Postoperative Hematocrit and Adverse Outcomes in Pediatric Cardiac Surgery Patients: A Cross-Sectional Study From the Society of Thoracic Surgeons and Congenital Cardiac Anesthesia Society Database Collaboration : Anesthesia & Analgesia

Secondary Logo

Journal Logo

Featured Articles: Original Clinical Research Report

Postoperative Hematocrit and Adverse Outcomes in Pediatric Cardiac Surgery Patients: A Cross-Sectional Study From the Society of Thoracic Surgeons and Congenital Cardiac Anesthesia Society Database Collaboration

Long, Justin B. MD*; Engorn, Branden M. MD; Hill, Kevin D. MD; Feng, Liqi MS§; Chiswell, Karen PhD§; Jacobs, Marshall L. MD; Jacobs, Jeffrey P. MD; Goswami, Dheeraj MD#

Author Information
doi: 10.1213/ANE.0000000000005416

Abstract

See Article, p 1074

KEY POINTS

  • Question: Is high postoperative hematocrit associated with worse outcomes in pediatric cardiac surgical patients?
  • Findings: High hematocrit on arrival to intensive care unit (ICU) is associated with increased postcardiac surgical risk in pediatric patients.
  • Meaning: Extremely high hematocrit targets have generally been favored in congenital cardiac surgery, but concerns remain about the association between high hematocrits and the association with morbidity and mortality in pediatric postcardiac surgical patients.

The majority of children undergoing cardiac surgical procedures with cardiopulmonary bypass (CPB) will receive at least 1 transfusion of red blood cells.1,2 The trigger, or threshold, for transfusion is a hematocrit value that prompts transfusion, and the target for transfusion is the desired hematocrit following transfusion.3 The target and trigger for transfusion are influenced by a multitude of factors which include the type of procedure, remaining postoperative cardiac lesions, current physiologic parameters, anticipated physiologic changes, ongoing bleeding, and the age of the patient.4 A prospective, observational study of pediatric postcardiac surgical patients in 30 North American hospitals revealed that transfusion practices varied greatly among the studied centers and that the transfusion triggers varied between and within acyanotic and cyanotic patient groups.5

Blood transfusions have been associated with increased risk of infection, length of intensive care unit (ICU) and hospital stay, duration of mechanical ventilation, and mortality in the pediatric population.6–8 Historically, pediatric providers use higher hematocrit triggers to prompt transfusion when compared to adult providers, especially for children with cyanotic congenital heart disease. Subgroup analyses performed on Transfusion Requirements in the Pediatric Intensive Care Unit (TRIPICU) data and other studies showed that similar clinical outcomes can be achieved with a more conservative transfusion trigger in pediatric postcardiac surgical patients.9–12 While gaps exist in transfusion trigger research, even less research exists on hematocrit targets. A recent retrospective study of 80 infants showed that higher postoperative hematocrit values were independently associated with an increased risk of early systemic to pulmonary artery shunt occlusion.13 Yet, the tolerable upper limit of the hematocrit in pediatric postcardiac surgical patients remains poorly defined.

Given the potential for adverse outcomes associated with blood transfusion and suboptimal hematocrit targets, a greater understanding of appropriate transfusion triggers and targets in the pediatric postcardiac surgical population is needed. In this cross-sectional study of data from The Society of Thoracic Surgeons (STS) Congenital Heart Surgery Database (CHSD), our primary objective was to examine potential associations between postoperative hematocrit values and postoperative complications or mortality. Secondary outcomes include duration of mechanical ventilation and hospital length of stay. We hypothesized that there exists a range for postoperative hematocrit in which the risk of major postoperative complications or mortality is increased.

METHODS

The STS-CHSD is a repository of demographic and clinical data collected on congenital and pediatric cardiac surgical operations in North America.14 Coding for this database is completed by clinicians and support staff and is audited rigorously for completeness and accuracy as previously described.15 The Congenital Cardiac Anesthesia Society (CCAS) has collaborated with the STS-CHSD Task Force to develop an anesthesia-specific module of the STS-CHSD (often referred to as the STS-CCAS database) and has collected information on the conduct of anesthesia and ICU arrival data for congenital heart surgery cases since 2010. The Duke Clinical Research Institute (DCRI) serves as the data warehouse and analytic center for the STS-CHSD. The Children’s Healthcare of Atlanta Institutional Review Board confirmed exempt status of this study. The STS Access and Publications Task Force of the STS Workforce on Research Development approved the study and this article. A data analysis and statistical plan was filed with DCRI before data access. No statistical power calculation was performed before the study.

The data were obtained from the STS-CHSD for this retrospective, cross-sectional study. The STS-CHSD was queried for index cardiac surgery operations with CPB between 1 January 2014 and 31 December 2018 on patients <18 years old with data included in the STS-CCAS module of the CHSD. Outcomes data were included through 30 June 2019. The index operation is the first cardiovascular surgical operation of the episode of care. A total of 62,240 index cases from 64 centers were potentially eligible for inclusion. The time period and sample size were selected based on available data in the STS-CHSD for the primary exposure (ie, postoperative hematocrit) and procedures with missing data for the primary exposure were excluded. Specific case exclusion criteria were applied to ensure rigorous data quality with 27,462 cases included in the analysis (Supplemental Digital Content 1, Table 1, https://links.lww.com/AA/D385).

Among the primary considerations regarding transfusion triggers is the expected oxygen saturation of the patient following the operation. For this analysis, we have considered the group expected on the basis of their circulatory physiology to have normal oxygen saturation (≥92%) postoperatively (ie, acyanotic patients) separately from the group expected to have lower oxygen saturation (<92%) postoperatively (ie, cyanotic patients). Each of the individual procedure types was reviewed to determine the best categorization of the patient postoperatively into either the cyanotic group (eg, bidirectional cavopulmonary anastomosis and systemic to pulmonary shunt), acyanotic group (eg, Ross procedure, repair of Tetralogy of Fallot, and Fontan procedure), or excluded due to an inability to categorize the surgical procedure reliably (Supplemental Digital Contents 2–3, Tables 2–3, https://links.lww.com/AA/D386, https://links.lww.com/AA/D387). The presumed categorization was validated against initial pulse oximeter saturation data (ie, minimum, maximum, mean, median, and standard deviation) on arrival to the ICU and 4 authors agreed to the classification (J.B.L., B.M.E., K.D.H., and D.G.).

