Introduction

The concept of hemostatic resuscitation, which has focused on the empiric replacement of blood components has changed the management of trauma patients with ongoing hemorrhage. The foundation of this strategy is the early and aggressive utilization of fresh frozen plasma (FFP) for patients requiring massive packed blood red cell (PRBC) transfusion. The emerging literature suggesting improved survival for patients receiving high FFP:PRBC ratios both in the military and in the civilian populations has resulted in the widespread adoption of this strategy [1–18].

Although the research targeting resuscitation and coagulopathy reversal during exsanguination has been carried out predominantly in the trauma setting, the bleeding non-trauma patient presents similar challenges. Whether the concept of hemostatic resuscitation with balanced blood product replacement can be extrapolated to patients with major blood loss from non-traumatic etiologies is unclear and the impact of this strategy in massively transfused non-trauma patients remains unclear. We hypothesized that high FFP:PRBC transfusion ratios would be associated with improved survival in this population.

Methods

This study received IRB approval and has been performed in accordance with the ethical standards defined in the 1964 Declaration of Helsinki and its later amendments. All non-trauma patients treated by a surgical service receiving a massive transfusion at the Los Angeles County + University of Southern California (LAC + USC) Medical Center from November 2003 to September 2011 were identified. For the purpose of this study, massive transfusion was defined as 10 or more units of PRBC during the first 24 h after initiation of transfusion. Patient variables extracted included age, gender, need for uncrossmatched blood transfusion, time of initiation of transfusion, heart rate, systolic blood pressure, hematocrit, hemoglobin, INR, pH, pO2, pCO2 and base deficit at initiation of transfusion, total units of PRBCs and FFP received and time of transfusion of each blood component. The time to completion of massive transfusion was carefully documented. Ratios of FFP:PRBC transfused within 24 h of initiation of blood transfusion were subsequently calculated. In order to minimize the possibility of survival bias, which has been a significant concern when analyzing this time-dependent variable [14, 16, 19, 20], all patients that died in the first 24 h after initiation of transfusion were excluded. The primary outcome measure was in-hospital mortality.

Continuous variables were dichotomized using clinically relevant cut points: age (older than 55 versus 55 years or younger), hematocrit (<24 versus ≥24 %), hemoglobin (<8 versus ≥8 g/dL), INR (<1.3 versus ≥1.3) and base deficit (<−6 versus ≥−6). Continuous variables were compared using Student’s t test or Fisher’s exact test and differences between proportions were compared using Chi square test or Mann–Whitney U test. Patients were subsequently classified into four categories according to their FFP:PRBC ratio using the same criteria previously published for trauma patients [3]: low ratio (≤1:8), medium ratio (>1:8 and ≤1:3), high ratio (>1:3 and ≤1:2) and highest ratio (>1:2). To identify if the ratio was independently associated with mortality, a stepwise logistic regression model was created using all variables that were found on univariate analysis to be associated with mortality at p < 0.2. The FFP:PRBC ratio was entered into the regression model as an ordinal variable based on the classification previously described.

Plotting the mortality rates for each of the FFP:PRBC ratio groups, a significant trend towards decreased mortality was observed and confirmed with regression analysis reaching a plateau at a ratio of 1:3. Chi square was used to compare the different FFP: PRBC groups. When 1:3 was compared with higher ratios, no statistical difference was noted. Based on this finding, the study population was dichotomized into two groups according to this cut-off ratio. To explore the association of FFP:PRBC ratio and mortality, logistic regression analysis was performed adjusting for all factors that on univariate analysis were different between the two groups p < 0.05 as well as the independent predictors previously identified. All analyses were performed using SPSS for Windows 17 (Chicago, IL).

Results

During the 8-year study period, 13,682 surgical patients required a blood transfusion. Massive transfusion occurred in 908 (7 %) of the transfused patients. Among the massively transfused patients, 343 (38 %) were non trauma patients and 260 survived greater than 24 h, meeting the inclusion criteria (Fig. 1). The majority of patients were male (72 %) with a mean age of 44 ± 17 years. On admission, 17 % were hypotensive and 26 % were tachycardic (Table 1). The reason for transfusion was gastrointestinal (GI) bleeding in 56 % (145) of cases, followed by obstetric and postpartum bleeding in 23 % (60), unexpected bleeding during elective or emergency surgery in 11 % [29], and other causes in 10 % [26]. The overall mortality was 30 %. Out of a total of 145 patients that had a GI bleed, only 6 required surgery due to inability to control the bleeding endoscopically or with interventional radiology catheter based therapy. The majority (88 %, 53) of the postpartum patients required return to the operating room for definitive bleeding control.

