Secondary Logo

Journal Logo

Pediatric Anesthesia: Society for Pediatric Anesthesia

Factors Associated with Blood Loss and Blood Product Transfusions

A Multivariate Analysis in Children After Open-Heart Surgery

Williams, Glyn D. FFA(SA); Bratton, Susan L. MD, MPH; Ramamoorthy, Chandra FRCA

Author Information
doi: 10.1213/00000539-199907000-00011
  • Free


In children, bleeding and transfusion of blood products after cardiopulmonary bypass (CPB) can be substantial [1-4] and are important causes of morbidity and mortality [5]. The coagulopathy associated with pediatric open-heart surgery (OHS) is complex and is influenced by age, surgery, and disease-related factors [1,4-7]. Knowledge of the factors associated with excessive perioperative blood loss after pediatric OHS may identify patients at risk. However, the few prospective pediatric studies with multivariate analyses provide limited information [1,8], and conclusions from investigations in adults [9-11] may not be applicable because of the many unique aspects of pediatric OHS [6].

Since 1995, we have prospectively collected information relevant to the patient's perioperative coagulation status to better understand the effects of hemostasis on OHS in children. Analysis of data from 414 children demonstrated that blood loss and transfusion requirements were inversely related to age [4]. Because many possible risk factors for bleeding were found to vary with age, we did not attempt an analysis of risk factors for bleeding until the number of patients entered in our database had increased and was similar to the numbers enrolled in equivalent studies in adults [9].

The aim of our prospective cohort study of children undergoing OHS was to identify demographic and perioperative factors associated with perioperative blood loss and blood product transfusions.


After institutional review board approval, information pertaining to perioperative hemostasis and its management was obtained prospectively for 548 consecutive children undergoing OHS and was recorded in a database. Some of the patients have been included in other studies [4,12].

Perioperative management has been reported previously [12]. Briefly, anticoagulants were discontinued preoperatively (aspirin >or=to7 days, coumadin >or=to3 days, and heparin >or=to12 h before surgery). Anesthetic technique consisted primarily of fentanyl (25-100 [micro sign]g/kg), midazolam (0.1-0.4 mg/kg), and muscle relaxants (vecuronium 0.1 mg/kg and/or pancuronium 0.1 mg/kg), supplemented with volatile anesthetics as needed. Anticoagulation was established with an initial bolus (infants <1 yr: 400 U/kg; children >1 yr: 300 U/kg) of porcine heparin; additional heparin was administered during CPB to maintain celite activated clotting time >480 s. Anticoagulation was empirically reversed using an initial protamine dose of 3-5 mg/kg.

Nonpulsatile CPB was performed with a hollow fiber membrane oxygenator. The CPB circuit contained heparin (3.2 +/- 0.9 U/mL, dose depended on prime volume), and prime volumes ranged from 375 to 2050 mL depending on patient size. When necessary, blood was added to maintain a hematocrit (Hct) near 20% during CPB. Hypothermia was induced in all patients. Blood conservation techniques included modified veno-venous ultrafiltration (for infants) and red blood cell salvage.

Laboratory tests performed before surgery included prothrombin time (PT) and international normalized ratio (INR), activated partial thromboplastin time (PTT), platelet count, fibrinogen concentration, and Hct. Before CPB, thromboelastograph (TEG) and ACT were performed. Fifteen minutes after initiation of CPB, TEG (with in vitro addition of protamine), platelet count, fibrinogen concentration, Hct, and ACT were performed. Finally, on admission to the intensive care unit (ICU), PT, INR, PTT, platelet count, fibrinogen concentration, Hct, ACT, TEG, thrombin time, and D-dimers were performed. TEG values measured included reaction time (R), coagulation time (K), angle (alpha), maximal amplitude (MA), and whole blood clot lysis index at 30 min (LY30).

Intraoperative blood loss per kilogram of body weight was calculated from swab weights, discard suction volumes, chest tube (CT) output, and volume of salvaged washed red cells. Postoperative blood loss per kilogram of body weight was calculated as running totals from the CT output 0, 2, 6, 12, and 24 h after arrival in the ICU.

