Hemorrhage remains the most common preventable cause of combat death.1 Upon autopsy review, 80% to 85% of deaths in combat are classified as nonsurvivable as the result of devastating injury.1,2 Of the remaining 15% to 20% of survivable deaths, 66% to 80% of patients die from hemorrhage.1,2 Severe trauma, defined as an Injury Severity Score (ISS) of greater than 15,3 is common during combat, representing 20% of all trauma admissions at one U.S. Army combat support hospital in Baghdad in 2004.4 Recent combat operations in the Middle East have refocused attention on the early coagulopathy of trauma present on hospital admission.5–7 Trauma and shock severity correlate with the degree of coagulopathy,6–9 which in turn has been associated with mortality. The most severely injured 10% of patients10 may require massive transfusion, defined as the transfusion of 10 or more red blood cell (RBC) units in a 24-hour period.11–13 In civilian trauma centers, the incidence of patients with traumatic injuries requiring massive transfusion ranges between 1% and 3%,13–15 with an incidence of massive transfusion reported as high as 15% in patients with the most severe injuries.5,16 The majority of massive transfusion patients die within 6 to 12 hours of hospital arrival,4,17,18 mortality rates for patients requiring massive transfusion range between 20% and 50%,11,16,19 and massive transfusion has been shown to be independently associated with increased mortality.14 The high mortality risk in massive transfusion patients largely results from the well-described “lethal triad” characterized by hypothermia, acidosis, and coagulopathy.20–23
Damage control resuscitation addresses that lethal triad immediately at admission to the combat hospital or civilian trauma center.10 Intravascular volume resuscitation is accomplished using thawed human plasma as primary resuscitation fluid in a 1:1 or 1:2 ratio with RBCs.5,11,24–27 For patients who require continued resuscitation, a massive transfusion protocol is activated and the blood bank delivers an additional 6 units of RBCs, 6 packs of platelets, 6 units of plasma, and one 10-unit bag of cryoprecipitate to the operating room.10
Cryoprecipitate is a human blood product derived from the precipitate fraction of cold-thawed human plasma.28 It typically contains about 2.5 g of fibrinogen per 10-unit bag. Among all clotting factors in human plasma, fibrinogen is the first to decrease to pathophysiologic levels in hypocoagulable patients who have undergone major elective abdominal surgery with blood loss exceeding 20% of their calculated blood volume.29 This hypofibrinogenemia does not seem to be completely explained by blood loss and resuscitation alone.30,31 After moderate hemorrhagic shock in pigs, we reported that fibrinogen decreases as a result of constant production and increased consumption and hyperfibrinolysis, which results in prolonged blood clotting times.32,33 Additionally, fibrinogen replacement has been shown to restore normal coagulation times in both coagulopathic pigs in hemorrhagic shock34 and in diluted coagulopathic in vitro human blood.35
Our hypothesis was that early administration of fibrinogen to patients with severe injuries requiring a massive transfusion would decrease death from hemorrhage as a result of minimizing coagulopathy, excessive bleeding, and shock. Our objective in this retrospective study was to determine whether an increased ratio of fibrinogen to RBCs transfused would improve survival to hospital discharge in patients with severe traumatic injuries requiring massive transfusion at a combat support hospital by decreasing death from hemorrhage.
PATIENTS AND METHODS
The data presented here were obtained under a human use protocol that received Institutional Review Board approval through the Department of Clinical Investigation at Brooke Army Medical Center at Fort Sam Houston in San Antonio, TX. Using the Joint Theater Trauma Registry (JTTR) maintained at the U.S. Army Institute of Surgical Research (USAIR) at Fort Sam Houston in San Antonio, Texas, we performed a retrospective analysis of data for trauma patients admitted to two combat support hospitals in Iraq between January 2004 and October 2005. The JTTR is a database established by the Department of Defense to capture data prospectively from military hospitals for coalition and foreign national patients and from point of injury through rehabilitation for the U.S. casualties.36
The JTTR was queried for patients who received a massive transfusion, defined as 10 or more RBC units [including both RBCs and fresh whole blood (FWB) units] within a 24-hour period, between January 2004 to October 2005. Data analyzed from the JTTR in this study were age, combined ISS, recombinant Factor VIIa use, overall mortality at hospital discharge, mortality resulting from hemorrhage, and laboratory values and vital signs at admission to the combat support hospital (temperature, systolic blood pressure, hemoglobin, platelet level, international normalized ratio [INR], and base deficit) as well as the number in units and type of blood products administered within 24 hours of hospital admission. Numbers of units of blood products tabulated were packed red blood cell (RBC) units, fresh whole blood (FWB) units, fresh frozen plasma (FFP) units, cryoprecipitate (Cryo) 10-unit bags, and apheresis platelet (aPLT) units. Fibrinogen levels on hospital admission were not available in theater.
