Ann Thorac Surg 2009;88:95-100. doi:10.1016/j.athoracsur.2009.04.047
© 2009 The Society of Thoracic Surgeons
Original Articles: Adult Cardiac
The Effect on Long-Term Survival of Erythrocyte Transfusion Given for Cardiac Valve Operations
Milo Engoren, MDa,d,*,
Robert H. Habib, PhDc,b,e,
Jonathan Hadaway, BSf,
Anoar Zacharias, MDc,b,g,
Thomas A. Schwann, MDb,g,
Christopher J. Riordan, MDb,g,
Samuel J. Durham, MDb,g,
Aamir Shah, MDb,g
a Department of Anesthesiology, Yvonne Viens, SGM, Research Institute, St. Vincent Mercy Medical Center, Toledo, Ohio
b Department of Cardiothoracic Surgery, Yvonne Viens, SGM, Research Institute, St. Vincent Mercy Medical Center, Toledo, Ohio
c Department of Cardiovascular and Pulmonary Research, Yvonne Viens, SGM, Research Institute, St. Vincent Mercy Medical Center, Toledo, Ohio
d Department of Anesthesiology, University of Toledo College of Medicine, Toledo, Ohio
e Department of Pediatrics, University of Toledo College of Medicine, Toledo, Ohio
g Department of Cardiothoracic Surgery, University of Toledo College of Medicine, Toledo, Ohio
f University of Toledo College of Medicine, Toledo, Ohio
Accepted for publication April 14, 2009.
* Address correspondence to Dr Engoren, Department of Anesthesiology, St. Vincent Mercy Medical Center, 2213 Cherry St, Toledo, OH 43608 (Email: engoren{at}pol.net).
 |
Abstract
|
|---|
Background: Studies in patients undergoing coronary artery bypass grafting (CABG) have shown an increased long-term mortality rates associated with perioperative blood transfusions. However, some studies in other patient populations have shown no effect on death or even a lowered mortality rate in patients receiving blood transfusions, which suggests that the effects of blood transfusion may be disease-dependent.
Methods: Data of all patients who underwent valve operations with or without associated CABG between October 2, 1991, and November 14, 2007, were obtained from the department's database and analyzed using logistic regression for 30-day and Cox models for long-term mortality to determine the effects of transfusion on death. To control for the potential interaction between transfusion and complications and sicker patients being more likely to receive blood, we separately analyzed the data for the different valve populations and used propensity analysis to control for sicker patients being more likely to receive blood.
Results: Of 1823 patients who underwent valve operations, the operation was isolated in 993 and combined with CABG in 830. By 30 days, 125 patients (6.9%) had died, and 717 (39%) were dead at follow-up. After controlling for type of operation and factors that influenced the transfusion decision, transfusion was associated with increased death only in patients who had combined valve and CABG, and not in isolated valve operations.
Conclusions: Transfusion had no effect on the mortality rate after isolated valve operations but was associated with increased mortality when valve operations were combined with CABG.
|
Introduction
|
|---|
Aortic and mitral valve operations commonly cause acute anemia with the risk of hemodynamic instability and organ ischemia, which may lead to organ dysfunction and failure or death if not reversed. Transfusions may provide immediate benefits of increased hematocrit, expanded blood volume, increased oxygen carrying capacity, and perhaps increased oxygen delivery; however, their effects, if any, on death are not obvious.
Several studies have evaluated the effects of erythrocyte blood transfusion on short-term mortality rates and found divergent results. Increased mortality was found in generalized intensive care unit (ICU) patients [1, 2], decreased mortality was found in septic patients [3, 4], and no effect was found in patients with hip fractures [5]. Similarly conflicting results have been found in patients with myocardial infarction: One study showed improved survival and others showed worse survival [6–8].
Long-term results are similarly variable. Several studies have shown that patients who undergo coronary artery bypass grafting (CABG) and receive transfusions intraoperatively or postoperatively are at increased risk of death during the next several years [9–12]. The studies, however, are not as consistent for patients undergoing other procedures or who have other diseases. Two studies evaluated transfusions in patients after operative repair of hip fractures [13, 14]. One found an association between transfusion and increased long-term mortality, the other study found no association. A recent study found that in a generalized ICU population that excluded cardiac operations, transfusions were associated with improved long-term survival [15].