Blood components were administered to treat excessive microvascular (coagulopathic) bleeding. The decision to transfuse was based on measured blood loss and intraoperative visual assessment of the surgical field. Component therapy was guided by Hct and laboratory coagulation tests. During the immediate post-CPB period, platelet transfusion (1 U/10 kg) was administered if the platelet count on CPB was <100,000 mm-3, and fresh-frozen plasma (10-20 mL/kg) or cyroprecipitate transfusion (1 U/5 kg) was considered if the fibrinogen concentration on CPB was <100 mg/dL. Further administration of blood products was guided by posttransfusion coagulation tests and followed previously published recommendations [5]. Postoperative minimal acceptable Hct values ranged from 20% to 45% depending on the presence of cyanosis and complexity of the surgical repair. Whole blood (>48 h since donation) was used in the early post-CPB period; fresh whole blood [1] was not available. Packed red blood cells were transfused if whole blood was not available or if the volume of blood to be administered was a concern. The volume and units of blood products transfused intraoperatively (including blood added to the CPB prime) and during the first 72 h after surgery (per kilogram of body weight) were noted.

The following perioperative variables were evaluated: patient age, height, weight, gender, and blood type; preoperative anticoagulant therapy; congestive heart failure (CHF); complexity of surgery; previous thoracotomy; repeat sternotomy; surgeon; anesthesiologist; perfusionist; total heparin (U/kg); total protamine (mg/kg); prophylactic antifibrinolytic therapy (epsilon-aminocaproic acid), duration of CPB; aortic clamping and deep hypothermic circulatory arrest (DHCA); prime type (blood or crystalloid); degree of hemodilution during CPB; minimal core temperature during CPB; use of modified veno-venous ultrafiltration or cell salvage; perioperative amrinone therapy; core temperature; and CT output on arrival in the ICU (CT 0h). CHF was defined as the preoperative use of at least two of the following medications: digoxin, diuretics, vasodilators, or IV inotropic drugs. Preoperative anticoagulation therapy was defined as the administration of aspirin, heparin, or coumadin. Complex surgery was defined using Manno et al.'s [1] classification (modified) and included arterial switch operation, truncus arteriosus repair, stage 1 palliation for hypoplastic left heart syndrome, Fontan procedure, modified Glenn shunt, and Ross procedure. We previously demonstrated that blood loss correlates inversely with age [4]; hence, patients were divided into two age groups (<or=to1 yr, >1 yr of age). Age was also examined as a continuous variable and as the natural logarithm of age because it was not normally distributed before log transformation.

Independent continuous data in the age groups were evaluated by using independent t-test for normally distributed data (expressed as mean +/- SD) and the Mann-Whitney U-test for non-normally distributed data (expressed as median [25th-75th quartile]). A multivariate analysis of variance was used to analyze serial measurements of CT drainage. Perioperative variables and laboratory tests (at described sample times) were initially evaluated against intraoperative blood loss (mL/kg); CT output (mL/kg) 2, 6, 12, and 24 h postoperatively; and blood products transfused (mL/kg and U/kg) by using Pearson's correlation. Significance was defined as P < 0.05. Variables that were significantly related in the univariate analysis were then evaluated by multiple stepwise linear regression to CT output or component transfusions. CT drainage at multiple time points was also evaluated with a dichotomous age variable (<or=to1 yr and >1 yr) in the models to determine whether risk factors for bleeding differed for infants (<or=to1 yr of age) compared with older children. Models were tested with and without CT 0h as a variable. Criteria for variable inclusion in the stepwise regression analysis were entry if P <or=to 0.05 and exclusion if P > 0.1.

As suggested by Sumner and Stark [13], excessive blood loss was defined as measured intraoperative loss >or=to50% of patient's estimated blood volume (EBV) or postoperative CT drainage >or=to20% of EBV during the initial 2 h in the ICU (2-h interval), >or=to20% of EBV 2-6 h in the ICU (4-h interval), or >or=to30% during 6-12 h in the ICU (6-hour interval). Using this definition, multiple forward stepwise logistic regression analysis was performed for the variables that were significantly related to excessive intra- and postoperative blood loss in the univariate analysis. Criteria for variable inclusion were entry if P <or=to 0.05 and exclusion if P > 0.1. SPSS for Windows (SPSS Inc., Chicago, IL) was used for all calculations.