We calculated the F:R ratio (Fig. 1) according to standard amounts of fibrinogen contained in each blood component (Table 1). In the numerator of the ratio, each blood product is multiplied by a conversion factor that converts the units of that blood product transfused to grams of fibrinogen in that blood product (Fig. 1). Note that cryoprecipitate contains 2.5 g of fibrinogen per 10-unit bag, and FWB contains 1 g of fibrinogen per unit. The FWB appearing in the numerator of the F:R ratio thus actually represents grams of fibrinogen, not whole blood units transfused.
To analyze the effect of fibrinogen-to-red cell (F:R) ratios on mortality, patients were initially divided into eight ratio groups (Fig. 2A) and then into quartiles (Fig. 2B) based on the ratio of F:R units transfused. The first quartile consisted of patients with F:R ratios of 0 to ≤0.2 g/unit, the second 0.2 to ≤0.4 g/unit, the third 0.4 to ≤0.6 g/unit, and the fourth ≥0.6 g/unit. A sharp drop in mortality was noted at 0.2 g/unit between the first and second quartile, so the four ratio groups were combined into two groups, with a low F:R ratio of <0.2 g/unit and a high F:R ratio of ≥0.2 g/unit. We then plotted both mortality and death resulting from hemorrhage as a function of low F:R ratio (<0.2 g/unit) and high F:R ratio (≥0.2 g/unit; Fig. 3).
All continuous parametric data are described as mean (±SD). Comparisons were performed with Student’s t test for continuous data and with Fisher’s exact test for categorical data. Statistical analysis was performed with SPSS 14.0 (Chicago, IL). Significant differences were determined at p < 0.05. A logistic regression model was developed using multivariate logistic regression analysis to determine which sets of variables were independent predictors of mortality. Correlation analysis was performed using both Pearson’s and Spearman’s correlation analysis with mortality as the dependent correlation factor. Values with greater than 0.15 correlation coefficients were considered candidates for model development. Regression analysis was performed using a backward likelihood ratio logistic regression. Results were validated using a forward stepwise regression of the same candidate variables. Variables comprised of single blood products alone were excluded to eliminate any collinear effects on the model.
During the 22-month period between January 2004 and October 2005, the combat support hospital received 5,586 patients from both military and civilian populations with traumatic injuries. The JTTR identified 252 of those 5,586 casualties (4.5%) who required massive transfusions. The mean age of the patients studied was 26 (±7.5) years. The mean ISS was 21 (±9.8). The overall mortality was 30% (75 of 252). The mean (SD) F:R ratios transfused for the low and high groups were 0.1 g/unit (±0.06) and 0.48 g/unit (±0.2), respectively (p < 0.001). Mortality was 27 of 52 (52%) and 48 of 200 (24%) in the low and high F:R ratio groups, respectively (p < 0.001; Fig. 3). The incidence of death from hemorrhage was higher in the low F:R group, 23 of 27 (85%), compared with the high F:R group, 21 of 48 (44%) (p < 0.001; Table 2). Additional comparisons of variables between patients in the low F:R ratio (<0.2 g/unit) and the high F:R ratio (≥0.2 g/unit) groups are summarized in Table 2. Upon multiple logistic regression, base deficit, ISS, and high F:R ratio (≥0.2 g fibrinogen per red cell unit [RBC or FWB] transfused) were noted to be independently associated with survival (Table 3).
Correlation analysis of variables resulted in eight variables as candidates for further multivariate analysis. These included hemoglobin, platelets, INR, base deficit, FFP, ISS, F:R ratio (continuous), and F:R ratio (high/low). The F:R ratio (high/low) variable had a higher correlation coefficient than F:R ratio (continuous) variable (−0.247 vs. −0.178) and was, therefore, chosen for multivariate analysis over the continuous F:R ratio variable. The INR variable showed a significant positive correlation (0.290) to mortality, but was excluded for low sample size (168 of 252). Recombinant Factor VIIa use was not correlated with mortality in this analysis and was therefore not a candidate for multivariate logistic regression.
