Patient Blood Management Guidelines: Module 1

Critical Bleeding Massive Transfusion

| Appendix |

Appendix E - Evidence matrixes

Evidence matrix 1

Clinical question
In patients with critical bleeding requiring massive transfusion, what is the effect of variation of physiologic, biochemical and metabolic (including temperature) parameters on morbidity, mortality and transfusion rate?
Evidence statement
Hypothermia, metabolic acidosis, thrombocytopenia and coagulopathy may be independently associated with increased mortality. 15,61,63-68

Evidence base
Poor (D): Eight Level IV studies with high risk of bias. 15,61,63-68

Consistency
Satisfactory (C): Some inconsistency reflecting genuine uncertainty around clinical question. Studies investigating:

Hypothermia: Consistency in definition of outcome. Similar direction of effect (i.e. hypothermia associated with poor survival), although some inconsistency in statistical significance, with three studies showing similar results and two studies in which temperature was not a significant predictor of mortality.

pH: All five studies were consistent with regard to low pH/acidosis being significantly associated with poorer survival. The definition of the outcome and the threshold parameters for pH values were also consistent.

Base deficit: Consistency in outcome definition. There was some inconsistency in direction of effect – three studies indicated an association between increased base deficit and poor survival, and two indicated that base deficit was not a significant predictor of mortality.

INR: All four studies were consistent with regard to a higher international normalised ratio (INR) being associated with poor survival. The definition of the outcome and the threshold parameters for INR values were also consistent.

Prothrombin time: N/A. One study included for this outcome.

Partial thromboplastin time: Two studies identified were inconsistent with regard to the association between PTT and mortality.

Platelet count: Three studies were inconsistent with regard to the association between low platelet count and mortality, primarily because different definitions were used.


Clinical impact
Satisfactory (C): Moderate clinical impact. Studies investigating:

Hypothermia: Sample size was sufficiently large (ranged from n = 45 to n = 246). Three studies showed that reduced core body temperature was associated with increased mortality in patients who had critical bleeding and those who were transfused. Gonzalez et al (2007) 64 and Moore et al (2008) 68 showed that reduced body temperature was not significantly associated with mortality in patients who experienced shock resuscitation or haemorrhagic shock.

pH: All five studies showed that reduced pH was associated with increased mortality in patients with critical bleeding and those who were transfused.

Base deficit: Sample size was sufficiently large (ranged from n = 45 to n = 252). Three studies showed that an increase in base deficit was associated with an increased mortality in critically bleeding and transfused patients. Two studies showed base deficit was not significantly associated with mortality in patients who experience shock resuscitation or have haemorrhagic shock.

INR: Sample size was sufficiently large (ranged from n = 97 to n = 247). All studies showed that an increase in INR was associated with increased mortality in critically bleeding and transfused patients. 15,61,64,68 Gonzalez and colleagues stratified analysis by admission to emergency department and admission to an intensive care unit.64

Prothrombin time: Small sample size (n = 45).

Partial thromboplastin time: Two studies included are of limited clinical impact, each demonstrating different activated partial prothrombin time (APTT) parameters and statistical significance. Sample size is sufficient (n = 45 to n = 119).

Platelet count: Three studies included; the largest study reporting on this outcome found that non-survivors had lower platelet counts. Sample size is sufficient with the included studies (n = 45 to n = 174).

In all studies, consideration of adverse events was not applicable to this recommendation as the outcome is mortality.


Generalisability
Good (B): Studies investigating:

Hypothermia: All participants in three studies were trauma patients and one study analysed shock resuscitation patients. All patients were critically bleeding.

pH: Participants in four studies were trauma patients; one study analysed patients with haemorrhagic shock.

Base deficit: Participants in three studies were trauma patients, 61,65,66 one study was on haemorrhagic shock patients 68 and one study on shock resuscitation patients. 64 All patients were critically bleeding.

INR: Participants in two studies were trauma patients, 15,61 one haemorrhagic shock 68 and one study shock resuscitation patients. 64

Prothrombin time: One study included.

Partial thromboplastin time: Small sample size and contradictory results for the two studies included.

Platelet count: Contradictory results for all three studies included.

There were no study design restrictions as all studies were case series; hence, patients were in natural environments when the outcome was measured.


Applicability
Good (B): Studies investigating:

Hypothermia: Three United States based studies and one Australian based study.

pH: Four United States based studies and one Australian based study.

Base deficit: Five United States based studies.

INR: Three United States based studies and one Australian based study.

Prothrombin time: One United States based study.

Partial thromboplastin time: One United States based and one Australian based study.

Platelet count: Two United States based studies and one Australian based study.

This evidence base is applicable to the Australian setting and there are no organisational or cultural barriers.