Data from 548 patients were obtained, but 22 patients (4%) required reoperation for hemorrhage and were excluded from all analyses because inadequate surgical hemostasis may have contributed to blood loss in these cases. Patient demographic, blood loss, and transfusion profiles are summarized in Table 1 and Figure 1. Compared with children >1 yr of age, infants incurred significantly greater intra- and postoperative hemorrhage and were exposed to significantly greater volumes and units of blood components. Median donor exposure was 6 (4-8) U for infants and 2 (0-4) U for children >1 yr of age.

Table 1
Table 1:
Patient Demographics, Intraoperative Blood Loss, and Blood Products Transfused
Figure 1
Figure 1:
Median cumulative postoperative chest tube output for infants (<or=to1 yr of age), children >1 yr of age, and all study patients. At all time points, chest tube output for infants was significantly greater than that for children >1 yr old (*P < 0.0001, multivariate analysis of variance).

Variables independently associated with blood loss or products transfused are detailed in Table 2, and values are compared between the two patient age groups. Surgeon A performed most operations and operated on more infants than the other surgeons. Children >1 yr old were significantly more likely to undergo resternotomy and to have received anticoagulants preoperatively. Infants were significantly more likely to have CHF; to undergo complex operations with longer periods of CPB, aortic clamping, and DHCA; and to be cooled to lower core temperatures during CPB. Pre-CPB laboratory tests were within normal limits and showed age-associated differences [4]. During CPB, values for platelet count, TEG MA, and angle were significantly lower for infants; these patients were transfused to higher Hct values in the postoperative period.

Table 2
Table 2:
Selected Perioperative Variables and Laboratory Tests

Multiple linear regression analyses of CT output (Models 1-4) and blood products transfused (Models 5 and 6) are presented in Table 3. Cumulative adjusted coefficient of determination (R2) values sequentially decreased from Model 1 to Model 4, which indicates that the study variables better accounted for variations in CT output in the early postoperative period. For Models 1-4, R2 increased if CT 0h was included as a variable, and, in these models, it was the variable most significantly associated with CT output. When CT 0h was excluded, patient age became the variable most significantly associated with CT output.

Table 3
Table 3:
Linear Regression Analysis of Chest Tube (CT) Output and Blood Products Transfused with CT 0h Included and Excluded as a Variable

In Models 1-6, 9 variables were independently associated with outcome measures when CT 0h was included, 15 variables were associated when CT 0h was excluded, and 7 variables (patient age, DHCA, surgeon, Hct in ICU, Hct preoperatively, duration of CPB, complex surgery) were common to both analyses. CT 0h was removed from the models because CT output at each time point was cumulative. However, CT 0h demonstrated that patients with increased post-CPB intraoperative blood loss had a significant risk of bleeding in the postoperative period (Table 3). Of the 12 models in Table 3, the following variables were independently associated with bleeding or transfusion on more than one occasion (frequency indicated): patient age [10], CT 0h [6], DHCA [6], surgeon [5], platelet count during CPB [4], Hct in ICU [4], Hct preoperatively [3], TEG preoperatively [3], core temperature on arrival in the ICU [2], TEG angle during CPB [2], TEG angle preoperatively [2], TEG MA preoperatively [2], total protamine dose [2], duration of CPB [2], and complex surgery [2]. Antifibrinolytic therapy was not independently associated with blood loss or transfusion therapy and did not alter analysis results when added to the models.