Backward and forward regression analysis resulted in the base deficit, ISS, and F:R ratio (high/low), as the best predictors of mortality with a receiver operating characteristic curve area of 0.71 (95% confidence interval 0.63–0.79; Fig. 4 and Table 3). Hosmer and Lemeshow tests showed a good model fit with a χ2 of 3.17 (p = 0.92). The model correctly predicted 93% (10 of 147) of survivors and 29% (17 of 50) of deaths in the sample set using a 0.5 predictive positive cutoff. Mortality rate predicted increased to 56% (33 of 59) using a 0.3 positive cutoff value (survival rate decreased to 76%, correct predictions were obtained in 36 of 147).
Previous studies have shown that fibrinogen levels fall early in elective surgical patients with significant blood loss,29 and in pigs32,33 subjected to blood loss, but neither blood loss nor resuscitation30,31 completely explains this decrease in fibrinogen levels. Acidosis in a swine model caused a near twofold increase in fibrinogen degradation but no changes in fibrinogen synthesis.37 Furthermore, neither bicarbonate38 nor tris-hydroxymethyl-aminomethane39 pH neutralization in similar swine models immediately corrected an acidosis-induced coagulopathy. On the other hand, replenishment of fibrinogen restored normal coagulation times in coagulopathic pigs in hemorrhagic shock34 and in diluted coagulopathic in vitro human blood.35
Our results are consistent with the studies above in that they demonstrate that an increased amount of fibrinogen in relation to the amount of RBCs transfused was independently associated with survival in combat casualties requiring massive transfusions. This is important since the majority of preventable deaths in combat are due to hemorrhage and strategies that can decrease death from hemorrhage in combat will significantly impact survival in this population. To the best of our knowledge, this is the first article to demonstrate the association of high levels of transfused fibrinogen on the survival of casualties requiring massive transfusion.
High concentrations of fibrinogen are available not only from cryoprecipitate but also from plasma-derived fibrinogen concentrates. Fibrinogen is also now being produced using recombinant techniques and is available in a lyophilized powder form (Pharming Group, Leiden, The Netherlands). The ability to reconstitute recombinant fibrinogen from a lyophilized powder form surmounts the logistical hurdle that burdens FFP administration in military operations, i.e., the need to transport frozen human plasma thousands of miles from home station to military hospitals in the combat zone. Future trials comparing equal amounts of fibrinogen in FFP, cryoprecipitate, and fibrinogen concentrates for coagulopathic patients are needed.
Recombinant Factor VIIa (NovoSeven, NovoNordisk, Princeton, NJ) is another clotting factor available in lyophilized powder form that many trauma centers have recently integrated into their massive transfusion protocols. Recombinant Factor VIIa reduces the total number of red cell units transfused in combat casualties,36 decreases death in combat casualties requiring massive transfusion,4 and requires adequate levels of platelets and fibrinogen in place to achieve hemostasis.40,41 Administering recombinant Factor VIIa in the face of low fibrinogen levels may not produce the desired hemostatic effect; in addition, acidosis should be reversed and platelet deficiencies corrected either simultaneously with or before recombinant Factor VIIa administration.41
Although our data reveal a strong association between a high F:R transfusion ratio and survival, that association does not necessarily imply causation. During the data collection period from January 2004 through October 2005, platelet availability varied, and deployed U.S. military medical personnel shifted progressively away from crystalloid toward more plasma and whole blood in factor-specific targeted resuscitation of combat casualties. Progressively smaller amounts of crystalloid infused may also have contributed to these results.42,43
Although compelling, our results are limited by the fact that fibrinogen was administered as cryoprecipitate, plasma, platelets, and whole blood, with each product containing other coagulation factors in varying amounts. Could those nonfibrinogen clotting factors—especially in plasma, because 239 of 252 casualties in our study received plasma—be responsible for the lowered mortality seen in the high F:R ratio group? Unfortunately, our sample size of 252 massive transfusion patients is not large enough to make valid comparisons between fibrinogen sources: plasma, cryoprecipitate, whole blood, etc., or to ascertain precisely where the mortality benefit came from: fibrinogen or from fibrinogen in addition to other clotting factors. A prospective randomized trial on massive transfusion casualties using cryoprecipitate, plasma, platelets, and whole blood as sole fibrinogen sources in separate treatment arms would come closer to answering that question; the likelihood of such a trial being conducted, however, may be remote for ethical reasons.
