IJCS | Volume 32, Nº6, November / December 2019

567 Tagliari et al. Blood transfusion in cardiovascular surgery Int J Cardiovasc Sci. 2019;32(6):565-572 Original Article In order to identify variables associatedwith outcomes, univariate analyses were performed. Then, variables with statistical significance were added to a multivariable logistic regression model. A bilateral p-value < 0.05 was considered significant. Based on the study by Murphy et al., 16 in which the observed 30-daymortality was 5% in those who received RBC and 1% in those who did not, a sample size of 164 patients was estimated considering an α value of 0.05 and a desired power of 0.80. Results Of the 271 patients evaluated, 100 (37%) required transfusion of some blood product during transoperative [87 (32.1%)] and/or postoperative periods [53 (19.5%)]. Baseline characteristics are shown in Table 1. Transoperative bleeding ≥ 500 mL was observed in 11 patients who did not require transfusion (6.4%) and in 40 (40%) who required it. The other transoperative characteristics are presented in Table 2. The type of transfusion stratified by the operative period is displayed in Table 3. Predictors of RBC transfusion in univariate and multivariate analyses are shown in Table 4. Reintervention was indicated in 21 patients (7.7%), 11 (52.3%) of them due to increased postoperative bleeding or cardiac tamponade, all of them required blood transfusion. Patients who received blood transfusion presented higher early postoperative complication rates, as described in Table 5. Discussion Dilutional anemia, CPB-related thrombocytopenia, coagulation disorders due to medications, major surgery and hypothermia contribute to high rates of blood transfusion in cardiac surgery, despite a current trend for restrictive protocols. In our sample of 271 patients, prospectively and consecutively enrolled, we observed a general transfusion rate of 37%; 28.8% of them required RBC, 22.1% platelets and 14.4% FFP, values similar to those in the literature. 1,5,6,9,10 We emphasize that, in our hospital, decisions about transfusion are a jointly made by the surgical, anesthetic and intensive care teams, who try to follow the best current evidence, but also take into account their experience and understanding of each situation, making the numbers presented here closer to real life. We observed that some preoperative characteristics were transfusion predictors, such as anemia and anormal coagulation tests [activated partial thromboplastin time (aPTT) and international normalized ratio (INR)], although platelet count was not. Predictors of blood transfusion in our analyses were similar to the TRACS trial, which showed higher transfusion rates in patients with previous cardiac surgery (OR = 8.92; p = 0.04), longer duration of cardiopulmonary bypass (OR = 1.01; p = 0.03) and lower preoperative hemoglobin levels (OR = 0.51; p = 0.001), 9 and to Stevens et al., 17 who reported as predictors: CKD, previous cardiac surgery, urgency surgery, ejection fraction, type of surgery, CPB duration, age and low body mass index, although the last two was not confirmed in our study. We observed no difference in other infections rates, although higher bronchopneumonia rates have been observed in transfused patients, similarly to TRACS, TRITe2 and Horvath findings. 9,10,18 We also found longer ICU and hospital lengths of stay, prolonged need for mechanical ventilation, delirium, ARF, acute CKD, stroke and transient ischemic attack in transfused patients. A retrospective Brazilian study on patients with ischemic and valve diseases had already reported higher rates of respiratory infection (27.8% vs. 17.1%; p < 0.001), ARF (14.5% vs. 7,3%; p < 0.001), stroke (4.8% vs. 2.6%; p = 0.001) and longer hospital stay (13 ± 12.07 days vs. 9.72 ± 7.66 days; p < 0.001) in transfused patients, but no difference in mortality. 19 Even though the baseline characteristics were significantly different between the two groups, with transfused patients sicker than not transfused patients, when analyzing transfusion-related mortality, adjusted for the main confounding factors (EuroScore II, age, cardiopulmonary bypass time ≥ 90 min, emergency or urgency surgery and combined surgery), we noticed that transfusion remained an independent predictor of mortality inmultivariate logistic regression (OR 5.3; 95% CI 1.3 – 21; p < 0.001). Therefore, we emphasize the importance of rethinking the almost routine decision on transfusion in cardiac surgery, taking into account that, even one unit of packaged RBCs, canworsen the postoperative outcomes.