Introduction

In the setting of chronic hypoxemia, children with cyanotic congenital heart disease (CCHD) often develop an increased hematocrit. This process is thought to represent a compensatory mechanism to achieve adequate oxygen delivery to tissue beds. Based on this physiologic adaptation, most congenital heart centers have adopted a strategy of red blood cell (RBC) transfusion among children with cyanotic heart disease to achieve greater serum hemoglobin (Hgb) concentration among hospitalized patients. Although there are no published guidelines, RBC transfusion is typically undertaken to achieve serum Hgb levels >13 gm/dl.

However, studies have demonstrated increased risks associated with RBC transfusions in adults after cardiac surgery [8, 26]. Although a more liberal transfusion strategy has been shown to have greater rates of worsening organ failure and other complications in adults, a randomized study of conservative versus liberal RBC transfusion strategies in critically ill children without congenital heart disease has demonstrated no difference in adverse outcomes [15]. Recent prospective studies have reported that significant hemodilution during surgical repair of infants with congenital heart defects was associated with poorer neurodevelopmental outcome [13, 22, 37]. Another prospective trial suggested that low hematocrit during cardiopulmonary bypass may affect early postoperative lactate levels [24]. It is not known if there are potential benefits and risks of RBC transfusions outside of the immediate postoperative period in patients with CCHD. Analysis of the effect of RBC transfusion is complex because of the time-dependent nature of the decision to transfuse. The clinical decision to transfuse is largely based on bleeding complications and low Hgb concentrations occurring during the hospital course, which in turn are dependent on multiple clinical variables and therapies.

At our institution, we have generally pursued a strategy of “proactive transfusion” by maintaining serum Hgb levels >13 gm/dl in young children with CCHD. Although this strategy may improve overall oxygen delivery, the efficacy of this approach has not been evaluated. Our goal was to assess how transfusions impact regional oxygen saturation (rSO2), arterial saturation, and hemodynamic markers based on Hgb levels before blood is transfused.

Materials and Methods

This retrospective study was performed with the approval of the Institutional Review Board of Emory University School of Medicine and Children’s Healthcare of Atlanta. We reviewed the hospital records of patients <1 year of age with cyanotic single-ventricle lesions and arterial saturations <90% who were admitted to the cardiac intensive care unit at Children’s Healthcare of Atlanta at Egleston and who received a blood transfusion between December 2007 and April 2009. We included all patients who were admitted before and after surgery as well as and those who did not undergo cardiac surgery during the same hospitalization. At our institution, general criteria for blood transfusions include patients who are believed to have low Hgb concentration (usually we maintain Hgb concentration >13 gm/dl), low arterial saturations (saturation <70% in the setting of CCHD), low blood pressure (mean arterial pressure <40 mm Hg in the neonate or 50 mm Hg in children ≤1 year of age), active bleeding (sanguinous chest tube output >2 ml/kg/h), or by clinician preference. Exclusion criteria included blood transfusions given within 72 h of cardiac surgery or transfusions given to patients with sanguinous chest tube output >2 ml/kg/h. The primary outcome variable was change in cerebral rSO2 as measured by near-infrared spectroscopy (NIRS). This is routinely used at our institution on all infants with congenital heart disease in the cardiac intensive care unit, and trends are followed-up as a noninvasive proxy for mixed venous oxygen saturation (SVO2). Secondary outcome variables included blood pressure, arterial saturation, heart rate, and adverse events. Complete blood counts, creatinine, and lactic acid levels before and after transfusion were obtained. We divided the data points into terciles based on pretransfusion Hgb concentration. Heart rate, blood pressure, arterial saturation, and cerebral NIRS (cNIRS) values before transfusion, as well as at 1, 2, 4, 8, and 12 h after the start of the transfusion, were collected.

Data Analysis

Data are presented as means or medians and ranges where appropriate. Comparisons of different groups were made by χ2 or Fisher exact test. Changes in hemodynamic data were evaluated by analysis of variance with the values before transfusion and at 8 and 12 h after transfusion. Significance was determined at a p < 0.05.

Results

During the study period, 59 patients and 183 transfusions were identified (Tables 1, 2). Several patients had transfusions during different hospitalizations, and these were all included. Forty-five patients received blood transfusions after surgery during the same hospitalization as their cardiac surgery, which accounted for 133 transfusions. Eight patients received transfusions during readmissions after surgery. Two were readmitted for sternal wound infections, 2 for respiratory distress, 3 for decreased arterial saturations, and 1 for decreased oral intake. Four patients received a blood transfusion as an inpatient after catheterization.