The conflicting findings of improved survival, worse survival, or no effect on survival suggest that a variety of factors, including type of operation, may control the effect of transfusion on survival. We hypothesized that as a primary outcome, transfusion would be associated with an increase in the long-term mortality rate, and as a secondary outcome, transfusion would be associated with an increase in the short-term mortality rate. The purpose of this study was to determine the effect of erythrocyte transfusion on survival after aortic or mitral valve operations.
|
Patients and Methods
|
|---|
This study was approved on June 23, 2008, by the Institutional Review Board, which waived informed consent. The cardiac surgery database of St. Vincent Mercy Medical Center was queried for patients who underwent valve operations between October 2, 1991, and November 14, 2007. The database includes the Society of Thoracic Surgeons database, which includes demographic, comorbidities, risk factors, intraoperative, and postoperative data. The database was last updated for death in February 2008 using the Social Security Death Index.
All patients who underwent valve operations with or without CABG were included. Patients were divided into two groups: valve operations combined with CABG and valve operations without CABG. In each of the two groups, patients who received erythrocyte transfusions were compared with those who had not.
Because the Cox proportional hazard model requires that the risk of death be constant across the time analysis, a Kaplan-Meier plot was done of all patients to determine the time intervals for the succeeding analyses. The time of the elbow in the Kaplan-Meier curve, where the risk of dying changed abruptly, was used as the divider between early and late mortality. For the univariable analyses, groups were compared using the t test for normally distributed data, rank sum for nonnormally distributed data, and the 
2 and Fisher exact tests for categorical data. Binary logistic regression, used to analyze early mortality, was performed separately on each group using all variables with forward selection with p < 0.05 and 95% confidence interval (CI) excluding 1 to remain in the model and to denote statistical significance. Late mortality was evaluated using Cox proportional hazard models using all variables with forward selection with p < 0.05 and 95% CI excluding 1 to remain in the model and to denote statistical significance. If transfusion was not a statistically significant predictor of death, it was forced to remain in the models.
Transfusions have also been associated with an increased risk for complications such as lung injury, renal failure, and infections [16–18]. Because our database does not contain the temporal relationship between the development of complications and transfusions (ie, was transfusion given before or after the complication), and because the relative risk of early causative factors in a causal chain can be hidden by more proximal variables, we repeated all analyses, including the calculation of the propensity scores, excluding all complications [19]. Hence, both groups (valve only and valve plus CABG) were analyzed twice—once with all variables including complications and, again, without complications. In addition, the subpopulation of valve plus CABG patients who had no complications was analyzed separately.
Because transfusion is not a random event, but was determined by the physician, propensity scoring was used to compare transfused patients with nontransfused patients who were at similar likelihood of receiving a transfusion [20, 21]. Binary logistic regression was used to create a nonparsimonious model that gave a score between 0 and 1 to each patient, which is the likelihood that each particular patient received a transfusion. This propensity score for each patient was then included in the binary logistic regressions and Cox models as one of the variables examined [20, 21].
The power analysis was based on 50% mortality in the nontransfused group during the 0.1- to 16.2-year follow-up and that 50% of patients received a transfusion. To detect a hazard ratio (HR) of 1.3 while achieving a power of 80% with a two-sided significance level of 0.05, 786 patients were needed for analysis in each of the two groups [22].
|
Results
|
|---|
A total of 1823 patients underwent valve operations, of which 993 had an isolated valve operation and 830 had a valve operation combined with CABG. Kaplan-Meier plotting showed an elbow in the survival curve at 30 days (Fig 1); therefore, mortality was analyzed separately for the first 30 days (early mortality) and for time after 30 days (late mortality). By 30 days, 125 patients (6.9%) had died, and 717 (39%) were dead at follow-up (Appendix Table 1
*). Patients who died were older and more likely to have comorbidities, develop complications, and to receive transfusions (Appendix Table 2
* and Appendix Table 3
*). By univariable analysis, transfusion was associated with increased mortality at 30 days and at follow-up for patients having valve operations with or without CABG (Appendix Table 2
* and Appendix Table 3
*). Patients who received blood transfusions were older, smaller, more likely to have comorbidities, have longer cardiopulmonary bypass times, and to sustain complications (Appendix Table 1
*).