Fig. 1
figure 1

Study outline

Full size image
Table 1 Comparison of patient characteristics
Full size table

After the study population was classified into four different groups according to their FFP:PRBC ratio, a statistically significant decrease in mortality was observed with increasing ratios, reaching a plateau after a ratio >1:3 was achieved (Fig. 2).

Fig. 2
figure 2

Mortality according to increasing FFP:PRBC ratios

Full size image

After stepwise logistic regression to identify factors associated with mortality in this population, the ratio of FFP:PRBC was found to be the second strongest independent predictor of mortality, with an odds ratio (OR) [95 % confidence interval (CI)] of 0.52 [0.37, 0.74], p < 0.001 (Table 2). The area under the ROC curve for this regression model was 0.76 (95 % CI 0.69, 0.82). A scatterplot using the FFP:PRBC ratio and the probability of survival of each patient according to this regression model was built and demonstrated a linear increase in the probability of survival as the ratio of FFP:PRBC increased.

Table 2 Stepwise logistic regression for mortality
Full size table

The study population was then dichotomized into two groups (FFP:PRBC ratio ≤1:3 versus >1:3). The two groups were compared for differences in baseline characteristics using univariate analysis and the results are displayed in Table 1. Crude mortality was significantly lower for patients with ratio >1:3 (26 versus 38 %, p = 0.050). After multivariable analysis, adjusted mortality remained significantly lower for those patients with a ratio >1:3 [Adjusted OR (95 % CI) = 0.308 (0.118, 0.803), p = 0.016] (Table 3).

Table 3 Adjusted odds ratio for mortality, stratified by FFP:PRBC ratio cut-off
Full size table

In order to evaluate if achieving the highest ratio (FFP:PRBC > 1:2) would provide additional survival benefit, the study population was subsequently dichotomized using this value as a cutoff. After multivariate analysis, survival was significantly higher for those receiving a ratio >1:2 [Adjusted OR (95 % CI) = 3.984 (1.345, 11.764), p = 0.012].

Discussion

This analysis demonstrated that increasing ratios of FFP:PRBC were associated with a significantly improved survival for non-trauma patients undergoing massive transfusion. Patients that received an FFP:PRBC transfusion ratio of more than 1:3 were significantly more likely to survive, with an adjusted odds ratio for survival of 3.24 (95 % CI 1.24, 8.47, p = 0.016). Patients in the highest FFP:PRBC ratio (>1:2) group had a 44 % relative reduction in mortality compared to those in the low ratio (<1:8) group.

These findings are consistent with the data emerging from the trauma literature suggesting that aggressive use of FFP in injured patients undergoing massive transfusion is associated with improved outcomes [1–17, 21]. The past decade has been marked by a significant paradigm shift regarding resuscitation strategies for injured patients with massive blood loss. The primary therapeutic goals for an exsanguinating patient, whether the etiology of the bleeding is traumatic or not, is surgical or endovascular control, which should be pursued concomitantly with an evidence based resuscitation strategy. This includes the restoration of blood volume and coagulation profile, while optimizing tissue oxygen delivery. Traditionally, crystalloid and PRBC infusion have been the main interventions used to achieve this goal, resulting in worsening of coagulopathy secondary to pro-inflammatory immunomodulation and dilution of clotting factors [22]. For crystalloids, recent evidence suggests a beneficial role for limited and goal directed crystalloid infusion strategy [23, 24].

