Risk Factors for Mortality and Morbidity After the Neonatal Blalock-Taussig Shunt Procedure

Transcription

1 ORIGINAL ARTICLES: SURGERY: The Annals of Thoracic Surgery CME Program is located online at To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal. Risk Factors for Mortality and Morbidity After the Neonatal Blalock-Taussig Shunt Procedure Orlando Petrucci, MD, PhD, Sean M. O Brien, PhD, Marshall L. Jacobs, MD, Jeffrey P. Jacobs, MD, Peter B. Manning, MD, and Pirooz Eghtesady, MD, PhD Division of Cardiac Surgery, Faculty of Medical Science, State University of Campinas, UNICAMP, Campinas, São Paulo, Brazil; Duke Clinical Research Institute, Durham, North Carolina; Center for Pediatric and Congenital Heart Disease, Cleveland Clinic Foundation, Cleveland, Ohio; The Congenital Heart Institute of Florida, University of South Florida, All Children s Hospital and Children s Hospital of Tampa, Saint Petersburg, Florida; and The Heart Institute, Division of Cardiothoracic Surgery, Cincinnati Children s Hospital Medical Center and University of Cincinnati, Cincinnati, Ohio Background. Perioperative advances have led to significant improvements in outcomes after many complex neonatal open heart procedures. Whether similar improvements have been realized for the modified Blalock- Taussig shunt, the most common palliative neonatal closed-heart procedure, is not known. Methods. Data were abstracted from The Society of Thoracic Surgeons Congenital Heart Surgery Database (2002 to 2009). Inclusion criteria were all neonates who received a modified Blalock-Taussig shunt with or without cardiopulmonary bypass, and with or without concomitant ligation of a patent ductus arteriosus. Discharge mortality was the primary end point. A composite morbidity end point one or more of the following: postoperative extracorporeal membrane oxygenation, low cardiac output, or unplanned reoperation. Associations with patient and procedural variables were assessed with univariable and multivariable analyses. Results. The inclusion criteria were met by 1273 patients. The discharge mortality rate was 7.2%, and composite morbidity, as defined, was 13.1%. Primary diagnoses were classified as (1) those potentially amenable to biventricular repair (62%), (2) functionally univentricular hearts (22%), and (3) pulmonary atresia with intact ventricular septum (PA/IVS; 14%), and miscellaneous (2%). Discharge mortality stratified by primary diagnoses was PA/IVS (15.6%), functionally univentricular hearts (7.2%), and diagnoses potentially amenable to biventricular repair (5.1%). Need for preoperative ventilatory support, diagnosis of PA/IVS or functionally univentricular hearts, and any weight less than 3 kg, were risk factors for death. Preoperative acidosis or shock (resolved or persistent) and diagnosis of PA/IVS or functionally univentricular hearts were predictors of composite morbidity. Nearly 33% of the deaths occurred within 24 hours postoperatively, and 75% within the first 30 days. Conclusions. The mortality rate after the neonatal modified Blalock-Taussig shunt remains high, particularly for infants weighing less than 3 kg and those with the diagnosis of PA/IVS. (Ann Thorac Surg 2011;92:642 52) 2011 by The Society of Thoracic Surgeons Systemic-to-pulmonary shunt procedures remain indispensable in the management of select neonates and infants with complex cardiac heart defects [1, 2]. The modified Blalock Taussig shunt (MBTS) is the most common palliative operation performed on this subset of patients [1, 3]. Despite numerous improvements in diagnosis, intensive care, and intraoperative management in recent decades, the overall mortality rate after the MBTS appears to trail successes in other arenas of Accepted for publication Feb 7, Presented at the Fifty-seventh Annual Meeting of the Southern Thoracic Surgical Association, Orlando, FL, Nov 3 6, Address correspondence to Dr Eghtesady, The Heart Institute, Division of Cardiothoracic Surgery, Cincinnati Children s Medical Center, 3333 Burnet Ave, Cincinnati, OH ; pirooz.eghtesady@cchmc. org. congenital heart surgery. Prior reports extending from neonatal to even older patients show a mortality rate of 2.3% to 16.0% [4 7]. Several single-institution studies or case series have investigated risk factors for morbidity and mortality after the MBTS [4 7] and have identified age, weight, and underlying cardiac diagnosis as among the risk factors for death. Our study used the large multi-institutional Society of Thoracic Surgeons (STS) Congenital database to identify potential risk factors in a recent cohort of patients. Material and Methods Database The STS Congenital Heart Surgery Database collects procedurally related operative data on neonates, infants, 2011 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc doi: /j.athoracsur