View larger version (32K):
[in this window]
[in a new window]
|
Fig 1. Kaplan-Meier survival plot of 15-year follow-up in all patients shows an elbow in the survival curve at 30 days, where the risk of dying changes. (A) Valve-only patients vs valve and coronary artery bypass grafting (CABG). (B) Transfused (XFN) vs non-transfused patients (No XFN).
|
|
View this table:
[in this window]
[in a new window]
|
Appendix Table 1 Selected Demographics, Comorbidities, Perioperative Processes, and Complications in 1803 Patients Undergoing Valve Operations
|
|
View this table:
[in this window]
[in a new window]
|
Appendix Table 2 Factors Univariably
a
Associated With Increased Risk of Dying Within 30 Days Postoperatively for Patients Undergoing Valve Operations or Valve Combined With CABG
|
|
View this table:
[in this window]
[in a new window]
|
Appendix Table 3 Factors Univariably
a
Associated With Increased Risk of Dying After 30 Postoperatively for Patients Undergoing Only Valve Operations or Valve Combined With CABG
|
|
In the valve only group, transfusion was not associated with increased risk of 30-day death after correction for other risks of dying (Table 1A) and regardless of analysis with and without complications. This lack of association was present when analyzed with and without complications. In contrast, blood transfusion was associated with an increased risk of 30-day death in patients undergoing combined valve and CABG procedures; however, this association became nonsignificant when postoperative complications were included in the model (Table 1B).
View this table:
[in this window]
[in a new window]
|
Table 1A Predictors of Death Within 30 Days After Valve-Only Operations With Transfusions Forced to Remain in the Models
|
|
View this table:
[in this window]
[in a new window]
|
Table 1B Predictors of Death Within 30 Days After Valve and CABG Surgery With Transfusions Forced to Remain in the Models
|
|
Cox hazard modeling showed that transfusion was not a predictor of long-term mortality in patients undergoing isolated valve operations—regardless of whether complications were included or excluded in the models (Table 2A,
Fig 2A and 3A). In patients undergoing combined valve and CABG procedures, however, transfusion was associated with an increased risk of dying (HR, 1.443; 95% CI, 1.133 to 1.838; p = 0.003; Table 2B,
Fig 2B and 3B). When complications were included in the model, transfusion no longer was a significant predictor of death. To further explore this, we separately analyzed the 304 patients who underwent combined valve and CABG procedures and had no postoperative complications. Here, transfusion was associated with an increased risk of death (HR, 1.843; 95% CI, 1.207 to 2.815; p = 0.005; Table 2C,
Fig 4).
View this table:
[in this window]
[in a new window]
|
Table 2A Predictors of Late Death, by Cox Modeling, in Patients Who Underwent Isolated Valve Operations and Survived at Least 30 Days With Transfusions Forced To Remain In The Models
|
|
View larger version (28K):
[in this window]
[in a new window]
|
Fig 2. Cox plots, shown as the mean of the covariates, for transfused (XFN) vs nontransfused (no XFN) valve-only patients. (A) Complications are excluded from the analyses. (B) Complications are included in the analyses. The grey line is XFN patients; the black line is no XFN patients.
|
|
View larger version (26K):
[in this window]
[in a new window]
|
Fig 3. Cox plots, shown as the mean of the covariates, for transfused (XFN) vs non-transfused (no XFN) valve and coronary artery bypass grafting (CABG). (A) Complications are excluded from the analyses. (B) Complications are included in the analyses. The grey line is XFN patients, and the black line is no XFN patients.
|
|
View this table:
[in this window]
[in a new window]
|
Table 2B Predictors of Late Death, by Cox Modeling, in Patients Who Underwent Combined Valve and CABG Operations and Survived at Least 30 Days With Transfusions Forced to Remain in the Models
|
|
View this table:
[in this window]
[in a new window]
|
Table 2C Predictors of Late Death, by Cox Modeling, in Patients Who Underwent Combined Valve and CABG Operations and Survived at Least 30 Days Without Complications
|
|
View larger version (25K):
[in this window]
[in a new window]
|
Fig 4. Cox plot, shown as the mean of the covariates, for the subset of transfused (XFN) vs non-transfused (no XFN) valve and coronary artery bypass graft (CABG) patients who had no postoperative complications.
|
|
There were many differences between patients who received transfusions and those who did not (Appendix Table 1
*). Including the propensity score, calculated separately for each group, in the models for both 30-day and long-term mortality produced no substantive differences in the association between transfusion and mortality.
|
Comment
|
|---|
Transfusion was associated with an increased risk of death only when valve operations were combined with CABG. This association was present for both early and late deaths, but disappeared when complications were added to the models. This suggests that in the combined valve and CABG group, transfusion is linked to and may cause complications, which then contribute to increased death. We found no association between death and transfusion when valve operations were performed without CABG.