Evidence matrix 2

Clinical question
In patients with critical bleeding requiring massive transfusion, does the dose, timing and ratio (algorithm) of RBCs to blood component therapy (FFP, platelets, cryoprecipitate or fibrinogen concentrate) influence morbidity, mortality and transfusion rate? Part 1 – Algorithm
Evidence statement
In trauma patients with critical bleeding requiring massive transfusion, the use of a protocol that includes the dose, timing and ratio of blood component therapy is associated with reduced mortality. 4-5

Evidence base
Poor (D). One Level III study with a high risk of bias; 4 one Level III study with a moderate risk of bias. 5

Consistency
Good (B): The studies were mostly consistent in their findings and inconsistency may be explained.

Clinical impact
Good (B). Substantial clinical impact. Studies included predominantly small sample sizes for an assessment of mortality differences, but the clinical impact was significant, with an absolute difference in mortality of approximately 10%.

Generalisability
Good (B). Both studies included patients with critical bleeding requiring massive transfusion.

Applicability
Satisfactory (C). Both studies were conducted in United States health-care settings.

Evidence matrix 3

Clinical question
In patients with critical bleeding requiring massive transfusion, does the dose, timing and ratio (algorithm) of RBCs to blood component therapy (FFP, platelets, cryoprecipitate or fibrinogen concentrate) influence morbidity, mortality and transfusion rate? Part 2 – Algorithm
Evidence statement
In trauma patients with critical bleeding requiring massive transfusion, a ratio of ≤ 2:1:1 of red blood cells:fresh frozen plasma:platelets is associated with reduced mortality.5,70,71 However, due to the possibility of survivor bias, it is not possible to recommend a target ratio of RBC:FFP:platelets.

Evidence base
Poor (D). Two Level III studies with a high risk of bias; 70,71 one Level III study with a moderate risk of bias. 5 Survivor bias is likely to have affected results.

Consistency
Excellent (A). All studies were consistent in their findings.

Clinical impact
Satisfactory(C). Moderate clinical impact. Studies included predominantly small sample sizes.

Generalisability
Satisfactory(C). All studies included patients with critical bleeding requiring massive transfusion; however, the definition of massive transfusion in Cinat et al (1999) 70 was ≥ 50 units of RBC or whole blood in 48 hours.

Applicability
Satisfactory (C). All studies were conducted in United States health-care settings.

Evidence matrix 4

Clinical question
In patients with critical bleeding requiring massive transfusion, does the dose, timing and ratio (algorithm) of RBCs to blood component therapy (FFP, platelets, cryoprecipitate or fibrinogen concentrate) influence morbidity, mortality and transfusion rate? Part 2 – Timing.
Evidence statement
In trauma patients with critical bleeding requiring massive transfusion, early transfusion of fresh frozen plasma and platelets is associated with reduced mortality and subsequent red blood cell requirements. 85,89

Evidence base
Poor (D). Two Level IV studies with a high risk of bias. 85,89

Consistency
Excellent (A). The studies were consistent in their findings.

Clinical impact
Satisfactory (C). More than 400 patients were reviewed in each study.

Generalisability
Good (B). Both studies included patients with critical bleeding requiring massive transfusion and the populations were civilian.

Applicability
Satisfactory (C). Both studies were conducted in United States health-care setting.

Evidence matrix 5

Clinical question
In patients with critical bleeding requiring massive transfusion, what is the effect of RBC transfusion on patient outcomes?
Evidence statement
In trauma patients with critical bleeding requiring massive transfusion, an increased volume of transfused red cells may be independently associated with increased mortality. 91,92.

Evidence base
Satisfactory (C): Two Level III studies with a moderate risk of bias. 91,92

Consistency
Good (B): Results of the two studies were consistent, although the different reference group in the studies make comparisons not completely clear.

Clinical impact
Poor (D): The studies are underpowered, with confidence interval values that cross 1.0 (odds ratio); thus, the likely clinical impact is unclear.

Generalisability
Satisfactory (C): Some generalisability to the target population.

Applicability
Satisfactory (C): Both studies were completed in the United States.

Evidence matrix 6

Clinical question
In patients with critical bleeding requiring massive transfusion, what is the effect of RBC transfusion on patient outcomes?
Evidence statement
In trauma patients with critical bleeding requiring massive transfusion, an increased volume of transfused red cells is independently associated with acute respiratory distress syndrome. 91-92

Evidence base
Satisfactory (C): Two Level III studies with a moderate risk of bias. 91, 92

Consistency
Good (B): Results of the two studies were consistent, although the different reference group in the studies make comparisons not completely clear.

Clinical impact
Satisfactory (C): The studies report a moderate clinical impact.

Generalisability
Satisfactory (C): Some generalisability to the target population.

Applicability
Satisfactory (C): Both studies were completed in the United States.