Demographic data included the weight, postnatal age, year of surgery, and sex of the patient. Preoperative data included the presence or absence of any syndromes, chromosomal abnormality, noncardiac congenital anatomic abnormality, gestational age, persistent shock, renal failure, ventilatory support, neurologic deficits, whether the patient had any prior CPB operations, whether the patient had any prior cardiothoracic operations, the STS-CHSD mortality risk model variables,16 and other preoperative factors. Operative data included the operation performed, STS – European Association for Cardio-Thoracic Surgery Congenital Heart Surgery (STAT) Mortality Categories,17,18 CPB time, and the hematocrit post-CPB, postprotamine. Postoperative data included the first hematocrit after transfer to the ICU,14 whether there was delayed sternal closure, number of days the chest remained open,19 total hours of mechanical ventilation for the patient care episode, postoperative length of stay, and total length of stay. The method for measurement of the postoperative hematocrit is not defined by the CHSD, but the most common technique is conductivity testing utilizing a blood gas analyzer.

We evaluated 2 primary outcomes, the first being operative mortality, defined by STS-CHSD as14:

  • All deaths, regardless of cause, occurring during the hospitalization in which the operation was performed, even if after 30 days (including patients transferred to other acute care facilities) and
  • All deaths, regardless of cause, occurring after discharge from the hospital, but before the end of the 30th postoperative day.

The second primary outcome is the occurrence of a major complication, defined as the occurrence of any one or more of the following postoperative events in the major complication composite20: renal failure, neurologic deficit persisting at discharge, extracorporeal membrane oxygenation (ECMO) or ventricular assist device (VAD), unplanned operative or cardiac catheterization reintervention, unplanned cardiac arrest, major infection (eg, endocarditis, mediastinitis, wound infection, and sepsis), or thrombotic complication (eg, intracardiac, central vein, deep vein, systemic to pulmonary artery shunt, or pulmonary artery thrombus).

Statistical Analysis

Plots of the log odds ratio (OR) for both primary outcomes (mortality and major postoperative complications) versus hematocrit were assembled for both the cyanotic and acyanotic groups, with curves estimated using restricted cubic spline functions with knots placed at the fifth, 35th, 65th, and 95th percentiles of the distribution within each subgroup (Figure 1).

F1
Figure 1.:
Graphs illustrating the relationship between relative odds of occurrence of the primary outcomes versus changes in hematocrit relative to the population median (hematocrit of 38%). The relative odds estimated by modeling hematocrit using restricted cubic spline functions are presented by the solid blue line. The blue shaded area represents the 95% CI; therefore, where the shaded area is wide, there is less data available for the analysis and the CI is widened. A, Acyanotic group association between the major complication composite and hematocrit. B, Cyanotic group association between the major complication composite and hematocrit. C, Acyanotic group association between operative mortality and hematocrit. D, Cyanotic group association between operative mortality and hematocrit. CI indicates confidence interval.

For presentation of descriptive summaries of patient characteristics and outcomes across the range of hematocrit values within each subgroup, the authors selected cut points for low, medium, and high hematocrit subgroups based on the observed nonlinear relationships between hematocrit and primary outcomes. The cyanotic group was split into <36%, 36%–42%, and >42%. The acyanotic group was split into <32%, 32%–40%, and >40%. As a sensitivity analysis, alternative cut points were selected based on the plots but rounded to reasonable, clinically relevant values. Patient and preoperative characteristics were summarized for the cyanotic and acyanotic subgroups and cumulatively for the overall cohort using medians [25th, 75th percentiles] for continuous variables as well as frequencies and percentages for categorical variables.

The relationship between hematocrit and each primary outcome was assessed using multivariable generalized linear mixed-effects modeling with a logit link function (logistic regression modeling with random effect terms). The variables included in the final model were age, weight, history of prematurity, primary procedure, STAT mortality categories 1–5, preoperative ventilation support, preoperative persistent shock, preoperative renal failure or need for dialysis, preoperative neurological deficit, any other preoperative factors, previous cardiothoracic operations, presence or absence of specific noncardiac congenital abnormalities (ie, tracheoesophageal fistula, congenital diaphragmatic hernia, omphalocele, gastroschisis, intestinal malrotation, anal atresia, and Hirschprung disease), and chromosomal abnormality or underlying syndrome risk categories 0–5 (Table 1).16 In the main analysis, hematocrit was modeled as a continuous variable, with ORs presented for 5% increments in hematocrit. The observed nonlinear relationship between hematocrit and the log odds was modeled using a 2-piece linear spline, with the knot set at 42% for cyanotic patients and at 38% in acyanotic patients to capture the main features of the nonlinearity. To adjust for case-mix confounding, the preoperative and demographic variables listed above from the published STS risk model were included as covariates.18 The model includes adjustment for the primary procedure of the index operation. Due to the potential for small sample sizes, these procedures are included in the model as random effects with a separate effect estimated within each age group. In this analysis we also included year of surgery as an additional adjustment covariate and included a random intercept for each center to adjust for within-center clustering. For each outcome, all estimates were calculated by fitting a single model including both cyanotic and acyanotic patients. This model included a variable adjusting for cyanotic versus acyanotic subgroup, the linear spline variables for hematocrit, and the interaction between subgroup and the spline variables. This allowed for estimation of the hematocrit ORs within each subgroup. As a secondary analysis, we estimated ORs comparing the high versus medium, high versus low, and low versus medium hematocrit subgroups defined by the cut points.

