- Email: [email protected]
Perioperative Hyperglycemia: Effect on Outcome After Infant Congenital Heart Surgery
William M. DeCampli, MD, PhD, Monica C. Olsen, BS, CCP, Hamish M. Munro, MD, FRCA, and Donald E. Felix, MD Congenital Heart Institute, Arnold Palmer Hospital for Children and University of Central Florida College of Medicine, Orlando, Florida
Background. Studies demonstrate that cardiopulmonary bypass (CPB) causes intraoperative and postoperative hyperglycemia. Hyperglycemia has been associated with morbidity and mortality after infant cardiac surgery. We studied the effects on early postoperative outcomes of glucose (GLU) changes during and after pediatric cardiac surgery. Methods. The records of 144 infants less than 10 kg who underwent CPB for a variety of congenital cardiac procedures were reviewed. The GLU values (at multiple intervals during and after surgery), age, weight, CPB time, ultrafiltration volume, and risk adjustment for congenital heart surgery (RACHS-1) score were recorded. Univariate and multivariate linear and binary logistic regression were used to examine the dependence of the
composite outcome mortality or postoperative infection, the mechanical ventilation time (VENT time), and the length of stay (LOS), on these variables. Results. The RACHS-1 score was the only significant predictor of the composite variable “mortality or infection” (p ⴝ 0.008). Glucose at any time was not a significant factor predicting this outcome. Lower pre-CPB GLU, younger age, and higher RACHS-1 score were significant predictors of greater LOS and VENT time. Conclusions. In this study, post-CPB and postoperative hyperglycemia were not risk factors for postoperative morbidity and mortality after infant cardiac surgery. (Ann Thorac Surg 2010;89:181– 6) © 2010 by The Society of Thoracic Surgeons
H
GLU between 125 mg/dL and 150 mg/dL may cause glycosuria and dehydration [15], and values exceeding 250 mg/dL are associated with worse outcome after brain injury [16]. More recent published literature [17] suggests that maintaining the GLU level between 110 mg/dL and 126 mg/dL is associated with better outcome after complex cardiac surgery. Furthermore, intensive insulin therapy during cardiopulmonary bypass has been shown to attenuate the systemic inflammatory response after infant cardiac surgery [18]. Given the conflicting conclusions of the aforementioned studies, there is a need for additional evidence as to the clinical effects of hyperglycemia during and after congenital cardiac surgery. We analyzed whether GLU values during and after congenital cardiac surgery were associated with mortality or three other measures of morbidity in the postoperative period.
yperglycemia is associated with cardiopulmonary bypass. Some of the postulated mechanisms include the following: hyperoxia [1, 2], anesthetic agents [3], hypothermia, insulin and catecholamine derangements [4, 5], and utilization of dextrose containing solutions [6]. Hyperoxia is defined as the utilization of 100% fraction of inspired oxygen (Fio2) through the pump oxygenator to increase partial pressure of oxygen, arterial (Pao2) above the “normal” range of 150 to 250 mm Hg [7] during cardiopulmonary bypass (CPB). Animal studies have demonstrated that hyperoxia, with or without CPB, causes an increase in plasma glucose that reverses after return to normoxia [1]. Some clinical reports have linked hyperglycemia to increased morbidity and mortality [5, 6, 8] while others have found no relation to adverse neurodevelopmental outcome [9, 10]. In fact, studies in piglets and cats have reported that hyperglycemia is associated with better maintenance of high energy metabolites and better preservation of brain mitochondria respiratory capacity [11, 12]. Studies of the effects of using dextrose containing solutions during infant cardiac surgery show conflicting results [6, 13]. In adult cardiac surgery, evidence shows that maintaining glucose (GLU) less than180 mg/dL is associated with better outcome [14]. Conversely, the GLU threshold for adverse outcome in children is less clear. In children, Accepted for publication Aug 25, 2009. Address correspondence to Dr DeCampli, Congenital Heart Institute, Arnold Palmer Hospital for Children, 50 W Sturtevant St, Orlando, FL 32806; e-mail: [email protected].
© 2010 by The Society of Thoracic Surgeons Published by Elsevier Inc
Patients and Methods Patient Population We reviewed the charts of 144 consecutive patients weighing 10 kg or less who underwent cardiopulmonary bypass between January 2006 and October 2007 at a single institution. We obtained Institutional Review Board approval with waiver of parental consent for this study.
Anesthesia and CPB Protocol We induced and maintained anesthesia using midazolam, fentanyl, pancuronium, and sevoflurane. Propofol 0003-4975/10/$36.00 doi:10.1016/j.athoracsur.2009.08.062
PEDIATRIC CARDIAC
Perioperative Hyperglycemia: Effect on Outcome After Infant Congenital Heart Surgery
182
DeCAMPLI ET AL HYPERGLYCEMIA IN INFANT HEART SURGERY
Ann Thorac Surg 2010;89:181– 6
PEDIATRIC CARDIAC
peroxia (Fio2 ⫽ 100%, Pao2 ⬎ 400 mm Hg, and venous saturations ⬎75%) was used and acid-base management included the use of pH stat techniques for patients cooled below 28°C and alpha stat for all others. We used continuous ultrafiltration in all cases. We did not administer supplemental dextrose during CPB and glucose was not a component of the cardioplegia solution. The cardioplegia dose was 20 mL/kg (minimum of 100 mL) induction followed by maintenance doses of 10 mL/kg (minimum of 50 mL).
Abbreviations and Acronyms ASO ⫽ arterial switch operation AVC ⫽ atrioventricular canal BDG ⫽ bidirectional Glenn B-T ⫽ Blalock-Taussing CICU ⫽ cardiac intensive care unit CPB ⫽ cardiopulmonary bypass DKS ⫽ Damus-Kaye-Stansel ECG ⫽ electroencephalograph GLU ⫽ glucose LOS ⫽ length of stay MORTINF ⫽ composite for mortality ⫹ incidence of infection RACHS-1 ⫽ risk adjustment for congenital heart surgery ROC ⫽ receiver operating curve TOF ⫽ tetralogy of Fallot VENT ⫽ mechanical ventilation VSD ⫽ ventricular septal defect
Data Collected
was used in patients expected to be extubated shortly after surgery. We gave dexamethasone (1 mg/kg) upon anesthesia induction before CPB was instituted. Insulin was not used in any patient during the operative period. We treated plasma GLU lower than 60 mg/dL with dextrose 25% (1 cc/kg) during the operative period. Hyperglycemia was sporadically treated with insulin (2 subjects) in the cardiac intensive care unit (CICU). The CPB circuit consisted of the Baby RX oxygenator and venous reservoir (Terumo, Ann Arbor, MI), Capiox AF02 arterial line filter (Terumo), ¼ inch tubing arterialvenous loop, and pediatric Bio-pump BP-50 (Medtronic, Minneapolis, MN). Our CPB circuit prime volume used for patients up to 10 kg is approximately 330 cc. Prime solution consisted of whole blood or a combination of washed pack red blood cells and fresh frozen plasma depending on the availability of the former. Plasma-Lyte A (Baxter International Inc, Deerfield, IL), sodium bicarbonate (15 mEq), heparin (2,000 units), calcium chloride (200 mg), and aprotinin (2 cc/kg) were also added. Hy-
Glucose was recorded at the following times: pre-CPB, at CPB initiation, near completion of CPB, post-CPB, upon arrival to the CICU, and last value obtained within the first and second 24 hour periods (last GLU day 1 and last GLU day 2, respectively. For each patient, age, operation, pump prime glucose concentration (prime GLU), CPB time, ultrafiltration volume, serum lactate, risk adjustment for congenital heart surgery (RACHS-1) category, incidence of electroencephalograph (ECG)-evident seizures, duration of mechanical ventilation (VENT time), length of stay (LOS), evidence of tracheal, blood, urine and wound infection, and in-hospital or 30 day mortality were recorded.