Evidence matrix 7

Clinical question
In patients with critical bleeding requiring massive transfusion, what is the effect of recombinant activated factor VII (rFVIIa) (prophylaxis or treatment) on morbidity, mortality and transfusion rate?
Evidence statement
In trauma patients with critical bleeding requiring massive transfusion, administration of rFVIIa has no effect on 48-hour or 30-day mortality. 6

Evidence base
Good (B): One good quality Level II study. 6

Consistency
Not applicable (NA): Only one study.

Clinical impact
Poor (D): There is no clinical impact from rFVIIa.

Generalisability
Satisfactory (C): The studies seem to be generalisable to critical bleeding patients resulting from blunt or penetrating trauma; however, the additional exclusion criteria need to be taken in to consideration before considering the results generalisable to all critically bleeding patients.

Applicability
Good (B): Study samples from 32 hospitals throughout Australia, Canada, France, Germany, Israel, Singapore, South Africa and the United Kingdom. Although only one hospital was in Australia the Canadian and United Kingdom settings are comparable to Australia.

Evidence matrix 8

Clinical question
In patients with critical bleeding requiring massive transfusion, what is the effect of recombinant activated factor VII (rFVIIa) (prophylaxis or treatment) on morbidity, mortality and transfusion rate?
Evidence statement
In patients with critical bleeding requiring massive transfusion, there is insufficient evidence to determine any association between rFVIIa and thromboembolism. 6

Evidence base
Good (B): One good quality Level II study.6

Consistency
Not applicable (NA). Only one study.

Clinical impact
Poor (D): The low incidence of the thromboembolic events and consequent lack of statistical power mean that the data are insufficient to draw any conclusions.

Generalisability
Satisfactory (C): The studies seem to be generalisable to a critically bleeding population resulting from blunt or penetrating trauma; however, the additional exclusion criteria need to be taken into account before considering the results generalisable to all critically bleeding patients.

Applicability
Good (B): Study samples from 32 hospitals throughout Australia, Canada, France, Germany, Israel, Singapore, South Africa and the United Kingdom. Although only one hospital was in Australia, the Canadian and United Kingdom settings are comparable to Australia.

Evidence matrix 9

Clinical question
In patients with critical bleeding requiring massive transfusion, what is the effect of recombinant activated factor VII (rFVIIa) (prophylaxis or treatment) on morbidity, mortality and transfusion rate?
Evidence statement
In patients with blunt trauma and critical bleeding requiring massive transfusion, administration of recombinant activated factor VII (rFVIIa) is associated with reduced red blood cell (RBC) transfusion requirements and the incidence of acute respiratory distress syndrome (ARDS).6 In patients with penetrating trauma and
critical bleeding requiring massive transfusion, administration of rFVIIa has no effect on morbidity.6

Evidence base
Good (B): One good quality Level II study.6

Consistency
Not applicable (NA). Only one study.

Clinical impact
Satisfactory (C): Moderate clinical impact. In blunt trauma patients administration of rFVIIa is associated with reduced RBC transfusion requirements and the incidence of ARDS. In patients with penetrating trauma administration of rFVIIa has no effect on morbidity.

Generalisability
Satisfactory (C): The studies seem to be generalisable to a critical bleed population resulting from blunt or penetrating trauma; however, the additional exclusion criteria need to be taken in to consideration before considering the results generalisable to all critically bleeding patients.

Applicability
Good (B): Study samples from 32 hospitals throughout Australia, Canada, France, Germany, Israel, Singapore, South Africa and the United Kingdom. Although only one hospital was in Australia the Canadian and United Kingdom settings are comparable to Australia.

Evidence matrix 10

Clinical question
In patients with critical bleeding requiring massive transfusion, what is the effect of FFP, cryoprecipitate, fibrinogen concentrate, and/or platelet transfusion on patient outcomes?
Evidence statement
In trauma patients with critical bleeding requiring massive transfusion, a red blood cell (RBC):FFP ratio of ≤ 2:1 is associated with reduced mortality.61,75

Evidence base
Poor (D): Two Level III studies with a high risk of bias.61,75

Consistency
Good (B). Both studies looked at several different outcomes. Where similar outcomes were reported findings were generally consistent.

Clinical impact
Satisfactory (C). Moderate clinical impact. One study has n = 246,61 the other has n = 135.,75 A RBC:FFP ratio of ≤ 2:1 was reported to be associated with reduced mortality, but there is uncertainty about whether this is related to survivor bias or the effect of the intervention.

Generalisability
Satisfactory (C). Both studies included patients with critical bleeding who required massive transfusion. One study was conducted in a military war zone setting which is not directly generalisable to a civilian setting. 61

Applicability
Satisfactory (C). One study was in a United States military hospital and the other75 was in the United States health-care setting.61