of coagulopathy (McGilvray 2001).

Preoperative screening for people with coagulopathy historically involved measurements of activated partial thromboplastin time (aPTT) and prothrombin time (PT) (van Veen 2011), both of which measure the time for blood to clot, and are affected by the function of different clotting factors. The PT is often presented as the international normalised ratio (INR), which controls for variations in PT measurements due to sensitivity differences among the commercial reagents (Rand 2005).

Clinicians have used abnormal laboratory coagulation results as a marker of coagulopathy, and these abnormalities have formed the rationale for replacing coagulation factors through transfusion of human plasma prophylactically (in the absence of bleeding) before invasive procedures or surgery (Stanworth 2007).

Description of the intervention

Human plasma is the non‐cellular component of blood, containing proteins that help the blood to clot (procoagulants) such as fibrinogen and factors II, V, VII, VIII, IX, X, and XI; anticoagulant proteins C and S and antithrombin; and immunoglobulins, water, albumin, and acute phase proteins (Desborough 2015). Plasma is collected either from a single whole blood donation following separation from red cell and platelet components, or from plasmapheresis.

One unit contains a variable volume of plasma (Desborough 2012), typically 200 mL to 300 mL (Benjamin 2012), and different preparations are available. Fresh frozen plasma (FFP) is frozen within eight hours of collection and contains greater concentrations of temperature‐labile factors V and VIII than frozen plasma (FP), which is frozen within 24 hours (Benjamin 2012). FFP is stored typically at ‐30°C for up to 36 months (Norfolk 2013; Stanworth 2007). Once thawed to 1°C to 6°C, FFP retains overall coagulation factor content for up to five days, although factors V and VIII undergo the greatest degradation during this time (Stanworth 2007).

There is also variability of factor concentrations in pathogen‐inactivated preparations, such as solvent‐detergent treated FFP, which contains reduced fibrinogen, factor VIII, and protein S (Norfolk 2013), or methylene blue‐treated FFP, which contains reduced fibrinogen and factor VIII (Pamphilon 2000).

Given the variability of clotting factor levels in healthy donors, together with processing, storage, and preparation differences (Stanworth 2007), the potency of coagulation factors in plasma can vary between pooled units from 50% to 150% of pooled standardised controls (Benjamin 2012), and even more between units of single donors. Indeed, mean factor VIII concentration has been the only quality‐controlled measure for the specification of plasma in the European Union (Stanworth 2007).

Risks associated with the intervention

Plasma transfusion has the potential to cause life‐threatening complications and carries higher risks compared with transfusion of other blood components (Khan 2007; MacLennan 2006). These include transfusion‐related acute lung injury (TRALI) (Eder 2007; Holness 2004), transfusion‐associated circulatory overload (TACO) (Narick 2012), anaphylaxis or acute allergic reactions ‐ common in 1% to 3% of transfusions (Desborough 2015) ‐ ABO incompatibility‐induced haemolysis (Norfolk 2013), multi‐organ failure (Watson 2009), and transfusion‐transmitted infection. Plasma transfusion is also independently associated with nosocomial infection and sepsis (Karam 2013; Sarani 2008).

Globally, a significant difference exists in the risk of transfusion‐transmitted infections between high‐income and low‐income countries (Dhingra 2013). World Health Organization (WHO) data from 2013 show the incomplete ability of 13 of 173 reporting countries to screen all collected blood for one or more of four transmissible infections ‐ HIV, hepatitis B, hepatitis C, or syphilis ‐ with limited access to test kits representing one such barrier to screening. Furthermore only 66% of donations in low‐income countries were tested following basic quality‐assured procedures (WHO 2017). Blood shortages and an unreliable donor base have historically encouraged the use of paid donors or transfusion without prior testing (WHO 2008). Evidence for increased prevalence and transmission of infections such as HIV in the commercial plasma‐donor population has been demonstrated (Volkow 2005; Wu 2001), and this has remained a concern (Abolghasemi 2010). Although an increase of 10.7 million blood donations from voluntary non‐renumerated donors between 2008 and 2013 was reported from 159 reporting countries, 71 of 178 countries in 2013 remained dependent on family/replacement and paid donations for more than 50% of their blood supplies (WHO 2017).

How the intervention might work

Plasma is a source of procoagulant factors, and a current practice exists to transfuse plasma prophylactically (in the absence of bleeding), based on the rationale that replacing clotting factors through plasma transfusion will correct a coagulopathy and reduce perioperative bleeding risk (Desborough 2012; Rutherford 2008; Stanworth 2007). Reducing this risk has clinical importance given that perioperative bleeding and blood transfusion have been associated with increased morbidity and mortality (Glance 2011; Shander 2007).

Although further research into exact mechanisms is needed, more recent research has demonstrated that plasma transfusion is associated with protective and restorative effects on the integrity of the lining of blood vessel walls (vascular endothelial glycocalyx layer) (Kozar 2011; Peng 2013; Potter 2015; Rahbar 2015). The lining of blood vessels plays a fundamental role in the initiation and regulation of coagulation, and it is easily damaged by haemorrhagic shock, hypovolaemia, or trauma (Schott 2016).

Prophylactic administration of plasma is often based on mildly deranged laboratory tests (Luk 2002; Palo 2006; Stanworth 2011a; Triulzi 2015), despite evidence that coagulation factors at an INR less than two remain at concentrations adequate to support haemostasis (Deitcher 2002). Furthermore, the degree to which plasma transfusion corrects mildly abnormal coagulation tests is poor (Abdel‐Wahab 2006; Holland 2006a; Stanworth 2011a; Williamson 1999).

Second, the underlying premise that abnormal coagulation tests are associated with an increased bleeding risk should be treated with caution (Desborough 2012). Studies suggest no difference in bleeding risk between people with normal or abnormal PT or aPTT undergoing a range of interventions, including spinal surgery (Schramm 2001), angiography (Darcy 1996), liver biopsy (McGill 1990; McVay 1990), thoracocentesis (Puchalski 2013), and abdominocentesis (McVay 1991). These laboratory tests may be prolonged for a variety of reasons (Stanworth 2007), and they are not validated in non‐bleeding individuals (Dzik 2004). Their poor predictive value for bleeding risk ‐ as reported in Chee 2008 and Segal 2005 ‐ and their role as a poor marker for haemostasis ‐ as discussed in Desborough 2012 and Stanworth 2007 ‐ are not surprising in light of the complexity of haemostatic mechanisms in vivo involving interplay of the endothelium, inflammatory mediators, procoagulant and anticoagulant factors, platelets, and fibrinolysis.

An alternative approach to transfusing plasma based on an INR or PT threshold (which detects only low coagulation factor levels) involves using a test such as rotational thromboelastometry (ROTEM) or thromboelastography (TEG) that assesses how well a blood clot forms in whole blood (haemostasis) (Kinard 2013). ROTEM and TEG not only assess coagulation factor function but also platelet function, strength of the clot, and whether the clot is rapidly broken down (Whiting 2014). TEG may reduce prophylactic transfusions without increasing bleeding complications in people with liver disease undergoing invasive procedures (De Pietri 2016).

These issues may place people with presumed or confirmed coagulopathy undergoing prophylactic FFP transfusion before surgery at risk of potentially life‐threatening transfusion‐related complications (Khan 2007; MacLennan 2006), without clear evidence that the intervention has benefit.

Furthermore, plasma transfusion is expensive and, together with other blood product use, incurs costs of collecting the product, along with substantial administration costs pertaining to laboratory and staff utilisation, product wastage, and management of transfusion‐related complications. These processes generate significant expense for the healthcare provider (Shander 2016; Stokes 2018).

Alternatives to prophylactic plasma transfusion include the following.

No treatment or placebo.
Intravenous fluids, including:
- crystalloids, such as saline, dextrose, or balanced electrolyte solutions; or
- colloids, which contain a suspension of macromolecules such as starches, gelatins, or dextrans (Lira 2014).
Other prohaemostatic agents such as:
- prothrombin complex concentrate, which is produced from plasma and contains a rich source of the vitamin K‐dependent factors II, VII, IX, and X in a more concentrated volume compared with plasma;
- cryoprecipitate, which is produced from plasma and is a rich source of fibrinogen, factor VIII, and von Willebrand factor in a concentrated volume and can increase fibrinogen levels with lower transfusion volumes compared with plasma (Norfolk 2013);
- cryosupernatant, which is cryoprecipitate‐depleted plasma that is used as an alternative to plasma for individuals with thrombotic thrombocytopenia purpura (O'Shaughnessy 2004);
- fibrinogen concentrate, which contains the substrate converted to fibrin during the final step in the coagulation cascade and formation of a fibrin clot, and may reduce surgical bleeding when administered preoperatively;
- antifibrinolytics, which increase clot strength by inhibiting the body's mechanism for lysis of formed clots; or
- recombinant factor VIIa (rFVIIa), which is licensed for congenital factor VII deficiency, haemophilia, and inhibitory allo‐antibodies but is also used off‐licence in the setting of uncontrolled haemorrhage refractory to other treatments (Desborough 2016a).

Why it is important to do this review

There is significant variation in plasma transfusion and prescribing practice (Toumi 2015; Whitaker 2016), together with evidence of inappropriate use (Ejaz 2015; Luk 2002; Moylan 2008; Pahuja 2012; Palo 2006; Prathiba 2001; Stanworth 2011a; Stanworth 2011b; Tinmouth 2013; Triulzi 2015). Furthermore, evidence shows variation in dosing (Stanworth 2011a; Tinmouth 2013; Triulzi 2015), as well as coagulopathy thresholds for transfusion (Stanworth 2011b), with evidence of significant transfusions occurring for mild INR/PT derangement (Ejaz 2015; Triulzi 2015). These variations highlight inconsistencies in management strategies between clinicians.

