One of the recent advances in cardiac surgery has been the recognition that perioperative transfusion of blood and blood products is associated with an increased risk of postoperative morbidity and mortality [1–8]. As a result, many cardiac surgery programs have implemented blood conservation protocols designed to reduce the number of transfusions that patients receive in the perioperative period . Furthermore, the Society of Thoracic Surgeons (STS) and the Society of Cardiovascular Anesthesiologists (SCA) recently published guidelines, which highlighted the rationale and safety of blood conservation strategies in routine cardiac surgery . Despite the growing evidence supporting the benefit of reducing blood transfusions in routine surgical procedures, little data exist regarding the applicability of these principles to more complex cardiac surgical procedures, such as aortic surgery. Recently, a blood conservation program in place for cardiac patients undergoing revascularization and/or valvular surgery at our institution was extended to include all patients, including those undergoing complex aortic reconstruction. Here we report the usage of blood for a heterogeneous cohort of patients undergoing aortic surgery and its correlation with outcomes in these patients.
Materials and Methods
Information for 63 patients undergoing scheduled and emergent open aortic procedures between March 1, 2010 and October 1, 2011 were collected; these cases were limited to those of a single surgeon (A.D.) at our institution in order to eliminate the potential confounder of multiple operators. This retrospective review included all patients who underwent surgical repair; no patients were excluded from the analysis. As the institution of the blood conservation program occurred simultaneously with the initiation of an aortic practice of the single surgeon at this institution, no comparison was attempted for those patients undergoing surgery prior to the conservation program. The data analyzed included patient demographics and preoperative characteristics, intraoperative data including cardiopulmonary bypass (CPB) information and packed red blood cell (pRBC) usage, and postoperative outcomes including pRBC transfusions, major complications (respiratory failure, reoperation for bleeding, renal failure, sepsis, infection), and death. The focus of this study was in red blood cell usage, so transfusion of platelets, cryoprecipitate, and fresh frozen plasma was not addressed.
Our multidisciplinary approach to blood conservation included pre-, intra-, and postoperative optimization. Preoperative management included avoidance of medications (e.g., Clopidogrel, low-molecular-weight heparin) known to alter the bleeding profile of patients, as well as (when possible) optimization of the red cell mass with iron, folate, and, when indicated and possible, recombinant erythropoietin. Data relating to red cell mass optimization were not collected. Intraoperatively, we relied on meticulous operative hemostasis, intraoperative hemoconcentration, and tolerance of perioperative anemia (hemoglobin (Hg) ≥ 7 mg/dL). An antifibrinolytic (Amicar; Xanodyne Pharmaceuticals, Newport, KY, USA) was used in each case. Perfusion techniques of retrograde autologous priming and small priming volumes were incorporated. All cases utilized cell saver technology. Additionally, recognizing that a significant portion of aortic procedures require some degree of hypothermia with or without circulatory arrest, we chose to remain on bypass until normothermia (venous blood temperature > 37°C) was achieved. Patients underwent various arterial and venous cannulation approaches, depending on the planned procedure and aortic anatomy/pathology, and hypothermic circulatory arrest was used when deemed appropriate. In the postoperative period we continued with minimizing crystalloid usage. Blood product usage, including platelet transfusions, was based on objective evidence of the physiological need for transfusion rather than relying on automatic triggers or empirical or prophylactic tactics, but we would transfuse pRBCs for Hg < 7 mg/dL. Point-of care testing to quantify coagulation profiles was not performed. Reflexive transfusion of platelets for low platelet counts in patients who were not bleeding was avoided. Patients undergoing endovascular surgery for aortic disease were not included in this analysis. The protocol differed from previous practices in a variety of ways, but the most significant differences were tolerance of anemia, attempting to achieve normothermia, and minimizing crystalloid usage. Statistical analyses were performed with SPSS v19.0 software (SPSS Inc, Chicago, IL, USA). Values are expressed as mean ± standard deviation and percentages.
This retrospective study was approved by the New York University School of Medicine Institutional Review Board with a specific waiver of individual patient consent. Institutional Review Board exemption for deidentified patient outcome data analysis was obtained and the requirement for written informed consent was waived.