Table 1. - Patient Characteristics and Data
Variable Level Overall (N = 27,462) Acyanotic (n = 22,553) Cyanotic (n = 4909)
Patient characteristics
 Age at surgery (d) 274 [99, 1635] 364 [114, 1991] 155 [10, 747]
 Age group Neonate 4680 (17.0%) 3119 (13.8%) 1561 (31.8%)
Infant 10,079 (36.7%) 8163 (36.2%) 1916 (39.0%)
Child 12,703 (46.3%) 11,271 (50.0%) 1432 (29.2%)
 Sex Male 15,193 (55.3%) 12,254 (54.3%) 2939 (59.9%)
Female 12,264 (44.7%) 10,294 (45.6%) 1970 (40.2%)
Ambiguous 5 (0.02%) 5 (0.02%) 0 (0.00%)
 Weight (kg) 7.6 [4.6, 16.2] 8.3 [4.8, 18.4] 6.1 [3.5, 10.8]
 Gestational age (wk) 38 [37, 39] 38 [37, 39] 39 [38, 39]
Preoperative data
 Prematurity (<37 wk completed gestation) Missing 1128 (4.1%) 1079 (4.8%) 49 (1.0%)
No 21,819 (79.5%) 17,559 (77.9%) 4260 (86.8%)
Yes 4515 (16.4%) 3915 (17.4%) 600 (12.2%)
 Mechanical ventilatory support No 25,101 (91.4%) 20,743 (92.0%) 4358 (88.8%)
Yes 2361 (8.6%) 1810 (8.0%) 551 (11.2%)
 Any syndrome Missing 14 (0.1%) 8 (0.04%) 6 (0.1%)
No 21,313 (77.6) 17,313 (76.8) 4000 (81.5%)
Yes 6135 (22.3%) 5232 (23.2) 903 (18.4)
 Noncardiac congenital anomaly Missing 38 (0.1%) 21 (0.1%) 17 (0.4%)
No 22,532 (82.1%) 18,595 (82.5%) 3937 (80.2%)
Yes 4892 (17.8%) 3937 (17.5%) 955 (19.5%)
 Any chromosomal abnormality Missing 7 (0.03%) 5 (0.02%) 2 (0.04%)
No 23,582 (85.9%) 18,867 (83.7%) 4715 (96.1%)
Yes 3873 (14.1%) 3681 (16.3%) 192 (3.9%)
 Redo sternotomy No 19,624 (71.5%) 17,091 (75.8%) 2533 (51.6%)
Yes 7838 (28.5%) 5462 (24.2%) 2376 (48.4%)
 Shock, persistent at time of surgery No 27,300 (99.4%) 22,442 (99.5%) 4858 (99.0%)
Yes 162 (0.6%) 111 (0.5%) 51 (1.0%)
 Renal failure No 27,177 (99.0%) 22,322 (99.0%) 4855 (99.0%)
Yes 285 (1.0%) 231 (1.0%) 54 (1.0%)
 Neurologic deficit No 26,325 (95.9%) 21,662 (96.1%) 4663 (95.0%)
Yes 1137 (4.1%) 891 (4.0%) 246 (5.0%)
 Any other preoperative factors No 21,846 (79.6%) 18,318 (81.2%) 3528 (71.9%)
Yes 5616 (20.5%) 4235 (18.8%) 1381 (28.1%)
 Congenital anomaly/syndrome risk category 0 21,671 (78.9%) 17,568 (77.9%) 4103 (83.6%)
1 3023 (11.0%) 2857 (12.7%) 166 (3.4%)
2 1656 (6.0%) 1361 (6.0%) 295 (6.0%)
3 687 (2.5%) 461 (2.0%) 226 (4.6%)
4 415 (1.5%) 306 (1.4%) 109 (2.2%)
5 10 (0.04%) 0 (0.0%) 10 (0.2%)
Operative data
 CPB time (min) 98 [66, 145] 97 [65, 144] 101 [71, 148]
 Hematocrit postprotamine 35 [31, 39] 34 [30, 38] 38 [34, 42]
 The Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Category Missing 71 (0.3%) 42 (0.2%) 29 (0.6%)
1 8441 (30.7%) 8204 (36.4%) 237 (4.8%)
2 8119 (29.6%) 5688 (25.2%) 2431 (49.5%)
3 4020 (14.6%) 3929 (17.4%) 91 (1.9%)
4 5412 (19.7%) 4556 (20.2%) 856 (17.4%)
5 1399 (5.1%) 134 (0.6%) 1265 (25.8%)
Postoperative data
 Hematocrit on ICU arrival 37.0 [32.9, 41.0] 36.0 [32.0, 40.0] 40.7 [36.6, 45.0]
 Postoperative length of stay (d) 7 [4, 14] 6 [4, 11] 12 [7, 29]
 Total length of stay (d) 8 [4, 18] 7 [4, 15] 15 [8, 36]
 Total ventilation time (h) 15.8 [5.7, 52.0] 14.6 [5.4, 46.8] 24.4 [7.6, 115.9]
 Delayed sternal closure No 24,667 (89.8%) 20,839 (92.4%) 3828 (78.0%)
Yes 2795 (10.2%) 1714 (7.6%) 1081 (22.0%)
 If delayed sternal closure, number of days chest is open (surgery date inclusive) 4 [3, 5] 3 [3, 5] 4 [3, 5]
Outcomes data
 Operative mortality Missing 59 (0.2%) 35 (0.2%) 24 (0.5%)
No 26,832 (97.7%) 22,216 (98.5%) 4616 (94.0%)
Yes 571 (2.1%) 302 (1.3%) 269 (5.5%)
 Major complication Missing 149 (0.5%) 121 (0.5%) 28 (0.6%)
No 23,139 (84.3%) 19,743 (87.5%) 3396 (69.2%)
Yes 4174 (15.2%) 2689 (11.9%) 1485 (30.3%)
 Renal failure No 27,200 (99.1%) 22,396 (99.3%) 4804 (97.9%)
Yes 262 (1.0%) 157 (0.7%) 105 (2.1%)
 Neurologic deficit persisting to discharge No 27,294 (99.4%) 22,456 (99.6%) 4838 (98.6%)
Yes 168 (0.6%) 97 (0.4%) 71 (1.4%)
 Extracorporeal membrane oxygenation or ventricular assist device No 26,930 (98.1%) 22,280 (98.8%) 4650 (94.7%)
Yes 532 (1.9%) 273 (1.2%) 259 (5.3%)
 Unplanned reintervention No 24,268 (88.4%) 20,545 (91.1%) 3723 (75.8%)
Yes 3194 (11.6%) 2008 (8.9%) 1186 (24.2%)
 Unexpected cardiac arrest during or after procedure Missing 149 (0.5%) 121 (0.5%) 28 (0.6%)
No 26,504 (96.5%) 21,956 (97.4%) 4548 (92.7%)
Yes 809 (3.0%) 476 (2.1%) 333 (6.8%)
 Composite major infection Missing 149 (0.5%) 121 (0.5%) 28 (0.6%)
No 26,231 (95.5%) 21,751 (96.4%) 4480 (91.3%)
Yes 1082 (3.9%) 681 (3.0%) 401 (8.2%)
 Composite thrombotic complicationa Missing 30 (0.2%) 23 (0.2%) 7 (0.2%)
No 15,961 (97.2%) 13,267 (98.0%) 2694 (93.3%)
Yes 429 (2.6%) 243 (1.8%) 186 (6.4%)
Data are presented as count (percentage) or median [25th percentile, 75th percentile].
Abbreviations: CHSD, Congenital Heart Surgery Database; CPB, cardiopulmonary bypass; ICU, intensive care unit; STS, Society of Thoracic Surgeons.
aData only available in newer versions of the STS-CHSD; therefore, total N for this observation = 16,420.