Although numerous guidelines for plasma usage exist, there is significant variation in the guidance or quality of the evidence base. Some guidelines specifically highlight the absence of high‐quality evidence (O'Shaughnessy 2004; Roback 2010; Szczepiorkowski 2013; Yaddanapudi 2014). Another ‐ NICE 2015 ‐ recommends plasma transfusion for people undergoing surgery with coagulopathy and risk of significant bleeding based on very low‐quality randomised controlled trial (RCT) evidence and the opinion of the Guideline Development Group (Padhi 2015). One guideline recommends that prophylactic transfusion should be avoided (Liumbruno 2009). Another recommends prophylactic transfusion dependent on severity of the coagulation test derangement (Wong 2007), and another recommends seeking specialist advice for consideration of transfusion for people with coagulopathy undergoing intracranial, intraocular, or neuraxial procedures (National Blood Authority 2012). Other guidelines make no mention of prophylactic use in non‐cardiac surgery (American Society of Anesthesiologists 2006; American Society of Anesthesiologists 2015).

These variations in FFP usage, dosing, and thresholds, together with variations in guidance, highlight the need to review high‐quality RCT evidence for the role of plasma. A systematic review first published in 2004 and updated in 2012 demonstrated lack of consistent evidence for prophylactic use across a range of clinical settings (Stanworth 2004; Yang 2012). Although a Cochrane Review has been performed to examine the role of FFP in cardiovascular surgery (Desborough 2015), as yet no Cochrane Review has targeted non‐cardiovascular surgery. In the cardiovascular review, 14 trials compared prophylactic use versus no FFP associated with cardiac surgery. Overall, these trials were small and were not powered to determine changes in mortality as a primary outcome. Review authors recommended that large studies are required to assess the therapeutic effects of FFP on clinical outcomes following bleeding.

This Cochrane Review is needed to update previous reviews with recent RCT evidence (Stanworth 2004; Yang 2012), specifically targeting prophylactic use of plasma in non‐cardiac surgery or invasive procedures, given that its role is currently uncertain, whereas transfusions carry risk of harm. This review will examine the evidence for FFP compared with no plasma or alternative pro‐haemostatic agents. It will also examine coagulopathy thresholds for transfusion and will include studies that utilise classical laboratory measurements (PT, INR, aPTT).

Objectives

To determine the clinical effectiveness and safety of prophylactic plasma transfusion for people with coagulation test abnormalities (in the absence of inherited bleeding illnesses or use of anticoagulant medication) requiring non‐cardiac surgery or invasive procedures.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs), with no restriction on language or publication status.

Types of participants

We included people of all ages with laboratory confirmed or presumed abnormal coagulation (as defined by the study) undergoing non‐cardiac surgery or invasive procedures.

We excluded:

people with clinical evidence of bleeding before enrolment (as they would likely have received blood products); and
people with inherited bleeding disorders or using anticoagulants (e.g. warfarin, rivaroxaban, apixaban).

We used a broad definition of the term 'surgery' to capture as comprehensive an evidence base as possible. We used the definition created by the American College of Surgeons in 2007 (ACS 2007), given that it encompasses not only the definition of incision and destruction of tissues, but also diagnostic and therapeutic treatments using a variety of instruments including probes or needles (Appendix 1).

We also used the section "Classification of procedures" from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9‐CM 2011), which classifies operations for guidance. Examples such as oesophago‐gastro‐duodenoscopy and liver biopsy, or tracheostomy and chest drain insertion, are registered under the group of operative procedures of the digestive system or the respiratory system, respectively.

We excluded studies that assessed only central line insertion because these studies are already included in another Cochrane Review (Hall 2016).

Types of interventions

We included RCTs comparing three types of plasma transfusion regimens.

Prophylactic plasma transfusion before surgery/invasive procedures versus no prophylactic plasma transfusion before surgery/invasive procedures (colloid, crystalloid, placebo, or no treatment).
Plasma transfusion before surgery/invasive procedures compared to alternative pro‐haemostatic agents (prothrombin complex concentrate, cryosupernatant, fibrinogen concentrate; antifibrinolytics, and rFVIIa).
Different haemostatic thresholds for administering a prophylactic plasma transfusion before surgery/invasive procedures (INR, PT, thromboelastography variables).

If sufficient data were available, we would have performed separate meta‐analyses for these three comparisons, and we would have assessed age, type and dose of plasma components, and procedure type in subgroup analyses for each of these. We will do this in future updates should the opportunity arise.

Types of outcome measures

Primary outcomes

All‐cause mortality (up to 24 hours, and up to 30 days)
Major bleeding within 24 hours and within seven days as defined by the study, or by the following (based on Schulman 2010).

- Fatal bleeding.
- Intracranial/intraspinal/pericardial/intraocular/retroperitoneal, into a non‐operated joint, or intramuscular causing compartment syndrome.
- Surgical/invasive procedure site bleeding requiring a second intervention or reoperation.
- Surgical/invasive procedure site bleeding that causes a haematoma or haemarthrosis of sufficient size to delay mobilisation or wound healing.
- Surgical/invasive procedure site bleeding that is unexpected and prolonged or causes haemodynamic instability (as defined by the study) and is associated with a 20‐g/L drop in haemoglobin (Hb), or requiring two or more units of whole blood/red cells within 24 hours of bleeding.
- Extrasurgical/invasive procedure site bleeding associated with a 20‐g/L drop in Hb, or requiring two or more units of whole blood/red cells within 48 hours of bleeding.

Secondary outcomes

Transfusion requirements (within seven days of surgery/invasive procedure).
- Number of individuals requiring a transfusion.
- Mean number of transfusions per participant.
Use of haemostatic agents (within seven days of surgery/invasive procedure).
Volume of blood loss (within seven days of surgery/invasive procedure).
Serious adverse events (as defined in Appendix 2) due to:
- plasma transfusion (e.g. TRALI, TACO, transfusion‐related infection, transfusion‐related dyspnoea, acute transfusion reaction) within 24 hours; or
- surgery/invasive procedure (e.g. delayed wound healing, infection) within 30 days after the operation/invasive procedure.
Resource use: hospital/intensive treatment unit (ITU) length of stay, operating time, return to theatre for management of bleeding.
Venous and arterial thromboembolism (including deep vein thrombosis; pulmonary embolism; stroke; myocardial infarction) (within 30 days of surgery/invasive procedure, and within 90 days of surgery/invasive procedure).
Coagulation test abnormalities PT, INR, aPTT, or as defined by the study (within 24 hours of surgery/invasive procedure).
Quality of life, as defined by individual studies.

Search methods for identification of studies

The Systematic Review Initiative’s Information Specialist working in collaboration with the Cochrane Haematological Malignancies Group devised the search strategy. These are listed in the appendices.

Electronic searches

We searched the following databases with no limitation on dates or language or publication status. Before review submission, we reran the search and, if additional studies were identified, we would have incorporated these into the review and would have updated findings as required. This strategy served to avoid missing new studies completed during the review process.

Databases searched included the following.

Cochrane Central Register of Controlled Trials (CENTRAL), in the Cochrane Library (www.cochranelibrary.com/) (Appendix 3).
MEDLINE (OvidSP, 1946 to 28 January 2019) (Appendix 4).
Embase (OvidSP, 1974 to 28 January 2019) (Appendix 5).
Cumulative Index to Nursing and Allied Health Literature (CINAHL) (EBSCOHost) (1937 to 28 January 2019) (Appendix 6).
PubMed (e‐publications and in‐process citations ahead of print only) (www.ncbi.nlm.nih.gov/pubmed) (Appendix 7).
Transfusion Evidence Library (1950 to 28 January 2019) (www.transfusionevidencelibrary.com) (Appendix 8).
Latin American Caribbean Health Sciences Literature (LILACS) (1982 to 28 January 2019) (lilacs.bvsalud.org/en/) (Appendix 9).
Web of Science: Conference Proceedings Citation Index‐Science (CPCI‐S) (Thomson Reuters, 1990 to 28 January 2019) (Appendix 10).

Ongoing trial databases included:

ClinicalTrials.gov (clinicaltrials.gov) (Appendix 11); and
WHO International Clinical Trials Registry Search Platform (ICTRP) (apps.who.int/trialsearch/) (Appendix 12).

Searching other resources

We conducted handsearches of the reference lists of included studies and any relevant systematic reviews to identify further relevant studies. We made contact with the lead authors of relevant studies to identify any unpublished material, missing data, or information regarding ongoing studies.

Data collection and analysis

Selection of studies

We managed study selection with reference to Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). Two review authors (JH, KW) had planned to independently screen titles and abstracts identified by the search of databases for relevance against the eligibility criteria and immediately excluded clearly irrelevant studies. Given the large numbers of titles and abstracts identified, an additional review author (LE) assisted with the screening process. Two review authors (JH, KW) screened all titles and abstracts independently. We retrieved full‐text papers for all references for which a decision on eligibility could not be made from title and abstract alone.

Three review authors (JH, KW, LE) then assessed the references for relevance from full text. Two (of the three review authors) assessed full text independently. They were not blinded to individual study meta‐data such as author, institution, or publication journal. We requested additional information from study authors as necessary to assess the eligibility of individual studies for inclusion.

We used Covidence software to perform simultaneous independent screening and to assist with discrepancy resolution (Covidence 2016). We resolved disagreement between review authors regarding a study’s eligibility through discussion and consensus, and through consultation with the third review author as necessary.

We reported the search results and the screening and selection process using a PRISMA flow diagram (Liberati 2009). We recorded the reasons for excluding studies based on full‐text assessment and added those to the Characteristics of excluded studies table.

We collated multiple reports of one study so that the study, and not the report, was the unit of analysis.

Data extraction and management

As recommended in Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a), two review authors (JH, LE) independently extracted data using Covidence onto standardised pre‐piloted forms, and performed a cross‐check for agreement of data. These review authors were not blinded to names of authors, institutions, journals, or study outcomes. We reported characteristics of the included studies in the Characteristics of included studies table.

We exported data from Covidence into Cochrane's systematic review software Review Manager 5 (RevMan 2014).

