Study design and population
All patients who underwent isolated coronary artery bypass graft surgery (CABG) or combined CABG and aortic valve replacement (AVR) from June 2011 until September 2014 in St. Antonius Hospital, Nieuwegein, The Netherlands, were eligible for inclusion in this historical cohort study. Local Medical Research Ethics Committee approval was obtained with a waiver for patient informed consent (MEC-U; Research and Development Department St. Antonius Hospital, trial number V15.020).
Data collection and potential confounding factors
Data regarding medical history and preoperative drug therapy were registered at the outpatient anesthesia clinic during routine preoperative screening. Standard laboratory hematology and coagulation parameters were retrieved from the hospital electronic patient records. Information regarding surgery, CPB, transfusion products and postoperative blood loss were collected from computerized perioperative medical records (Metavision Suite 5.46.44, iMDsoft®, Düsseldorf, Germany).
Prior to analysis, variables that were potentially associated with RBC transfusion were specified based on previous literature reports and biological plausibility including: age, sex, body mass index (BMI), preoperative Hb concentration, preoperative creatinine concentration, smoking, chronic obstructive pulmonary disease (COPD), hypertension, diabetes, peripheral artery disease, previous stroke, left ventricular ejection fraction (LVEF), previous cardiac surgery, unstable angina, emergency surgery, duration of operation time, autologous red blood cell transfusion, use of an internal mammary artery and type of surgery [11, 12].
Outcome measures
The primary outcome measure was postoperative mortality within 30 days after surgery, further denominated as mortality. Registration of mortality was conducted in the context of a national registry of cardiac interventions in The Netherlands (Netherlands Heart Registry) [13].
Clinical management
Anesthesia management and cardiopulmonary bypass
During the study period antiplatelet therapy (APT) was routinely discontinued 5 to 10 days prior to an elective surgical procedure. Routine perioperative anesthesia care included induction of anesthesia with midazolam, propofol, fentanyl and pancuronium and maintenance of anesthesia with propofol and fentanyl or remifentanyl. Vasoactive medications e.g. norepinephrine, dopamine, enoximone and nitroglycerine were used by discretion of the attending anesthetist. All patients received antimicrobial prophylaxis (cefazolin) at induction of anesthesia followed by additional cefazolin every 4 h for the duration of surgery and for 48 h in case of valve replacement. Intraoperative TEE as part of hemodynamic monitoring was routinely performed during valve surgery. In isolated CABG, use of TEE monitoring was left to the discretion of the attending anesthetist.
CPB consisted of a miniaturized extracorporeal circulation (MECC) or a conventional extracorporeal circulation (ECC) according to the preference of the surgeon and type of procedure. For CPB, non-pulsatile perfusion was used with a flow of 2.0 to 2.4 L/min/m2. All patients received heparin before CPB to achieve an adequate kaolin activated clotting time (ACT > 300 s for MECC and > 400 s for ECC). Additional heparin was administered when needed to keep ACT on target. After aortic cross-clamping, cardiac arrest was initiated using a cold crystalloid or blood (in case of MECC) cardioplegia solution. During CPB patients were cooled to a rectal temperature of 32 °C to 34 °C. Patients were weaned from CPB after rectal temperature reached 35.5 °C. In general, intraoperative management targeted a SvO2 of 65% and a MAP of 50 mmHg during CPB. According to institutional recommendations, heparin was reversed with protamine sulphate, 1 mg for every 100 U of previously administered heparin.
After surgery patients were transferred to the ICU and weaned from mechanical ventilation after exhibiting complete recovery from anesthesia, hemodynamically stability with no evidence of significant bleeding, core temperature > 36 °C and normal blood gas values. Patients were discharged from the ICU the following morning after meeting institutional discharge criteria.
Blood transfusion management
Blood product transfusion was performed according to a local transfusion protocol. The trigger for intraoperative RBC transfusion was a hematocrit < 20% during CPB or < 25% after separation from CPB. After surgery, RBC transfusion was at the discretion of the attending ICU physician and based on postoperative hemodynamics and chest tube output. A postoperative Hb value of 4.5 mmol/l (7.3 g/dL) was an absolute trigger for RBC transfusion.
Intraoperative plasma transfusion was based on the amount of blood loss; i.e. the number of transfused cell saver units or clinical signs of coagulopathy after protamine administration. Platelet transfusion depended on clinical signs of coagulopathy in combination with low platelet count (< 100 × 109/l) or continuation of APT during surgery. The final decision for blood product transfusion was at the discretion of the attending physician.
Administration of factor concentrates was not part of routine blood management. Point-of-care hemostatic monitoring was implemented in Q2 of 2015 as a part of routine blood management. Intraoperative tranexamic acid use was left to the discretion of the attending anesthetist. Intraoperative cell salvage was routinely used.
Statistical analysis
Categorical data are stated as number and percentages. Chi-square test or Fisher’s exact test were used to compare dichotomous variables between groups. Continuous data are described as median and interquartile range (IQR) as all continuous variables were non-normally distributed. The P-P plot was used to check for normality in distribution. Mann-Whitney U test or Student’s t-test was used to compare independent continuous variables between groups. To estimate the odds ratio for mortality in patients receiving intraoperative RBC transfusion a logistic regression analysis was performed using a propensity score (PS) as single confounder [14]. To define high-risk subgroups patients were stratified into four groups of equal size (quartiles) based on the sample distribution of the PS. Differences among quartiles were analysed using Mann-Whitney U test or Student’s t-test. A subgroup analysis was performed in the group of patients at highest risk of RBC transfusion (Q4) using multivariate regression analysis. A p value < 0.05 was considered statistically significant. For statistical analysis IBM SPSS version 22 (IBM Corp. Released 2013, SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp.) was used.
Propensity score
The PS was constructed beforehand and represented the likelihood of receiving RBC transfusion based on pre- and intraoperative characteristics. Variables that were potentially associated with RBC transfusion were included in a multivariable logistic regression analysis and included age, sex, BMI, smoking, COPD, hypertension, diabetes, peripheral artery disease, previous stroke, LVEF, previous cardiac surgery, unstable angina, emergency surgery, additive EuroSCORE, preoperative Hb concentration, duration of operation time, autologous RBC transfusion, use of an internal mammary artery and type of surgery.