Association between major complication subtypes and the hematocrit sub group on ICU arrival was assessed using descriptive summaries (ie, frequencies and percentages) and χ2 tests. Kruskal-Wallis tests were used to compare the distribution of continuous secondary outcomes by hematocrit subgroup. A 2-tailed P value of <.01 was used as the trigger for statistical significance because this study has a large number of observations and multiple comparisons were tested. All analyses were performed using SAS v9.4 software (SAS Institute Inc, Cary, NC).

RESULTS

Patient Characteristics: Cyanotic and Acyanotic Group Comparisons

There were 27,462 cases included for analysis with 4909 (17.9%) classified as cyanotic and 22,553 (82.1%) as acyanotic. Patient characteristics and variables included in the multiple regression model, separated by acyanotic and cyanotic groups, are found in Table 1. The median [25th percentile, 75th percentile] hematocrit on arrival to the ICU in the cyanotic group was 40.7 [36.6, 45.0] percent vs 36.0 [32.0, 40.0] percent in the acyanotic group. Operative mortality occurred more often in the cyanotic versus the acyanotic group, 5.5% vs 1.3%. Likewise, major postoperative complications were more common in the cyanotic versus the acyanotic group, 30.3% vs 11.9%. Each of the complications in the major postoperative complications composite were more common in the cyanotic versus the acyanotic group (Table 1).

Cyanotic Group Analysis

Patient characteristics and variables included in the multivariable regression model for the cyanotic group, subdivided by low, medium, and high hematocrit subgroups are found in Supplemental Digital Content 4, Table 4, https://links.lww.com/AA/D388. Descriptive summaries of the primary and secondary outcomes for the cyanotic group, subdivided by low, medium, and high hematocrit subgroups are found in Table 2. After adjusting for case-mix, odds of mortality and odds of a major complication did not vary significantly with changes in hematocrit <42% (mortality: OR 0.94 per 5% increment in hematocrit, 95% CI, 0.78-1.13; complications: OR 1.01, 95% CI, 0.92-1.11). However, for each 5% incremental increase in hematocrit over 42%, the odds, demonstrated by the adjusted odds ratio (aOR), of operative mortality increased by 1.31 fold (aOR = 1.31; 95% CI, 1.10-1.55; P = .003) and the odds of a major complication increased by 1.22 fold (aOR = 1.22; 95% CI, 1.10-1.36; P < .001; Table 3). When comparing the low, medium, and high hematocrit subgroups defined by cut points, differences for mortality did not meet thresholds of statistical significance, but the high versus medium hematocrit groups did differ significantly for odds of major complication (Supplemental Digital Content 5, Table 5, https://links.lww.com/AA/D389).

Table 2. - Cyanotic Group Outcomes by Hematocrit Subgroup
Variable Level Overall (N = 4909) Low (<36) (N = 983) Med (36–42) (N = 2004) High (>42) (N = 1922) P value P value
Low versus med
P value
Med versus high
P value
Low versus high
Postoperative length of stay (d) 12 [7, 29] 11 [7, 25] 12 [7, 27] 14 [8, 33] <.001 .519 <.001 <.001
Total length of stay (d) 15 [8, 35.5] 14 [7, 32] 14 [8, 33] 17 [8, 39] <.001 .787 <.001 <.001
Total ventilation (h) 24.4 [7.6, 115.9] 24.2 [7.0, 99.3] 19.7 [7.1, 98.3] 29.0 [9.1, 133.9] <.001 .111 <.001 <.001
Delayed sternal closure No 3828 (78.0%) 778 (79.2%) 1626 (81.1%) 1424 (74.1%) <.001 .197 <.001 .003
Yes 1081 (22.0%) 205 (20.9%) 378 (18.9%) 498 (25.9%)
If delayed sternal closure, number of days chest is open (surgery date inclusive) 4 [3, 5] 4 [3, 5] 3.5 [2, 5] 4 [3, 6] .004 .841 .003 .016
Outcomes
 Operative mortality Missing 24 (0.5%) 4 (0.4%) 7 (0.4%) 13 (0.7%) .320 .556 .300 .160
No 4616 (94.0%) 932 (94.8%) 1891 (94.4%) 1793 (93.3%)
Yes 269 (5.5%) 47 (4.8%) 106 (5.3%) 116 (6.0%)
 Major complication Missing 28 (0.6%) 3 (0.3%) 5 (0.3%) 20 (1.0%) <.001 .935 <.001 <.001
No 3396 (69.2%) 707 (71.9%) 1445 (72.1%) 1244 (64.7%)
Yes 1485 (30.3%) 273 (27.8%) 554 (27.6%) 658 (34.2%)
 Renal failure No 4804 (97.9%) 973 (99.0%) 1961 (97.9%) 1870 (97.3%) .012 .028 .254 .003
Yes 105 (2.1%) 10 (1.0%) 43 (2.2%) 52 (2.7%)
 Neurologic deficit persisting to discharge No 4838 (98.6%) 972 (98.9%) 1976 (98.6%) 1890 (98.3%) .492 .529 .494 .249
Yes 71 (1.5%) 11 (1.1%) 28 (1.4%) 32 (1.7%)
 Extracorporeal membrane oxygenation or ventricular assist device No 4650 (94.7%) 950 (96.6%) 1907 (95.2%) 1793 (93.3%) <.001 .062 .012 <.001
Yes 259 (5.3%) 33 (3.4%) 97 (4.8%) 129 (6.7%)
 Unplanned reintervention No 3723 (75.8%) 779 (79.3%) 1560 (77.8%) 1384 (72.0%) <.001 .382 <.001 <.001
Yes 1186 (24.2%) 204 (20.8%) 444 (22.2%) 538 (28.0%)
 Unexpected cardiac arrest during or after procedure Missing 28 (0.6%) 3 (0.3%) 5 (0.3%) 20 (1.0%) .003 .710 .004 .008
No 4548 (92.7%) 925 (94.1%) 1880 (93.8%) 1743 (90.7%)
Yes 333 (6.8%) 55 (5.6%) 119 (5.9%) 159 (8.3%)
 Composite major infection Missing 28 (0.6%) 3 (0.3%) 5 (0.3%) 20 (1.0%) .337 .883 .163 .326
No 4480 (91.3%) 903 (91.9%) 1845 (92.1%) 1732 (90.1%)
Yes 401 (8.2%) 77 (7.8%) 154 (7.7%) 170 (8.8%)
 Composite thrombotic complicationa Missing 7 (0.2%) 2 (0.4%) 2 (0.2%) 3 (0.3%) .050 .343 .093 .025
No 2694 (93.3%) 547 (94.8%) 1100 (93.9%) 1047 (92.0%)
Yes 186 (6.4%) 28 (4.9%) 70 (6.0%) 88 (7.7%)
All tests treat the column variable as nominal. Data are presented as count (percentage) or median [25th percentile, 75th percentile]. P values are based on Pearson χ2 tests for all categorical row variables. P values were calculated by comparing only nonmissing row values. P values are based on χ2 rank–based group means score statistics for all continuous/ordinal row variables which is equivalent to Kruskal-Wallis tests.
Abbreviations: CHSD, Congenital Heart Surgery Database; STS, Society of Thoracic Surgeons.
aData only available in newer versions of the STS-CHSD.