Data collected included:

source: study ID, report ID, study author ID, citation and contact details, date of extraction;
general study details and eligibility: eligibility for inclusion confirmed, reason for exclusion, funding source, conflict of interest declared, references to other relevant studies;
study methods: location and country, clinical setting, number of centres, study design/type, recruitment dates and study duration, length of follow‐up, power calculation, stopping rules, method of sequence generation, method of allocation sequence concealment, method of blinding, bias concerns;
participant characteristics: age, gender, study population, primary diagnosis and/or operation, baseline laboratory measures of coagulopathy (PT, INR, aPTT, thromboelastography variables, platelet count, Hb) or evidence of presumed coagulopathy, total number screened, number included, number excluded, arm sample size, number analysed, number who received treatment, dropout rate, protocol violations, missing data;
intervention characteristics: number of study arms, description of arms, type of plasma, control product (e.g. crystalloid, colloid, placebo, alternative pro‐haemostatic agent), no treatment, haemostatic threshold for administering transfusion, dose of intervention/control; and
outcomes and results: all‐cause mortality within 24 hours and 30 days, major bleeding within 24 hours and seven days, transfusion requirements or number of patients requiring transfusion, use of haemostatic agents, blood loss volume, serious adverse events due to transfusion within 24 hours or surgery within 30 days, operating time, return to theatre for haemostatic control, hospital/critical care length of stay, venous and arterial thromboembolism, change in laboratory measures of coagulation (PT, INR, aPTT, thromboelastography variables) within 24 hours of plasma transfusion, estimate of effects with confidence intervals, key conclusions from study authors, miscellaneous comments from review authors, correspondence with study authors required.

Assessment of risk of bias in included studies

We assessed the risk of bias for all included RCTs using the Cochrane ‘Risk of bias’ tool according to Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). Two review authors (JH, LE) worked independently to assess each domain of potential bias listed below as ‘high’, ‘unclear’, or ‘low’ risk of bias. We reported a brief description of the judgement statements upon which we assessed potential bias in the Characteristics of included studies table. We ensured that a consensus on the degree of risk of bias was met through comparison of review authors’ statements. If necessary, we had planned to consult with a third review author (KW). We used the Cochrane tool for assessing risk of bias, which included the following domains.

Selection bias.
Performance bias.
Detection bias.
Attrition bias.
Reporting bias.
Other bias.

Measures of treatment effect

For continuous outcomes, we recorded mean, standard deviation, and total number of participants in both treatment and control groups. We were unable to perform analyses given the small number of included studies.

In the future, we will perform analyses using the mean difference (MD) with 95% confidence intervals (CIs) for continuous outcomes using the same scale, and standardised mean difference (SMD) when the scales are different. If available, we will extract and report hazard ratios (HRs) for mortality data. If HRs are not available, we will make every effort to estimate the HR as accurately as possible using available data and a purpose‐built method based on the Parmar and Tierney approach (Parmar 1998; Tierney 2007). If sufficient studies provide HRs, we will use HRs in favour of risk ratios (RRs) in a meta‐analysis, but for completeness, we will also perform a separate meta‐analysis of data from studies providing only RRs for the same outcome.

For dichotomous data, we recorded the number of events and the total number of participants in both treatment and control groups. We were unable to pool data due to the small number of included studies.

In future updates, we will report the pooled RR with 95% CI, or when the number of observed events is small (< 5% of sample per group) and when trials include balanced treatment groups, we will report the Peto odds ratio (OR) with 95% CI (Deeks 2011). We will report the number needed to treat for an additional beneficial outcome (NNTB) and the number needed to treat for an additional harmful outcome (NNTH), and we will perform quantitative measurements, or will provide a narrative report, as appropriate. If data allow, we will undertake quantitative assessments using Review Manager 5.

Unit of analysis issues

We did not include any clustered or cross‐over trials. In the future, if clustered or cross‐over trials will be included, we anticipate that unit of analysis issues may arise with recurring events or multiple treatment events. We would follow guidance from Chapter 16 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011c). We would report adverse event outcomes as groups of transfusion‐related and surgery‐related adverse events, as well as venous or arterial events. However often this is not possible due to duplicate counting of the same participant who may have experienced more than one adverse event of the same category (e.g. more than one transfusion‐related adverse event). In this case, we would report subgroup categories of adverse events separately and would report the 99% CI of the pooled RR to allow for multiple statistical testing. If this is not possible, we would provide a narrative summary.

Dealing with missing data

We recorded participants lost to follow‐up for each study. We contacted the authors of two trials by email for further information and are currently awaiting responses (Boyd 1996; De Pietri 2016). We utilised the assistance of Cochrane Russia for translation of one study to assess eligibility (Tseĭmakh 2008).

Assessment of heterogeneity

Given the small number of included studies, we were unable to combine studies to perform a meta‐analysis. In the future, if clinical and methodological characteristics of individual studies are sufficiently homogeneous, we would combine the data to perform a meta‐analysis. We would evaluate the extent of heterogeneity by visually inspecting forest plots as well as by utilising statistical methods. We would assess statistical heterogeneity of treatment effects between studies using a Chi² test (with P < 0.1). We would quantify heterogeneity using the I² statistic and would classify it as low (I² ≤ 50%), moderate (50% to 80%), or considerable (> 80%) (Deeks 2011). We would use a random‐effects model for low to moderate statistical heterogeneity given that we anticipate different but related effects across studies. If statistical heterogeneity is considerable, we would not report the overall summary statistic. We would explore potential causes of heterogeneity by performing sensitivity and subgroup analyses as appropriate (Deeks 2011).

Assessment of reporting biases

We were unable to complete a formal analysis of publication bias, given the small number of included studies. In the future, when at least 10 studies are incorporated into a meta‐analysis, we will explore potential publication bias (small‐trial bias) by generating a funnel plot and performing an appropriate test for asymmetry as recommended in Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2011).

Data synthesis

We planned to perform meta‐analyses based on recommendations from Chapter 9 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011). However, we identified an insufficient number of studies to do this.

In the future, if further studies are included, and provided they are sufficiently homogenous in their study design, we will conduct separate meta‐analyses (and subgroup analyses) for the three intervention comparisons. In the event of limited quantitative data for statistical analysis and synthesis, we will report findings through qualitative narrative summaries and tables. If sufficient data are available for meta‐analysis, we will perform this using Review Manager 5. One review author will enter data into the software programme, which will be independently checked for errors by a second review author. Given the likely variation in intervention practice, we will use a random‐effects model in the first instance. We will use the Mantel‐Haenszel method for dichotomous data and the inverse variance method for continuous data. We will use the Peto method when event numbers are small.

If heterogeneity is found to be above 80%, and if we identify a cause for the heterogeneity, we will explore this by performing subgroup analyses. If we cannot find a cause for the heterogeneity, we will not perform a meta‐analysis but will comment on the results as a narrative while presenting results from all studies in tables.

Subgroup analysis and investigation of heterogeneity

We planned to carry out separate subgroup analyses. However given the small number of included studies, it was not possible to do this. In the future, if sufficient data are available, we will carry out separate subgroup analyses for the three intervention comparisons to assess heterogeneity for the following.

Age of individual (neonate, infant, child, adult).
Type of procedure.
Plasma type.
Plasma dose.

We will categorise control interventions into three groups.

No prophylactic plasma transfusion before surgery (colloid, crystalloid, placebo, or no treatment).
Alternative pro‐haemostatic agents (prothrombin complex concentrate, cryosupernatant, fibrinogen concentrate; antifibrinolytics, and rFVIIa).
Different haemostatic thresholds for administering a prophylactic plasma transfusion before surgery (INR, PT, thromboelastography variables).

Sensitivity analysis

We were unable to perform sensitivity analyses due to the small number of included studies. If possible, in the future, we will perform sensitivity analyses to examine the robustness of our findings, by considering only:

studies with low risk of bias; or
studies with a low dropout rate (< 20%).

‘Summary of findings’ table

We used GRADEproGDT and the guidance provided in Chapters 11 and 12 of the Cochrane Handbook for Systematic Reviews of Interventions to produce a 'Summary of findings' table for each of the three intervention comparisons (GRADEpro 2015; Schünemann 2011a; Schünemann 2011b). We will utilise the GRADE approach, which defines the quality of the body of evidence as ‘high’, ‘moderate’, ‘low’, or ‘very low’, based on the following five considerations: design and implementation limitations causing risk of bias, indirectness of evidence, inconsistency or imprecision of results, and risk of publication bias. These tables will include the following outcomes.

Mortality within 30 days.
Major bleeding within 24 hours.
Transfusion requirements measured by mean number of transfusions per participant.
Transfusion requirements measured by number of individuals requiring a transfusion.
Serious adverse events measured by plasma transfusion‐related complications within 24 hours.
Serious adverse events measured by surgery‐related complications within 30 days.
Quality of life, as defined by individual studies.

Results

Description of studies

Results of the search

A literature search, conducted by CD and current up to 28 January 2019, identified a total of 10,448 references plus 758 ongoing studies. After removing duplicates, we screened 6326 references and 706 ongoing studies. Any two of three review authors (JH, LE, or KW) excluded 6974 records independently on the basis of the abstract. We retrieved 58 full‐text references for independent assessment by the same review authors.

We identified 11 studies within 17 full‐text references as potentially eligible for inclusion: five completed studies (De Pietri 2016; Mannucci 1976; Müller 2015; NCT00953901; Veelo 2012); three ongoing studies (NCT02561026; NCT02637427; Smart 2017); and three studies awaiting classification (Boyd 1996; NCT02777424; Tseĭmakh 2008). (See the PRISMA study flow diagram in Figure 1.)

Figure 1

PRISMA study flow diagram.

Included studies

We included five studies in the review (see Characteristics of included studies).

We grouped the included trials by type of intervention. One trial compared prophylactic plasma transfusion before invasive procedures versus no prophylactic plasma transfusion before invasive procedures (Müller 2015). Another trial compared prophylactic transfusion with FFP or platelets or both with neither FFP nor platelets before invasive procedures (Veelo 2012); one trial compared plasma transfusion before an invasive procedure versus alternative pro‐haemostatic agents (prothrombin complex concentrate) (Mannucci 1976); and two trials compared different haemostatic thresholds for administering a prophylactic plasma transfusion before surgical/invasive procedures (INR, PT, thromboelastography variables) (De Pietri 2016; NCT00953901).

