TRANSFUSION PRACTICE Blood transfusion in cardiac surgery does increase the risk of 5-year mortality: results from a contemporary series of 1714 propensity-matched patients Richard E. Shaw,1* Christopher K. Johnson,1* Giovanni Ferrari,2 Mariano E. Brizzio,1 Kathleen Sayles,1 Nancy Rioux,1 Alex Zapolanski,1 and Juan B. Grau1,2*
BACKGROUND: Studies have found that cardiac surgery patients receiving blood transfusions are at risk for increased mortality during the first year after surgery, but risk appears to decrease after the first year. This study compared 5-year mortality in a propensity-matched cohort of cardiac surgery patients. STUDY DESIGN AND METHODS: Between July 1, 2004, and June 30, 2011, 3516 patients had cardiac surgery with 1920 (54.6%) requiring blood transfusion. Propensity matching based on 22 baseline characteristics yielded two balanced groups (blood transfusion group [BTG] and nontransfused control group [NCG]) of 857 patients (1714 in total). The type and number of blood products were compared in the BTG. RESULTS: Operative mortality was higher in BTG versus NCG (2.3% vs. 0.4%; p < 0.0001). Kaplan-Meier analysis of 5-year survival demonstrated no difference between groups in the first 2 years (BTG 96.3% and 93.0% vs. NCG 96.4% and 93.9%, respectively). There was a significant divergence during Years 3 to 5 (BTG 82.0% vs. NCG 89.3% at 5 years; p < 0.007). Five-year survival was significantly lower in patients who received at least 2 units of blood (79.6% vs. 88.0%; p < 0.0001). In multivariate Cox regression analyses, transfusion was independently associated with increased risk for 5-year mortality. Patients receiving cryoprecipitate products had a twofold mortality risk increase (adjusted hazard ratio, 2.106; p = 0.002). CONCLUSION: Blood transfusion, specifically cryoprecipitates, was independently associated with increased 5-year mortality. Transfusion during cardiac surgery should be limited to patients who are in critical need of blood products. [Correction added after online publication 02-Aug-2013: wording has been changed to improve clarity.]
TRANSFUSION Volume 54, April 2014
ardiovascular surgery accounts for 20% of blood transfusions that are given in the United States.1,2 Patients are more likely to be transfused during some combination of coronary artery bypass grafting (CABG) and valvular surgery compared to an isolated procedure, but there are increased complication and mortality rates in all transfused patients regardless of the type of surgery.3 Blood transfusion in cardiac surgery has been linked to increased risk for late mortality.3-7 The administration of 1 to 2 units of blood to patients under the age of 80 increases the risk for lowoutput heart failure by 27%, and both short- and long-term survival are decreased.3,7-9 The effect of blood transfusions on outcomes in older patients is not clear.10,11 The transfusion of blood adds significant risks for cardiac events, infection, and morbidity postoperatively even in patients who were deemed“low risk” preoperatively.12 In addition to transfusion-transmitted infections there are various noninfectious complications that can occur, including the acute transfusion reactions transfusion-related acute lung injury and transfusion-associated circulatory overload.13-15 ABBREVIATIONS: BTG = blood transfusion group; CABG = coronary artery bypass grafting; CHF = congestive heart failure; NCG = nontransfused control group; STS = Society of Thoracic Surgeons. From 1The Valley Columbia Heart Center, Columbia University College of Physicians and Surgeons, Ridgewood, New Jersey; and the 2University of Pennsylvania School of Medicine, Glenolden, Pennsylvania. Address reprint requests to: Juan B. Grau, MD, The Valley Columbia Heart Center, 223 North Van Dien Avenue, Ridgewood, NJ 07450; e-mail: [email protected]
*These authors contributed equally to this work. This work was presented as an oral presentation at the 49th Annual Meeting of the STS held in Los Angeles, CA, January 26-30, 2013. Received for publication April 16, 2013; revision received June 24, 2013, and accepted June 26, 2013. doi: 10.1111/trf.12364 TRANSFUSION 2014;54:1106-1113.