Of the patients who bled excessively in the ICU, 64% fulfilled criteria for excessive CT output [13] within 2 h, and 94% fulfilled these criteria by 6 h. Results of the logistic regression analyses of patients with excessive bleeding are presented in Table 4 (Models 7 and 8). Using mean values from the two age groups to illustrate risk, a decrease in platelet count during CPB from 135,000 to 74,000 mm-3 increased the risk of bleeding by 6.2-fold. Similarly, a decrease in age from 79.7 to 3.6 mo doubled the risk of bleeding. Linear regression analyses of the two age groups (Table 5, Models 9-15) showed that the following variables were independently associated with the outcome measures on more than one occasion (frequency indicated): infants (<or=to1 yr of age)-minimal core temperature during CPB [5] and patient weight [3]; children >1 year of age-patient age [4], duration CPB [4], preoperative CHF [3], patient weight [2], complex surgery [2], platelet count during CPB [2], and resternotomy [2]. The variable identified as most significantly associated with bleeding or transfusion was minimal core temperature during CPB for infants and patient age for children >1 yr of age.

Table 4
Table 4:
Logistic Stepwise Regression: Variables Independently Associated with Excessive Intra- and Postoperative Bleedinga
Table 5
Table 5:
Linear Regression Analysis of Perioperative Blood Loss and Transfusions for Children Aged <or=to1 Year and Children >1 Year of Age


The hazards of multiple and massive transfusions with allogenic blood products are well recognized and warrant a critical approach to hemostasis management after CPB [9,11,14]. Because perioperative bleeding is multifactorial in etiology, multivariate discriminant analysis may identify variables associated with hemorrhage. In our study, we evaluated 61 perioperative variables and found that young patient age was the variable most significantly associated with bleeding and blood product transfusions. High preoperative Hct, low platelet count during CPB, and duration of DHCA also were significant risk factors. Factors associated with bleeding and transfusions varied with patient age. Patients who bled excessively after CPB during chest closure were at risk of excessive postoperative bleeding in the ICU.

Adjusted R2 values for CT 2h (0.757) and units transfused per kilogram (0.735) were greater than those reported for similar adult studies (R2 = 0.36) [9], which suggests that the more extreme hemostatic derangement encountered in children [6,7] enhances the relationships between risk factors and bleeding. Excessive hemorrhage often began intraoperatively or early in the postoperative period: 64% of patients who bled met the criteria for excessive CT output within 2 h after arrival in the ICU; when included in analyses, CT 0 h was the variable most significantly related to CT output later in the postoperative period. In children, CT 0 h may be a useful advance warning of probable excessive postoperative hemorrhage, as has been demonstrated in adults [9,10].

The results of linear and logistic regression analyses were similar. Patient age was shown to be the most significant variable and is an easy and practical method for identifying children at risk of excessive hemorrhage. Patient height, weight, and age are closely interrelated, and other pediatric cardiac studies have reported age or weight to be independently associated with CT output [1,4,8] and component therapy [3,4]. We previously described the association between patient age and bleeding [4]. In the present study, we further define this relationship by demonstrating that age continues to be an important independent predictor of bleeding even when other risk factors for hemorrhage are included in the analyses.

Other variables consistently associated with blood loss and transfusions were duration of DHCA, platelet count during CPB, preoperative Hct, complex surgery, and surgeon. DHCA results in extreme perturbation of patient physiology, and correlation between DHCA and a requirement for platelet transfusions has been described [2]. In a previous investigation, we found platelet count during CPB to be the laboratory test most significantly associated with CT output and blood products transfused in children, probably reflecting the consequences of hemodilution during CPB [12]. Similarly, platelet count after protamine administration has been shown to correlate with CT drainage in children [8] and adults [9]. TEG angle is affected by platelet function (and soluble coagulation factor activity) and measures the rate of clot formation. It was independently associated with blood loss in several of our models and could be useful during CPB, especially when the platelet count during CPB borders on inadequate values. TEG variables measured after CPB have been reported to predict excessive postoperative bleeding in adults [15] and children [8,12,16]. The positive correlation between preoperative Hct and blood loss (also noted in children >1 year of age) has been reported [2] and is consistent with the described coagulopathic effects of cyanotic heart disease [17]. In summary, the present study provides new information by quantifying the predictive value of perioperative coagulation tests relative to other risk factors for bleeding and transfusions.

Patients undergoing complex surgery often had multiple characteristics associated with bleeding (young age, deep hypothermia, high Hct, CHF, and resternotomy), and four of the complex operations involved extensive aortic suture lines. Complex cardiac surgery has been associated with bleeding in some [1,3], but not all, pediatric studies [2]. Surgeon identity is an independent risk factor for bleeding during adult cardiac surgery [18], but in our study, the unequal distribution of cases between surgeons confounded interpretation.