Although the sample size in this study is small, our data demonstrate that transfusion of a high F:R ratio (≥0.2 g of fibrinogen per red cell unit transfused) was independently associated with survival to hospital discharge, primarily by decreasing death from hemorrhage. One 15-mL cryoprecipitate bag containing 250 mg of fibrinogen can be transfused per unit of RBCs to achieve this ratio; plasma and whole blood seem to be just as capable of supplying the needed fibrinogen. Clinicians can meet this requirement by (1) transfusing 1 unit of FFP for every 2 units of red cells transfused, or (2) transfusing 1 unit of whole blood for every 4 units of red cells transfused (Table 1), or (3) transfusing one 10-unit bag of cryoprecipitate for every 10 units of red cells transfused. Whether the survival benefit of the increased F:R ratio is a result of fibrinogen alone or to fibrinogen working with the other clotting factors present in plasma and whole blood is unknown. More prospective studies are needed to evaluate the best source of fibrinogen and the optimal empiric ratio of fibrinogen to RBCs in patients requiring massive transfusion.
We thank Ms. Lindsey Stinger for assistance with data collection and Ms. Amy Newland for support, helpful discussions, and critical evaluation of this article.
1. Bellamy RF. The causes of death in conventional land warfare: implications for combat casualty care research. Mil Med.
2. Holcomb JB, Caruso J, McMullin NR, et al. Causes of death in Special Operations Forces on the modern battlefield: 2001–2006. Ann Surg.
3. Demetriades D, Martin M, Salim A, et al. Relationship between American College of Surgeons trauma
center designation and mortality in patients with severe trauma
(ISS >15). J Am Coll Surg.
4. Spinella PC, Perkins JG, McLaughlin DF, et al. The effect of recombinant activated factor VII on mortality in combat-related casualties with severe trauma
and massive transfusion
. J Trauma.
5. Borgman M, Spinella PC, Perkins JG, et al. The ratio of blood products transfused affects mortality in patients receiving massive transfusions at a combat support hospital. J Trauma.
6. Brohi K, Singh J, Heron M, Coats T. Acute traumatic coagulopathy
. J Trauma.
7. McLeod JB, Lynn M, McKenney MG, Cohn SM, Murtha M. Early coagulopathy
predicts mortality in trauma
. J Trauma.
8. Brohi K, Cohen MJ, Ganter MT, Matthay MA, Mackersie RC, Pittet JF. Acute traumatic coagulopathy
: initiated by hypoperfusion: modulated through the protein C pathway? Ann Surg.
9. Gonzalez EA, Moore FA, Holcomb JB, et al. Fresh frozen plasma should be given earlier to patients requiring massive transfusion
. J Trauma.
10. Holcomb JB, Jenkins D, Rhee P, et al. Damage control resuscitation: directly addressing the early coagulopathy
. J Trauma.
11. Malone DL, Holcomb J, Fingerhut A. Massive transfusion
practices around the globe and a suggestion for a common massive transfusion
protocol. J Trauma.
12. Phillips TF, Soulier G, Wilson RF. Outcome of massive transfusion
exceeding two blood volumes in trauma
and emergency surgery. J Trauma.
13. Wudel JH, Morris JA Jr, Yates K, Wilson A, Bass SM. Massive transfusion
: outcome in blunt trauma
patients. J Trauma.
14. Como JJ, Dutton RP, Scalea TM, Edelman BB, Hess JR. Blood transfusion rates in the care of acute trauma
15. Malone DL, Dunne J, Tracy JK, Putnam AT, Scalea TM, Napolitano LM. Blood transfusion, independent of shock severity, is associated with worse outcome in trauma
. J Trauma.
16. Huber-Wagner S, Qvick M, Mussack T, et al. Massive blood transfusion and outcome in 1062 polytrauma patients: a prospective study based on the Trauma
Registry of the German Trauma
Society. Vox Sang.
17. Peng R, Chang C, Gilmore D, Bongard F. Epidemiology of immediate and early trauma
deaths at an urban level 1 trauma
center. Am Surg.
18. Demetriades D, Murray J, Charalambides K, et al. Trauma
fatalities: time and location of hospital deaths. J Am Coll Surg.