2 Ann Thorac Surg PETRUCCI ET AL 2011;92: MORTALITY AND MORBIDITY AFTER NEONATAL BTS 643 children, and adults undergoing operations for congenital heart anomalies. As of 2009, 86 institutions have contributed data. Data collected in the Congenital Heart Surgery Database include basic demographic information, anatomic diagnosis, associated noncardiac abnormalities, preoperative risk factors, intraoperative details, surgical procedure performed, postoperative complications, and operative death. The Duke University Institutional Review Board reviews research performed on the STS Database at the Duke Clinical Research Institute. Because the data used for research represent a limited data set (no direct patient identifiers) that was originally collected for nonresearch purposes, and the investigators do not know the identity of individual patients, the analysis of these data was declared by the Duke University Health System Institutional Review Board to be research not involving human subjects. Patient Population The data source was the STS Congenital Heart Surgery Database. Inclusion criteria were (1) surgery in years 2002 through 2009 inclusive, (2) age at operation younger than 30 days, (3) MBTS index procedure, with or without cardiopulmonary bypass (CPB), allowable concomitant procedure of surgical closure of patent ductus arteriosus (PDA), (3) weight at operation 1.5 kg or heavier. Any concomitant procedure other than surgical closure of PDA was grounds for exclusion, including pulmonary arterioplasty. Data from participants with more than 15% of data missing for any of the following fields were also excluded: discharge mortality, postoperative length of stay, preoperative risk factors, noncardiac abnormalities. Also excluded were patients with a primary diagnosis of hypoplastic left heart syndrome or conduit failure. End Points The primary end point was discharge mortality. The secondary end point was composite morbidity. Composite morbidity was defined as occurrence of any one or more of the following: postoperative need for extracorporeal membrane oxygenation (ECMO) or mechanical circulatory support (MCS), postoperative low cardiac output, and unplanned reoperation after MBTS. Current definitions of the individual variables captured in the composite end point can be found at the STS Web site ( congenital-heart-surgery-database). Risk Factors and Explanatory Variables Patient preoperative characteristics included in the analysis were cardiac diagnosis (categorized as single ventricle, double ventricle, pulmonary atresia/intact ventricular septum diagnosis, and other), presence of any noncardiac abnormality, preoperative MCS, preoperative mechanical ventilatory support, arrhythmia, acidosis or shock, renal failure with or without dialysis, and weight. Operative characteristics included use of CPB, duration of CPB, and whether or not a PDA was ligated. Continuous variables were analyzed in their original continuous form and as categories. Categories for display purposes were age into four categories (0 to 6, 7 to 13, 14 to 20, and 21 days or older), weight into five categories (1500 to 1999, 2000 to 2499, 2500 to 2999, 3000 to 3499, and 3500 grams or heavier), and CPB time into five categories (0 to 59 seconds, 1 to 29 minutes, 30 to 59, 60 to 89, and 90 minutes or longer). The average annual volume of MBTS procedures was calculated for each center by dividing its total number of MBTS procedures by the number of months of participation in the database and multiplying by 12. Statistical Analysis Preoperative and operative characteristics were summarized as frequencies and percentages. The frequency of each end point was compared across levels of each explanatory variable using the Pearson 2 test for unordered categoric variables and the Mantel 2 test for ordered categories. In each case, the null hypothesis was that occurrence of the end point would be unrelated to the candidate factor. The unadjusted association between weight as a continuous variable and discharge mortality was estimated using nonparametric smoothing splines as implemented in the S-PLUS (TIBCO Software Inc, Palo Alto, CA) generalized additive model function. The estimated probability of in-hospital