| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Pediatric Anaesthesia and Intensive Care, Gottsegen György Hungarian Institute of Cardiology, Budapest, Hungary
b Department of Pediatric Cardiac Surgery, Gottsegen György Hungarian Institute of Cardiology, Budapest, Hungary
c Department of Pediatric Cardiology, Gottsegen György Hungarian Institute of Cardiology, Budapest, Hungary
d School of PhD Studies, Semmelweis University, Budapest, Hungary
e Department of Cardiology, Semmelweis University, Budapest, Hungary
f First Department of Pediatrics, Semmelweis University, Budapest, Hungary
g Uzsoki Street Hospital of the Budapest Municipality, Budapest, Hungary
Accepted for publication September 30, 2008.
* Address correspondence to Dr Székely, Gottsegen György Hungarian Institute of Cardiology, Haller u. 29, Budapest, H-1096, Hungary (Email: szekelya{at}kardio.hu).
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentReferences |
|---|
Methods: We retrospectively analyzed a prospectively collected database (January 2002 to December 2003) of 657 consecutive pediatric patients undergoing open heart procedures in a tertiary pediatric cardiac center. Risk models were calculated for each blood product and for the total amount of blood transfused during the operation and in the first 24 hours. Postoperative adverse events were investigated after propensity score adjustment.
Results: During the postoperative period, 30 patients (4.6%) died, 80 (12.2%) sustained nonvascular pulmonary complications, and 113 (17.2%) had infection. The risk model for the total amount of blood transfusion included weight, preoperative creatinine clearance, preoperative mechanical ventilation, duration of operation and cross-clamp, surgeon, delayed chest closure, inotropic dose, and nitric oxide administration. Univariate analyses demonstrated significant associations between blood transfusion and occurrence of every complication except of neurologic events. After adjustment for propensity score and disease severity, the total amount of blood transfusion was independently associated with an increased risk for infections (odds ratio, 1.01; 95% confidence interval, 1.002 to 1.02; p = 0.01). Transfusion of platelets was associated with lower incidence of nonvascular pulmonary complications (odds ratio, 0.89; 95% confidence interval, 0.79 to 0.99; p = 0.049).
Conclusions: The amount of blood transfusion is independently associated with infections but not with mortality.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Despite continuous technical and medical improvements in pediatric cardiac surgery, transfusion still remains unavoidable in most operations. Neonatal age, cyanotic heart disease, and pediatric cardiopulmonary bypass (CPB) circuits are known factors of increased need for transfusion [3]. Significant transfusion variation has also been noted with fresh frozen plasma (FFP), platelets, and cryoprecipitate in adults [4].
We have thought that the role of physicians and intraoperative factors must also be investigated in the use of different blood products during the perioperative period. In addition, we have assumed that the various blood products are not equally related to the development of complications. Therefore, we tried to build a risk model for determining the predictors of blood products transfusion, including total amount, red blood cells (RBC), FFP, and platelets. We also aimed to determine, using our prospectively collected database, whether the amount of blood products transfused is a predictor of major outcomes in pediatric patients undergoing open heart procedures.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Clinical Practice
Anesthesia was performed with fentanyl, midazolam, propofol, pancuronium, and inhalation of sevoflurane or isoflurane. CPB was performed with a non-heparin-coated circuit; a membrane, hollow-fiber oxygenator (body weight to 15 kg: Dideco D-901, D-902, Sorin Group Italia, Mirandola Modena, Italy; and body weight > 15 kg: Terumo SX-10, SX-18 and SX-25, Terumo Medical Corporation, Somerset, NJ), and roller pump (Jostra, Lund, Sweden) using standard technique according to age, weight, and hematocrit. The membrane oxygenator was primed with crystalloid and colloid solutions. Mannitol, albumin, or blood products were added to the circuit as needed.