Patient demographics and characteristics are outlined in Table 1. Of the 63 patients, 31 patients were operated on for isolated aortic aneurysms which involved the aortic root (n = 7), and ascending (n = 13), arch (n = 6), descending (n = 3), and thoracoabdominal aorta (n = 2). Thirty-two patients had aortic dissections including 15 acute and 17 chronic dissections. The acute dissections were primarily Type A dissections (n = 14), whereas the chronic dissections were more evenly divided between Type A (n = 7) and Type B (n = 8). The one acute Type B dissection had radiographic evidence of a contained posterior leak (confirmed at operation), prompting emergency surgical intervention. Mean CPB time was 161 ± 75 min and mean aortic cross-clamp time was 97 ± 60 min. Thirty-one patients underwent a period of full circulatory arrest with a mean time of 25 ± 20 min and a mean core temperature of 16.0 ± 2.8°C. Of the 31 patients who required a period of hypothermic circulatory arrest, 18 underwent selective antegrade cerebral perfusion for a mean time of 18.0 ± 13.2 min; in the remaining 13 patients cerebral protection was afforded by hypothermia alone, with a mean cerebral ischemic time of 23.1 ± 9 min. Of the 63 patients, only 1 patient (undergoing emergency repair of a ruptured thoracoabdominal aortic aneurysm) received recombinant activated Factor VII (rFVIIa; NovoSeven, NovoNordisk, Copenhagen, Denmark).
The in-hospital mortality rate was 11.1% (7 patients) and the respiratory failure rate was 11.1% (7 patients). There were six (9.5%) reoperations for bleeding within the first 36 hours, and four patients (6.3%) developed renal failure requiring dialysis. Additional complications included gastrointestinal bleeding in two patients (3.2%), cerebrovascular accidents in two patients (3.2%), and sepsis in one patient (1.6%). The mean postoperative length of stay was 11.6 days (range 1-52 days). There were no sternal wound infections or myocardial infarctions.
Table 2 outlines the utilization of pRBCs and separates the transfusions by timing, i.e., either intraoperatively or postoperatively. The majority of patients in the intraoperative period did not require a pRBC transfusion, using our criteria. Although not included in the statistical analysis, the average platelet, fresh frozen plasma (FFP), and cryoprecipitate transfusion was (mean [range]) 1.03 U [0-10], 1.37 U [0-20], and 0.13 U [0-2], respectively. In addition, more than one third did not require pRBC transfusion in either the intraoperative or postoperative period. For those patients who were transfused, most required only 1 or 2 U. With this conservative approach, there was a noticeable decrease in mean hemoglobin and hematocrit levels with time; these data are presented in Table 3. A paired t test comparison of preoperative and discharge mean hemoglobin (12.1 versus 8.7, P < 0.001) and hematocrit (35.5 versus 26.3, P < 0.001) demonstrated a statistically significant decrease.
|0 U||44 (74%)||34 (54%)||24 (39%)|
|1 U||8 (13%)||10 (16%)||10 (16%)|
|2 U||2 (3%)||9 (14%)||11 (17%)|
|3 U||4 (6%)||4 (6%)||7 (11%)|
|4+ U||5 (4%)||6 (10%)||11 (17%)|
|Hemoglobin (mean ± SD)||12.1 ± 2.4||10.5 ± 1.7||8.7 ± 1.4*|
|Hematocrit (mean ± SD)||35.5 ± 6.8||30.3 ± 6.3||26.3 ± 4.4*|
Patients were subsequently divided into two groups based on their transfusion requirement (Table 4). Thirty-four (54%) patients required 0 or 1 U of pRBCs during their hospitalization, while 29 (46%) patients required two or more units of PRBC. Patients requiring 0-1 U of pRBCs were more likely to have suffered no complication (94.1% versus 59%, p < 0.001), and were less likely to have suffered a major complication [see Appendix A for definitions of complications], including respiratory failure, renal failure, sepsis, and mortality (0% versus 38%, P < 0.001). Further analysis for the patients receiving transfusions demonstrated that major complications were only found with transfusion of >2 U of pRBCs. Backward stepwise logistic regression analysis was performed to define predictive variables for blood transfusion requirements. The two significant variables were prolonged CPB time (P = 0.015) and low preoperative hematocrit (P = 0.042). Multivariate predictors of mortality included endocarditis (P = 0.021) and low preoperative hematocrit (P = 0.05).
|0-1 U of pRBCs (n = 34)||2 or more units of pRBCs (n = 29)||p-value|
|No complications||32 (94%)||17 (59%)||0.001|
|Major complications (respiratory/renal failure, sepsis, mortality)||0 (0%)||11 (38%)||<0.001|
Table 5 outlines a comparison between our aortic aneurysm procedures during the study period (n = 31) and available data from the STS database for 2010. In our cohort, 80.6% of patients required ≤ 1 intraoperative unit of PRBCs compared to 54.3% in STS benchmark data, which was statistically significant (P < 0.0001).