death was plotted along with 95% point-wise confidence intervals (CI) as a function of the patient s preoperative weight. The multivariable association between patient characteristics and each end point was assessed using logistic regression. Explanatory factors were identical to those described above (see Risk factors and explanatory variables) except for the exclusion of age (which highly correlated with weight), and three risk factors preoperative MCS, preoperative arrhythmia, preoperative renal failure that were too rare to analyze reliably. CPB usage was analyzed as a dichotomous (yes/no) variable without adjustment for CPB duration. Estimated odds ratios (ORs) were reported for each variable using a 95% CI. Regression coefficients were estimated using generalized estimating equations methodology to account for statistical dependence between patients from the same hospital. The center case volume and mortality association was assessed using logistic regression. Explanatory factors were the same as the previous paragraph with the addition of a single variable for average annual MBTS volume. The timing of in-hospital deaths was explored graphically by creating a histogram of the number of days between the operation and death. Because many hospitals did not monitor patients after discharge, patients who died after discharge were not included in this analysis. The characteristics of patients who died in the hospital within 48 hours vs after 48 hours were compared using Pearson 2 tests for unordered categoric variables and Mantel 2 tests for ordered categories. Missing data were rare, with more than 99% complete data for each binary risk factor and 100% complete data

3 644 PETRUCCI ET AL Ann Thorac Surg MORTALITY AND MORBIDITY AFTER NEONATAL BTS 2011;92: for other categoric and continuous risk factors. For descriptive analyses, patients with missing data were excluded from analyses involving the missing variable. For multivariable analyses, missing binary risk factors were imputed to the negative. The values for p were not adjusted for multiple comparisons. All analyses were performed using SAS 9.2 software (SAS Institute, Cary, NC) and S-PLUS 6.1 software. Results A total of 1604 patients from 82 hospitals met the inclusion criteria. Of these, 12 hospitals failed to meet the threshold of more than 85% complete data and were excluded, leaving 1280 patients from 70 hospitals. Finally, 5 patients with missing weight and 2 patients who weighed less than 1.5 kg were excluded, leaving a final population of 1273 operations from 70 hospitals. The mean age at operation was days, with 740 boys (58.13%) and 533 girls (41.87%). Overall, 91 patients did not survive to discharge after the MBTS (7.2% mortality). The composite morbidity, as defined, occurred in 166 patients (13.1%), including 97 patients (7.6%) with unplanned reoperation, 67 (5.3%) with postoperative low cardiac output, and 39 (3.1%) requiring ECMO or MCS in the postoperative period. The univariable analyses of preoperative and intraoperative variables for death and composite morbidity are reported in Table 1. We observed that death was significantly associated with preoperative MCS, renal failure with or without dialysis, mechanical ventilatory support, acidosis or shock (resolved or persistent), diagnosis (underlying congenital heart defect single ventricle, double ventricle, pulmonary atresia with IVS [PA/IVS] or other), and weight. Preoperative and patient factors that were significantly associated with composite morbidity included presence of noncardiac abnormality, preoperative ventilatory support, acidosis or shock (resolved or persistent), and diagnosis of underlying congenital heart disease. Of note, use or nonuse of CPB, and concomitant surgical closure of PDA did not appear to affect the risk of death or composite morbidity. The STS Database collected information from 731 patients related to status of prematurity and gestational age at the time of operation. Of these, 699 had complete data. The mortality rate was 10.5% for premature patients (12 of 114) vs 7.0% (41 of 585), but this apparent difference was not statistically significant (p 0.19). The composite morbidity rate was 15.8% for premature infants (18 of 114) vs 15.6% for full-term patients (91 of 585), which was similar (p 0.95). Because weight was a significant risk factor for death, we further explored the effect of weight as a continuous variable and as a categoric variable in increments of 500 grams. Figure 1A demonstrates the