The amount of RBCs was calculated by age, weight, and hematocrit value (target hematocrit value for neonates: 0.30; for infants between 5 and 10 kg body weight: 0.25; and for children above this weight: 0.20). FFP was used in the CPB circuit (10 mL/kg) solely in neonates. The flow rate was set to 2.0 to 2.4 L/min/m2, and hypothermia was induced according to the individual operation protocol. Intermittent anterograde cold crystalloid cardioplegia was used for myocardial protection. When necessary, deep hypothermic cardiac arrest was achieved by cooling to 20° to 28°C, depending on the anticipated duration of circulatory arrest.
Administration of aprotinin (Trasylol, Bayer AG, Germany) was at the discretion of the attending anesthesiologist. Further transfusion was indicated by the attending anesthesiologist. Unfractionated heparin was administered (300 to 400 IU/kg) and controlled with the activated clotting time, which had a target of 480 seconds. After termination of CPB, heparin was neutralized with protamine sulfate in a 1:1 ratio.
After separation from CPB, hematocrit concentrations were measured every 30 minutes in the operating room and every 3 hours after arrival at the ICU (and more frequently in bleeding or unstable patients). RBCs were transfused if the post-CPB hematocrit decreased to 0.35 in neonates, 0.30 in infants, and 0.25 in children. Leukocyte-reduced RBCs were used in patients aged younger than 3 months or if indicated by the hematologist. FFP was transfused in bleeding patients after neutralization of heparin. Indications for platelet transfusion included platelet count of less than 50 G/L, ongoing bleeding after reversal of heparin, or after prolonged CPB (> 120 min). All platelets were leukocyte-reduced.
At the ICU, bleeding was considered significant if blood loss exceeded 10 mL/kg/h or 4 mL/kg/h for 3 consecutive hours. RBCs were given if the hematocrit value was lower than 0.35 in neonates, 0.30 in infants, and 0.25 in children. In case of bleeding, FFP was administered (10 mL/kg initial dose) if the prothrombin value was less than 60% of the control value. If the bleeding persisted and was still critical, a pooled platelet infusion was transfused (10 mL/kg). During the study period, our institute used recommendations—but not protocols—for transfusion; thus, the final decision was left to individual discretion of the anesthesiologist or intensivist.
Dependent and Predictor Variables
Predictor variables included perioperative variables shown to be associated with perioperative blood loss (Table 1), as well as several variables related to adverse postoperative outcomes. Cardiac surgical procedures were graded as class 1 to 6 by complexity, applying the Risk Adjustment for Congenital Heart Surgery, version 1 (RACHS) category [6]. The cumulative index of inotropic support was quantified as the total inotropic dose proposed by Wernovsky and colleagues [7]. The score was adapted after the introduction of milrinone [8], and the modified score was the following: dopamine (µg/kg/min) + dobutamine (µg/kg/min) + 100 x epinephrine (µg/ kg/min) + 100 x norepinephrine (µg/kg/min) + 20 x milrinone (µg/kg/min).
|
Death was determined as in-hospital death after arrival at the ICU. Death at the transferred hospital was also recorded. Postoperative low output syndrome was defined as previously described [9]. Renal failure was defined as the need for peritoneal or hemodialysis support. Pulmonary failure was defined as noninfectious, nonvascular oxygenation problems including atelectasias, interstitial and alveolar edema, pleural effusions, pneumonia, or ratio of partial pressure of arterial oxygen and fraction of inspired oxygen (PaO 2/FIO 2) of less than 150 mm Hg after complete repair of circulation. Infection was defined as catheter-related, deep sternal wound infection, positive blood culture, or sepsis positive. Significant neurologic events were convulsion without prior history, hemorrhage, or infarction demonstrated on cranial imaging. All patient outcomes included in the database were determined separately by an intensivist and a cardiologist upon patient discharge from the hospital, except death. The two physicians were blinded for the purposes of the study.