Design

Four trials were published in English between 1976 and 2016. The remaining trial, whose full‐text report has not been published, was started in 2006 and was stopped in 2008, after enrolling two participants (NCT00953901). Four studies were parallel‐group two‐arm studies (De Pietri 2016; Müller 2015; NCT00953901; Veelo 2012). In the fifth trial, after a two‐arm study was completed, a third additional arm was added, in which both interventions from the two separate arms were combined (Mannucci 1976).

Sample size

The trials recruited 236 participants, ranging from 2 in NCT00953901 to 81 in Müller 2015. Three trials were stopped early due to poor recruitment (Müller 2015; NCT00953901; Veelo 2012). The initial plan was to recruit 400 participants in Müller 2015, 188 participants in NCT00953901, and 152 participants in Veelo 2012. Reasons reported for poor recruitment included short time frame for the opportunity to recruit (due to procedure urgency), refusal of consent (Müller 2015), and physician preference for management leading to resistance to recruit (Müller 2015; Veelo 2012).

Setting

All included studies were set in high‐income countries according to the World Bank classification (WB 2017). They were conducted in three countries: two in Italy (De Pietri 2016; Mannucci 1976); two in the Netherlands (Müller 2015; Veelo 2012); and one in the United States (NCT00953901). One was a multi‐centre trial incorporating four centres (Müller 2015).

Type of procedure

Four studies investigated solely bedside invasive procedures (such as chest drain insertion, needle liver biopsy, or abdominal paracentesis) (Mannucci 1976; Müller 2015; NCT00953901; Veelo 2012). One study investigated invasive procedures, which included bedside, radiologically interventional procedures (such as transjugular intrahepatic porto‐systemic shunt), other invasive procedures (such as oesophago‐gastro‐duodenoscopy or colonoscopy), and major surgically operative procedures (17% of study participants, including hepatic resection, thoracotomy, or other abdominal surgery) (De Pietri 2016). No study solely investigated major operative surgical procedures.

Participants

Four trials included only adults (De Pietri 2016; Müller 2015; NCT00953901; Veelo 2012), and one study did not report the age of participants (Mannucci 1976). Two trials included only participants with chronic liver disease (De Pietri 2016; Mannucci 1976), and two trials included only participants in intensive care (Müller 2015; Veelo 2012).

One study excluded participants taking anticoagulants or antiplatelets at the time of, or within seven days of, enrolment (De Pietri 2016). One study excluded participants with a bleeding time greater than seven minutes (Mannucci 1976). Another study excluded participants with an INR greater than three, and those taking vitamin K antagonists, activated protein C, abciximab, tirofiban, ticlopidine, or prothrombin complex concentrates, although heparin and low molecular weight heparin were not exclusion criteria provided they were discontinued for an 'appropriate period' (Müller 2015). One study excluded participants with prothrombin time greater than 20 seconds and those receiving clopidogrel (Veelo 2012).

Interventions

One study compared solely the use of FFP versus no FFP (Müller 2015). Another study compared the use of FFP or platelets or both with neither FFP nor platelets (Veelo 2012). One study compared use of FFP versus use of factors II, IX, and X followed by VII (Mannucci 1976). Two studies compared use of FFP versus different coagulopathy test thresholds (De Pietri 2016; NCT00953901).

Studies comparing FFP vs no FFP

One study compared FFP transfusion versus no FFP transfusion (Müller 2015). The other study compared transfusion with FFP or platelets (if low count and/or acetylsalicylic acid used) or both with neither FFP nor platelet transfusion (Veelo 2012). In this study, of the 35 participants randomised to the correction arm, 18 (51%) underwent FFP transfusion, and 23 (66%) underwent platelet transfusion. Given that more than 50% of participants in the group received FFP, and given the paucity of trials available for inclusion, we currently included this study and its data in the review. However, we also contacted the first author to request data related only to participants who received FFP. We are grateful for her acknowledgement of this request, and we are awaiting further correspondence.

Studies comparing FFP vs alternative pro‐haemostatic agents

One study compared FFP versus factors II, IX, and X, followed by VII (Mannucci 1976). In this study, after completing recruitment of participants to the two arms, researchers recruited a further cohort of participants to each receive a combination of FFP and prothrombin complex concentrate (PCC), thereby subsequently generating a third (non‐parallel) arm.

Studies comparing different transfusion thresholds

One study compared transfusing FFP using a thromboelastography R‐time threshold > 40 minutes as a transfusion trigger versus using an INR > 1.8 as a transfusion trigger (De Pietri 2016). Another study compared liberal (to keep INR < 1.6) versus restrictive (INR 1.6 to 3) FFP transfusion regimens (NCT00953901).

Outcomes

No studies measured all primary outcomes defined by the review. Two studies measured all‐cause mortality up to 24 hours ‐ Müller 2015 ‐ or at 30 days ‐ Veelo 2012 ‐ as defined in the review criteria above. One study measured 90‐day mortality (De Pietri 2016). We contacted study authors for information regarding 24‐hour and 30‐day mortality, and we are currently awaiting a response.

Four studies measured major bleeding within 24 hours (De Pietri 2016; Mannucci 1976; Müller 2015; Veelo 2012), and one study within seven days (De Pietri 2016).

No studies measured all secondary outcomes defined by the review.

Ongoing studies

We identified three ongoing studies (NCT02561026; NCT02637427; Smart 2017). Both NCT02561026 and NCT02637427 are pilot studies determining feasibility for a large trial (Characteristics of ongoing studies). NCT02561026 is an open‐label parallel‐group two‐arm trial conducted in Canada to compare prophylactic FFP transfusion versus no transfusion before an invasive procedure. This multi‐centre trial aimed to recruit 80 participants across three hospitals in an intensive care unit setting. Recruitment was expected to be complete by August 2018.

NCT02637427 is a parallel‐group two‐arm, single‐blinded (to outcome assessor) trial conducted in the United States to compare prophylactic FFP transfusion versus no transfusion before an invasive procedure outside of the operating room. This multi‐centre trial aimed to recruit 110 participants over two locations. It is expected to complete recruitment by April 2020.

Smart 2017 is a parallel‐group, two‐arm, prospective, randomised controlled trial comparing blood product use, bleeding events, and costs during and after endoscopic procedures in participants with liver cirrhosis with transfusion as guided by thromboelastometry versus conventional coagulation tests. An abstract has been published, and a full study report has not been published.

Studies awaiting classification

Three studies are awaiting classification (Boyd 1996; NCT02777424; Tseĭmakh 2008). The timing of the intervention in Boyd 1996 requires further clarification. Following contact with the study author, we are awaiting further information. In NCT02777424, the need for 'surgery' is not an absolute requirement to meet the study's inclusion criteria for enrolment. Given that the review's inclusion criteria require that participants undergo either surgery or invasive procedures, further assessment of this study will be needed following its completion to determine whether a subgroup of participants undergoing surgery (or invasive procedures) can be included in the review. Tseĭmakh 2008 was written in Russian, and we are awaiting a translation of the full text by Cochrane Russia. See Characteristics of studies awaiting classification for further information.

Excluded studies

We excluded 32 studies within 41 full‐text references after assessing their eligibility through a full‐text review (see Characteristics of excluded studies).

We excluded:

13 studies as the comparator did not meet our eligibility criteria (ChiCTR‐INR‐17013901; Freeman 1998; Galganski 2017; Laine 2003; Lerner 1997; Mintz 2006; NCT00235183; NCT03700723; Palmieri 2013; Ramies 2002; Tinmouth 2008; Wieding 1999; Williamson 1999);
eight studies as timing of the intervention did not meet our eligibility criteria (Hedstrand 1987; Liu 1994; NCT02352181; NCT00994045; Pieters 2015; Sommoggy 1990; Urwyler 2009; Wang 2010);
four studies as the participant population did not meet our eligibility criteria (Cao 2016; Jilma‐Stohlawetz 2011; Lance 2012; NCT00302965);
three studies as the study design did not meet our eligibility criteria (Gazzard 1975; Hildebrandt 2007; Kerner 2008);
two studies as the intervention did not meet our eligibility criteria (Bauer 1986; NCT00656396);
one study as the indication for the intervention did not meet our eligibility criteria (Rocha 2017); and
one study as the study was withdrawn before participant enrolment was complete (NCT00233246).

Risk of bias in included studies

See Figure 2 and Figure 3 for visual representations of ’Risk of bias’ assessments across all included trials and for each individual item in the included trials. See the Characteristics of included studies section ’Risk of bias’ table for further information about bias identified within individual trials.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

One study was completed in 2008 after recruiting two patients (NCT00953901). No report has been published, thereby precluding an analysis.

Allocation

Two trials were at low risk of selection bias due to adequate methods of sequence generation and allocation concealment (Müller 2015; Veelo 2012).

Random sequence generation

Three trials were at low risk of bias due to random sequence generation because they used computer‐generated ‐ Müller 2015 and Veelo 2012 ‐ or electronically generated randomisation (De Pietri 2016).

One trial was at unclear risk of bias because it did not provide any details about the method of sequence generation (Mannucci 1976).

Allocation concealment

Two trials were at low risk of bias due to allocation concealment because they used either a web‐based permuted block system ‐ Müller 2015 ‐ or consecutively numbered, opaque, sealed envelopes (Veelo 2012).

Two trials were at unclear risk of bias due to allocation concealment. Mannucci 1976 used a system of sealed envelopes, but it is unclear from the report whether opaque, sequentially numbered, sealed envelopes were used. De Pietri 2016 provided no information about the method of allocation concealment.

Blinding

Blinding of participants and personnel

One study was at unclear risk of performance bias because information in the study report was insufficient to make an assessment (Mannucci 1976).

Three studies were at high risk of performance bias (except for all‐cause mortality) because of their open‐label design (De Pietri 2016; Müller 2015; Veelo 2012).

Blinding of outcome assessors

One study was at low risk of detection bias, given its blinded endpoint evaluation design, which blinded the outcome assessor from the intervention during assessment of bleeding (Müller 2015).

One study was at unclear risk of detection bias given absence of information in the study report to permit assessment (Mannucci 1976).

Two studies were at high risk of bias (except for all‐cause mortality) because of their open‐label design (De Pietri 2016; Veelo 2012).