relationship between weight as a continuous variable and the risk of death after MBTS, and Table 2 reports the risk relative to weight in increments of 500 grams. We next used multivariable analysis to examine the effect of the various risk factors on mortality and morbidity. Once again, preoperative ventilatory support and weight of less than 3 kg were predictive for in-hospital death (Table 2). For the preoperative diagnosis, patients with the single-ventricle and double-ventricle diagnoses were less likely to die compared with the reference category of PA/IVS (Table 2). In contrast, only preoperative acidosis or shock diagnoses were predictive of composite morbidity (Table 2). Of note, 24 of the 39 patients (62%) who required postoperative ECMO or MCS did not survive, nor did 43 of 97 patients (44%) who required reoperation. Analysis of the interaction between diagnosis and morbidity showed that for patients who experienced the composite morbidity, the single-ventricle and PA/IVS patients had a higher mortality rate. The other diagnoses, however, even in the setting of composite morbidity, were not associated with increased death (Table 3). Because we observed a 13% rate of composite morbidity, we evaluated the relation of composite morbidity and death. Overall, the composite morbidity was associated with a higher mortality rate (OR, 9.0; 95% CI, 5.8 to 14.2). We observed 166 patients with composite morbidity, and nearly 30% died in the hospital. To assess the effect of center experience with MBTS with the outcome measures, we looked at center case volume as well as MBTS specific volume. Neither, however, was associated with mortality, with an OR per 10-unit increase in average MBTS volume of 0.98 (95% CI, 0.85 to 1.13; p 0.78). We next examined the time of death after MBTS to determine whether there was a period of highest risk. Figure 1B shows the histogram of deaths observed vs time after the operation, and nearly 33% of deaths occurred within 24 hours and 75% within 30 days. The relationship between the various risk factors and the timing of death is reported in Table 4 and Table 5. In the univariable analysis, use of CPB for performance of MBTS and need for postoperative MCS were associated with timing of death when those patients who died within 48 hours of their operation were compared with those who died later. We observed a higher incidence of death after 48 hours postoperatively when any MCS was necessary. Conversely, the patients who underwent MBTS with CPB were more likely to die within 48 hours of the operation. Duration of CPB, however, did not correlate with timing of death in the same analysis. Comment The creation of systemic-to-pulmonary shunts is an effective palliative therapy for some patients with congenital heart defects [4, 7 9]. A number of single-center reports (Table 6) have evaluated the outcome of the MBTS. Most of these were from an earlier era, before many improvements occurred in perioperative monitoring and care. These studies reported a mortality range of 3.7% to 14%. Among these, Gold and colleagues [10] reported the lowest mortality rate, but the patients in their series were 3 months old at operation, and about

4 Ann Thorac Surg PETRUCCI ET AL 2011;92: MORTALITY AND MORBIDITY AFTER NEONATAL BTS 645 Table 1. Univariable Analysis of Preoperative and Intraoperative Factors for Death and Composite Morbidity Composite Morbidity Death Variable Totals No. (%) No. Rate p Value No. Rate p Value Any noncardiac abnormality No 863 (67.79) Yes 402 (31.58) Pre-op mechanical circulatory support No 1259 (98.90) Yes 2 (0.16) Age categorized by days to (52.71) to (29.22) to (11.00) (7.07) Pre-op mechanical ventilatory support No 822 (64.57) Yes 439 (34.49) Pre-op arrhythmia 0.11 No 1244 (97.72) Yes 17 (1.34) Pre-op acidosis/shock (persistent or resolved) No 1205 (94.66) Yes 56 (4.40) Premature birth a No 585 ( Yes 114 (16.31) Pre-op renal failure with or without dialysis No 1259 (98.90) Yes 2 (0.16) Weight, grams (3.14) (15.08) (28.28) (32.68) (20.82) Cardiopulmonary bypass used No 1115 (87.59) Yes 158 (12.41) Cardiopulmonary bypass time, min and 1 5 (3.16) and (13.29) and (41.14) and (24.68) (15.19) PDA ligation performed No 640 (50.27) Yes 633 (49.73) Pre-op diagnosis categorized Single-ventricle 277 (21.76) Double-ventricle 790 (62.06) PA/IVS 179 (14.06) Other diagnoses 27 (2.12) a Premature birth was defined as gestational age of less than 37 weeks. IVS intact ventricular septum; PA pulmonary atresia; PDA patent ductus arteriosus.