Table 3. - Association Between Hematocrit on ICU Arrival and Primary Outcomes Stratified by Acyanotic and Cyanotic Procedures
Unadjusted Adjusted
Outcomes Variable Levels Odds ratio (95% CI) P value Global P value Odds ratio (95% CI) P value Global P value
Operative mortality Acyanotic Per 5% increment, for HCT <38 1.00 (0.84-1.19) .985 <.001 0.88 (0.74-1.05) .150 <.001
Per 5% increment, for HCT ≥38 1.84 (1.64-2.07) <.001 1.45 (1.28-1.65) <.001
Cyanotic Per 5% increment, for HCT <42 0.93 (0.78-1.10) .393 <.001 0.94 (0.78-1.13) .481 .009
Per 5% increment, for HCT ≥42 1.53 (1.31-1.79) <.001 1.31 (1.10-1.55) .003
Major complication composite Acyanotic Per 5% increment, for HCT <38 1.10 (1.03-1.17) .003 <.001 0.97 (0.91-1.04) .353 <.001
Per 5% increment, for HCT ≥38 1.51 (1.43-1.59) <.001 1.21 (1.14-1.29) <.001
Cyanotic Per 5% increment, for HCT <42 1.03 (0.94-1.12) .562 <.001 1.01 (0.92-1.11) .842 <.001
Per 5% increment, for HCT ≥42 1.37 (1.24-1.50) <.001 1.22 (1.10-1.36) <.001
Hematocrit is treated as a set of continuous variables in the models. The association between hematocrit and outcome was nonlinear and the model includes 2 odds ratios describing the relationship—one odds ratio for 5% increment in hematocrit when hematocrit is <38% for acyanotic patients and <42% for cyanotic patients as well as a second odds ratio for 5% increment in hematocrit when hematocrit is ≥38% for acyanotic patients and ≥42% for cyanotic patients.
Abbreviations: CI, confidence interval; HCT, hematocrit; ICU, intensive care unit.

F2
Figure 2.:
Bar chart of the incidence of categorical secondary outcomes. The category for neurologic deficits represents new neurologic deficits persisting to discharge and very few were reported in either group; therefore, the incidence is close to 0 in these charts. A, Acyanotic group complications subdivided by low (blue), medium (green), and high (orange) hematocrit subgroups. B, Cyanotic group complications subdivided by low (blue), medium (green), and high (orange) hematocrit subgroups.

With respect to secondary outcomes, the high hematocrit subgroup had a statistically significant longer postoperative length of stay, total length of stay, total ventilation hours, and higher risk of delayed sternal closure with longer total number of days with open chest versus the medium hematocrit subgroup. There was no statistically significant difference between the medium hematocrit and low hematocrit subgroups for the same outcomes (Table 2). The risk of unplanned reoperation was higher in the high hematocrit subgroup versus the medium hematocrit subgroup (28.0% vs 22.2%; P < .001; Table 2; Figure 2). The risk of unexpected cardiac arrest during or following the procedure was also higher in the high hematocrit subgroup versus the medium hematocrit subgroup (8.3% vs 5.9%; P = .004; Table 2; Figure 2). The differences between the cyanotic subgroups for the remainder of the components of the major complication composite did not reach statistical significance.

Acyanotic Group Analysis

Patient characteristics and variables included in the multiple regression model for the acyanotic group, subdivided by low, medium, and high hematocrit subgroups are found in Supplemental Digital Content 6, Table 6, https://links.lww.com/AA/D390. Descriptive summaries of the primary and secondary outcomes for the cyanotic group, subdivided by low, medium, and high hematocrit subgroups are found in Table 4. After adjusting for case-mix, odds of mortality and odds of a major complication did not vary significantly with changes in hematocrit <38% (mortality: OR 0.88 per 5% increment in hematocrit, 95% CI, 0.74-1.05; complications: OR 0.97, 95% CI, 0.91-1.04). However, for each 5% incremental increase in hematocrit over 38%, the odds of operative mortality increased 1.45 fold (aOR = 1.45; 95% CI, 1.28-1.65; P < .001) and odds of a major complication increased by 1.21 fold (aOR = 1.21; 95% CI, 1.14-1.29; P < .001; Table 3). When comparing the low versus medium hematocrit subgroups defined by the cut points, the adjusted risks of mortality and of major complications did not differ significantly between low and medium hematocrit groups. However, the adjusted risk of each outcome other than neurologic outcomes was significantly higher for the high compared to the medium hematocrit subgroups (Table 4).