Incomplete outcome data

Three studies were at a low risk of attrition bias. In one study, all randomised participants were included in an intention‐to‐treat (ITT) analysis (De Pietri 2016). In another study, all participants who were randomised and underwent a procedure were included (Müller 2015). In the third study, a clear CONSORT diagram was reported, demonstrating an equal number of participants in each study arm who did not undergo the procedure after randomisation (Veelo 2012).

One study was at an unclear risk of attrition bias because information was insufficient to permit assessment (Mannucci 1976).

Selective reporting

One study was at unclear risk of bias as no protocol or trial registration was found, thereby precluding assessment (Mannucci 1976).

Three studies were at high risk of bias (De Pietri 2016; Müller 2015; Veelo 2012). One study recognised limitations in an outcome assessment method, leading to risk of under‐reporting (Müller 2015). In another study, the time span for the primary outcome was not clearly reported (De Pietri 2016), whereas in yet another study, the primary outcomes were not defined in the report (Veelo 2012). Two studies reported an outcome that was not predefined in their protocol ‐ De Pietri 2016 ‐ or public registry (Veelo 2012). All three studies did not report an outcome that had been predefined in their protocol ‐ De Pietri 2016 and Müller 2015 ‐ or in the public registry (Veelo 2012).

Other potential sources of bias

Balance of baseline characteristics

One study was at low risk of bias because it reported and demonstrated similar baseline characteristics between arms (De Pietri 2016).

Two studies were at unclear risk of bias. The first study did not report participant baseline characteristics (Mannucci 1976). The second reported baseline characteristics but did not highlight nor discuss a statistically significant difference (among patients with expected prolonged duration of mechanical ventilation) between arms (Veelo 2012). The material risk of this difference is unclear.

One study was at high risk of bias because it reported a statistically significant imbalance in participants with liver disease between groups (Müller 2015).

Early termination of study

One study was at unclear risk of bias because it was terminated early after randomising only 47% of its recruitment target due to increasing clinician resistance to transfusing FFP following recognition of a low incidence of bleeding in the no FFP group (Veelo 2012).

One study was at high risk of bias because it was terminated early after randomising only 20% of its recruitment target owing to a slow rate of inclusion (Müller 2015). As a result, study authors could not demonstrate non‐inferiority between the two arms.

Effects of interventions

See: Summary of findings for the main comparison Prophylactic plasma transfusion before surgery/invasive procedures compared to no prophylactic plasma transfusion before surgery/invasive procedures (colloid, crystalloid, placebo, or no treatment) for patients undergoing non‐cardiac surgery or invasive procedures; Summary of findings 2 Prophylactic plasma transfusion before surgery/invasive procedures compared to alternative haemostatic agents for patients undergoing non‐cardiac surgery/invasive procedures; Summary of findings 3 Thromboelastography threshold compared to standard of care (laboratory parameters) for patients undergoing non‐cardiac surgery or invasive procedures

See summary of findings Table for the main comparison, summary of findings Table 2, and summary of findings Table 3 for the main comparisons.

Studies comparing fresh frozen plasma (FFP) vs no FFP

One study compared FFP transfusion versus no FFP transfusion (Müller 2015). Another study compared FFP or platelet transfusion or both versus neither FFP nor platelet transfusion (Veelo 2012).

Primary outcomes

Mortality up to 24 hours and up to 30 days

One study reported all‐cause mortality up to 24 hours (Müller 2015). We are uncertain whether there is any difference between the two groups (1 trial, 76 participants; risk ratio (RR) 0.70, 95% confidence interval (CI) 0.48 to 1.03; low‐quality evidence) (Analysis 1.1).

Another study reported intensive care unit (ICU) mortality up to 30 days (Veelo 2012). We are very uncertain whether there is any difference between the two groups (1 trial, 72 participants; RR 0.38, 95% CI 0.13 to 1.10; very low‐quality evidence) (Analysis 1.1).

Major bleeding within 24 hours and within seven days

Two studies reported major bleeding at 24 hours (Müller 2015; Veelo 2012).

In the study comparing FFP transfusion versus no transfusion (Müller 2015), we are very uncertain whether there is any difference in major bleeding between the two arms (1 trial, 76 participants; RR 0.33, 95% CI 0.01 to 7.93; very low‐quality evidence) (Analysis 1.2). A single bleeding event (haemothorax) was reported following the invasive procedure (chest drain insertion) in a participant assigned to the non‐transfusion group, which fulfilled the above criteria of major bleeding (requiring 3 units of red blood cells (RBCs), 3 units of FFP, and 1 unit of platelets). Haemostasis was achieved following this treatment.

In the study comparing FFP or platelet transfusion or both versus neither FFP nor platelet transfusion, we are very uncertain whether there is any difference in major bleeding between the two groups (1 trial, 72 participants; RR 1.59, 95% CI 0.28 to 8.93; very low‐quality evidence) (Analysis 1.2) (Veelo 2012).

No study reported bleeding within seven days.

Secondary outcomes

Transfusion requirements within seven days

One study reported the number of participants requiring transfusion (Veelo 2012), although the time scale for this outcome was not reported. No RBC transfusion occurred in either group (1 trial, 72 participants; very low‐quality evidence).

One study reported transfusion requirements as median number (and interquartile range (IQR)) of transfusions per participant in the first 24 hours only (Müller 2015; Table 1). Study authors reported no difference in RBCs, FFP, or platelet transfusion between groups (1 trial, 76 participants; RBC P = 0.91, FFP P = 0.06, platelets (PLTs) P = 0.43; very low‐quality evidence).

Use of haemostatic agents within seven days

No study reported this outcome.

Volume of blood loss within seven days

One study reported this outcome as median (grams) blood loss (with IQR) per participant, although the duration of assessment for this outcome is unclear from the report (Veelo 2012). Study authors reported no difference between the two study arms (1 trial, 72 participants; median 3.0 grams, IQR 1.0 to 6.0 in the intervention group vs median 3.0 grams, IQR 2.0 to 6.0 in the no intervention group (P = 0.96; Table 1).

Serious adverse events from plasma transfusion (within 24 hours) or from surgery/intervention (within 30 days)

One study reported a lung injury score at 24 hours after randomisation as median and IQR (Müller 2015). Study authors reported that a difference between the two groups "did not reach statistical significance" (1 trial, 76 participants; median 2, IQR 0.8 to 2.5 in the intervention arm vs median 1.25, IQR 0.4 to 2.4 in the no intervention arm (P = 0.28; Table 1). Study authors did not provide quantitative data regarding baseline lung injury scores for the two arms nor changes in score between baseline and 24 hours post procedure. They qualitatively describe in the report's narrative that baseline scores were increased in both groups.

No study reported surgical or procedure‐related adverse events (excluding bleeding).

Resource use

Hospital length of stay

No study reported this outcome.

ICU length of stay

Two studies reported ICU length of stay as median and IQR (Table 1) (Müller 2015; Veelo 2012). Study authors from both studies showed no evidence of a difference between the two groups (Müller 2015: 76 participants, P = 0.13; Veelo 2012: 72 participants, P = 0.21).

Duration of operation/intervention

No study reported this outcome.

Return to theatre/intervention room

One study reported a single major bleeding event within 24 hours (Müller 2015). However study authors did not report the need to transfer/return to the theatre/intervention room for intervention to achieve haemostasis, although the follow‐up period was not described in the study report. Therefore no events were reported for this outcome (1 trial, 76 participants).

Venous and arterial thromboembolism within 30 days and 90 days

No study reported this outcome.

Coagulation test abnormalities

One study reported this outcome as median international normalised ratio (INR) and IQR before and after transfusion (Table 1) (Müller 2015). Study authors reported a reduction in INR following FFP transfusion (1 trial, 38 participants; median 1.8, IQR 1.5 to 2.5 before FFP transfusion vs median 1.4, IQR 1.3 to 1.63 after FFP transfusion; P < 0.001). However only 21 of 38 participants (54%) demonstrated a reduction in INR < 1.5 after transfusion.

Quality of life

No study reported this outcome.

Studies comparing FFP vs other pro‐haemostatic alternatives

Only one study compared FFP versus factors II, IX, X, and VII (Mannucci 1976).

Primary outcomes

Mortality up to 24 hours and up to 30 days

Mannucci 1976 did not report this outcome.

Major bleeding within 24 hours and within seven days

Mannucci 1976 reported this outcome, although the time span for its measurement is unclear from the report. The study followed up with participants for 12 months. No major bleeding events were reported in either group (1 trial, 21 participants; very low‐quality evidence). We are very uncertain of the significance of this result, given the very low quality of evidence for this outcome.

Secondary outcomes

Transfusion requirements within seven days

Mannucci 1976 did not report this outcome.

Use of haemostatic agents within seven days

Mannucci 1976 did not report this outcome.

Volume of blood loss within seven days

Mannucci 1976 did not report this outcome.

Serious adverse events from plasma transfusion (within 24 hours) or from surgery/intervention (within 30 days)

Mannucci 1976 reported serious adverse events from plasma transfusion (defined by the criteria in Appendix 2). We are very uncertain whether there was any difference in these events between the two groups (1 trial, 21 participants; RR 9.82, 95% CI 0.59 to 162.24; very low‐quality evidence) (Analysis 2.1). During the study period, following four events in the FFP group, there was a practice change whereby all subsequent participants in the FFP group received prophylactic steroids, which may have reduced the rate of, or clinical detection of, future events, and may have masked their true incidence. The study was also likely underpowered to detect a difference between groups.

Mannucci 1976 did not report serious adverse events related to the procedure within 30 days.

Resource use

Mannucci 1976 did not report hospital length of stay, ICU length of stay, procedure duration, or need to return to the theatre/procedure room.

Venous and arterial thromboembolism within 30 days and 90 days

We are very uncertain whether there is a difference in thromboembolism with the use of FFP. Mannucci 1976 reported no cases of thromboembolism in the groups, although the timing of assessment of this outcome was not reported (1 trial, 21 participants; very low‐quality evidence). Participants were followed up for 12 months.