5 646 PETRUCCI ET AL Ann Thorac Surg MORTALITY AND MORBIDITY AFTER NEONATAL BTS 2011;92: Fig 1. (A) Estimated mortality risk with weight as continuous variable. (B) Time to death in days after the modified Blalock- Taussig shunt. 50% of their patients had some variant of tetralogy of Fallot, thus belonging to our double-ventricle category (mortality of 5%). Lamberti and colleagues [7] showed a mortality rate of 2.3%, with 1 death in a 1.7-kg neonate with PA; this series, however, consisted of patients with a mean age of 6 months (vs 8.3 days in the current series). Williams and colleagues [9] report the largest series of Blalock-Taussig cases examined 2000 cases over 6 decades with an overall mortality of 14%; in their series, however, patients were a mean age of 8.3 years and a mean weight of 18.9 kg for the first Blalock-Taussig procedure [9]. Our study examined the risk factors for mortality, morbidity, and timing of death during the in-hospital stay in a relatively contemporary group of neonates undergoing MBTS. Further, by using the multiinstitutional STS Congenital Database, our findings likely represent a larger and more heterogenous practice pattern. Remarkably, the data suggest that MBTS continues to be a relatively high-risk procedure, with an overall mortality of 7.2%, despite being performed most commonly as a closed-heart procedure without CPB. This contrasts with a mortality risk in the range of many neonatal open heart procedures, excluding stage 1 palliation of hypoplastic left heart syndrome [11]. For example, data from a 2010 STS report show the MBTS is essentially riskier than the arterial switch operation with ventricular septal defect closure (6.9% mortality) and is a little safer than repair of interrupted aortic arch (8.1% mortality) or truncus arteriosus repair (12.1%). Considering that our report only focuses on in-hospital mortality, and many have reported on continued risk of death after discharge (about 10% to 15%), patients undergoing the MBTS Table 2. Multivariable Analysis of In-Hospital Mortality and Composite Morbidity a In-Hospital Mortality Composite Morbidity Variable OR (95% CI) p Value OR (95% CI) p Value Any noncardiac abnormality 1.11 ( ) (0.93, 2.34) 0.10 Pre-op ventilation support 1.51 ( ) ( ) 0.21 Pre-op acidosis or shock 1.98 ( ) ( ) 0.01 Weight in kg 3.0 to ( ) ( ) to ( ) ( ) to ( ) ( ) to ( ) ( ) 0.42 CPB used 1.31 ( ) ( ) 0.34 PDA ligation performed 0.94 ( ) ( ) 0.91 PA/IVS Reference Pre-op diagnosis category Single-ventricle 0.4 ( ) ( ) 0.11 Double-ventricle 0.25 ( ) ( ) 0.01 Other 0.58 ( ) ( ) 0.60 a Defined as any of the following: postoperative extracorporeal membrane oxygenation or mechanical circulatory support, low cardiac output, or unplanned reoperation. CI confidence interval; CPB cardiopulmonary bypass; IVS intact ventricular septum; OR odds ratio; PA pulmonary atresia; PDA patent ductus arteriosus.

6 Ann Thorac Surg PETRUCCI ET AL 2011;92: MORTALITY AND MORBIDITY AFTER NEONATAL BTS 647 Table 3. Overall and Categorized Diagnoses Association of Composite Morbidity and Mortality Death, No. (%) Overall No Yes Total OR (95% CI) Composite morbidity No 1059 (95.92) 45 (4.08) ( ) Yes 120 (72.29) 46 (27.71) 166 Single-ventricle diagnoses No 232 (95.87) 10 (4.13) ( ) Yes 25 (71.43) 10 (28.57) 35 Double-ventricle diagnoses No 676 (97.13) 20 (2.87) ( ) Yes 71 (78.02) 20 (21.98) 91 PA/IVS diagnoses No 133 (91.10) 13 (8.90) ( ) Yes 18 (54.55) 15 (45.45) 33 Other diagnoses No 18 (90.00) 2 (10.00) ( ) Yes 6 (85.71) 1 (14.29) 7 CI confidence interval; IVS intact ventricular septum; OR odds ratio; PA pulmonary atresia. clearly represent one of the highest-risk population of patients, particularly when considering the relatively simpler nature of the intervention compared with other neonatal procedures, except coarctation repair. In this large population of neonates, we found that age, gender, and race did not affect mortality rates after the MBTS. Similar results were previously described [12 14]. Our data also demonstrate that patient weight continues to be a significant risk factor for death after MBTS, even with improvements of anesthesia, surgical techniques, shunt material, and postoperative care. In particular, our data suggest that this risk increases incrementally as patient weight decreases. Patients who weighed less than 2.5 kg showed a mortality rate of 