Table 4. - Acyanotic Group Outcomes by Hematocrit Subgroup
Variable Level Overall (N = 22,553) Low (<32) (N = 5201) Med (32–40) (N = 11,771) High (>40) (N = 5581) P value P value
Low versus med
P value
Med versus high
P value
Low versus high
Postoperative length of stay (d) 6 [4, 11] 5 [4, 8] 6 [4, 11] 8 [5, 16] <.001 <.001 <.001 <.001
Total length of stay (d) 7 [4, 15] 5 [4, 10] 6 [4, 14] 10 [6, 22] <.001 <.001 <.001 <.001
Total ventilation (h) 14.6 [5.4, 46.8] 8.3 [4.3, 26.8] 14.1 [5.4, 34.5] 26.7 [8.5, 77.0] <.001 <.001 <.001 <.001
Delayed sternal closure No 20,839 (92.4%) 4904 (94.3%) 11,007 (93.5%) 4928 (88.3%) <.001 .053 <.001 <.001
Yes 1714 (7.6%) 297 (5.7%) 764 (6.5%) 653 (11.7%)
If delayed sternal closure, number of days chest is open (surgery date inclusive) 3 [3, 5] 3 [2, 4] 3 [3, 5] 4 [3, 5] <.001 .529 <.001 <.001
Outcomes
 Operative mortality Missing 35 (0.2%) 6 (0.1%) 18 (0.2%) 11 (0.2%) <.001 .083 <.001 <.001
No 22,216 (98.5%) 5153 (99.1%) 11,624 (98.8%) 5439 (97.5%)
Yes 302 (1.3%) 42 (0.8%) 129 (1.1%) 131 (2.4%)
 Major complication Missing 121 (0.5%) 22 (0.4%) 45 (0.4%) 54 (1.0%) <.001 <.001 <.001 <.001
No 19,743 (87.5%) 4727 (90.1%) 10,468 (88.9%) 4548 (81.5%)
Yes 2689 (11.9%) 452 (8.7%) 1258 (10.7%) 979 (17.5%)
 Renal failure No 22,396 (99.3%) 5174 (99.5%) 11,706 (99.5%) 5516 (98.8%) <.001 .787 <.001 <.001
Yes 157 (0.7%) 27 (0.5%) 65 (0.6%) 65 (1.2%)
 Neurologic deficit persisting to discharge No 22,456 (99.6%) 5182 (99.6%) 11,726 (99.6%) 5548 (99.4%) .104 .868 .055 .091
Yes 97 (0.4%) 19 (0.4%) 45 (0.4%) 33 (0.6%)
 Extracorporeal membrane oxygenation or ventricular assist device No 22,280 (98.8%) 5156 (99.1%) 11,662 (99.1%) 5462 (97.9%) <.001 .700 <.001 <.001
Yes 273 (1.2%) 45 (0.9%) 109 (0.9%) 119 (2.1%)
 Unplanned reintervention No 20,545 (91.1%) 4870 (93.6%) 10,841 (92.1%) 4834 (86.6%) <.001 <.001 <.001 <.001
Yes 2008 (8.9%) 331 (6.4%) 930 (7.9%) 747 (13.4%)
 Unexpected cardiac arrest during or after procedure Missing 121 (0.5%) 22 (0.4%) 45 (0.4%) 54 (1.0%) <.001 .066 <.001 <.001
No 21,956 (97.4%) 5107 (98.2%) 11,517 (97.8%) 5332 (95.5%)
Yes 476 (2.1%) 72 (1.4%) 209 (1.8%) 195 (3.5%)
 Composite major infection Missing 121 (0.5%) 22 (0.4%) 45 (0.4%) 54 (1.0%) <.001 .069 <.001 <.001
No 21,751 (96.4%) 5069 (97.5%) 11,422 (97.0%) 5260 (94.3%)
Yes 681 (3.0%) 110 (2.1%) 304 (2.6%) 267 (4.8%)
 Composite thrombotic complicationa Missing 23 (0.2%) 11 (0.4%) 7 (0.1%) 5 (0.2%) <.001 .498 <.001 <.001
No 13,267 (98.0%) 3090 (98.4%) 6870 (98.5%) 3307 (96.9%)
Yes 243 (1.8%) 40 (1.3%) 101 (1.5%) 102 (3.0%)
All tests treat the column variable as nominal. Data are presented as count (percentage) or median [25th percentile, 75th percentile]. P values are based on Pearson χ2 tests for all categorical row variables. P values were calculated by comparing only nonmissing row values. P values are based on χ2 rank–based group means score statistics for all continuous/ordinal row variables which is equivalent to Kruskal-Wallis tests.
Abbreviations: CHSD, Congenital Heart Surgery Database; STS, Society of Thoracic Surgeons.
aData only available in newer versions of the STS-CHSD.

For the secondary outcomes, the high hematocrit subgroup had a statistically significant longer postoperative length of stay, total length of stay, total ventilation hours, and higher risk of delayed sternal closure with longer total number of days with open chest versus the medium hematocrit subgroup (Table 4). The low hematocrit subgroup had a statistically significant shorter length of stay, total length of stay, and total ventilation hours versus the medium hematocrit subgroup while there was no difference in risk of delayed sternal closure or number of days the chest remained open (Table 4). The risk of renal failure, need for ECMO or VAD support, unplanned reoperation, unexpected cardiac arrest during or following the procedure, major infection, and thrombotic complications were all higher in the high hematocrit subgroup versus the medium hematocrit subgroup (Table 4; Figure 2). The risk of neurologic deficit was not different among the hematocrit subgroups (Table 4; Figure 2). The risk of unplanned reoperation was lower in the low hematocrit subgroup versus the medium hematocrit subgroup (Table 4; Figure 2).

DISCUSSION

The most significant finding in this investigation is the association between high hematocrit on arrival to the ICU and increase in major complications and operative mortality. For acyanotic patients, incremental increases in hematocrit >38% are associated with increased odds of a major complication and operative mortality. For cyanotic patients, incremental increases in hematocrit >42% are associated with increased odds of a major complication and operative mortality. The high hematocrit subgroups, when compared to medium hematocrit subgroups, had a statistically significant longer postoperative length of stay, total length of stay, total ventilation hours, and higher risk of delayed sternal closure with longer total number of days with open chest. Overall, the results of this study show that a range for hematocrit values on arrival to the ICU exist, beyond which, the overall risk of major complications or operative mortality increases linearly.

This study is important, as the results of this study call for a reevaluation of prevalent practices with respect to transfusion targets. In this study, 25% of cyanotic patients were transfused to hematocrits >45% and 25% of acyanotic patients were transfused to hematocrits >41%. This study supports what other studies have concluded regarding the safety of lower transfusion triggers since no association was observed between the low hematocrit range and worse outcomes versus the medium hematocrit range in this investigation.10–12 The higher risk observed with high hematocrit postoperatively and the high frequency of high postoperative hematocrit observed in this study suggest a reevaluation of prevalent practice is necessary with respect to fluid management and transfusion in the CPB and post-CPB period intraoperatively.

A Cochrane review from 2014 suggested that a lower transfusion trigger may prolong the ICU stay for cyanotic patients while a higher transfusion trigger may increase the incidence of infection for acyanotic patients.21 This contrasts with this study, where postoperative length of stay, total length of stay, and total ventilation hours were not prolonged in the low hematocrit range versus the medium hematocrit range for cyanotic patients (Table 2). Postoperative length of stay, total length of stay, and total ventilation hours were decreased in the low hematocrit range versus the medium hematocrit range for acyanotic patients (Table 4). The difference in major infections between the low hematocrit range and medium hematocrit range were not statistically significant in both acyanotic and cyanotic patients (Tables 2 and 4).