Statistical Methods Univariate regression was carried out to determine single variable associations between each of the independent variables (seven GLU values, age, BYAGE, CPB time, prime GLU, ultrafiltration volume, and RACHS category) and the outcome variables (mechanical ventilation time, length of stay, and the composite of mortality ⫹ incidence of infection [MORTINF]). The latter composite score was chosen because a power analysis showed that the mortality rate itself was too low to determine its risk factors within the limits chosen for type I and II errors (␣ ⫽ 0.05 and  ⫽ 0.1), respectively, for the sample size of 144. We treated age as a dichotomous variable (BYAGE) (ⱕ30 days, ⬎30 days) but also checked for any significant differences when using age as a continuous variable. All risk factors with a p value less than 0.1 were then entered
Table 1. Demographics and Outcome Measures by RACHS-1 Category RACHS-1 Category Age (days) Ages, median (range) Weight (kg) CPB time (minutes) Ultrafiltration (mL) Mortality Ventilation time (days) Length of stay (days) Incidence of seizures Incidence of infection
2 (n ⫽ 53)
3 (n ⫽ 43)
4 (n ⫽ 34)
6 (n ⫽ 14)
181.09 ⫾ 172.89 148.5 (14–1095) 5.56 ⫾ 1.73 110.82 ⫾ 40.22 668.86 ⫾ 421.21 1 1.56 ⫾ 2.0 6.97 ⫾ 3.16 0 7
156 ⫾ 193.94 98 (3–930) 5 ⫾ 2.26 171.47 ⫾ 61.42 954.41 ⫾ 625.85 1 4.20 ⫾ 5.46 13.65 ⫾ 11.62 1 7
14.32 ⫾ 29.57 5.5 (1–164) 3.14 ⫾ 0.86 202.32 ⫾ 69.64 978.67 ⫾ 584.26 1 8.70 ⫾ 11.65 21.24 ⫾ 18.18 3 6
31.85 ⫾ 101.09 5.0 (2.0–385) 3.35 ⫾ 1.04 235.78 ⫾ 72.38 1237.14 ⫾ 783.43 2 21.07 ⫾ 31.33 50.07 ⫾ 54.68 0 5
Data expressed as mean ⫾ standard deviation. CPB ⫽ cardiopulmonary bypass;
RACHS-1 ⫽ risk adjustment for congenital heart surgery.
Table 2. Procedures by RACHS-1 Category Category 2 B-T shunt (1) VSD closure (9) BDG (17) VSD closure plus other procedure (8) Repair of TOF (14) Left ventricular outflow tract resection (1) Mitral valve repair (1) Atrial septal defect closure (2) Category 3 Left ventricle to pulmonary artery conduit (2) Right ventricular outflow tract reconstruction (5) AVC (12) Pulmonary atresia/TOF (5) B-T shunt with additional procedures (5) Central shunt plus pulmonary vein repair (2) ASO (9) DKS revision (1) Bidirectional pulmonary artery plasty (1) BDG plus tricuspid valvuloplasty (1) Category 4 Arch reconstruction (19) ASO with VSD closure (9) TOF plus unifocalization (1) Truncus arteriosus repair (2) Total anomalous pulmonary venous return repair (2) Starne’s procedure (1) Category 6 Norwood procedure (12) DKS (1) Norwood with Rastelli (1) ASO ⫽ arterial switch operation; AVC ⫽ atrioventricular canal; B-T ⫽ Blalock-Taussing; BDG ⫽ bidirectional Glenn; DKS ⫽ DamusKaye-Stansel; RACHS-1 ⫽ risk adjustment for congenital heart surgery; TOF ⫽ tetralogy of Fallot; VSD ⫽ ventricular septal defect.
into a multivariate (linear regression) analysis for continuous dependent variables (ventilator time, length of stay), or binary logistic regression for the categoric variable MORTINF to calculate the final determinates of risk. For continuous dependent variables, the tolerance (T) of each covariant was calculated to determine important multicollinearity, with T less than 0.4 taken to indicate the latter possibility. For logistic regression the receiver operating curve (ROC) was used to estimate the Cstatistic, or predictive power for the solution. Univariate regression, Student t test, and Pearson or Spearman correlation were used to compare other pairs of variables. We used SPSS version 14.0 (SPSS Inc, Chicago, IL) as our statistical analysis package. Measurements are expressed as mean and standard deviation and a p value less than 0.05 was considered statistically significant.