15.6%, nearly five times the risk of death for patients heavier than 3.5 kg. Earlier publications reported weight at operation was predictive of death [1, 3, 6, 15]; some have attributed this to lack of appropriately sized polytetrafluoroethylene grafts smaller than 3.0 mm diameter or a pulmonary artery size, or both. Recent reports also confirm this important risk factor for neonatal patients undergoing MBTS [3, 16, 17]. Alkhulaifi and colleagues [18] also showed in their recent work that weight was predictive for death, and the cutoff value in their series was 2.0 kg. Along these lines, Curzon and colleagues [11] used the STS Database to study the neonatal population who underwent any cardiac operation, and among 3022 patients, found 512 who weighed less than 2.5 kg. Their patients who weighed less than 2.5 kg had an increased mortality risk rate, varying from 1.5% to 8.8%, even when stratified by the Risk Adjustment in Congenital Heart Surgery-1 or Aristotle Basic Complexity scoring system. Taken together these data indicate lower weight is a risk factor for MBTS as it is for all cardiac surgical procedures in neonates. Some have advocated that this risk can be mitigated with continued prostaglandin support while awaiting the growth of the child to a theoretical threshold of 3 kg, especially if performance of the MBTS does not impact timing of hospital discharge. Our study cannot answer that question; a prospective randomized study, likely multiinstitutional in nature, may be the only way to accurately answer such a question. Some reports have raised the question of whether center case volume affects outcomes, especially with increasing the case complexity [19]. In the present series, we did not observe any association of mortality with center case volume This finding suggests that perhaps technical issues or system factors are not the primary reason for the observed outcomes; rather, patientspecific factors play a more important role. Our data also show that patients with the diagnosis of PA/IVS continue to have the highest risk of death after the MBTS, with an overall rate of 15.6%, which is similar to prior reports [20, 21]. Clearly, outcomes for this diagnosis continue to be quite poor, especially compared with other single-ventricle anomalies. Because this risk may be related to the prevalence of ventricular-coronary artery fistulas or sinusoids, our results prompt the question of whether current diagnostic modalities inadequately identify the sinusoids or are the poor results a reflection of poor judgment to proceed with the MBTS despite such knowledge. The prematurity at the time of the MBTS procedure might have an important effect on death. Pulmonary vascular resistance is likely higher in a 2.0 kg baby at 28-weeks of gestation vs a similar 2.0 kg (small-forgestational age) 40-week newborn. We did observe a propensity toward a higher mortality rate with prematurity, but our results did not reach statistical significance, likely related to the limited data. We observed

7 648 PETRUCCI ET AL Ann Thorac Surg MORTALITY AND MORBIDITY AFTER NEONATAL BTS 2011;92: Table 4. Relationship Between Preoperative Risk Factors and Timing of Death Occurrence of Death Post-Op Preoperative Variable and Level Overall 48 hours 48 hours (n 91) (n 20) (n 71) No. (%) No. (%) No. (%) p Value a Primary diagnosis categorized Single-ventricle diagnoses 20 (21.98) 5 (25.00) 15 (21.13) 0.82 Double-ventricle diagnoses 40 (43.96) 7 (35.00) 33 (46.48) PA/IVS diagnoses 28 (30.77) 7 (35.00) 21 (29.58) Other diagnoses 3 (3.30) 1 (5.00) 2 (2.82) Weight categorized, grams b (5.49) 1 (5.00) 4 (5.63) (32.97) 7 (35.00) 23 (32.39) (31.87) 6 (30.00) 23 (32.39) (18.68) 5 (25.00) 12 (16.90) (10.99) 1 (5.00) 9 (12.68) Any noncardiac abnormality reported Missing 1 (1.10) 0 (0.00) 1 (1.41) 0.38 No 57 (62.64) 11 (55.00) 46 (64.79) Yes 33 (36.26) 9 (45.00) 24 (33.80) Mechanical circulatory support No 90 (98.90) 20 (100.00) 70 (98.59) 0.60 Yes 1 (1.10) 0 (0.00) 1 (1.41) Mechanical ventilatory support No 47 (51.65) 10 (50.00) 37 (52.11) 0.87 Yes 44 (48.35) 10 (50.00) 34 (47.89) Acidosis or shock No 82 (90.11) 17 (85.00) 65 (91.55) 0.39 Yes 9 (9.89) 3 (15.00) 6 (8.45) Renal failure with or without dialysis No 90 (98.90) 20 (100.00) 70 (98.59) 0.60 Yes 1 (1.10) 0 (0.00) 1 (1.41) a Pearson X 2 test for unordered categoric variables; b categorized CPB