Table 1 Characteristics of study population
Full size table
Table 2 Clinical status of patients at the time of transfusion
Full size table

Pretransfusion Hgb concentration in group A was 7.8 to 12.3 gm/dl, in group B was 12.4 to 13.2 gm/dl, and in group C was 13.3 to 15.7 gm/dl. Further patient characteristics of the three groups are listed in Table 3. There was no statistical difference among the groups with regard to those receiving mechanical ventilation or vasoactive drips (Fig. 1). Sixty controls were obtained. These were periods when the patients had an Hgb value >13.3 gm/dl and did not receive a blood transfusion. They were matched with the study groups in terms of mechanical ventilation and inotropic support. There was no significant change of any parameter during 12 h. Mean heart rate decreased by 1 beat/minute; mean systolic blood pressure decreased by 2.5 mm Hg; mean diastolic pressure decreased by 1 mm Hg; mean arterial saturation was unchanged; and mean NIRS value decreased by 1.5.

Table 3 Subgroups of patients who received blood transfusions
Full size table
Fig. 1
figure 1

Subgroups of patients on mechanical ventilation or vasoactive drips

Full size image

After RBC transfusion, there was no statistically significant change in heart rate regardless of Hgb concentration (Figs. 2, 3, 4, 5, and 6). There was, however, significant improvement in diastolic blood pressure, arterial saturation, and cNIRS when an RBC transfusion was given at Hgb < 12.3 gm/dl (group A). There was an almost 8% increase in cNIRS. Transfusions given in group B also resulted in statistically significant improvement in diastolic blood pressure and arterial saturation. However, there was a less robust response in cNIRS. There were little to no statistically significant improvements when transfusions were given at Hgb > 13.2 gm/dl (Group C).

Fig. 2
figure 2

Heart rate changes after transfusion. *Data points are reported as mean values. p-Values represent change from hours 0 to 12

Full size image
Fig. 3
figure 3

Systolic blood pressure changes after transfusion

Full size image
Fig. 4
figure 4

Diastolic blood pressure changes after transfusion

Full size image
Fig. 5
figure 5

Arterial saturation changes after transfusion

Full size image
Fig. 6
figure 6

Cerebral NIRS changes after transfusion

Full size image

Because the decision to transfuse RBCs in some centers may relate more to clinical status than baseline Hgb, we also examined whether responses varied by whether subjects received ventilatory support or inotropic medications. The improvement in cNIRS was similar in both patients who were ventilated and those who were breathing spontaneously (5.9% vs. 6.2%, p = 0.55). Those subjects on inotropic support demonstrated an increase in cNIRS of 6.7% compared with 7.1% for those not on inotropic medications, p = 0.80.

Fourteen adverse events occurring within 48 h of a blood transfusion were identified: 5 from group A, 4 from group B, and 5 from group C. These events are listed in Table 4. They occurred in 11 patients (one patient had 3 events, and 1 patient had 2 events). There were no occurrences of acute renal failure or significant changes in lactic acid levels after transfusion. In group C, there was an increase in median lactic acid concentration after blood transfusions, but this did not approach statistical significance (p = 0.27), and the levels all remained clinically low. Three deaths occurred during the hospitalization in which a blood transfusion was received, but all occurred >48 h after transfusion. One patient who had undergone a Norwood procedure died after cardiopulmonary arrest 3 days after receiving a blood transfusion at postoperative day 56. The patient also had long-standing renal failure resulting in anuria. A second patient died after cardiopulmonary arrest at 7 days after the last blood transfusion, and a third patient died 1 month after the last blood transfusion because care was withdrawn.

Table 4 Adverse events within 48 h of transfusion
Full size table

Discussion

The rationale for RBC transfusion is to achieve improvement of oxygen transport and, ultimately, tissue oxygenation. In the setting of hypoxemia, increasing the Hgb concentration will increase arterial oxygen content and oxygen delivery (DO2) at a given cardiac output. Oxygen consumption (VO2) can remain constant despite decrease in oxygen delivery until a critical value is reached, at which point tissue hypoxia develops [19]. This critical level varies among individuals, but it is important to keep oxygen delivery above this level. In patients with single-ventricle physiology, their physiology limits their arterial saturation and oxygen content.

Our analysis demonstrated that RBC transfusions do appear to improve a number of important physiologic parameters. In those subjects with pretransfusion Hgb values <12.3 gm/dl, RBC transfusions favorably impacted blood pressure, arterial saturation, and, most importantly, rSO2. Not surprisingly, at greater Hgb levels, transfusion of RBCs appeared to have less of an impact. These findings should be of value in devising a uniform RBC transfusion strategy for this population.

A recent prospective study compared a restrictive versus liberal RBC-transfusion strategy in the immediate postoperative period in children undergoing either a Glenn anastomosis or Fontan completion [4]. The strategies were divided to maintain an Hgb concentration either >9 gm/dl or >13 gm/dl, respectively. Cholette et al. demonstrated no difference in arterial lactate levels or in arteriovenous and arteriocerebral oxygen content during the first 48 h after surgery. These data suggests that patients with palliated single-ventricle physiology may not require as high an Hgb concentration as previously believed.