Our finding of increased death with transfusion only when valve operations are combined with CABG suggests that the mechanism that causes increased death in isolated CABG [9–11] is not applicable in isolated valvular procedures. Although the putative mechanism for the increased mortality rate in CABG found in association with transfusion is generally ascribed to immunomodulation, the details of this have not been elucidated. It is possible that transfusion accelerates graft thrombosis, fibrointimal hyperplasia, or atherosclerosis, thus promoting earlier graft failure while having no effect on the valves. Transfusion stimulates the release of proinflammatory cytokines, and the resultant systemic inflammation may contribute to atherosclerotic plaque rupture and acute coronary syndrome [23, 24]. Randomized prospective studies of transfusion have shown opposite effects on death in generalized patients in the ICU compared with patients with sepsis [1, 2]. Further study is needed to explicate the harmful or beneficial effects of transfusion on outcome, any dependence on hemoglobin levels, and its effects in different diseases.
Although all other studies that evaluated the effect of transfusion on death in isolated CABG [9–11] or in predominantly CABG with a few valvular surgical patients [12]—without separate analyses for the valve patients—found that transfusion was associated with increased mortality, we found no such association with isolated valve operations. We could not find any other studies that evaluated the effect of transfusion on mortality after valve procedures.
This study has several limitations. First, this is a single-center study. Our results may not be generalizable to other institutions or other countries where the transfusion practices are different. Second, we do not have hemoglobin levels; thus, we cannot make any determinations about the relative effects of transfusion vs anemia on death, nor can we make recommendations about the appropriate threshold level of hemoglobin at which to transfuse. Further research should include prospective randomized studies that control for hemoglobin levels and other risk factors.
This study has several strengths. First, this study was conducted using the same methods and during roughly the same time period as our previous study that determined perioperative transfusion was associated with an increased mortality rate after isolated CABG [9]. Although we did not evaluate the effect of red blood cell storage time (which information was not available to us) on death, the blood bank policies and perioperative transfusion practices did not differ between patients undergoing isolated CABG, isolated valve operations, or CABG combined with valve operations. This ensures that the lack of effect of transfusion on death in isolated valve operations was not caused by shorter storage times of transfused blood to the valve patients and patients undergoing isolated CABG or CABG combined with valve operations receiving older, damaged blood. Our finding of a lack of a similar association for valve operations suggests that any effect of preoperative transfusion on death may depend on the type of operation and this may help to find the etiology of the transfusion-associated increase in death seen with CABG [9–12].
Another strength was separate models that included complications and excluded complications (Tables 1 and 2). Transfusions may be given because of the development of intermediate outcomes, such as respiratory or renal failure, or they may be given before their development and contribute to their genesis. By creating models both with and without complications, we were able to better separate out the effects of transfusion from the effects of the complications on death.
In summary, we found that transfusion had no effect on mortality rates in patients undergoing isolated valve operations, but was associated with an increased mortality rates when valve operations were combined with CABG.
|
Appendix
|
|---|
|
Acknowledgments
|
|---|
This study was supported solely by institutional funds.
|
Footnotes
|
|---|
The Appendix is available only online. To access it, please visit: http://ats.ctsnetjurnals.org and search for the article by Engoren, Vol. 88, pages 95–100.
* See note at end of article regarding e-only Appendix. 