Several variables had less significant associations with bleeding or products transfused. The finding that preoperative anticoagulant therapy was a significant variable in only one model is supported by adult studies that showed no increase in blood loss with aspirin [19] and warfarin [20] therapy. The efficacy of prophylactic antifibrinolytic therapy during pediatric OHS remains controversial [21]. In our study, epsilon-aminocaproic acid had no statistically demonstrable effect on transfusion requirements, but the dose regimen used (150-mg/kg load, 30-mg [middle dot] kg-1 [middle dot] h-1 infusion) may have been inadequate [22].

The significant differences in intraoperative blood loss, postoperative CT output, and blood products transfused between infants and older children (>1 year of age) are consistent with previous reports [1,2,4,8] and suggest that risk factors for bleeding are age-dependent. Lower weight (neonates) and colder CPB were associated with increased blood loss and product transfusions among infants. A retrospective study of 73 infants similarly noted an association between minimal temperature during CPB and requirements for platelet transfusions [2]. For children >1 year of age, patient age was the most significant variable for bleeding, but CPB duration, preoperative CHF, complex surgery, resternotomy, and platelet count during CPB were also significant factors for blood loss. An association between CPB duration and bleeding has been reported for both pediatric [8,23] and adult cardiac surgery patients [9,10,24]. Resternotomy in adults has been found to correlate with blood loss [9,24], but a relationship has not previously been demonstrated in children [5]. Abnormal hemostasis is associated with hypoxia, liver congestion, and low cardiac output [25]-conditions that may be present in children with CHF.

The findings of this study enabled us to refine protocols for the pre-, intra-, and postoperative management of hemostasis in children undergoing OHS. Blood conservation and transfusion strategy planning is possible because important risk factors such as patient age, polycythemia, complexity of surgery, resternotomy, and use of DHCA are known before surgery. During CPB, risk indicators such as minimal core temperature, platelet count, and TEG angle allow coagulation therapy to be individualized to each patient's requirements. Increased CT 0h identifies children who will bleed excessively postoperatively.

Study limitations should be considered. Although associations can be demonstrated by multivariate analyses, a cause and effect relationship cannot be assumed because there may be unmeasured confounding variables [9]. For both children [1,2,8] and adults [11], blood loss and transfusion practices after cardiac surgery vary widely among institutions, and the spectrum of pediatric variables associated with perioperative bleeding may also differ among institutions. Adjusted R2 for volume of blood products transfused (mL/kg) was markedly less than adjusted R (2) for units of blood products transfused (U/kg). It is possible that physicians were more concerned about donor exposure than volume transfused, and that some patients received a volume of blood products greater than was required to control hemostasis.

In summary, the overwhelming impact of CPB often disrupted hemostasis in children, with blood loss and transfusions showing strong associations with a small number of perioperative variables. Young patient age was the variable most significantly associated with bleeding and blood product transfusions. Other important factors included the use of DHCA and deep hypothermia, complex surgery, high preoperative Hct, and low platelet count during CPB. In older children, resternotomy, preoperative CHF, and prolonged CPB became significant. These variables may assist preoperative formulation of hemostasis strategies and guide intraoperative coagulation management [12].