In adult cardiac surgery patients, several parameters have been suggested to help determine the need for blood transfusion. These include relative tachycardia, relative hypotension, decrease in VO2 > 10% from baseline, oxygen extraction ratio (VO2/DO2) > 50%, SVO2 < 50%, and a mixed venous oxygen partial pressure (PvO2) < 32 mm Hg [30]. However, these suggested guidelines may not be applicable for clinicians caring for children with CCHD. SVO2 is thought to be a single statistically independent variable that accurately reflects the adequacy of systemic perfusion [1] and may improve survival when used as goal-directed therapy [36]. Studies have looked at the use of cNIRS to monitor continuous cerebral oxygen saturation as an estimation of SVO2 in patients with CCHD. Although a correlation is present, cNIRS is more useful to follow as a trend than taken as an absolute measurement [2, 12, 16, 21]. Furthermore, a positive correlation has also been shown between rSO2 and Hgb concentration in children with and without congenital heart disease [2, 17]. Low rSO2 and SVO2 in patients after a Norwood procedure has an association with worse outcomes in terms of organ failure [35], neurodevelopment [7, 11], and mortality [25].

However, there is no agreement as to the appropriate threshold to maintain Hgb concentration in patients with single-ventricle physiology. Polycythemia leads to an increase in O2 content, but this is offset by a decrease in cardiac output, leading to no change or even a decrease in O2 delivery. This has been previously attributed to hyperviscosity of blood and increased afterload [6, 9, 23], but newer evidence suggests that this may be due to regulation of oxygen delivery and not solely on blood viscosity [18].

There is growing literature that stored RBCs have a diminished ability for oxygen delivery compared with native RBCs. There is some debate as to the efficacy of newer versus older stored RBCs in terms of O2 transport [30, 33], and it has been suggested that older stored blood may increase the risk of postoperative complications [14]. An animal model has suggested that circulation of stored RBCs result in significantly malperfused and underoxygenated microvasculature not detected at the systemic level [34]. Storage of RBCs affects cell deformability and depletes nitric oxide, which impairs the vasodilatory activity of RBCs due to the decrease in S-nitroso-Hgb concentration [27].

Transfusions also carry risk of adverse effects. Blood product exposure has been demonstrated as an independent risk factor for the development of central line-associated bloodstream infection [5]. Sepsis due to contaminated blood has been documented at 1/250,000 transfusions, but this is probably underreported as a complication [10]. Immune modulation may be a more important mechanism for sepsis than contaminated blood. Exposure to blood products is thought to decrease natural killer cell function, affect antigen presentation, decrease helper-to-suppressor T-lymphocyte ratio, and decrease cell-mediated immunity [3, 29]. Studies have demonstrated increased morbidity, mortality, and cost associated with RBC transfusions due to infections, acute respiratory distress syndrome, and organ failure after cardiac surgery [20, 32]. We identified 14 adverse events in 11 patients. These patients received a mean of 5.7 transfusions each within the study period. Due to their high number of transfusions, they likely were more critically ill and more prone to the effects of immune modulation. We had 4 incidents of culture-positive sepsis (2.5%) but no acute increases in serum creatinine or lactic acid levels. There is also some evidence that greater levels of Hgb is associated with increased incidence of systemic–to–pulmonary shunt thrombosis [28].

There are alternatives to the use of RBC transfusion in patients with evidence of hypovolemia. They can include crystalloid fluid, albumin, and hydroxyethyl starch. A prospective, randomized trial using hydroxyethyl starch or albumin in the setting of noncardiac surgery demonstrated effective hemodynamic stabilization without adverse impact on coagulation or other safety parameters in both groups [31]. Using these fluids may improve cardiac output without the detriment of increasing blood viscosity or decreasing DO2 to a level at which tissue hypoxia develops. Patients may have sufficient oxygen-carrying capacity at lower Hgb concentrations than initially expected. Children with normal two-ventricle circulation do not demonstrate an increase in adverse events with maintaining Hgb as low as 7 gm/dl [15]. This threshold is likely greater in patients with single-ventricle physiology.

Limitations

Limitations of this study include its retrospective design and lack of randomization. There may be selection bias in terms of which patients received blood transfusions because our institution does not have a set protocol for the administration of blood products. The decision for blood transfusion is strictly clinician derived. More critically ill patients, especially those with a more tenuous postoperative course, are more likely to receive blood. Patients who clinically demonstrate improvement with previous RBC transfusions are also more likely to receive further transfusions. In addition, adverse events are listed but may not be solely linked to transfusions.

Although liberal versus restrictive transfusion strategies have been examined in relatively stable patients in a pediatric intensive care unit [15] and after cavopulmonary connections [4], we believe there is a role for larger prospective studies looking at these strategies to optimize oxygen delivery and demonstrate survival benefit and improved clinical outcome measures in patients with CCHD.

Conclusion

In summary, blood transfusions improve markers of oxygen delivery, such as arterial saturation and cNIRS, when given to patients with Hgb concentration <13.2 gm/dl. Further studies are needed to determine optimal Hgb levels in hospitalized infants with single-ventricle physiology. Interventions to augment Hgb above this level may be of limited benefit.