run time in minutes. IVS intact ventricular septum; PA pulmonary atresia. similar findings related to the composite morbidity, with no association between prematurity and increased morbidity, perhaps for the same limitations noted earlier. Lastly, the need for preoperative mechanical ventilatory support was a significant risk factor for death and composite morbidity in our study. This likely reflects the precarious condition of these neonates, because most patients who require MBTS do not require ventilatory support, unless for presumptive prostaglandin-induced apnea. One could also hypothesize that mechanical ventilation further compromises the circulation of these infants, perhaps making it more difficult to manage the ratio of pulmonary-to-systemic blood flow, resulting in a higher prevalence of suboptimal outcomes. Our data, however, do not allow the dissection of such hypotheses. The present article also evaluates potential factors that could predict postoperative challenges of unplanned reoperation, low cardiac output, and need for ECMO or MCS. These end points are important not only from a clinical perspective but also potentially affect hospital cost and quality of life. Although univariable analysis showed any noncardiac abnormality, preoperative acidosis or shock diagnoses, MCS and diagnosis as predictive for significant morbidity, only preoperative acidosis was significant in the multivariable analysis. In part, this likely reflects an inherent selection bias a child would have to survive to develop the highlighted postoperative morbidity. These data suggest, however, that significant postoperative morbidity is likely to develop in neonates with preoperative acidosis or shock should they survive the procedure. Further, nearly 62% of those in our study who required ECMO support after MBTS did not survive to hospital discharge. These data are consistent with a recent report from Toronto group [22] that examined the outcome of ECMO support for single-ventricle lesions; only 1 of 4 patients in their group who required ECMO after MBTS survived to hospital discharge. In essence, ECMO support appears to delay the inevitable death in these

8 Ann Thorac Surg PETRUCCI ET AL 2011;92: MORTALITY AND MORBIDITY AFTER NEONATAL BTS 649 Table 5. Relationship Between Intraoperative and Postoperative Risk Factors and Timing of Death Occurrence of Death Post-op Variable and Level Overall 48 hours 48 hours (n 91) (n 20) (n 71) No. (%) No. (%) No. (%) p Value a Intraoperative CPB used No 77 (84.62) 14 (70.00) 63 ( Yes 14 (15.38) 6 (30.00) 8 (11.27) CPB time categorized, min b 0 and 1 1 (7.14) 0 (0.00) 1 (12.50) and 30 3 (21.43) 0 (0.00) 3 (37.50) 30 and 60 3 (21.43) 3 (50.00) 0 (0.00) 60 and 90 4 (28.57) 0 (0.00) 4 (50.00) 90 3 (21.43) 3 (50.00) 0 (0.00) PDA ligation performed No 44 (48.35) 11 (55.00) 33 (46.48) 0.50 Yes 47 (51.65) 9 (45.00) 38 (53.52) Postoperative Mechanical circulatory support c Missing 1 (1.10) 0 (0.00) 1 (1.41) 0.01 No 66 (72.53) 19 (95.00) 47 (66.20) Yes 24 (26.37) 1 (5.00) 23 (32.39) Low cardiac output No 70 (76.92) 16 (80.00) 54 (76.06) 0.79 Yes 20 (21.98) 4 (20.00) 16 (22.54) Re-op during this admission (unplanned) No 67 (73.63) 17 (85.00) 50 (70.42) 0.22 Yes 23 (25.27) 3 (15.00) 20 (28.17) a Pearson X 2 test for unordered categoric variables; b categorized weight in grams; c includes intraaortic balloon pump, ventricular assistance device, extracorporeal membrane oxygenation, and cardiopulmonary support. CPB cardiopulmonary bypass; PDA persistent ductus arteriosus. patients and likely is a reflection of perhaps further patient or technical factors beyond the perioperative stress of the operation (ie, not just low cardiac output). In our study, similarly, 43 of 97 of those who required a reoperation did not survive. We are currently analyzing the nature of these reoperations. Intraoperative comple- Table 6. Mortality and Risk Factors From Historical Series First Author Year Mortality (%) Patients (N) Risk Factors Mean Age (Average Age) Lamberti [7] mon Gold [10] (3 mon) Gladman [8] a 58 days Alkhulaifi [18] Pulmonary hypoplasia 11.5 days Pre-op mechanical ventilation Shunt size Weight 2000 grams Rao et al. [24] Restrictive ASD 14.1 days Univentricular physiology Post-op reintervention Williams [9] years Ahmad [12] Age 11.2 days a All patients had tetralogy of Fallot. ASD atrial septal defect.