|
References
|
|---|
- Hebert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care N Engl J Med 1999;340:409-417.[Medline]
- Vincent J-L, Sakr Y, Sprung C, et al. Are blood transfusions associated with greater mortality rates?. Results of the sepsis occurrence in acutely ill patients study. Anesthesiology 2008;108:31-39.[Medline]
- Vincent JL, Baron JF, Reinhart K, et al. Anemia and blood transfusion in critically ill patients JAMA 2002;288:1499-1507.[Abstract/Free Full Text]
- Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the treatment of severe sepsis and septic shock N Engl J Med 2001;345:1368-1377.[Medline]
- Carson JL, Duff A, Berlin JA, et al. Perioperative blood transfusion and postoperative mortality JAMA 1998;279:199-205.[Abstract/Free Full Text]
- Wu WC, Rathore SS, Wang Y, Radford MJ, Krumholz HM. Blood transfusion in elderly patients with acute myocardial infarction N Eng J Med 2001;345:1230-1236.[Medline]
- Rao SV, Jollis JG, Harrington RA, et al. Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes JAMA 2004;292:1555-1562.[Abstract/Free Full Text]
- Jani SM, Smith DE, Share E, et al. Blood transfusion and in-hospital outcomes in anemic patients with myocardial infarction undergoing percutaneous coronary intervention Clin Cardiol 2007;30:II49-II56.[Medline]
- Engoren MC, Habib RH, Zacharias A, Schwann TA, Riordan CJ, Durham SJ. Effect of blood transfusion on long-term survival after cardiac operation Ann Thorac Surg 2002;74:1180-1186.[Abstract/Free Full Text]
- Koch CG, Li L, Duncan AI, et al. Transfusion in coronary artery bypass grafting is associated with reduced long-term survival Ann Thorac Surg 2006;81:1650-1657.[Abstract/Free Full Text]
- Kuduvalli M, Oo AY, Newall N, et al. Effect of peri-operative red blood cell transfusion on 30-day and 1-year mortality following coronary artery bypass surgery Eur J Cardiothorac Surg 2005;27:592-598.[Abstract/Free Full Text]
- Murphy GJ, Reeves BC, Rogers CA, Rizvi SIA, Culliford L, Angelini GD. Increased mortality, postoperative morbidity, and cost after red blood cell transfusion in patients having cardiac surgery Circulation 2007;116:2544-2552.[Abstract/Free Full Text]
- Johnston P, Wynn-Jones H, Chakravarty D, Boyle A, Parker MJ. Is perioperative blood transfusion a risk factor for mortality or infection after hip fracture? J Orthop Trauma 2006;20:675-679.[Medline]
- Engoren M, Mitchell E, Perring P, Sferra J. The effect of erythrocyte blood transfusions on survival after surgery for hip fracture J Trauma 2008;65:1411-1415.[Medline]
- Engoren M, Arslanian-Engoren C. The effect of erythrocyte blood transfusion in the intensive care unit on long term survival Am J Crit Care 2009;18124–13.
- Schoenfeld DA. Sample-size formula for the proportional-hazards regression model Biometrics 1983;39:499-503.[Medline]
- Zacharias A, Habib RH. Factors predisposing to median sternotomy complications. Deep vs superficial infection. Chest 1996;110:1173-1178.[Abstract/Free Full Text]
- Iscimen R, Cartin-Ceba R, Yilmaz M, et al. Risk factors for the development of acute lung injury in patients with septic shock: an observational cohort study Crit Care Med 2008;36:1518-1522.[Medline]
- Weitkunat R, Wildner M. Exploratory causal modeling in epidemiology: are all factors created equal? J Clin Epidem 2002;55:436-444.[Medline]
- Rosenbaum PR, Rubin DB. The bias due to incomplete matching Biometrics 1985;41:103-116.[Medline]
- Rubin DB. Estimating causal effects from large data sets using propensity scores Ann Int Med 1997;127:757-763.[Abstract/Free Full Text]
- Koch CG, Li L, Sessler DI, et al. Duration of red-cell storage and complications after cardiac surgery N Engl J Med 2008;358:1229-1239.[Medline]
- Fransen E, Maessen J, Dentener M, Senden N, Buurman W. Impact of blood transfusions on inflammatory mediator release in patients undergoing cardiac surgery Chest 1999;116:1233-1239.[Abstract/Free Full Text]
- Libby P, Ridker PM, Maseri A. Inflammation, atherosclerosis, and coronary artery disease Circulation 2002;105:1135-1143.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
M. Ranucci, S. Aronson, W. Dietrich, C. M. Dyke, A. Hofmann, K. Karkouti, M. Levi, G. J. Murphy, F. W. Sellke, L. Shore-Lesserson, et al.
Patient blood management during cardiac surgery: Do we have enough evidence for clinical practice?
J. Thorac. Cardiovasc. Surg.,
August 1, 2011;
142(2):
249.e1 - 249.e32.
[Full Text]
[PDF]
|
|
|
|
|
|
|
M. P. Zalunardo, C. Thalmann, B. Seifert, J. D'Cunja, W. Weder, A. Boehler, and D. R. Spahn
Impact of preoperative right-ventricular function and platelet transfusion on outcome after lung transplantation
Eur J Cardiothorac Surg,
April 1, 2011;
39(4):
538 - 542.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