Results Demographic and mean outcome data stratified by RACHS-1 score are shown in Table 1. Operations are
DeCAMPLI ET AL HYPERGLYCEMIA IN INFANT HEART SURGERY
183
shown in Table 2. Peak GLU generally occurred intraoperatively post-CPB and then decreased during the next 48 hours toward normal range (Fig 1). The number and percentage of patients with GLU greater than 150 mg/dL at each time were as follows: near CPB completion (n ⫽ 72, 50%), post-CPB (n ⫽ 81, 56.2%), upon arrival to the CICU (n ⫽ 61, 42.3%), end of first day postoperative (n ⫽ 21, 14.5%), and end of second postoperative (n ⫽ 10, 6.94%). Analysis of the cohort by age was performed to ascertain if age less than 30 days was a factor affecting GLU trends. Significant difference in GLU values between the two age groups (A⫽ ⱕ30 days old, n ⫽ 63; B ⫽ ⬎30 days old, n ⫽ 81) was observed at the following periods: near CPB completion (A 133.73 ⫾ 35.62; B 160.33 ⫾ 32.24; p ⬍ 0.05); post-CPB (A 144.61 ⫾ 40.34; B 161.23 ⫾ 37.76; p ⬍ 0.05); upon arrival to the CICU (A 151.38 ⫾ 43.81; B 142.20 ⫾ 40.08; p ⬍ 0.05); end of day 1 (A 130.40 ⫾ 35.54; B 114.32 ⫾ 28.42; p ⬍ 0.05); and end of day 2 (A 111.48 ⫾ 34.60; B 106.01 ⫾ 28.60; p ⬍ 0.05). Thus, neonates demonstrated a lower GLU intraoperatively compared with non-neonates, but higher GLU level upon arrival to the CICU and through the 48 hours postoperative than nonneonates. The GLU at CPB initiation correlated weakly with prime GLU (Spearman r ⫽ 0.26, p ⫽ 0.001) and was even more weakly correlated with subsequent GLU measurements. Neither post-CPB GLU nor GLU on arrival to CICU correlated with CPB time (Spearman r ⫽ ⫺0.06 and 0.07, respectively). The GLU changes did not correlate with lactate changes in this study (Pearson r ⬍ 0.5 at all phases studied). The incidence of hyperlactemia (lactate ⬎ 3 mmol/L) exhibited a downward trend with time, as follows: near CPB end (n ⫽ 41; 28.4%); post-CPB (n ⫽ 41; 28.4%); upon arrival to the CICU (n ⫽ 31; 21.5%); end of postoperative day 1 (n ⫽ 5; 3.47%); and end of postoperative day 2 (n ⫽ 6, 4.16%). Four patients presented with EEG-evident seizures in the postoperative period. Risk factors for seizures in these patients included preoperative diagnosis of periventricular leukomalacia (1 patient) and chromosomal abnormalities (3 patients). The most common infection site was tracheal aspirate in 21 subjects. Other sites included blood (17 subjects), urine (2 subjects), wound (2 subjects), and others (2 subjects). Univariate statistics for the outcome variable MORTINF are shown in Table 3. Only BYAGE, age, and RACHS-1
Fig 1. Glucose trends. Glucose concentration (mg/dL) at various intraoperative and postoperative intervals as defined in the text. Glucose levels are expressed as mean and ⫾ standard deviation.
PEDIATRIC CARDIAC
Ann Thorac Surg 2010;89:181– 6
184
DeCAMPLI ET AL HYPERGLYCEMIA IN INFANT HEART SURGERY
Table 3. Univariate Statistics for Regression of MORTINF Variable PEDIATRIC CARDIAC
BYAGE (⬍30 and ⬎30 days) Age (days) Prime GLU (mg/dL) CPB time (minutes) Ultrafiltration vol. (mL) RACHS-1 Pre CPB GLU (mg/dL) GLU at CPB initiation GLU near end CPB Post-CPB GLU GLU upon arrival CICU Last GLU day 1 Last GLU day 2
p Value 0.09 0.08 0.85 0.80 0.40 0.008 0.21 0.79 0.20 0.16 0.14 0.16 0.98
BYAGE ⫽ age as a dichotomous variable; CICU ⫽ cardiac intensive care unit; CPB ⫽ cardiopulmonary bypass; GLU ⫽ glucose; MORTINF ⫽ composite for mortality ⫹ incidence of infection; RACHS-1 ⫽ risk adjustment for congenital heart surgery.
score were significant univariate determinants. In particular, no GLU value was a significant predictor of mortality or infection. In the multivariate logistic regression only RACHS-1 score was a significant determinant (p ⫽ 0.008). The odds ratio for death or infection was 1.47 (95% confidence interval 1.1 to 2.0) for every increase by one in RACHS-1 score. The C-statistic, or area under the ROC for this solution (indicative of its predictive power) was 0.64 ⫾ 0.05 (p ⫽ 0.01). Univariate predictors of LOS and of VENT time included age, BYAGE, CPB time, RACHS-1 score, and pre-CPB GLU. By multivariate analysis, BYAGE, RACHS-1, and pre-CPB GLU were significant predictors of LOS (p ⬍ 0.015). (A model with the continuous variable “age” instead of the dichotomous variable BYAGE produced a very similar result.) The model is shown in Table 4. The T (tolerance)statistic for age, RACHS-1, and pre-CPB GLU was 0.46, 0.50 and 0.85, respectively, and thus there was not significant multicollinearity. Only RACHS-1 score remained a significant predictor of VENT time in the multivariate analysis (p ⫽ 0.001).
Comment The highest GLU elevation occurred during the immediate post-CPB period, followed, in general, by a steady decline toward more normal ranges by the end of 48 hours in the CICU. In addition to the CPB-induced systemic inflammatory response, it is possible that other factors such as hyperoxia, prime glucose value, or the use of corticosteroids could have resulted in hyperglycemia. Interestingly, however, post-CPB GLU was not correlated with duration of CPB nor to the glucose value in the prime solution, and the latter only weakly correlated with glucose at initiation of CPB. Published studies differ in the apparent effect of intraoperative hyperglycemia on clinical outcomes [8, 9, 10, 19]. Some reports suggest hypoglycemia rather than
Ann Thorac Surg 2010;89:181– 6
hyperglycemia to be related to electroencephalographic seizures, slower electroencephalographic recovery, and an increase risk of adverse events [9, 19]. In other studies hyperglycemia has not been associated with adverse neurodevelopmental outcome [9, 10]. One study found an association between hyperglycemia and morbidity and mortality [8]. A recent report found GLU less than 109 mg/dL and greater than 143 mg/dL associated with mortality and the duration of time in which GLU greater than 126 mg/dL during the first 72 hours postoperative associated with greater LOS [17]. The present study fails to show that GLU, either before, during, or after CPB, are predictors of mortality plus infection or increase duration of ventilation time. The pre-CPB glucose was a determinant of hospital LOS, with lower glucose values correlated with increased LOS. Despite the fact that the LOS solution did not suffer from significant multicollinearity, the independent variables age and pre-CPB GLU were correlated to some extent (Pearson r ⫽ 0.39). This raises the possibility that, for the patients with longest duration hospital stay, the lower glucoses were due largely to the fact that these patients were also younger. Indeed, from a mechanistic viewpoint, it is unlikely that lower pre-CPB glucose would cause longer hospital stay. However, the sample size of this cohort does not allow us to statistically prove it. This study has several limitations. First, it is a nonblinded retrospective study. Second, we examined only short-term clinical outcome. Third, the relative small sample size did not allow us to discriminate small differences in morbidity or mortality, to analyze subgroups, or to consider additional possible risk factors. For example, it is obvious that associated noncardiac anomalies or intervening complications later in the hospital stay should influence hospital length of stay. It is unlikely, however, that the inclusion of such risk factors would cause perioperative GLU values to reenter the multivariate solution as significant covariates (risks). In conclusion we have demonstrated that intraoperative hyperglycemia commonly occurs during infant cardiac surgery. Intraoperative hyperglycemia per se does not appear to be associated with adverse early outcomes. Hyperglycemia after infant cardiac surgery peaks early after CPB and decreases to normal values during the next 48 hours postsurgery unless other complications intervene. Thus, we find no evidence, for example, to support the aggressive use of insulin to manage early postoperative hyperglycemia after infant cardiac surgery. Table 4. Multivariate Regression Model for Log (LOS) Variable Constant RACHS-1 BYAGE Pre CPB GLU(mg/dL)
Coefficient
p Value
1.26 0.107 ⫺0.002 ⫺0.002
0.004 0.004 0.003 0.014
BYAGE ⫽ age as a dichotomous variable; CPB ⫽ cardiopulmonary bypass; GLU ⫽ glucose; LOS ⫽ length of stay; RACHS-1 ⫽ risk adjustment for congenital heart surgery.