Given the frequency of blood transfusions in pediatric cardiac surgical patients,2 increased frequency of adverse reactions in blood transfusion in children versus adults,6 higher volume of blood transfused in children versus adults on a per kg basis, known risks of blood transfusion, and the results presented herein, it is crucial that providers consider the target for blood transfusion just as carefully as the trigger for transfusion is considered. Postoperative blood transfusion has been previously independently associated with postoperative adverse events in this patient population.7 Increased volume of blood transfusion postoperatively has been suggested to be associated with mortality in this patient population.22 Transfusion may be associated with tissue iron deposition, induction of the inflammatory cascade, and immune suppression which partially explains the association consistently observed between transfusion and increased morbidity and mortality.23–25 Additional studies will be needed to understand the optimal hematocrit at the time of ICU arrival, whether number of donor exposures or volume of blood exposure affects the outcomes more, and what the optimal regimen in lieu of blood might be for intraoperative volume resuscitation.

LIMITATIONS

As with any retrospective study, there are several potential sources of bias. It is possible that there are unmeasured confounders associated with both increased hematocrit and untoward outcomes. Therefore, this study describes the association between increased hematocrit at the time of arrival to the ICU in pediatric postcardiac surgical patients with increased morbidity and mortality but cannot determine causation. This study cannot evaluate the cause of higher hematocrit on ICU arrival which may include management of hematocrit on CPB, physician preference, hemoconcentration, transfusion, and crystalloid management.26 An important consideration in interpretation of this study must be what causes, beyond high hematocrit, could account for a high associated morbidity and mortality and may also lead to high postoperative hematocrit. One possibility is that there was high intraoperative blood loss or there is high ongoing blood loss which has prompted transfusion, this could account for high hematocrit and increased morbidity and mortality as early postoperative bleeding has been previously associated with increased morbidity and mortality in infants following cardiac surgical procedures.22 The volume of transfusion and motivations or target for transfusion cannot be established from the available data, but are an important consideration for further study of this topic.

An attempt was made to decrease information bias by excluding any records without data for the primary exposure as well as eliminating all data from centers with ≥15% missing data for hematocrit (Supplemental Digital Content 1, Table 1, https://links.lww.com/AA/D385). This method of exclusion, used in prior publications from the STS database, did result in an exclusion of 24,352 of the 62,240 potential cases which should have reduced information bias but may have increased selection bias by excluding some reporting centers. One concern with leaving centers with high missing data for the primary exposure in the final cohort is that centers with high missingness were checking hematocrit only in certain subsets of patients on arrival to the ICU and the motivation for such cannot be discerned within the database (eg, hematocrit was only checked in unstable patients). Selection bias is potentially increased within each center’s data so a center-level random effect was also added in each model to adjust for within-center clustering. This study cannot account for the potential selection bias associated with centers that do not report to the anesthesia module of the STS-CHSD.

Thrombosis as a composite complication category is only available in newer versions of the STS database; thus, this was not evaluated as thoroughly as other complications in the database. The incidence of thrombotic complications was greater in the high hematocrit, cyanotic patient population, but this did not reach statistical significance. There was an increased risk of thrombotic complications in the high hematocrit, acyanotic patient population. Thrombosis may be an important outcome to consider in future studies as has been previously suggested in the literature.13,27

The data contained within this investigation should be interpreted cautiously. There is not enough granularity for any given single procedure to make broad recommendations regarding intraoperative hematocrit management. Additionally, variations likely exist between various age groups (eg, neonates and adolescents) that would affect the target hematocrit for each patient population. While the auditing process for the STS database is rigorous and each patient is carefully followed-up for complications, the sensitive nature of some complications could lead to under-reporting which would affect the results of this study.28 Despite the limitations of self-reporting, the results remain compelling.

CONCLUSIONS

Overall, this study shows that the hematocrit on ICU arrival is highly variable and there exists a range of higher hematocrit values on ICU arrival that are linearly associated with an increased risk of negative outcomes for acyanotic and cyanotic pediatric postcardiac surgical patients. As the post-CPB hematocrit is largely controllable, this study suggests opportunities exist for providers to consider the desired target for transfusion. The motivation for providers to transfuse to high post-CPB hematocrits, the effect of high post-CPB hematocrits, and the optimal target for post-CPB hematocrit remains unclear.

ACKNOWLEDGMENTS

The authors thank Courtney McCracken, PhD, and Michael Kelleman, MSPH, Biostatistics Core, Emory University, Atlanta, GA, for development of the statistical analysis plan in support of the funding and data application for this project. The authors also thank David Vener, MD, Professor of Pediatrics and Anesthesiology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, for review of the application to Society of Thoracic Surgeons for funding and data access.

DISCLOSURES

Name: Justin B. Long, MD.

Contribution: This author helped with study design, conduct of the study, data analysis, first draft, literature search, manuscript preparation, and approved the final manuscript.

Name: Branden M. Engorn, MD.

Contribution: This author helped with study design, conduct of the study, literature search, manuscript preparation, and approved the final manuscript.

Name: Kevin D. Hill, MD.

Contribution: This author helped with study design, conduct of the study, data analysis, manuscript preparation, and approved the final manuscript.

Name: Liqi Feng, MS.

Contribution: This author helped with study design, conduct of the study, data analysis, figure design, table preparation, manuscript preparation, and approved the final manuscript.

Name: Karen Chiswell, PhD.

Contribution: This author helped with study design, conduct of the study, data analysis, table preparation, manuscript preparation, and approved the final manuscript.

Name: Marshall L. Jacobs, MD.

Contribution: This author helped with study design, conduct of the study, literature search, manuscript preparation, and approved the final manuscript.

Name: Jeffrey P. Jacobs, MD.

Contribution: This author helped with study design, conduct of the study, manuscript preparation, and approved the final manuscript.

Name: Dheeraj Goswami, MD.

Contribution: This author helped with study design, conduct of the study, literature search, manuscript preparation, and approved the final manuscript.

This manuscript was handled by: James A. DiNardo, MD, FAAP.