For blood components, the evolving concept of hemostatic resuscitation strategy originated from a shift in transfusion practices during recent conflicts, with the use of fresh whole blood [25, 26] and the demonstration of improved survival with high FFP:PRBC ratios transfusion [2]. Multiple studies investigating the impact of FFP utilization in the outcomes of injured patients undergoing massive transfusion both in the military and in the civilian sector have been published and suggest that high FFP:PRBC ratios are associated with improved survival [1–17, 21]. The optimal blood component ratios for patients undergoing massive blood transfusion however remains subject of debate and whether or not the use of fixed-ratio blood component resuscitation is beneficial is still unresolved. Holcomb et al. compared two high-ratio transfusion strategies in a prospective randomized trial and found no significant early or delayed mortality in massively transfused patients with 1:1:1 or 1:1:2 ratios [18]. Nascimento et al. demonstrated that although feasible, the institution of a fixed-ratio protocol resulted in increased waste of plasma units [27]. More recently, the use of viscoelastic assay thrombelastography (TEG) to guide the use of blood products during massive transfusion for traumatic hemorrhage was found to be associated with a survival benefit compared to conventional coagulation studies [28]. This finding suggests that goal-directed transfusion therapy using TEG may be a superior benchmark against which to compare fixed-ratio blood component resuscitation.

The issue of hemostatic resuscitation for non-trauma patients requiring massive transfusion has been previously investigated in the subset of patients with ruptured abdominal aortic aneurysms (AAA). In a retrospective review of 128 patients requiring massive transfusion for ruptured AAA, Mell et al. demonstrated a statistically significant survival benefit for patients receiving FFP:PRBC ratios greater than 1:2, with an odds ratio of 4.23 (95 % CI 1.23–14.49, p < 0.003) [29]. Similarly, Johansson et al. found that for patients with ruptured AAA, a transfusion protocol that included proactive transfusion of FFP and platelets, with a 1:1 ratio of FFP:PRBC, was associated with improved survival and reduced the need for PRBC transfusion [30.] Those findings are in contrast to a study from Kauvar et al. investigating the impact of FFP utilization in the outcomes of 89 patients receiving massive transfusion for ruptured AAA, which demonstrated that the use of autotransfusion but not FFP was associated with increased survival [31.] In that study, patients with FFP:PRBC ratio >1:2 had comparable mortality compared to those with FFP:PRBC ≤1:2 (49 versus 40 %, p = 0.39) [31]. As data accumulates supporting the empiric and more aggressive utilization of FFP both in trauma and non-trauma patients with major bleeding, shifting to such a resuscitation strategy may increase the risk of potential complications associated with plasma transfusion, as has been demonstrated in non massively transfused trauma patients [32–34]. The optimal ratios, trigger for initiation and timing of replacement are unknown and warrant further investigation.

This study shares the limitations observed in many of the studies investigating the issue of plasma utilization in the resuscitation of injured patients. The retrospective nature of the study limits our ability to fully evaluate the relationship of these dynamic time-dependent variables and the outcome, as well as the possible contribution of ratio variability that may occur during the transfusion process. Several studies have highlighted the possibility of survival bias when investigating the FFP:PRBC ratios during massive transfusion [14, 19, 20]. The potential of survival bias exists because patients who survive longer are more likely to eventually receive higher quantities of FFP transfusion. This happens because traditionally FFP transfusion is started only after several units of PRBCs have been given. As a result, those patients that die early receive comparatively less FFP overall. Time-dependent covariate analysis has been used with success to address the issue of survival bias [14]. That type of analysis is only possible if hourly FFP:PRBC ratios are available, which was not the case in our population. Despite data suggesting that the association with survival is stronger than the potential survival bias [16], we attempted to minimize this possibility using a conservative approach and only including patients who survived the initial 24 h. Additional limitations to the present study include the inability to account for other unmeasured possible confounders such as the presence of liver cirrhosis, withdrawal of care, and ratio variation during the resuscitation period. More importantly, the study included a broad and heterogeneous group of patients with bleeding from gastrointestinal as well as gynecological and obstetrical bleeding. It is also important to highlight that cardiac surgery patients, which frequently require massive transfusion, were not included in the study as this is not a group of patients frequently treated at our institution.

Conclusion

In non-trauma patients undergoing a massive transfusion, an increasing plasma to red blood cell ratio was associated with improved survival. Achieving an FFP:PRBC ratio greater than 1:3 significantly improved the probability of survival. Further prospective validation is warranted.