The authors thank the Orlando Health Foundation for their generous support of this study.
References 1. Bandali KS, Belanger MP, Wittnich C. Is hyperglycemia seen in children during cardiopulmonary bypass a result of hyperoxia? J Thorac Cardiovasc Surg 2001;122:753– 8. 2. Bandali KS, Belanger MP, Wittnich C. Does hyperoxia affect glucose regulation and transport in the newborn. J Thorac Cardiovasc Surg 2003;126:1730 –5. 3. Ellis DJ, Steward DJ. Fentanyl dosage is associated with reduced blood glucose in pediatric patients after hypothermic cardiopulmonary bypass. Anesthesiology 1990;72:812–5. 4. Benzing G III, Francis PD, Kaplan S, Helmsworth JA, Sperling MA. Glucose and insulin changes in infants and children undergoing hypothermic open-heart surgery. Am J Cardiol 1983;52:133– 6. 5. Anand KJS, Phil D, Hansen DD, Hickey PR. Hormonalmetabolic stress responses in neonates undergoing cardiac surgery. Anesthesiology 1990;73:661–70. 6. Bell C, Hughes CW, Oh TH, Donielson DW, O’Connor T. The effect of intravenous dextrose infusion on postbypass hyperglycemia in pediatric patients undergoing cardiac operations. J Clin Anesth 1993;5:381–5. 7. Brown DM, Holt DW, Edwards JT, Burnett RJ III. Normoxia vs. hyperoxia: impact of oxygen tension strategies on outcome for patients receiving cardiopulmonary bypass for routine cardiac surgical repair. J Extra Corpor Technol 2006; 38:241– 8. 8. Yates AR, Dyke PC II, Taeed R, et al. Hyperglycemia is a marker for poor outcome in the postoperative pediatric cardiac patient. Pediatr Crit Care Med 2006;7:351–5. 9. de Ferranti S, Gauvreau K, Hickey PR, et al. Intraoperative hyperglycemia during infant cardiac surgery is not associ-
DeCAMPLI ET AL HYPERGLYCEMIA IN INFANT HEART SURGERY
10.
11. 12. 13. 14. 15. 16. 17.
18.
19.
185
ated with adverse neurodevelopmental outcome at 1, 4, and 8 years. Anesthesiology 2004;100:1345–52. Ballweg JA, Wernovsky G, Ittenbach RF, et al. Hyperglycemia after infant cardiac surgery does not adversely impact neurodevelopmental outcome. Ann Thorac Surg 2007;84: 2052– 8. Laptook AR, Cobertt RJT, Arencibia-Mireles O, Ruley J. Glucose-associated alterations in ischemic brain metabolism of neonatal piglet. Stroke 1992;23:1504 –11. Wagner KR, Myers RE. Hyperglycemia preserves brain mitochondrial respiration during anoxia. J Neurochem 1986; 47:1620 – 6. Aouifi AA, Neidecker J, Vedrinne C, et al. Glucose versus lactated ringer’s solution during pediatric cardiac surgery. J Cardiothorac Vasc Anesth 1997;11:411– 4. Bojar RM. Early postoperative care. In: Manual of perioperative care in adult cardiac surgery. 4th ed. Maden, MA: Blackwell Publishing, Inc; 2005:243. Gregory GA. Monitoring during surgery. In: Pediatric anesthesia. 4th ed. Philadelphia, PA: Churchill Livingstone; 2002: 260. Vavilala M. Pediatric trauma and burn management. In: Litman RS. Pediatric anesthesia. 1st ed. Philadelphia, PA: Elsevier Mosby; 2004:307. Polito A, Thiagarajan RR, Laussen PC, et al. Association between intraoperative and early postoperative glucose levels and adverse outcomes after complex congenital heart surgery. Circulation 2008;118:2235– 42. Chun-Hu G, Qin C, Yun-Ya W, et al. Effects of insulin therapy on inflammatory mediators in infants undergoing cardiac surgery with cardiopulmonary bypass. Cytokine 2008; 44:96 –100. Rossano JW, Taylor MD, Smith EO, et al. Glycemic profile in infants who have undergone the arterial switch operation: hyperglycemia is not associated with adverse events. J Thorac Cardiovasc Surg 2008;135:739 – 45.
INVITED COMMENTARY This article is a descriptive observation study that attempted to determine if glucose level was associated with adverse outcome in a population of 144 pediatric cardiac surgical patients [1]. In the final version and analysis of this study, glucose was treated as a continuous variable, and in fact, no level of glucose predicted an increased risk for the primary composite mortality and infection end point (MORTINF). Only prebypass glucose, risk adjustment for congenital heart surgery (RACHS-1) level, and being a neonate predicted increased length of stay. The authors conclude that hyperglycemia occurs commonly after infant cardiac surgery and that hyperglycemia was not associated with adverse outcome. Interestingly, in no patient did serum glucose level rise above 200 mg/dL in the study period, which contrasts with other reports in which perioperative glucose often rises above 200 mg/dL. Glucose levels were somewhat arbitrarily selected throughout the postoperative period as the “last glucose day 1 and 2.” In fact, the significance of the “postbypass glucose level” is questionable because continuous ultrafiltration was used throughout the bypass period and the ultrafiltration process removes glucose. The study was not appropriately powered for mortality as a primary end point, so the authors chose to combine © 2010 by The Society of Thoracic Surgeons Published by Elsevier Inc
mortality and infections as a composite end point. This choice appears to be based on the assumption that hyperglycemia may be associated with increased risk of perioperative infection. Although this association seems reasonably sound in the adult cardiac surgical population, there is no definitive support for this in the pediatric cardiac surgical population. This association has not been clearly determined in existing literature, and the MORTINF composite outcome has not been validated in previous studies. It has recently been shown that occurrence of hyperglycemia in adult cardiac surgical patients is not benign. Several studies have shown that tight glycemic control in the perioperative period decreases mortality after cardiac procedures and reduces the incidence of postoperative adverse events, including infectious complications (pneumonia and mediastinitis), myocardial complications (low cardiac output, atrial fibrillation, recurrent ischemia), and prolonged mechanical ventilation and intensive care unit length of stay. The evidence for benefit of tight glycemic control in the pediatric cardiac surgical population is both insufficient and conflicting. A recent article by Ballweg and colleagues [2] suggests that hyperglycemia after cardiopulmonary bypass did not result in increased risk of neuro0003-4975/10/$36.00 doi:10.1016/j.athoracsur.2009.11.006
PEDIATRIC CARDIAC
Ann Thorac Surg 2010;89:181– 6
Background. Studies demonstrate that cardiopulmonary bypass (CPB) causes intraoperative and postoperative hyperglycemia. Hyperglycemia has been associated with morbidity and mortality after infant cardiac surgery. We studied the effects on early postoperative outcomes of glucose (GLU) changes during and after pediatric cardiac surgery. Methods. The records of 144 infants less than 10 kg who underwent CPB for a variety of congenital cardiac procedures were reviewed. The GLU values (at multiple intervals during and after surgery), age, weight, CPB time, ultrafiltration volume, and risk adjustment for congenital heart surgery (RACHS-1) score were recorded. Univariate and multivariate linear and binary logistic regression were used to examine the dependence of the