19. Sauaia A, Moore FA, Moore EE, et al. Epidemiology of trauma
deaths: a reassessment. J Trauma.
20. Engstrom M, Schott U, Romner B, Reinstrup P. Acidosis impairs the coagulation: a thromboelastographic study. J Trauma.
21. Cosgriff N, Moore EE, Sauaia A, Kenny-Moynihan M, Burch JM, Galloway B. Predicting life-threatening coagulopathy
in the massively transfused trauma
patient: hypothermia and acidoses revisited. J Trauma.
22. DeLoughery TG. Coagulation defects in trauma
patients: etiology, recognition, and therapy. Crit Care Clin.
23. Ferrara A, MacArthur JD, Wright HK, Modlin IM, McMillen MA. Hypothermia and acidosis worsen coagulopathy
in the patient requiring massive transfusion
. Am J Surg.
24. McMullin NR, Holcomb J, Sondeen J. Hemostatic resuscitation. In: Vincent J, ed. Yearbook of Intensive Care and Emergency Medicine.
New York: Springer; 2006:265–278.
25. Hirshberg A, Dugas M, Banez EI, Scott BG, Wall MJ Jr, Mattox KL. Minimizing dilutional coagulopathy
in exsanguinating hemorrhage: a computer simulation. J Trauma.
26. Ho AM, Dion PW, Cheng CA, et al. A mathematical model for fresh frozen plasma transfusion strategies during major trauma
resuscitation with ongoing hemorrhage. Can J Surg.
27. Ketchum L, Hess JR, Hiippala S. Indications for early fresh frozen plasma, cryoprecipitate
, and platelet transfusion in trauma
. J Trauma.
28. Kelley DL. Update on plasma and cryoprecipitate
transfusion. Trans Med Update.
29. Hiippala ST, Myllyla GJ, Vahtera EM. Hemostatic factors and replacement of major blood loss with plasma-poor red cell concentrates. Anesth Analg.
30. Armand R, Hess JR. Treating coagulopathy
patients. Transfus Med Rev.
31. Collins JA. The pathophysiology of hemorrhagic shock. Prog Clin Biol Res.
32. Martini WZ, Chinkes DL, Pusateri AE, et al. Acute changes in fibrinogen
metabolism and coagulation after hemorrhage in pigs. Am J Physiol Endocrinol Metab.
33. Martini WZ, Chinkes DL, Sondeen J, Dubick MA. Effects of hemorrhage and lactated Ringer’s resuscitation on coagulation and fibrinogen
metabolism in swine. Shock.
34. Fries D, Krismer A, Klinger A, et al. Effect of fibrinogen
on reversal of dilutional coagulopathy
: a porcine model. Br J Anaesth.
35. Fries D, Innerhofer P, Reif C, et al. The effect of fibrinogen
substitution on reversal of dilutional coagulopathy
: an in-vitro model. Anesth Analg.
36. Perkins J, Schreiber M, Wade C, Holcomb J. Early versus late recombinant factor VIIa (rFVIIa) in combat trauma
patients requiring massive transfusion
. J Trauma.
37. Martini WZ, Holcomb JB. Acidosis and coagulopathy
: the differential effects on fibrinogen
synthesis and breakdown in pigs. Ann Surg.
38. Martini WZ, Dubick MA, Pusateri AE, et al. Does bicarbonate correct coagulation function impaired by acidosis in swine? J Trauma.
39. Martini WZ, Dubick MA, Wade CE, Holcomb JB. Evaluation of tris-hydroxymethylaminomethane on reversing coagulation abnormalities caused by acidosis in pigs. Crit Care Med.
40. Mannucci PM, Levi M. Prevention and treatment of major blood loss. NEJM.
41. Dempfle CE, Borggrefe M. Acidosis and impaired blood coagulation: what and how to correct before using recombinant human factor VIIa. Crit Care Med.
42. Cotton BA, Guy JS, Morris JA Jr, et al. The cellular, metabolic and systemic consequences of aggressive fluid resuscitation strategies. Shock.
43. Rhee P, Wang D, Ruff P, et al. Human neutrophil activation and increased adhesion by various resuscitation fluids. Crit Care Med.