Coagulation test abnormalities

Mannucci 1976 reported the number of participants with corrected indices of coagulation. We are very uncertain whether there is a difference between the groups of participants with persistently deranged indices (1 trial, 21 participants; aPTT: RR 2.75, 95% CI 0.68 to 11.13; PT: RR 0.44, 95% CI 0.2 to 0.96; Normotest: RR 0.43, 95% CI 0.21 to 0.88; very low‐quality evidence) (Analysis 2.2).

Quality of life

Mannucci 1976 did not report this outcome.

Studies comparing different transfusion thresholds

Two studies compared different transfusion thresholds. One study compared a liberal (to keep INR < 1.6) versus restrictive (INR 1.6 to 3) FFP transfusion regimen (NCT00953901). However this study ceased recruitment after enrolling two participants and reported no outcomes.

Another study compared FFP transfusion using a thromboelastography R‐time threshold > 40 minutes as a transfusion trigger vs a transfusion trigger with INR > 1.8 (De Pietri 2016).

Primary outcomes

Mortality up to 24 hours and up to 30 days

No study reported 24‐hour or 30‐day mortality. One study reported all‐cause mortality up to 90 days (De Pietri 2016). We are very uncertain whether there is a difference between groups (1 trial, 60 participants; RR 1.14, 95% CI 0.47 to 2.75; very low‐quality evidence) (Analysis 3.1).

Major bleeding within 24 hours and within seven days

One study reported data for this outcome (De Pietri 2016). One participant (1/30) developed major bleeding according to the above criteria on the day following the invasive procedure (paracentesis), requiring packed red cells and FFP transfusions. This participant had been randomised to the standard of care group and had received FFP before undergoing the procedure. There were no events in the thromboelastography (TEG) group (0/30). We are very uncertain whether there is a difference between the groups (1 trial, 60 participants; RR 0.33, 95% CI 0.01 to 7.87; very low‐quality evidence) (Analysis 3.2).

Secondary outcomes

Transfusion requirements within seven days

One study reported data for this outcome (De Pietri 2016). The number of individuals requiring overall blood products in the TEG group may be reduced versus the number in the standard of care (SOC) group (1 trial, 60 participants; RR 0.18, 95% CI 0.08 to 0.39; low‐quality evidence). However we are very uncertain whether there was a reduction in participants within the TEG group requiring either FFP alone (1 trial, 60 participants; RR 0.03, 95% CI 0.00 to 0.48; very low‐quality evidence), PLTs alone (1 trial, 60 participants; RR 0.20, 95% CI 0.05 to 0.84; very low‐quality evidence), or both FFP and PLTs (1 trial, 60 participants; RR 0.75, 95% CI 0.18 to 3.07; very low‐quality evidence) (Analysis 3.3).

No study reported the mean number of transfusions per patient.

Use of haemostatic agents within seven days

No study reported this outcome.

Volume of blood loss within seven days

No study reported this outcome.

Serious adverse events from plasma transfusion (within 24 hours) or from surgery/intervention (within 30 days)

One study reported an allergic reaction during FFP transfusion in the SOC group (1/30) compared to none in the TEG group (0/30) (De Pietri 2016). We are very uncertain whether there was a difference between groups (1 trial, 60 participants; RR 0.33, 95% CI 0.01 to 7.87; very low‐quality evidence) (Analysis 3.4).

No study reported surgical or procedure‐related serious adverse events (excluding bleeding, as already described above and in Analysis 3.2).

Resource use

No study reported hospital length of stay, ICU length of stay, or procedure duration. Studies did not describe the necessity to return to the theatre or the intervention room to control bleeding in the participant with a post‐procedural bleeding event.

Venous and arterial thromboembolism within 30 days and within 90 days

No study reported this outcome.

Coagulation test abnormalities

No study reported this outcome.

Quality of life

No study reported this outcome.

Discussion

The objective of this review was to determine the clinical effectiveness and safety of prophylactic plasma transfusion for people with confirmed or presumed coagulopathy requiring non‐cardiac surgery or invasive procedures.

Five trials were eligible for inclusion in this systematic review (De Pietri 2016; Mannucci 1976; Müller 2015; NCT00953901; Veelo 2012). One study was terminated after recruiting only two patients; this study report has not been published (NCT00953901). No results can be ascertained from this study.

Of the four completed studies, all were carried out in high‐income countries between the years 1976 and 2016; they involved a total of 234 participants, all of whom were > 18 years of age.

Two trials studied participants in an intensive care unit (ICU) setting (153 participants) (Müller 2015; Veelo 2012). Two trials studied only procedures carried out by participants' bedside (153 participants; in Müller 2015 and Veelo 2012, although seven participants in Veelo 2012 subsequently underwent surgical tracheotomy as opposed to percutaneous dilatational tracheotomy due to landmark recognition difficulties). Only one trial studied participants who underwent surgical procedures in an operating theatre such as hepatic resection, other abdominal surgery, or thoracotomy (10 participants out of a total of 60) (De Pietri 2016).

Only one trial studied fresh frozen plasma (FFP) as a single intervention compared with no intervention (Müller 2015). Another study compared FFP or platelet transfusion or both versus neither FFP nor platelet transfusion (Veelo 2012). A single trial studied the use of FFP versus the use of a combination of alternative pro‐haemostatic agents ‐ Prothromplex (containing factors II, IX, and X), followed by factor VII concentrate (Mannucci 1976). A single study compared different haemostatic thresholds for FFP transfusion, reporting a thromboelastography (TEG)‐measured R‐time > 40 minutes compared with an international normalised ratio (INR) >1.8 (De Pietri 2016).

Summary of main results

Given the very low‐quality evidence, we are very uncertain whether there is a difference in:

mortality with the use of prophylactic FFP versus no FFP. No trials studying other eligible comparators reported mortality up to 30 days;
major bleeding within 24 hours, whether compared with no FFP or alternative pro‐haemostatic agents, or using different transfusion thresholds. Few or no events in the studies and small study sample sizes have contributed to this uncertainty;
transfusion requirements, in the form of number of transfusions per participant (one study comparing FFP transfusion vs no transfusion) or number of participants requiring a transfusion (one study comparing FFP or platelet transfusion or both vs neither FFP nor platelet transfusion);
serious adverse events from FFP transfusion versus no transfusion or other pro‐haemostatic agents, or using different thresholds for transfusion;
surgical/interventional adverse events in participants transfused with FFP using a TEG‐guided transfusion trigger versus standard tests for coagulation. No trials comparing FFP use versus no FFP use or use of an alternative pro‐haemostatic product reported this outcome; or
arterial or thromboembolism rate. Only one study, which compared participants receiving FFP versus factors II, IX, X, and VII, reported this outcome (Mannucci 1976).

For participants receiving FFP transfusion via a TEG‐guided transfusion trigger versus standard tests of coagulation (INR), the number of participants subsequently requiring transfusion (for overall blood products) via a TEG‐guided trigger may have been reduced. Our confidence is limited due to the low quality of evidence for this outcome. However we are very uncertain whether there is a difference in individual product use (FFP, platelets, or both), given that the evidence is of very low quality.

No studies reported the outcome looking at quality of life.

Overall completeness and applicability of evidence

This review aimed to provide the most up‐to‐date and comprehensive assessment of the clinical effectiveness and safety of prophylactic plasma transfusion for people with confirmed or presumed coagulopathy requiring non‐cardiac surgical or invasive procedures.

To facilitate development of wide and comprehensive criteria for inclusion, we used a broad definition of surgical or surgically interventional procedures. We used the definition of the American College of Surgeons, which includes consideration of the use of probes or needles. We also used the classification system from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9‐CM 2011), which codifies various surgically oriented or interventional procedures within its operative classification.

Our search in any language has resulted in consideration of a foreign language trial, resulting in collaboration with Cochrane Russia for its translation (Tseĭmakh 2008).

We sought contact with the author of one study for further information required to determine its eligibility for inclusion based on our criteria (Boyd 1996). A second study awaiting classification is still ongoing (NCT02777424). Both studies compare use of FFP versus an alternative pro‐haemostatic agent. Two ongoing studies are comparing FFP use versus no transfusion (NCT02561026; NCT02637427), and one ongoing study is comparing transfusion of blood products guided by thromboelastometry versus conventional coagulation tests (Smart 2017).

Several limitations may affect the strength and applicability of our review.

Two studies (153 participants) of the four included in this review (234 participants) were set in the intensive care unit (Müller 2015; Veelo 2012). Given that these studies represent 65% of participants included in this review on which review findings are based, and given that individuals in the ICU represent more severe acute illness or physiological derangement compared with other individuals with a coagulopathy requiring surgery or interventional procedures, caution is required in extrapolating or generalising these findings to other person cohorts.
Of the four included studies, two trials studied bedside interventional procedures (Müller 2015; Veelo 2012), and one trial studied percutaneous needle liver biopsy (likely carried out in an interventional area or theatre, although the locality is not clear from the report) (Mannucci 1976). Only a single trial included major operative surgery such as solid organ (liver) resection, other abdominal surgery, or thoracotomy (together with other invasive procedures) (De Pietri 2016). Major general surgery was therefore under‐represented by the included studies, and no included studies represented major head and neck, neurological, or orthopaedic surgery.
No trials eligible for inclusion in the review studied children.
Significant variability is evident in methods used by the four studies for assessing, grading, and reporting bleeding. One study did not report its approach to assessment of bleeding (Mannucci 1976), another study ‐ Müller 2015 ‐ used the HEmorrhage MEasurement (HEME) tool (Arnold 2007), another study ‐ De Pietri 2016 ‐ used the World Health Organization (WHO) bleeding score (Miller 1981), and the authors of still another study generated a scale of clinically irrelevant bleeding, minor bleeding, and major bleeding (Veelo 2012).
The dose of FFP given to individuals in the four studies was variable. In two studies, 12 mL/kg was used (Mannucci 1976; Müller 2015); in one study, 10 mL/kg ideal body weight was used (De Pietri 2016); and in another study, 300 to 600 mL was used (dependent on prothrombin time (PT) result), equating to 4.3 to 8.6 mL/kg for a 70‐kg individual (Veelo 2012).
It is questionable whether the sample size for important primary outcomes including mortality and bleeding, as well as other important outcomes such as adverse events or thromboembolism, was sufficiently large to detect a difference. Lack of any events for some outcomes precluded meaningful analysis. Two of the four studies terminated recruitment early due to a slow rate of recruitment (81 participants; Müller 2015), or due to increasing resistance to FFP transfusion by physicians owing to observations of a low bleeding rate in an open‐label trial (72 participants; Veelo 2012). One trial recruited 21 patients (Mannucci 1976).
Although three of the four studies were conducted over the last decade, one study was conducted in 1976 (Mannucci 1976). Given the development of strategies to reduce adverse events from plasma transfusion, including pathogen reduction techniques or use of male donors (Rogers 2015), how applicable these historical data remain to modern clinical practice remains questionable.
Although one of the included studies utilised thromboelastography (De Pietri 2016), accessibility of such a modality for use in global clinical practice will depend on availability at individual institutions, which requires significant cost, equipment maintenance, quality control procedures, and user training (Subramanian 2018). Furthermore, there may be intra‐operator and inter‐operator variability in generated results (Quarterman 2014), although newer systems that utilise an automated process for sample and reagent mixing and activation may reduce the scope for operator‐dependent variability and the requirement for training (Hartmann 2018).