composite outcome mortality or postoperative infection, the mechanical ventilation time (VENT time), and the length of stay (LOS), on these variables. Results. The RACHS-1 score was the only significant predictor of the composite variable “mortality or infection” (p ⴝ 0.008). Glucose at any time was not a significant factor predicting this outcome. Lower pre-CPB GLU, younger age, and higher RACHS-1 score were significant predictors of greater LOS and VENT time. Conclusions. In this study, post-CPB and postoperative hyperglycemia were not risk factors for postoperative morbidity and mortality after infant cardiac surgery. (Ann Thorac Surg 2010;89:181– 6) © 2010 by The Society of Thoracic Surgeons
H
GLU between 125 mg/dL and 150 mg/dL may cause glycosuria and dehydration [15], and values exceeding 250 mg/dL are associated with worse outcome after brain injury [16]. More recent published literature [17] suggests that maintaining the GLU level between 110 mg/dL and 126 mg/dL is associated with better outcome after complex cardiac surgery. Furthermore, intensive insulin therapy during cardiopulmonary bypass has been shown to attenuate the systemic inflammatory response after infant cardiac surgery [18]. Given the conflicting conclusions of the aforementioned studies, there is a need for additional evidence as to the clinical effects of hyperglycemia during and after congenital cardiac surgery. We analyzed whether GLU values during and after congenital cardiac surgery were associated with mortality or three other measures of morbidity in the postoperative period.
yperglycemia is associated with cardiopulmonary bypass. Some of the postulated mechanisms include the following: hyperoxia [1, 2], anesthetic agents [3], hypothermia, insulin and catecholamine derangements [4, 5], and utilization of dextrose containing solutions [6]. Hyperoxia is defined as the utilization of 100% fraction of inspired oxygen (Fio2) through the pump oxygenator to increase partial pressure of oxygen, arterial (Pao2) above the “normal” range of 150 to 250 mm Hg [7] during cardiopulmonary bypass (CPB). Animal studies have demonstrated that hyperoxia, with or without CPB, causes an increase in plasma glucose that reverses after return to normoxia [1]. Some clinical reports have linked hyperglycemia to increased morbidity and mortality [5, 6, 8] while others have found no relation to adverse neurodevelopmental outcome [9, 10]. In fact, studies in piglets and cats have reported that hyperglycemia is associated with better maintenance of high energy metabolites and better preservation of brain mitochondria respiratory capacity [11, 12]. Studies of the effects of using dextrose containing solutions during infant cardiac surgery show conflicting results [6, 13]. In adult cardiac surgery, evidence shows that maintaining glucose (GLU) less than180 mg/dL is associated with better outcome [14]. Conversely, the GLU threshold for adverse outcome in children is less clear. In children, Accepted for publication Aug 25, 2009. Address correspondence to Dr DeCampli, Congenital Heart Institute, Arnold Palmer Hospital for Children, 50 W Sturtevant St, Orlando, FL 32806; e-mail: [email protected].
© 2010 by The Society of Thoracic Surgeons Published by Elsevier Inc
Patients and Methods Patient Population We reviewed the charts of 144 consecutive patients weighing 10 kg or less who underwent cardiopulmonary bypass between January 2006 and October 2007 at a single institution. We obtained Institutional Review Board approval with waiver of parental consent for this study.
Anesthesia and CPB Protocol We induced and maintained anesthesia using midazolam, fentanyl, pancuronium, and sevoflurane. Propofol 0003-4975/10/$36.00 doi:10.1016/j.athoracsur.2009.08.062
PEDIATRIC CARDIAC
Perioperative Hyperglycemia: Effect on Outcome After Infant Congenital Heart Surgery
182
DeCAMPLI ET AL HYPERGLYCEMIA IN INFANT HEART SURGERY
Ann Thorac Surg 2010;89:181– 6
PEDIATRIC CARDIAC
peroxia (Fio2 ⫽ 100%, Pao2 ⬎ 400 mm Hg, and venous saturations ⬎75%) was used and acid-base management included the use of pH stat techniques for patients cooled below 28°C and alpha stat for all others. We used continuous ultrafiltration in all cases. We did not administer supplemental dextrose during CPB and glucose was not a component of the cardioplegia solution. The cardioplegia dose was 20 mL/kg (minimum of 100 mL) induction followed by maintenance doses of 10 mL/kg (minimum of 50 mL).
Abbreviations and Acronyms ASO ⫽ arterial switch operation AVC ⫽ atrioventricular canal BDG ⫽ bidirectional Glenn B-T ⫽ Blalock-Taussing CICU ⫽ cardiac intensive care unit CPB ⫽ cardiopulmonary bypass DKS ⫽ Damus-Kaye-Stansel ECG ⫽ electroencephalograph GLU ⫽ glucose LOS ⫽ length of stay MORTINF ⫽ composite for mortality ⫹ incidence of infection RACHS-1 ⫽ risk adjustment for congenital heart surgery ROC ⫽ receiver operating curve TOF ⫽ tetralogy of Fallot VENT ⫽ mechanical ventilation VSD ⫽ ventricular septal defect
Data Collected
was used in patients expected to be extubated shortly after surgery. We gave dexamethasone (1 mg/kg) upon anesthesia induction before CPB was instituted. Insulin was not used in any patient during the operative period. We treated plasma GLU lower than 60 mg/dL with dextrose 25% (1 cc/kg) during the operative period. Hyperglycemia was sporadically treated with insulin (2 subjects) in the cardiac intensive care unit (CICU). The CPB circuit consisted of the Baby RX oxygenator and venous reservoir (Terumo, Ann Arbor, MI), Capiox AF02 arterial line filter (Terumo), ¼ inch tubing arterialvenous loop, and pediatric Bio-pump BP-50 (Medtronic, Minneapolis, MN). Our CPB circuit prime volume used for patients up to 10 kg is approximately 330 cc. Prime solution consisted of whole blood or a combination of washed pack red blood cells and fresh frozen plasma depending on the availability of the former. Plasma-Lyte A (Baxter International Inc, Deerfield, IL), sodium bicarbonate (15 mEq), heparin (2,000 units), calcium chloride (200 mg), and aprotinin (2 cc/kg) were also added. Hy-
Glucose was recorded at the following times: pre-CPB, at CPB initiation, near completion of CPB, post-CPB, upon arrival to the CICU, and last value obtained within the first and second 24 hour periods (last GLU day 1 and last GLU day 2, respectively. For each patient, age, operation, pump prime glucose concentration (prime GLU), CPB time, ultrafiltration volume, serum lactate, risk adjustment for congenital heart surgery (RACHS-1) category, incidence of electroencephalograph (ECG)-evident seizures, duration of mechanical ventilation (VENT time), length of stay (LOS), evidence of tracheal, blood, urine and wound infection, and in-hospital or 30 day mortality were recorded.