Statistical Methods
The 
2 test was used for categoric factors such as pulmonary hypertension or reoperation. The Mann-Whitney U test was used for continuous data such as the duration of CPB or amount of blood transfusion. Correlation was assessed by the Spearman 
test.
We analyzed the predictors of transfusion of the different blood products with linear regression analysis. Square root transformation was used to normalize the distribution of the volume of blood products (RBC, FFP, and platelets), and the total amount of the intraoperative and postoperative 0- to 24-hour blood transfusion. The normalized sum of intraoperative and postoperative 0- to 24-hour blood product transfusion was the dependent variable, and separate univariate linear regression analyses were done with each preoperative and postoperative variable listed in Table 1. Then we entered variables with p < 0.2 in the multivariate linear regression model. A backward elimination procedure was used to remove covariates with adjusted values of p > 0.05 (Table 2). In the multivariate regression analyses, we assessed the coefficient of determination (R 2) and the role of multicollinearity using the variance inflation factor. Every variance inflation factor was less than 5, indicating an absence of severe multicollinearity.
|
Logistic regression analyses were performed with the presence of each major outcome (ie, death, need for dialysis, low cardiac output syndrome, infection, nonvascular pulmonary failure and neurological complications) as dependent variable and the volume of the different blood products as an independent variable adjusted for its propensity score in separate models (Table 3). Sensitivity analyses were performed to investigate the effect of unobserved covariates. The discriminative power of the propensity score was quantified by measuring receiver operation characteristic (ROC) area under the curve (C index).
|
|
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Of the patients studied, 21.5% did not receive any blood transfusion, and 13.2% received more than 100 mL/kg. These patients were younger, had more severe preoperative conditions, and underwent more complex operations with longer CPB and operation times. Their postoperative condition was characterized with delayed chest closure, positive fluid balance, need for nitric oxide treatment, and a five-times-higher dose of inotropic support. The median amount of total transfused blood products during the operation and the first 24 hours postoperatively was 33.2 mL/kg (interquartile range, 10 to 75 mL/kg). Figure 1 shows the distribution of blood transfusion during the operation and the first 24 hours postoperatively. Of the patients with more than 50 mL/kg of intraoperative blood transfusion, 50% required a large quantity of blood products in the postoperative period as well. Cyanotic patients required more transfusions throughout the study period (data are not shown). Patients undergoing repeat sternotomy received more FFP (p < 0.001) and platelets (p < 0.001) but not RBC (p = 0.158) compared with patients without previous operations.
|
show the distribution of transfused blood products according to age and weight categories. Each patient aged younger than 1 year received RBCs. FFP was given to 75.8% of the neonates, 43.4% of the infants (30–364 days), and to 29.7% of the patients aged older then 1 year. Those who received FFP had lower preoperative prothrombin values (p < 0.001). Platelets were transfused to 37.1% of the neonates, 15.6% of the infants, and 12.6% of the children aged older than 1 year. Patients with platelet transfusion had lower preoperative platelet count (p = 0.02) and prothrombin value (p = 0.002). The amounts of transfused blood products showed only weak correlation to the preoperative coagulation variables. Amounts of total blood transfusion correlated inversely to the platelet count on the day after the operation and to the change in platelet count from the preoperative value (r = –0.55 and r = –0.52, respectively; p < 0.001 for both).
|
|
Postoperative Complications
During the postoperative period, 30 patients (4.6%) died, 121 (18.4%) had low output syndrome, 41 (6.2 %) required dialysis, 80 (12.2%) sustained nonvascular pulmonary complications, and 113 (17.2%) had infection. Univariate analysis showed that every outcome variable, except for neurologic complications, was associated with the amount of blood transfusion. After propensity score adjustment, death, low output syndrome, nonvascular pulmonary failure, renal failure, and prevalence of infection exhibited an independent relationship to the total amount of blood transfusion (Table 3). Transfusion of RBC was also associated with the occurrence of every complication except for neurologic complications, whereas transfusion of FFP was associated with death, pulmonary failure, and infection.