Dr. Myung S. Park (Wilford Hall Medical Center, San Antonio, TX): I would like to thank Dr. Stinger and his co-authors for tackling the much debated issue of how much of which blood components do we give to an exsanguinating trauma patient requiring massive transfusion. In their review of 252 patients who received massive transfusions (10 or more RBC units within 24-hour period) at the combat support hospital in Baghdad, they found that the ratio of fibrinogen per unit of RBCs transfused (F:R ratio) differed significantly between survivors and nonsurvivors. Specifically, patients who received <200 mg of fibrinogen per unit of RBC had a mortality rate of 52% versus 24% in those who received >200 mg of fibrinogen per unit of RBC. The percentage of death by hemorrhage was even more striking with mortality rate of 85% versus 44%, respectively. Furthermore, in multiple logistic regression, they found that the high fibrinogen-to-RBC ratio (>200 mg/unit) was an independent additive predictor of survival. Thus far, this is the first clinical study that has shown the survival benefit of fibrinogen during massive transfusions. This study supports the important concept of damage control hemostatic resuscitation as currently practiced in military combat support hospitals where plasma (containing 400 mg fibrinogen per unit) is given in 1:1 ratio to RBCs.
Based on this study, the authors recommend that we give at least 200 mg fibrinogen per unit of RBC transfused. As the authors note, “Although compelling, our results are limited by the fact that fibrinogen was administered as cryoprecipitate, plasma, platelets, and whole blood, with each product containing other coagulation factors in varying amounts.” Because of their small sample size of 252 patients, comparisons between different fibrinogen sources were not made. Despite this shortcoming, I think this study delineates the importance of adequate fibrinogen supplementation in massive transfusions.
I have two questions: (1) Significant number of patients in this study received rFVIIa, especially in those who received the high F:R ratio. How can you attribute the improvement in survival to the latter when rFVIIa may be a contributor to increased survival? It is likely that both are important factors in improving the outcome of the exsanguinating trauma patients, but it is hard to discern the two based on the study presented. (2) Will the authors look into any mortality difference between patients who received fibrinogen from different sources, i.e., cryoprecipitate, plasma, whole blood, etc. using a larger cohort of patients?
Again, I would like to commend the authors for sharing with us their data that support fibrinogen is one of the key components in the resuscitation strategy.
Dr. Harry K. Stinger (Brooke Army Medical Center, Fort Sam Houston, TX): Thanks for your comments, Dr. Park. This analysis documented rFVIIa use versus nonuse in the 252 massive transfusion patients; it did not assess the timing or the dose of rFVIIa administered. In our analysis, the rate of rFVIIa use was similar (between survivors [51 of 177, 29%] and nonsurvivors [22 of 75, 29%]). However, a much higher percentage of patients in the high F:R ratio group (77 of 200, 38.5%) received rFVIIa than in the low F:R ratio group (5 of 52, 9.6%; p < 0.001). This issue is one limitation mentioned in the discussion section above, that is, was it the fibrinogen in the various blood products that positively impacted survival, or was it the other clotting factors present in those blood products—especially in plasma and in whole blood—that conferred the survival advantage? Recombinant Factor VIIa is a clotting factor that, along with fibrinogen, may have conferred a survival benefit.
At least in theory, rFVIIa needs adequate serum levels of fibrinogen and platelets in a normothermic, nonacidotic patient to work. I suspect that at least part of the survival benefit in the high F:R ratio category was the synergistic result of high serum levels of rFVIIa, plasma clotting factors, platelets and fibrinogen all working together to achieve hemostasis, and not the result of any single clotting factor operating alone. Separating out the effects of each factor individually is a challenge that would require a prospective randomized trial of massive transfusion casualties, hypothetically by assigning the various fibrinogen sources to separate treatment arms. Recombinant Factor VIIa use could then be evaluated by further subdividing each treatment arm into Factor VIIa use and nonuse categories, and looking at the effect of mortality in each group. Again, as stated in the article, the likelihood of such a trial being conducted is low for ethical reasons.
With regard to fibrinogen administration, based on these preliminary findings, we recommend transfusing a minimum of 200 mg of fibrinogen per red cell unit. Clinicians can meet this requirement by (1) transfusing 1 unit of FFP for every 2 units of red cells, (2) transfusing 1 unit of whole blood for every 4 units of red cells, or (3) transfusing one 10-unit bag of cryoprecipitate initially and then just before every subsequent 10th unit of red cells. Keep in mind that this is a preliminary recommendation based on this analysis only; we plan to analyze larger JTTR data sets in the near future to further refine this recommendation.