Statistical Methods Univariate regression was carried out to determine single variable associations between each of the independent variables (seven GLU values, age, BYAGE, CPB time, prime GLU, ultrafiltration volume, and RACHS category) and the outcome variables (mechanical ventilation time, length of stay, and the composite of mortality ⫹ incidence of infection [MORTINF]). The latter composite score was chosen because a power analysis showed that the mortality rate itself was too low to determine its risk factors within the limits chosen for type I and II errors (␣ ⫽ 0.05 and  ⫽ 0.1), respectively, for the sample size of 144. We treated age as a dichotomous variable (BYAGE) (ⱕ30 days, ⬎30 days) but also checked for any significant differences when using age as a continuous variable. All risk factors with a p value less than 0.1 were then entered
Table 1. Demographics and Outcome Measures by RACHS-1 Category RACHS-1 Category Age (days) Ages, median (range) Weight (kg) CPB time (minutes) Ultrafiltration (mL) Mortality Ventilation time (days) Length of stay (days) Incidence of seizures Incidence of infection
2 (n ⫽ 53)
3 (n ⫽ 43)
4 (n ⫽ 34)
6 (n ⫽ 14)
181.09 ⫾ 172.89 148.5 (14–1095) 5.56 ⫾ 1.73 110.82 ⫾ 40.22 668.86 ⫾ 421.21 1 1.56 ⫾ 2.0 6.97 ⫾ 3.16 0 7
156 ⫾ 193.94 98 (3–930) 5 ⫾ 2.26 171.47 ⫾ 61.42 954.41 ⫾ 625.85 1 4.20 ⫾ 5.46 13.65 ⫾ 11.62 1 7
14.32 ⫾ 29.57 5.5 (1–164) 3.14 ⫾ 0.86 202.32 ⫾ 69.64 978.67 ⫾ 584.26 1 8.70 ⫾ 11.65 21.24 ⫾ 18.18 3 6
31.85 ⫾ 101.09 5.0 (2.0–385) 3.35 ⫾ 1.04 235.78 ⫾ 72.38 1237.14 ⫾ 783.43 2 21.07 ⫾ 31.33 50.07 ⫾ 54.68 0 5
Data expressed as mean ⫾ standard deviation. CPB ⫽ cardiopulmonary bypass;
RACHS-1 ⫽ risk adjustment for congenital heart surgery.
Table 2. Procedures by RACHS-1 Category Category 2 B-T shunt (1) VSD closure (9) BDG (17) VSD closure plus other procedure (8) Repair of TOF (14) Left ventricular outflow tract resection (1) Mitral valve repair (1) Atrial septal defect closure (2) Category 3 Left ventricle to pulmonary artery conduit (2) Right ventricular outflow tract reconstruction (5) AVC (12) Pulmonary atresia/TOF (5) B-T shunt with additional procedures (5) Central shunt plus pulmonary vein repair (2) ASO (9) DKS revision (1) Bidirectional pulmonary artery plasty (1) BDG plus tricuspid valvuloplasty (1) Category 4 Arch reconstruction (19) ASO with VSD closure (9) TOF plus unifocalization (1) Truncus arteriosus repair (2) Total anomalous pulmonary venous return repair (2) Starne’s procedure (1) Category 6 Norwood procedure (12) DKS (1) Norwood with Rastelli (1) ASO ⫽ arterial switch operation; AVC ⫽ atrioventricular canal; B-T ⫽ Blalock-Taussing; BDG ⫽ bidirectional Glenn; DKS ⫽ DamusKaye-Stansel; RACHS-1 ⫽ risk adjustment for congenital heart surgery; TOF ⫽ tetralogy of Fallot; VSD ⫽ ventricular septal defect.
into a multivariate (linear regression) analysis for continuous dependent variables (ventilator time, length of stay), or binary logistic regression for the categoric variable MORTINF to calculate the final determinates of risk. For continuous dependent variables, the tolerance (T) of each covariant was calculated to determine important multicollinearity, with T less than 0.4 taken to indicate the latter possibility. For logistic regression the receiver operating curve (ROC) was used to estimate the Cstatistic, or predictive power for the solution. Univariate regression, Student t test, and Pearson or Spearman correlation were used to compare other pairs of variables. We used SPSS version 14.0 (SPSS Inc, Chicago, IL) as our statistical analysis package. Measurements are expressed as mean and standard deviation and a p value less than 0.05 was considered statistically significant.
Results Demographic and mean outcome data stratified by RACHS-1 score are shown in Table 1. Operations are
DeCAMPLI ET AL HYPERGLYCEMIA IN INFANT HEART SURGERY
183
shown in Table 2. Peak GLU generally occurred intraoperatively post-CPB and then decreased during the next 48 hours toward normal range (Fig 1). The number and percentage of patients with GLU greater than 150 mg/dL at each time were as follows: near CPB completion (n ⫽ 72, 50%), post-CPB (n ⫽ 81, 56.2%), upon arrival to the CICU (n ⫽ 61, 42.3%), end of first day postoperative (n ⫽ 21, 14.5%), and end of second postoperative (n ⫽ 10, 6.94%). Analysis of the cohort by age was performed to ascertain if age less than 30 days was a factor affecting GLU trends. Significant difference in GLU values between the two age groups (A⫽ ⱕ30 days old, n ⫽ 63; B ⫽ ⬎30 days old, n ⫽ 81) was observed at the following periods: near CPB completion (A 133.73 ⫾ 35.62; B 160.33 ⫾ 32.24; p ⬍ 0.05); post-CPB (A 144.61 ⫾ 40.34; B 161.23 ⫾ 37.76; p ⬍ 0.05); upon arrival to the CICU (A 151.38 ⫾ 43.81; B 142.20 ⫾ 40.08; p ⬍ 0.05); end of day 1 (A 130.40 ⫾ 35.54; B 114.32 ⫾ 28.42; p ⬍ 0.05); and end of day 2 (A 111.48 ⫾ 34.60; B 106.01 ⫾ 28.60; p ⬍ 0.05). Thus, neonates demonstrated a lower GLU intraoperatively compared with non-neonates, but higher GLU level upon arrival to the CICU and through the 48 hours postoperative than nonneonates. The GLU at CPB initiation correlated weakly with prime GLU (Spearman r ⫽ 0.26, p ⫽ 0.001) and was even more weakly correlated with subsequent GLU measurements. Neither post-CPB GLU nor GLU on arrival to CICU correlated with CPB time (Spearman r ⫽ ⫺0.06 and 0.07, respectively). The GLU changes did not correlate with lactate changes in this study (Pearson r ⬍ 0.5 at all phases studied). The incidence of hyperlactemia (lactate ⬎ 3 mmol/L) exhibited a downward trend with time, as follows: near CPB end (n ⫽ 41; 28.4%); post-CPB (n ⫽ 41; 28.4%); upon arrival to the CICU (n ⫽ 31; 21.5%); end of postoperative day 1 (n ⫽ 5; 3.47%); and end of postoperative day 2 (n ⫽ 6, 4.16%). Four patients presented with EEG-evident seizures in the postoperative period. Risk factors for seizures in these patients included preoperative diagnosis of periventricular leukomalacia (1 patient) and chromosomal abnormalities (3 patients). The most common infection site was tracheal aspirate in 21 subjects. Other sites included blood (17 subjects), urine (2 subjects), wound (2 subjects), and others (2 subjects). Univariate statistics for the outcome variable MORTINF are shown in Table 3. Only BYAGE, age, and RACHS-1
Fig 1. Glucose trends. Glucose concentration (mg/dL) at various intraoperative and postoperative intervals as defined in the text. Glucose levels are expressed as mean and ⫾ standard deviation.