We have analyzed the effect of the total amount of transfused blood products and the amount of the different blood products after adjustment for perioperative risk factors (Table 4). The amount (mL/kg) and the units of RBCs were independently associated with increased occurrence of renal failure requiring dialysis and occurrence of low cardiac output syndrome. These associations were not significant after adjustment for propensity. The total amount of transfused blood products (in mL/kg and in units) was associated with an increased incidence of infections (Table 4). The amount of FFP (measured in units) was related to increased incidence of pulmonary and infectious complications. Transfusion of platelets (measured in mL/kg) was associated with lower incidence of nonvascular pulmonary complications.
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Predictors of Blood Transfusion
The amount of perioperative blood transfusion cannot be solely determined by blood loss or preoperative hemoglobin levels, particularly in the pediatric cardiac population. The amount of blood transfusion in the first postoperative 24 hours is also not negligible beyond the intraoperative amount. The predictors of blood transfusion can be divided into patient-related, procedure-related, and process-related factors [12].
Several factors that might have contributed to increased blood loss were associated with age [13]. Neonates and infants have small body size and low body weight; therefore, they require a relative higher volume for the priming circuit [14]. Furthermore, the results of coagulation tests are more likely to be abnormal, and polycythemia further worsens the coagulation profile [15]. One study nevertheless reported that two-thirds of the patients between the age of 1 month and 1 year could be treated without any other postoperative transfusion except for RBC supplementation [16]. In addition, and probably as a confounding factor of blood transfusion, these patients often present with preoperative congestive heart failure and they undergo more complex operations, with consequently longer CPB duration.
Cyanotic patients with a hematocrit greater than 50% have prolonged coagulation time, a decreased level of fibrinogens, and thrombocytopenia [17]. At the same time, drugs with platelet inhibitor properties such as prostaglandin E1 and aspirin are frequently administered to cyanotic patients.
The preoperative stay in the ICU or angiographic procedure also leads to preoperative anemia and increased perioperative transfusion [18]. The presence of preoperatively diminished cardiac function and mechanical ventilation might trigger rapid transfusion to achieve optimal coagulation and optimal oxygen delivery as early as possible.
Deep hypothermic circulatory arrest, hypothermia, hemodilution, long duration of CPB, or a combination of these, are well-known procedure-related factors for bleeding [14, 15, 19]. Repeat sternotomy, urgency, and complexity of the operation can further increase the need for transfusion [20]. In our heterogenic pediatric cardiac population, predictors were very similar to those of the adults who need massive blood transfusion [21]. In accordance with other adult findings, the model predicting RBC and FFP transfusion performed better, and the model predicting platelet transfusion performed the worst [22]. The withdrawal of aprotinin from the market might also further increase the need for transfusion.
Blood transfusion differs between institutions and countries [23]. The individual approach in recognition and decision making in cases of excessive bleeding also has influence on blood loss. Despite the recent techniques in monitoring tissue oxygenation, low hemoglobin levels remained the trigger for transfusion [24]. Transfusion guidelines have been strengthened in recent decades in adult cardiac surgery [12], but the latest pediatric recommendations [25] still suggest that high hemoglobin levels are desirable, particularly in cyanotic patients [26]. A recent large, pediatric multicenter study confirmed that a decreased hemoglobin threshold of 7 g/dL resulted in a 96% reduction in the number of patients receiving transfusions and a 44% decrease in the number of administered RBC transfusions, without increasing the rates of multiorgan dysfunction syndrome [27]. Nevertheless, this multicenter study did not include patients with cardiovascular problems, and low hematocrit levels were reported to lead to a higher rate of neurodevelopmental problems 1 year later [28].