Quality of the evidence

All five included studies were parallel‐group randomised controlled trials (RCTs). Two were open‐label trials, with high risk of both performance bias and detection bias (De Pietri 2016; Veelo 2012). One study used an open‐label blinded‐endpoint design with high risk of performance bias (Müller 2015). In one study, details of study design were unclear (Mannucci 1976). Another study, which ceased recruitment after enrolling two participants, was a double‐blinded trial (NCT00953901). No published report is available for review of its methods or analysis of its methodological quality.

Our judgements for risk of bias of included studies can be found in the risk of bias tables in the Characteristics of included studies section of the review, and they are summarised in Figure 2 and Figure 3

We judged three studies to be at high risk of bias in three out of seven domains (De Pietri 2016; Müller 2015; Veelo 2012).
We judged one study to be at low risk of bias in three out of seven domains (De Pietri 2016).
We judged two studies to be at low risk of bias in four out of seven domains (Müller 2015; Veelo 2012).

We assessed the quality of evidence using the GRADE approach, finding that quality ranged from low to very low.

With regard to the main outcomes of this review, we assessed the quality of evidence as low for the following.

Number of individuals requiring transfusion, overall use of products within seven days, and comparison of FFP use versus different transfusion thresholds: we downgraded the evidence due to serious risk of bias and indirectness in these studies.

We assessed the quality of evidence as very low for the following.

ICU mortality at 30 days: we downgraded the evidence for very serious risk of indirectness due to inclusion of a single trial of 76 participants, conducted in an intensive treatment unit (ITU) setting in patients undergoing a single procedure (percutaneous tracheostomy), which reported this outcome, thereby limiting applicability of study findings to the review's wider surgical setting. We also downgraded the evidence for serious risk of imprecision, given the wide confidence interval that crossed the line of no difference.
Major bleeding within 24 hours: we downgraded the evidence due to serious risk of bias in some trials, serious or very serious risk of indirectness in some trials, and serious risk of imprecision in all trials.
Number of transfusions per participant within seven days: we downgraded the evidence due to serious risk of bias, indirectness, and imprecision in some trials.
Number of individuals requiring a transfusion within seven days in the setting of FFP transfusion versus no transfusion: we downgraded the evidence due to the trial's serious risk of bias, indirectness, and imprecision.
Serious adverse events measured by plasma transfusion‐related complications within 24 hours: we downgraded the evidence due to serious risk of bias in some trials, serious risk of indirectness in all trials, and serious or very serious risk of imprecision in all trials.
Serious adverse events measured by surgery‐related complications within 30 days in the setting of FFP use based on different transfusion thresholds: we downgraded the evidence due to serious risk of bias, indirectness, and imprecision in these trials.

Potential biases in the review process

This review reflects no obvious bias. A wide and comprehensive search was performed with no restriction on language, country, or publication status. We are grateful to Cochrane Russia for translation of a Russian text (Tseĭmakh 2008). We contacted authors of two trials for further information and are currently awaiting response (Boyd 1996; De Pietri 2016).

We used Covidence software to permit two of three review authors (JH, LE, KW) to independently and carefully screen each reference, abstract, or full text for relevance. Given the large number of references initially found through the wide search (7032 results), three review authors carried out this task. Two review authors (JH and LE) independently extracted and agreed on data from included studies. Two review authors (PF and JH) independently extracted and checked for agreement of data from ongoing studies or studies awaiting classification. Given the small number of included trials, construction of a funnel plot for identification of publication bias was not possible. In one included trial (NCT00953901), which was terminated after two participants were enrolled, no result data were published, thereby precluding assessment. Given early termination and the small patient sample size (two participants), publication bias is unlikely.

We pre‐specified all reported outcomes and subgroups for analysis in advance; these were published in the review protocol before the search was conducted.

Agreements and disagreements with other studies or reviews

This review has shown lack of clear evidence demonstrating benefit for the use of prophylactic FFP before non‐cardiac surgical or invasive procedures across a range of outcomes. This is consistent with two previous systematic reviews, which indicated lack of effectiveness within a range of different clinical settings (Stanworth 2004; Yang 2012).

However this review has highlighted the uncertainty of our confidence in the evidence and in the results due to low‐ or predominantly very low‐quality evidence for important patient‐centred outcomes. This is also consistent with findings of the aforementioned systematic reviews, which found limited evidence for most clinical situations ‐ Stanworth 2004 ‐ and noted little improvement in methodological quality over the years ‐ Yang 2012.

Moreover, the small number of trials pertaining to non‐cardiac surgery, the large number of trials in which only a small number of participants were enrolled, and the low incidence of certain outcome events (such as bleeding or adverse events), as highlighted by previous reviews, represent a consistent theme over the past 15 years and within this review.

The limited quality of evidence in this review is also consistent with that of a recent Cochrane Review on plasma use in the alternative setting of cardiovascular surgery (Desborough 2015), which judged the quality of evidence as moderate or low for most outcomes.

Figure 1

PRISMA study flow diagram.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Analysis 1.1

Comparison 1 Prophylactic plasma transfusion before surgery versus no prophylactic plasma transfusion before surgery (colloid, crystalloid, placebo, or no treatment), Outcome 1 All‐cause mortality.

Analysis 1.2

Comparison 1 Prophylactic plasma transfusion before surgery versus no prophylactic plasma transfusion before surgery (colloid, crystalloid, placebo, or no treatment), Outcome 2 Major bleeding.

Analysis 2.1

Comparison 2 Prophylactic plasma transfusion before surgery versus alternative haemostatic agents, Outcome 1 Serious adverse event due to plasma transfusion.

Analysis 2.2

Comparison 2 Prophylactic plasma transfusion before surgery versus alternative haemostatic agents, Outcome 2 Coagulation test abnormalities.

Analysis 3.1

Comparison 3 Thromboelastography threshold versus standard of care (laboratory parameters), Outcome 1 All‐cause mortality.

Analysis 3.2

Comparison 3 Thromboelastography threshold versus standard of care (laboratory parameters), Outcome 2 Major bleeding.

Analysis 3.3

Comparison 3 Thromboelastography threshold versus standard of care (laboratory parameters), Outcome 3 Number of individuals requiring a transfusion.

Analysis 3.4

Comparison 3 Thromboelastography threshold versus standard of care (laboratory parameters), Outcome 4 Serious adverse event due to plasma transfusion.

Prophylactic plasma transfusion before surgery/invasive procedures compared to no prophylactic plasma transfusion before surgery/invasive procedures (colloid, crystalloid, placebo, or no treatment) for patients undergoing non‐cardiac surgery or invasive procedures
Patient or population: patients undergoing non‐cardiac surgery or invasive procedures Setting: individuals in ICU undergoing invasive procedures. Studies conducted in The Netherlands Intervention: prophylactic plasma transfusion before surgery/invasive procedures Comparison: no prophylactic plasma transfusion before surgery/invasive procedures (colloid, crystalloid, placebo, or no treatment)
Outcomes	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**			Certainty of the evidence (GRADE)
		Without prophylactic plasma transfusion before surgery/invasive procedures	With prophylactic plasma transfusion before surgery/invasive procedures	Difference
All‐cause mortality up to 30 days № of participants: 72 (1 RCT)	RR 0.38 (0.13 to 1.10)	Study population			⊕⊝⊝⊝ Very low^a,b
		297 per 1000	113 per 1000 (39 to 327)	184 fewer per 1000 (259 fewer to 30 more)
Major bleeding within 24 hours № of participants: 148 (2 RCTs)	Two studies; reported separately^c FFP group events 0/38; no FFP group events 1/38; RR 0.33, 95% CI 0.01 to 7.93 (Müller 2015) FFP or platelets or both group events 3/35; no transfusion group events 2/37; RR 1.59, 95% CI 0.28 to 8.93 (Veelo 2012)				⊕⊝⊝⊝ Very low^b,d,e
Number of transfusions per participant within 7 days № of participants: 76 (1 RCT)	Study authors reported no difference ("P = 0.91 (RBC), P = 0.06 (FFP), P = 0.43 (PLT)") between the 2 groups (76 participants; Müller 2015). See Table 1				⊕⊝⊝⊝ Very low^d,f,g
Number of individuals requiring a transfusion within 7 days № of participants: 72 (1 RCT)	Study authors reported "no... transfusion of packed red cells for blood loss during or after [the procedure]" (72 participants; Veelo 2012)				⊕⊝⊝⊝ Very low^d,g,h
Serious adverse events measured by plasma transfusion‐related complications within 24 hours № of participants: 76 (1 RCT)	Study authors reported that a difference between the 2 groups "did not reach statistical significance" (Müller 2015). See Table 1				⊕⊝⊝⊝ Very low^d,f,i
Serious adverse events measured by surgery or procedure‐related complications within 30 days ‐ not reported	‐	‐	‐		‐
Quality of life ‐ not reported	‐	‐	‐		‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; FFP: fresh frozen plasma; ICU: intensive care unit; PLT: platelet; RBC: red blood cell; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded two points due to indirectness (single trial, ICU setting, single bedside procedure; Veelo 2012). ^bDowngraded one point due to imprecision (wide confidence intervals, which cross the line of no difference). ^cGiven only two studies, small numbers of participants, low event count, and very poor‐quality evidence by GRADE assessment with heterogenous groups and wide confidence intervals that cross the line of no difference, we have not pooled the results, as doing so would be neither statistically nor clinically meaningful. ^dDowngraded one point due to high risk of bias over multiple domains. ^eDowngraded one point due to indirectness (2 trials, ICU setting, participants undergoing bedside procedures). ^fDowngraded one point due to indirectness (ICU setting, participants undergoing bedside procedures; Müller 2015). ^gDowngraded one point due to imprecision (low or absent event incidence). ^hWould be downgraded two points due to very serious indirectness. However downgraded only one point as already downgraded two points in other domains (maximum 3 downgrade points allowed by GRADE method). ⁱDowngraded one point for imprecision (single study, 76 participants, terminated early, likely under‐powered to demonstrate an effect; Müller 2015).