PEDIATRIC CARDIAC
Ann Thorac Surg 2010;89:181– 6
184
DeCAMPLI ET AL HYPERGLYCEMIA IN INFANT HEART SURGERY
Table 3. Univariate Statistics for Regression of MORTINF Variable PEDIATRIC CARDIAC
BYAGE (⬍30 and ⬎30 days) Age (days) Prime GLU (mg/dL) CPB time (minutes) Ultrafiltration vol. (mL) RACHS-1 Pre CPB GLU (mg/dL) GLU at CPB initiation GLU near end CPB Post-CPB GLU GLU upon arrival CICU Last GLU day 1 Last GLU day 2
p Value 0.09 0.08 0.85 0.80 0.40 0.008 0.21 0.79 0.20 0.16 0.14 0.16 0.98
BYAGE ⫽ age as a dichotomous variable; CICU ⫽ cardiac intensive care unit; CPB ⫽ cardiopulmonary bypass; GLU ⫽ glucose; MORTINF ⫽ composite for mortality ⫹ incidence of infection; RACHS-1 ⫽ risk adjustment for congenital heart surgery.
score were significant univariate determinants. In particular, no GLU value was a significant predictor of mortality or infection. In the multivariate logistic regression only RACHS-1 score was a significant determinant (p ⫽ 0.008). The odds ratio for death or infection was 1.47 (95% confidence interval 1.1 to 2.0) for every increase by one in RACHS-1 score. The C-statistic, or area under the ROC for this solution (indicative of its predictive power) was 0.64 ⫾ 0.05 (p ⫽ 0.01). Univariate predictors of LOS and of VENT time included age, BYAGE, CPB time, RACHS-1 score, and pre-CPB GLU. By multivariate analysis, BYAGE, RACHS-1, and pre-CPB GLU were significant predictors of LOS (p ⬍ 0.015). (A model with the continuous variable “age” instead of the dichotomous variable BYAGE produced a very similar result.) The model is shown in Table 4. The T (tolerance)statistic for age, RACHS-1, and pre-CPB GLU was 0.46, 0.50 and 0.85, respectively, and thus there was not significant multicollinearity. Only RACHS-1 score remained a significant predictor of VENT time in the multivariate analysis (p ⫽ 0.001).
Comment The highest GLU elevation occurred during the immediate post-CPB period, followed, in general, by a steady decline toward more normal ranges by the end of 48 hours in the CICU. In addition to the CPB-induced systemic inflammatory response, it is possible that other factors such as hyperoxia, prime glucose value, or the use of corticosteroids could have resulted in hyperglycemia. Interestingly, however, post-CPB GLU was not correlated with duration of CPB nor to the glucose value in the prime solution, and the latter only weakly correlated with glucose at initiation of CPB. Published studies differ in the apparent effect of intraoperative hyperglycemia on clinical outcomes [8, 9, 10, 19]. Some reports suggest hypoglycemia rather than
Ann Thorac Surg 2010;89:181– 6
hyperglycemia to be related to electroencephalographic seizures, slower electroencephalographic recovery, and an increase risk of adverse events [9, 19]. In other studies hyperglycemia has not been associated with adverse neurodevelopmental outcome [9, 10]. One study found an association between hyperglycemia and morbidity and mortality [8]. A recent report found GLU less than 109 mg/dL and greater than 143 mg/dL associated with mortality and the duration of time in which GLU greater than 126 mg/dL during the first 72 hours postoperative associated with greater LOS [17]. The present study fails to show that GLU, either before, during, or after CPB, are predictors of mortality plus infection or increase duration of ventilation time. The pre-CPB glucose was a determinant of hospital LOS, with lower glucose values correlated with increased LOS. Despite the fact that the LOS solution did not suffer from significant multicollinearity, the independent variables age and pre-CPB GLU were correlated to some extent (Pearson r ⫽ 0.39). This raises the possibility that, for the patients with longest duration hospital stay, the lower glucoses were due largely to the fact that these patients were also younger. Indeed, from a mechanistic viewpoint, it is unlikely that lower pre-CPB glucose would cause longer hospital stay. However, the sample size of this cohort does not allow us to statistically prove it. This study has several limitations. First, it is a nonblinded retrospective study. Second, we examined only short-term clinical outcome. Third, the relative small sample size did not allow us to discriminate small differences in morbidity or mortality, to analyze subgroups, or to consider additional possible risk factors. For example, it is obvious that associated noncardiac anomalies or intervening complications later in the hospital stay should influence hospital length of stay. It is unlikely, however, that the inclusion of such risk factors would cause perioperative GLU values to reenter the multivariate solution as significant covariates (risks). In conclusion we have demonstrated that intraoperative hyperglycemia commonly occurs during infant cardiac surgery. Intraoperative hyperglycemia per se does not appear to be associated with adverse early outcomes. Hyperglycemia after infant cardiac surgery peaks early after CPB and decreases to normal values during the next 48 hours postsurgery unless other complications intervene. Thus, we find no evidence, for example, to support the aggressive use of insulin to manage early postoperative hyperglycemia after infant cardiac surgery. Table 4. Multivariate Regression Model for Log (LOS) Variable Constant RACHS-1 BYAGE Pre CPB GLU(mg/dL)
Coefficient
p Value
1.26 0.107 ⫺0.002 ⫺0.002
0.004 0.004 0.003 0.014
BYAGE ⫽ age as a dichotomous variable; CPB ⫽ cardiopulmonary bypass; GLU ⫽ glucose; LOS ⫽ length of stay; RACHS-1 ⫽ risk adjustment for congenital heart surgery.