Adverse Events and Blood Transfusion
Despite differences among the guidelines and institutional protocols, a consistent finding is that excessive blood loss and blood transfusion are associated with increased risk of adverse complications. Perioperative RBC transfusion after adult cardiac procedures was associated with an increased risk of death during a 1-year follow-up [29]. Recently, the same relationship in critically ill children was also reported after adjustment for age, sepsis, and severity scores [30]. Our results did not confirm that, which can be explained by the low mortality rate in this population and the propensity score adjustment (ie, the adjustment for differences between the transfused and nontransfused patients); however, the total amount of blood products was related to postoperative complications.
Transfusion leads to substantial changes in the immune system and increases the occurrence of infections and recurrence of malignancies [11, 31]. Most "minor" transfusion reactions, such as fever, minor allergic reactions, and hypotension, occur frequently and remain underdiagnosed [32]. The transfusion of blood products is occasionally complicated by transfusion-related acute lung injury (TRALI) or a pulmonary leukoagglutinin reaction [33]. TRALI might be masked in this population because clinical signs such as low arterial oxygen tension can be confounded by incomplete repair, atelectasias, congestive heart failure, production of pleural fluid, or as a consequence of ischemia and reperfusion injury induced by CPB [34].
FFP transfusion calculated in units but not in mL/kg was independently associated with the occurrence of nonvascular pulmonary failure, whereas platelet transfusion calculated in mL/kg was associated with less pulmonary morbidity even after adjustment of other risk factors and propensity score. In accordance with our findings, platelet transfusion was not associated with increased mortality and morbidity after risk adjustment despite the higher incidence of complications in the transfused patients in adults who had undergone cardiac procedures [11, 21, 35]. Moreover, platelet co-transfusion was found protective for infection [11].
Our study may help to highlight the different results as a consequence of the calculation either in mL/kg or in units. We assumed that the calculation in units indicated the relationship between the increased number of blood donors and the risk of complications. Propensity score adjustment helped to balance the differences between the transfused and nontransfused population. In the presence of propensity score, RBC transfusion was no longer related to renal failure or low cardiac output syndrome, presuming that the patients who received a transfusion were sicker than those who did not. Therefore, the morbid events were associated with the preoperative and intraoperative variables and not with the transfusion per se.
The limitations and strength of this study deserve comments. First, it was a single-center study in a tertiary pediatric cardiac center. This limits the general application of the study for multidisciplinary pediatric care units or adult cardiac surgery. Only reoperations during the same hospital admission were excluded from our consecutive cohort. Transfusion practice was not strictly controlled and was left to the discretion of the attending anesthesiologist and intensivist. The use of cryoprecipitate, which is widely recognized to be very important to replace fibrinogen in significant bleeding cases, is not approved in our country. We investigated the exact amount and the transfused units of blood products during the operation and in the first 24 hours postoperatively. Unmeasured confounding variables may exist that may also influence the amount of transfused blood products.
Because data were retrospectively analyzed, causality cannot be determined. We applied a propensity score during the analysis of different outcomes to eliminate the bias resulting from the inconsistent distribution of transfusions [36]. We completed several tests without adjusting for multiplicity; thus, the nominal level of type I error might be greater than 5%. Further studies are necessary to confirm the findings of our exploratory study.
In conclusion, the total amount of transfused blood products was associated with increased occurrence of infections, but not with death, after adjusting for causes of blood transfusion and perioperative risk factors. Transfusion of platelets seemed to have a protective role in the incidence of nonvascular pulmonary failure. In the transfusion strategy, the independent role of FFPs and RBCs should be considered.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Related Article
This article has been cited by other articles:
|
|
 |
|
  T. J. Karsies, K. K. Nicol, M. E. Galantowicz, J. A. Stephens, and B. A. Kerlin Thrombotic Risk of Recombinant Factor Seven in Pediatric Cardiac Surgery: A Single Institution Experience. Ann. Thorac. Surg., February 1, 2010; 89(2): 570 - 576. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. M. Hoffman Invited Commentary Ann. Thorac. Surg., January 1, 2009; 87(1): 197 - 197. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