LONG-TERM OUTCOME OF BLOOD TRANSFUSION
There have been no randomized trials analyzing the effects of transfusion after cardiac surgery in the United States. In fact, the only prospective randomized clinical trial in cardiac surgery comes from Brazil. The Transfusion Requirements After Cardiac Surgery (TRACS) trial transfused patients to maintain hematocrit (Hct) levels of more than 30% versus more than 24%.16 This study found no difference in 30-day survival or complications between the two strategies; however, they found that the risk for complications increased with the number of red blood cell (RBC) units given. These results were not consistent with the results of our previous study, which demonstrated an increase in postoperative complications in patients receiving blood products across Hct levels.17 The current Society of Thoracic Surgeons (STS) and the Society of Cardiovascular Anesthesiologists clinical practice guidelines are largely based on the Transfusion Requirements in Critical Care (TRICC) trial.18-20 In this study, Hébert and colleagues18 randomized a series of Canadian critical care patients into a “liberal” or “restrictive” blood utilization strategy, this time using hemoglobin (Hb)-based transfusion triggers of 10 and 7 g/dL, respectively. They found that using the restrictive strategy was associated with increased survival while mortality was increased when patients with Hb levels of more than 10 g/dL were given blood products. However, it is important to note these findings were only significant in younger (75% stenosis of the carotid). Operative morRBCs + PLTs tality was defined as death within 30 days of surgery. Long15.5% term mortality was obtained from the Social Security Death Index. Pre- and postoperative Hct, Hb, PLT count, bleeding time (prothrombin time), and international normalized ratio were collected for both the transfused and the nontransfused patients. RBCs only RBCs+FFP 61.0% Categorical data are expressed as proportions. Uni6.0% variate statistical tests for continuous data included tests Fig. 1. Distribution of blood products used during of mean differences using the t test. Categorical variables hospitalization. were analyzed using the chi-square test. Univariate analysis and display of long-term survival was done with Kaplan-Meier survival curves and the Mantel-Cox log rank test. MulTABLE 1. Demographics and clinical characteristics for patients discharged and alive at 30 days* tivariate analyses to determine the Patient characteristics BTG (n = 837) NCG (n = 854) p value unique association of factors from the Sex (female %) 257 (30.7) 293 (34.3) 0.114 index hospitalization with late outcome Age (years) 69 ± 11 68 ± 11 0.114 was accomplished using Cox proporBody mass index 28.4 ± 5 28.7 ± 5 0.257 Smoker 467 (56.1) 460 (53.9) 0.425 tional hazards regression modeling. A Smoking current 116 (13.9) 130 (15.2) 0.363 value of p < 0.05 was used to determine Diabetes 250 (29.9) 264 (30.9) 0.641 the statistical significance of all tests Dyslipidemia 635 (74.1) 670 (78.2) 0.325 Renal failure 38 (4.5) 42 (4.9) 0.714 used. Analyses were performed using Dialysis 7 (0.8) 1 (0.1) 0.070 a statistical software package (SPSS, Hypertension 677 (79.0) 720 (84.0) 0.067 Version 19.0, IBM/SPSS, Inc., Chicago, Cerebrovascular accident 53 (6.3) 37 (4.3) 0.113 Cerebrovascular disease 111 (13.0) 79 (9.2) 0.214 IL).
Data collection and statistical analysis
Endocarditis Chronic lung disease Immunosuppressed Peripheral vascular disease Myocardial infarction Myocardial infarction within 7 days Unstable angina History of arrhythmia Heart failure Class III or IV Triple vessel disease Left main disease Ejection fraction Creatinine Preoperative anticoagulation STS operative mortality risk (%)
17 (2.0) 103 (12.0) 24 (2.8) 95 (11.4) 285 (34.1) 141 (16.5) 156 (18.2) 145 (17.3) 232 (27.7) 455 (53.1) 241 (28.8) 53.0 ± 12 1.11 ± 0.6 412 (49.2) 2.7
14 (1.6) 89 (8.9) 24 (2.8) 82 (9.6) 295 (34.5) 142 (16.6) 164 (19.1) 176 (20.6) 203 (23.8) 463 (54.0) 244 (28.6) 52.4 ± 12 1.05 ± 0.6 434 (50.8) 2.5
0.863 0.559 1.000 0.502 0.831 0.912 0.342 0.227 0.063 0.605 0.920 0.346 0.061 0.994 0.107
* Data are reported as number (%) or mean ± SD.
TABLE 2. Type of cardiac surgery performed* Surgery type Urgent or emergent operation Reoperation Isolated CABG Valve CABG + valve Other surgery
BTG (n = 837)
NCG (n = 854)
458 (53.4) 54 (6.5) 448 (52.5) 117 (14.0) 76 (9.1) 196 (23.4)
456 (54.7) 60 (7.0) 423 (49.4) 125 (14.6) 79 (9.3) 227 (26.6)
0.695 0.638 0.100 0.699 0.903 0.133
* Data are reported as number (%).
TRANSFUSION Volume 54, April 2014
RESULTS In the BTG group, most patients received RBCs, while a smaller percentage received fresh-frozen plasma (FFP), PLTs, or cryoprecipitate (Fig. 1). The overall rate of return to the operating room for bleeding was 3.3% and was significantly higher in the BTG (6.3% vs. 0.4%; p < 0.0001). Operative mortality occurred in 2.3% of BTG and 0.4% of NCG patients (p < 0.0001). Baseline characteristics and the type of surgery performed were similar between the 837 BTG and 854 NCG patients discharged alive and who survived to 30 days postsurgery (Tables 1 and 2). The discharge hematologic profiles showed no significant difference between BTG and NCG patients for Hct, Hb level, or prothrombin time, while PLT count was elevated in patients receiving
LONG-TERM OUTCOME OF BLOOD TRANSFUSION
TABLE 3. Admission and discharge hematologic profile* Blood characteristic Hct (%) Admission Discharge Hb (g/dL) Admission Discharge Prothrombin time (sec) Admission Discharge PLT count (×109/L) Admission Discharge International normalized ratio Admission Discharge
34.9 ± 6 28.9 ± 4
36.3 ± 5 29.2 ± 4