1. Manno CS, Hedberg KW, Kim HC, et al. Comparison of the hemostatic effects of fresh whole blood, stored whole blood, and components after open heart surgery in children. Blood 1991;77:930-6.
2. Petaja J, Lundstrom U, Leijala M, et al. Bleeding and the use of blood products after heart operations in infants. J Thorac Cardiovasc Surg 1995;109:524-9.
3. Chambers LA, Cohen DM, Davis JT. Transfusion patterns in pediatric open heart surgery. Transfusion 1996;36:150-4.
4. Williams GD, Bratton LS, Riley EC, Ramamoorthy C. Association between age and blood loss in children undergoing open-heart surgery. Ann Thorac Surg 1998;66:870-6.
5. Guay J, Rivard GE. Mediastinal bleeding after cardiopulmonary bypass in pediatric patients. Ann Thorac Surg 1996;62:1955-60.
6. Kern FH, Schulman SR, Greeley WJ. Cardiopulmonary bypass: techniques and effects. In: Greeley WJ, ed. Perioperative management of the patient with congenital heart disease. Baltimore: Williams & Wilkins, 1996:67-120.
7. Chan AK, Leaker M, Burrows FA, et al. Coagulation and fibrinolytic profile of paediatric patients undergoing cardiopulmonary bypass. Thromb Haemost 1997;77:270-7.
8. Miller BE, Mochizuki T, Levy JH, et al. Predicting and treating coagulopathies after cardiopulmonary bypass in children. Anesth Analg 1997;85:1196-202.
9. Despotis GJ, Filos KS, Zoys TN, et al. Factors associated with excessive postoperative blood loss and hemostatic transfusion requirements: a multivariate analysis in cardiac surgical patients. Anesth Analg 1996;82:13-21.
10. Liu B, Belboul A, Larsson S, Roberts D. Factors influencing haemostasis and blood transfusion in cardiac surgery. Perfusion 1996;11:131-43.
11. Belisle S, Hardy JF. Hemorrhage and the use of blood products after adult cardiac operations: myths and realities. Ann Thorac Surg 1996;62:1908-17.
12. Williams GD, Bratton LS, Riley EC, Ramamoorthy C. Coagulation tests during cardiopulmonary bypass correlate with blood loss in children undergoing cardiac surgery. J Cardiothorac Vasc Anesth. In press.
13. Sumner E, Stark J. Post-operative care. In: Stark J, de Leval M, eds. Surgery for congenital heart defects. Philadelphia: WB Saunders, 1994:225.
14. Gelb AB, Roth RI, Levin J, et al. Changes in blood coagulation during and following cardiopulmonary bypass: lack of correlation with clinical bleeding. Am J Clin Pathol 1996;106:87-99.
15. Spiess BD. Thromboelastography and cardiopulmonary bypass. Semin Thromb Hemost 1995;21:27-33.
16. Martin P, Horkay F, Rajah SM, Walker DR. Monitoring of coagulation status using thrombelastography during paediatric open heart surgery. Int J Clin Monit 1991;8:183-7.
17. Suarez CR, Menendez CE, Griffin AJ, et al. Cyanotic congenital heart disease in children: hemostatic disorders and relevance of molecular markers of hemostasis. Semin Thromb Hemost 1984;10:285-9.
18. Vander Salm TJ, Kaur S, Lancey RA, et al. Reduction of bleeding after heart operations through the prophylactic use of epsilon-aminocaproic acid. J Thorac Cardiovasc Surg 1996;112:1098-107.
19. Vuylsteke A, Oduro A, Cardan E, Latimer RD. Effect of aspirin in coronary artery bypass grafting. J Cardiothorac Vasc Anesth 1997;11:831-4.
20. Dietrich W, Dilthey G, Spannagl M, Richter JA. Warfarin pretreatment does not lead to increased bleeding tendency during cardiac surgery. J Cardiothorac Vasc Anesth 1995;9:250-4.
21. Pouard P. Review of efficacy parameters. Ann Thorac Surg 1998;65:S40-4.
22. Williams GD, Bratton LS, Riley EC, Ramamoorthy C. Effect of epsilon aminocaproic acid in children undergoing open-heart surgery. J Cardiothorac Vasc Anesth 1999;13:304-8.
23. Seear MD, Wadsworth LD, Rogers PC, et al. The effect of desmopression acetate (DDAVP) on postoperative blood loss after cardiac operations in children. J Thorac Cardiovasc Surg 1989;98:217-9.
24. Hardy JF, Perrault J, Tremblay N, et al. The stratification of cardiac surgical procedures according to use of blood products: a retrospective analysis of 1480 cases. Can J Anaesth 1991;38:511-7.
25. Mace S, Borkat G, Liebman J. Hepatic dysfunction and cardiovascular abnormalities. Am J Dis Child 1985;139:60-5.
© 1999 International Anesthesia Research Society