It has been appreciated that cardiac surgery is associated with significant usage of blood and blood products, comparable to that used in orthopedics and trauma surgery on a per patient basis. Recent evidence and consensus statements have advocated for a shift in the paradigm regarding transfusion practices in cardiac surgery, but there remains wide variability in such practice, even when comparisons are controlled and matched. For “routine” cases where bleeding (and, therefore, need for transfusion) would not be anticipated, reflexive practices may be engrained into clinical practice, such as transfusing to a hemoglobin of > 0 mg/dL or administering the traditional “round” of products (e.g., FFP, cryoprecipitate, platelets) for the postoperative patient who appears coagulopathic.
For procedures that have been characterized as more likely to result in bleeding, including complex aortic procedures, blood conservation practices have been questioned, despite growing evidence that transfusion and/or the need for transfusion is associated with worse outcomes. Cambria et al.  demonstrated that for repair of thoracoabdominal aortic aneurysms, pRBC transfusion requirement was one of two independent correlates of mortality (odds-ratio 1.4, P = 0.005). It is difficult to assess how much blood and blood products are given to any patient since studies often refer to intraoperative transfusion or postoperative chest tube output without a unifying benchmark. In a similar but smaller study, Rahe-Meyer et al.  found that with a thromboelastogram (TEG)-directed transfusion protocol, blood product utilization in the first 24 hours postoperatively decreased from 16.4 to 2.5 U; two thirds of the protocol patients did not require any product in the first postoperative day. The authors attributed this to TEG-directed intraoperative administration of fibrinogen concentrate.
In this study, we sought to determine the usage of blood for a heterogeneous cohort of patients undergoing aortic surgery. Our results are consistent with other reports of decreased postoperative morbidity associated with fewer pRBC transfusions. In this study, 94.1% of patients who required 0 or 1 U of pRBCs suffered no perioperative complications. While this retrospective study may not prove causation, it certainly highlights the relationship between transfusions and complications in this group of patients. As cardiac surgeons and the physicians and nurses who care for these patients perioperatively gain experience with blood product conservation in routine cases, we believe that these practices can be safely extended to more complicated cardiac surgeries, including complex aortic reconstruction. Presumably, the benefits afforded to patients who undergo “routine” cardiac surgery without transfusion would be extended to this patient group as well. As such, we would suggest that the type of procedure performed may not be an appropriate part of a transfusion algorithm. Likewise, the empiric use of blood-product derivatives as prophylaxis may not be necessary. Goksedef et al.  described their experience with complex aortic surgery, propensity matching patients treated with and without rFVIIa for those patients found to be coagulopathic following aortic surgery. In their study, there was a significant decrease in bleeding requiring exploration and need for blood transfusion in patients who received rFVIIa given for refractory bleeding. This reduction translated to a decrease from 9 ± 4 to 7 ± 2 U of pRBCs. In our current study, we had one patient with refractory bleeding.
In addition, the current study demonstrates that the application of the principles of a blood conservation program to patients undergoing complex procedures will decrease the number of transfusions these patients receive. In this challenging group of patients, only 10 patients (16%) required >2 U of pRBCs in the postoperative period. While intraoperative measures to conserve blood will reduce the need for postoperative transfusions, it is likely that the tolerance of perioperative anemia was the most significant factor, as evidenced by the fact that the mean discharge hemoglobin was 8.7 mg/dL. This change in practice requires a collaborative effort among all members of the surgical and intensive care unit team. In a different environment, without open communication between the surgeons, intensivists, nurses, and other team members, many of these patients likely would have received a transfusion prior to discharge.
As is inherent in most retrospective observational analyses, the current study is not without limitations. Perhaps most important is the difficulty in determining exactly why any given patient was transfused. Despite the implementation of the aforementioned blood conservation guidelines, invariably some patients will receive transfusions who may not have needed them. Likewise, some patients who were otherwise stable may not have been transfused despite a perioperative hemoglobin value that was below the standard threshold. With retrospective assessment of the available data, it is difficult to control for all of the variables that impact a clinical decision.
Perhaps the major lesson from reviewing data such as these is to remind ourselves of the potential harm of giving a transfusion to a patient both during and after cardiac surgery. In addition, we would caution that the indiscriminate transfusion of patients after complex aortic procedures may not be beneficial. Follow-up of this group of patients over the coming years will allow us to further evaluate the effect of perioperative transfusions on their long-term outcome.
Recent STS/SCA guidelines highlight the advantages of implementing a blood conservation program for routine cardiac surgery cases. This study shows that a previously implemented blood conservation program for routine cardiac surgery can be safely extended to complex aortic procedures. The implementation of such a program can decrease the number of transfusions in this patient population and is associated with improved outcomes in these complicated patients.