The authors thank the Orlando Health Foundation for their generous support of this study.
References 1. Bandali KS, Belanger MP, Wittnich C. Is hyperglycemia seen in children during cardiopulmonary bypass a result of hyperoxia? J Thorac Cardiovasc Surg 2001;122:753– 8. 2. Bandali KS, Belanger MP, Wittnich C. Does hyperoxia affect glucose regulation and transport in the newborn. J Thorac Cardiovasc Surg 2003;126:1730 –5. 3. Ellis DJ, Steward DJ. Fentanyl dosage is associated with reduced blood glucose in pediatric patients after hypothermic cardiopulmonary bypass. Anesthesiology 1990;72:812–5. 4. Benzing G III, Francis PD, Kaplan S, Helmsworth JA, Sperling MA. Glucose and insulin changes in infants and children undergoing hypothermic open-heart surgery. Am J Cardiol 1983;52:133– 6. 5. Anand KJS, Phil D, Hansen DD, Hickey PR. Hormonalmetabolic stress responses in neonates undergoing cardiac surgery. Anesthesiology 1990;73:661–70. 6. Bell C, Hughes CW, Oh TH, Donielson DW, O’Connor T. The effect of intravenous dextrose infusion on postbypass hyperglycemia in pediatric patients undergoing cardiac operations. J Clin Anesth 1993;5:381–5. 7. Brown DM, Holt DW, Edwards JT, Burnett RJ III. Normoxia vs. hyperoxia: impact of oxygen tension strategies on outcome for patients receiving cardiopulmonary bypass for routine cardiac surgical repair. J Extra Corpor Technol 2006; 38:241– 8. 8. Yates AR, Dyke PC II, Taeed R, et al. Hyperglycemia is a marker for poor outcome in the postoperative pediatric cardiac patient. Pediatr Crit Care Med 2006;7:351–5. 9. de Ferranti S, Gauvreau K, Hickey PR, et al. Intraoperative hyperglycemia during infant cardiac surgery is not associ-
DeCAMPLI ET AL HYPERGLYCEMIA IN INFANT HEART SURGERY
10.
11. 12. 13. 14. 15. 16. 17.
18.
19.
185
ated with adverse neurodevelopmental outcome at 1, 4, and 8 years. Anesthesiology 2004;100:1345–52. Ballweg JA, Wernovsky G, Ittenbach RF, et al. Hyperglycemia after infant cardiac surgery does not adversely impact neurodevelopmental outcome. Ann Thorac Surg 2007;84: 2052– 8. Laptook AR, Cobertt RJT, Arencibia-Mireles O, Ruley J. Glucose-associated alterations in ischemic brain metabolism of neonatal piglet. Stroke 1992;23:1504 –11. Wagner KR, Myers RE. Hyperglycemia preserves brain mitochondrial respiration during anoxia. J Neurochem 1986; 47:1620 – 6. Aouifi AA, Neidecker J, Vedrinne C, et al. Glucose versus lactated ringer’s solution during pediatric cardiac surgery. J Cardiothorac Vasc Anesth 1997;11:411– 4. Bojar RM. Early postoperative care. In: Manual of perioperative care in adult cardiac surgery. 4th ed. Maden, MA: Blackwell Publishing, Inc; 2005:243. Gregory GA. Monitoring during surgery. In: Pediatric anesthesia. 4th ed. Philadelphia, PA: Churchill Livingstone; 2002: 260. Vavilala M. Pediatric trauma and burn management. In: Litman RS. Pediatric anesthesia. 1st ed. Philadelphia, PA: Elsevier Mosby; 2004:307. Polito A, Thiagarajan RR, Laussen PC, et al. Association between intraoperative and early postoperative glucose levels and adverse outcomes after complex congenital heart surgery. Circulation 2008;118:2235– 42. Chun-Hu G, Qin C, Yun-Ya W, et al. Effects of insulin therapy on inflammatory mediators in infants undergoing cardiac surgery with cardiopulmonary bypass. Cytokine 2008; 44:96 –100. Rossano JW, Taylor MD, Smith EO, et al. Glycemic profile in infants who have undergone the arterial switch operation: hyperglycemia is not associated with adverse events. J Thorac Cardiovasc Surg 2008;135:739 – 45.
INVITED COMMENTARY This article is a descriptive observation study that attempted to determine if glucose level was associated with adverse outcome in a population of 144 pediatric cardiac surgical patients [1]. In the final version and analysis of this study, glucose was treated as a continuous variable, and in fact, no level of glucose predicted an increased risk for the primary composite mortality and infection end point (MORTINF). Only prebypass glucose, risk adjustment for congenital heart surgery (RACHS-1) level, and being a neonate predicted increased length of stay. The authors conclude that hyperglycemia occurs commonly after infant cardiac surgery and that hyperglycemia was not associated with adverse outcome. Interestingly, in no patient did serum glucose level rise above 200 mg/dL in the study period, which contrasts with other reports in which perioperative glucose often rises above 200 mg/dL. Glucose levels were somewhat arbitrarily selected throughout the postoperative period as the “last glucose day 1 and 2.” In fact, the significance of the “postbypass glucose level” is questionable because continuous ultrafiltration was used throughout the bypass period and the ultrafiltration process removes glucose. The study was not appropriately powered for mortality as a primary end point, so the authors chose to combine © 2010 by The Society of Thoracic Surgeons Published by Elsevier Inc
mortality and infections as a composite end point. This choice appears to be based on the assumption that hyperglycemia may be associated with increased risk of perioperative infection. Although this association seems reasonably sound in the adult cardiac surgical population, there is no definitive support for this in the pediatric cardiac surgical population. This association has not been clearly determined in existing literature, and the MORTINF composite outcome has not been validated in previous studies. It has recently been shown that occurrence of hyperglycemia in adult cardiac surgical patients is not benign. Several studies have shown that tight glycemic control in the perioperative period decreases mortality after cardiac procedures and reduces the incidence of postoperative adverse events, including infectious complications (pneumonia and mediastinitis), myocardial complications (low cardiac output, atrial fibrillation, recurrent ischemia), and prolonged mechanical ventilation and intensive care unit length of stay. The evidence for benefit of tight glycemic control in the pediatric cardiac surgical population is both insufficient and conflicting. A recent article by Ballweg and colleagues [2] suggests that hyperglycemia after cardiopulmonary bypass did not result in increased risk of neuro0003-4975/10/$36.00 doi:10.1016/j.athoracsur.2009.11.006
PEDIATRIC CARDIAC
Ann Thorac Surg 2010;89:181– 6