Table 1. Additional data: outcomes reported for FFP versus no FFP with median and IQR

Outcome	Prophylactic transfusion			No prophylactic transfusion			P value
Outcome	Median	IQR	Total number of participants in group	Median	IQR	Total number of participants in group	P value
Number of transfusions after intervention, per participant (within 24 hours)
Müller 2015
RBC	1	0 to 2	38	1	0 to 3	38	0.91
FFP	0	0 to 1	38	2	0 to 2	38	0.06
PLT	1	0 to 2	38	0	0 to 1	38	0.43

Volume of blood loss per participant (within 7 days)
Veelo 2012	3.0	1.0 to 6.0	35	3.0	2.0 to 6.0	37	0.96

Lung injury score (within 24 hours after randomisation)
Müller 2015	2	0.8 to 2.5	38	1.25	0.4 to 2.4	38	0.28

ICU length of stay (within 30 days)
Müller 2015	12	6 to 19	38	7	3 to 17	38	0.13
Veelo 2012	15	8 to 29	35	21	14 to 26	37	0.21

Coagulation test abnormalities (within 24 hours)
Müller 2015
INR before FFP intervention	1.8	1.5 to 2.5	38	‐	‐	‐
INR after FFP intervention	1.4	1.3 to 1.63	38	‐	‐	‐	< 0.001
FFP: fresh frozen plasma; ICU: intensive care unit; INR: international normalised ratio; IQR: interquartile range; PLT: platelets; RBC: red blood cells.

Table 1. Additional data: outcomes reported for FFP versus no FFP with median and IQR

Summary of findings 2. Prophylactic plasma transfusion before surgery/invasive procedures compared to alternative haemostatic agents for patients undergoing non‐cardiac surgery/invasive procedures

Prophylactic plasma transfusion before surgery compared to alternative haemostatic agents for patients undergoing non‐cardiac surgery or invasive procedures
Patient or population: patients undergoing non‐cardiac surgery/invasive procedures Setting: individuals with chronic liver disease undergoing liver needle biopsy. Study conducted in Italy in 1976 Intervention: prophylactic plasma transfusion before surgery/invasive procedures Comparison: alternative haemostatic agents
Outcomes	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**			Certainty of the evidence (GRADE)
Outcomes	Relative effect (95% CI)	Without prophylactic plasma transfusion before surgery/invasive procedures	With prophylactic plasma transfusion before surgery/invasive procedures	Difference	Certainty of the evidence (GRADE)
All‐cause mortality up to 30 days ‐ not reported	‐	‐	‐		‐
Major bleeding within 24 hours № of participants: 21 (1 RCT)	Study authors reported, "in none of the patients was there any clinical or laboratory evidence of bleeding" (21 participants; Mannucci 1976)				⊕⊝⊝⊝ Very low^a,b
Number of transfusions per participant within 7 days ‐ not reported	‐	‐	‐		‐
Number of individuals requiring a transfusion within 7 days ‐ not reported	‐	‐	‐		‐
Serious adverse events measured by plasma transfusion‐related complications within 24 hours № of participants: 21 (1 RCT)	RR 9.82 (0.59 to 162.24)	Study population			⊕⊝⊝⊝ Very low^d,e
		Low
		10 per 1000^c	98 per 1000 (6 to 1000)	88 more per 1000 (4 fewer to 1612 more)
		High
		100 per 1000^c	982 per 1000 (59 to 1000)	882 more per 1000 (41 fewer to 16124 more)
Serious adverse events measured by surgery or procedure‐related complications within 30 days ‐ not reported	‐	‐	‐		‐
Quality of life ‐ not reported	‐	‐	‐		‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded two points due to indirectness (single study, 21 participants, single bedside procedure; Mannucci 1976). ^bDowngraded one point due to imprecision (low or absent event incidence, small study). ^cAssumed risks taken from available drug product information by Baxter AG, Austria, for the use of Prothromplex TOTAL (HPRA 2018). Prothromplex TOTAL is a 4 factor concentrate (II, VII, IX, and X) with added protein C. Prothromplex, Immuno Vienna as used in Mannucci 1976 was a three‐factor concentrate (II, VII, IX) after which factor VII, Immuno, Vienna was administered. No published safety data found for Prothromplex, Immuno, Austria, as used in Mannucci 1976. ^dWould have been downgraded two points due to indirectness (differences in plasma preparation in modern practice limit applicability of results from a study conducted in 1976, and use of prophylactic steroids before plasma transfusion in the study is not recommended by modern guidelines; Mannucci 1976). However downgraded only one point as already downgraded two points in another domain (maximum three downgrade points allowed by GRADE method). ^eDowngraded two points due to imprecision (very wide confidence intervals, crossing the line of no difference, which could include both significant harm and benefit, and clinicians instituted a practice change in a single arm during the study, by giving participants prophylactic steroids before FFP transfusion, following four febrile or allergic reactions in the group).

Summary of findings 2. Prophylactic plasma transfusion before surgery/invasive procedures compared to alternative haemostatic agents for patients undergoing non‐cardiac surgery/invasive procedures

Summary of findings 3. Thromboelastography threshold compared to standard of care (laboratory parameters) for patients undergoing non‐cardiac surgery or invasive procedures

Thromboelastography threshold compared to standard of care (laboratory parameters) for patients undergoing non‐cardiac surgery or invasive procedures
Patient or population: patients undergoing non‐cardiac surgery/invasive procedures Setting: individuals with cirrhosis undergoing invasive procedures. Study conducted in Italy Intervention: thromboelastography threshold Comparison: standard of care (laboratory parameters)
Outcomes	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**			Certainty of the evidence (GRADE)
		Without thromboelastography threshold	With thromboelastography threshold	Difference
All‐cause mortality up to 30 days ‐ not reported	‐	‐	‐		‐
Major bleeding within 24 hours № of participants: 60 (1 RCT)	RR 0.33 (0.01 to 7.87)	Study population			⊕⊝⊝⊝ Very low^a,b,c
		33 per 1000	11 per 1000 (0 to 262)	22 fewer per 1000 (33 fewer to 229 more)
Number of transfusions per participant within 7 days ‐ not reported	‐	‐	‐		‐
Number of individuals requiring a transfusion within 7 days № of participants: 60 (1 RCT)	RR 0.18 (0.08 to 0.39)	Study population			⊕⊕⊝⊝ Low^a,b
		1000 per 1000	180 per 1000 (80 to 390)	820 fewer per 1000 (920 fewer to 610 fewer)
Serious adverse events measured by plasma transfusion‐related complications within 24 hours № of participants: 60 (1 RCT)	RR 0.33 (0.01 to 7.87)	Study population			⊕⊝⊝⊝ Very low^a,b,c
		33 per 1000	11 per 1000 (0 to 262)	22 fewer per 1000 (33 fewer to 229 more)
Serious adverse events measured by surgery or procedure‐related complications within 30 days ‐ not reported	‐	‐	‐		‐
Quality of life ‐ not reported	‐	‐	‐		‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one point due to high risk of bias across multiple domains (De Pietri 2016). ^bDowngraded one point due to indirectness (single trial, 60 participants with cirrhosis; De Pietri 2016). ^cWould be downgraded two points due to very serious imprecision (low event incidence, very wide confidence intervals including both serious harm and benefit). However downgraded only one point as already downgraded two points in other domains (maximum three downgrade points allowed by GRADE method).

Summary of findings 3. Thromboelastography threshold compared to standard of care (laboratory parameters) for patients undergoing non‐cardiac surgery or invasive procedures

Comparison 1. Prophylactic plasma transfusion before surgery versus no prophylactic plasma transfusion before surgery (colloid, crystalloid, placebo, or no treatment)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 All‐cause mortality Show forest plot	2		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

1.1 Up to 24 hours	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 Up to 30 days	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Major bleeding Show forest plot	2		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

2.1 Within 24 hours	2		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Comparison 1. Prophylactic plasma transfusion before surgery versus no prophylactic plasma transfusion before surgery (colloid, crystalloid, placebo, or no treatment)

Comparison 2. Prophylactic plasma transfusion before surgery versus alternative haemostatic agents

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Serious adverse event due to plasma transfusion Show forest plot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

1.1 Within 24 hours	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Coagulation test abnormalities Show forest plot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

2.1 Within 24 hours, correction of partial thromboplastin time	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 Within 24 hours, correction of prothrombin time	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.3 Within 24 hours, correction of Normotest	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Comparison 2. Prophylactic plasma transfusion before surgery versus alternative haemostatic agents

Comparison 3. Thromboelastography threshold versus standard of care (laboratory parameters)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 All‐cause mortality Show forest plot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

1.1 Up to 90 days	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Major bleeding Show forest plot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

2.1 Within 24 hours	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 Within 7 days	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Number of individuals requiring a transfusion Show forest plot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

3.1 Within 7 days, overall blood product use	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.2 Within 7 days, FFP use only	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.3 Within 7 days, PLT use only	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.4 Within 7 days, FFP and PLT use	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Serious adverse event due to plasma transfusion Show forest plot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

4.1 Within 24 hours	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Comparison 3. Thromboelastography threshold versus standard of care (laboratory parameters)