    REFERENCES

    1. Chambers LA, Cohen DM, Davis JT. Transfusion patterns in pediatric open heart surgery. Transfusion. 1996;36:150–154.
    2. Kartha VM, Jacobs JP, Vener DF, et al. National benchmarks for proportions of patients receiving blood transfusions during pediatric and congenital heart surgery: an analysis of the STS congenital heart surgery database. Ann Thorac Surg. 2018;106:1197–1203.
    3. Ejaz A, Spolverato G, Kim Y, Frank SM, Pawlik TM. Identifying variations in blood use based on hemoglobin transfusion trigger and target among hepatopancreaticobiliary surgeons. J Am Coll Surg. 2014;219:217–228.
    4. Markham C, Small S, Hovmand P, Doctor A. Transfusion decision making in pediatric critical illness. Pediatr Clin North Am. 2017;64:991–1015.
    5. Mazine A, Rached-D’Astous S, Ducruet T, Lacroix J, Poirier N; Pediatric Acute Lung Injury and Sepsis Investigators Network. Blood transfusions after pediatric cardiac operations: a North American multicenter prospective study. Ann Thorac Surg. 2015;100:671–677.
    6. Lavoie J. Blood transfusion risks and alternative strategies in pediatric patients. Paediatr Anaesth. 2011;21:14–24.
    7. Kneyber MC, Grotenhuis F, Berger RF, Ebels TW, Burgerhof JG, Albers MJ. Transfusion of leukocyte-depleted RBCs is independently associated with increased morbidity after pediatric cardiac surgery. Pediatr Crit Care Med. 2013;14:298–305.
    8. Iyengar A, Scipione CN, Sheth P, et al. Association of complications with blood transfusions in pediatric cardiac surgery patients. Ann Thorac Surg. 2013;96:910–916.
    9. Lacroix J, Hébert PC, Hutchison JS, et al.; TRIPICU Investigators; Canadian Critical Care Trials Group; Pediatric Acute Lung Injury and Sepsis Investigators Network. Transfusion strategies for patients in pediatric intensive care units. N Engl J Med. 2007;356:1609–1619.
    10. Cholette JM, Swartz MF, Rubenstein J, et al. Outcomes using a conservative versus liberal red blood cell transfusion strategy in infants requiring cardiac operation. Ann Thorac Surg. 2017;103:206–214.
    11. Willems A, Harrington K, Lacroix J, et al.; TRIPICU investigators; Canadian Critical Care Trials Group; Pediatric Acute Lung Injury and Sepsis Investigators (PALISI) Network. Comparison of two red-cell transfusion strategies after pediatric cardiac surgery: a subgroup analysis. Crit Care Med. 2010;38:649–656.
    12. de Gast-Bakker DH, de Wilde RB, Hazekamp MG, et al. Safety and effects of two red blood cell transfusion strategies in pediatric cardiac surgery patients: a randomized controlled trial. Intensive Care Med. 2013;39:2011–2019.
    13. Anderson BR, Blancha VL, Duchon JM, et al. The effects of postoperative hematocrit on shunt occlusion for neonates undergoing single ventricle palliation. J Thorac Cardiovasc Surg. 2017;153:947–955.
    14. The Society of Thoracic Surgeons. STS Congenital Heart Surgery Database Data Specifications Version 3.41. Updated October 24, 2018. Accessed August 20, 2020. https://www.sts.org/sites/default/files/documents/CongenitalDataSpecsV3_41.pdf.
    15. Clarke DR, Breen LS, Jacobs ML, et al. Verification of data in congenital cardiac surgery. Cardiol Young. 2008;18(suppl 2):177–187.
    16. Jacobs JP, O’Brien SM, Hill KD, et al. Refining the Society of Thoracic Surgeons congenital heart surgery database mortality risk model with enhanced risk adjustment for chromosomal abnormalities, syndromes, and noncardiac congenital anatomic abnormalities. Ann Thorac Surg. 2019;108:558–566.
    17. O’Brien SM, Clarke DR, Jacobs JP, et al. An empirically based tool for analyzing mortality associated with congenital heart surgery. J Thorac Cardiovasc Surg. 2009;138:1139–1153.
    18. Jacobs JP, Jacobs ML, Maruszewski B, et al. Initial application in the EACTS and STS congenital heart surgery databases of an empirically derived methodology of complexity adjustment to evaluate surgical case mix and results. Eur J Cardiothorac Surg. 2012;42:775–779.
    19. Nelson-McMillan K, Hornik CP, He X, et al. Delayed sternal closure in infant heart surgery-the importance of where and when: an analysis of the STS congenital heart surgery database. Ann Thorac Surg. 2016;102:1565–1572.
    20. Jacobs ML, O’Brien SM, Jacobs JP, et al. An empirically based tool for analyzing morbidity associated with operations for congenital heart disease. J Thorac Cardiovasc Surg. 2013;145:1046–1057.e1.
    21. Wilkinson KL, Brunskill SJ, Doree C, et al. Red cell transfusion management for patients undergoing cardiac surgery for congenital heart disease. Cochrane Database Syst Rev. 2014;2014:CD009752.
    22. Wolf MJ, Maher KO, Kanter KR, Kogon BE, Guzzetta NA, Mahle WT. Early postoperative bleeding is independently associated with increased surgical mortality in infants after cardiopulmonary bypass. J Thorac Cardiovasc Surg. 2014;148:631–6.e1.
    23. Hod EA, Zhang N, Sokol SA, et al. Transfusion of red blood cells after prolonged storage produces harmful effects that are mediated by iron and inflammation. Blood. 2010;115:4284–4292.
    24. Hod EA, Spitalnik SL. Stored red blood cell transfusions: iron, inflammation, immunity, and infection. Transfus Clin Biol. 2012;19:84–89.
    25. Theodoraki K, Markatou M, Rizos D, Fassoulaki A. The impact of two different transfusion strategies on patient immune response during major abdominal surgery: a preliminary report. J Immunol Res. 2014;2014:945829.
    26. Wypij D, Jonas RA, Bellinger DC, et al. The effect of hematocrit during hypothermic cardiopulmonary bypass in infant heart surgery: results from the combined Boston hematocrit trials. J Thorac Cardiovasc Surg. 2008;135:355–360.
    27. Manlhiot C, Menjak IB, Brandão LR, et al. Risk, clinical features, and outcomes of thrombosis associated with pediatric cardiac surgery. Circulation. 2011;124:1511–1519.
    28. Auroy Y, Benhamou D, Amaberti R. Risk assessment and control require analysis of both outcomes and process of care. Anesthesiology. 2004;101:815–817.

    Supplemental Digital Content

    Copyright © 2021 International Anesthesia Research Society