9 650 PETRUCCI ET AL Ann Thorac Surg MORTALITY AND MORBIDITY AFTER NEONATAL BTS 2011;92: tion or exit angiography, as proposed by some [23], may reduce the potential need for some of these reoperations and perhaps improve outcomes. We observed an overall 13% incidence of composite morbidity, which was associated with a higher mortality (OR, 9.02). Nearly 30% of 166 patients with morbidity died in the hospital. These findings have significant implications for counseling families and with regard to committed resources, particularly in the subset of single-ventricle patients or patients with PA/IVS. Prior studies have specifically looked at the effect of shunt size, weight, and age at the operation as predictive factors for shunt thrombosis or revision [12, 16, 25]. The nature of our data did not allow us to assess the role of these factors. Our study also specifically looked at the timing of death because this could perhaps shed further light on mechanisms leading to death. We observed a peak of death occurring within 24 hours after the operation; further, about 75% of deaths occurred during the first 30 days. Importantly, if the 24 patients whose death was simply delayed by ECMO support are counted among the at-risk group for early death, fully 48% or nearly half of all patients who died, died within 48 hours after the operation. This finding highlights that the immediate postoperative period is the highest risk period for these patients; further focused studies are warranted to assess the role of various factors that affect this early death. Among those patients who died in the hospital, use of CPB for the MBTS procedure was predictive of earlier death, whereas use of ECMO or MCS after MBTS, not surprisingly, was predictive of later death. Importantly, use of CPB was not a predictor of death by itself. Therefore, use of CPB may be a surrogate for other factors, such as the patient is too unstable to tolerate the MBTS procedure off-pump; or alternatively, the associated changes in the inflammatory cascade or in coagulation may increase the risk of death. Of note, our study specifically excluded patients who required concomitant pulmonary arterioplasty at the time of MBTS; therefore, the effect of CPB is likely not a reflection of a more complex lesion. This is consistent with the finding that the underlying diagnosis also did not impact the timing of death. The performance of PDA ligation at the time of the MBTS also did not influence the timing of death or the mortality rate in general. Our study has a number of limitations, including a retrospective design and the voluntary nature of database, which limit the thorough capture of all relevant data. It is possible that some data might be underreported, such as PDA ligation at the time of the MBTS procedure. Also, some data elements, such as low cardiac output, are potentially subject to subjective interpretation. In addition, some important periprocedural data are lacking, such as the anticoagulation regimen, which can significantly affect observed outcomes. Lastly, the timing of death was only considered during the hospital admission. In summary, our study demonstrates the continued significant risk of death in patients with low weight and those that require preoperative mechanical ventilatory support as well as those with certain diagnoses. Our study also highlights the need for further studies to evaluate the risk factors and outcomes of those patients who survive the MBTS procedure, because many continue to have significant morbidity. Considering that the MBTS continues to be one of the most commonly performed neonatal procedures, comprising 6.4% of all reported procedures in neonates vs 10.1% for the stage I Norwood according to the STS 2010 report (excluding delayed sternal closures and PDA ligations), our study highlights the need for further efforts, including quality improvement or multi-institutional initiatives, directed at improving outcomes in this group of patients. References 1. Moulton AL, Brenner JI, Ringel R, et al. Classic versus modified Blalock-Taussig shunts in neonates and infants. Circulation 1985;72:II Woolf PK, Stephenson LW, Meijboom E, et al. A comparison of Blalock-Taussig, Waterston, and polytetrafluoroethylene shunts in children less than two weeks of age. Ann Thorac Surg 1984;38: Maghur HA, Ben-Musa AA, Salim ME, Abuzakhar SS. The modified Blalock-Taussig shunt: a 6-year experience from a developing country. Pediatr Cardiol 2002;23: Kyger ER 3rd, Reul GJ Jr, Sandiford FM, Wukasch DC, Hallman GL, Cooley DA. Surgical palliation of tricuspid atresia. Circulation 1975;52: Donahoo JS, Gardner TJ, Zahka K, Kidd BS. Systemicpulmonary shunts inneonates and infants using microporous expanded polytetrafluoroethylene: Immediate and late results. Ann Thorac Surg 1980;30: Guyton RA, Owens JE, Waumett JD, Dooley KJ, Hatcher CR Jr, Williams WH. The Blalock-Taussig shunt. Low risk, effective palliation, and pulmonary artery growth. J Thorac Cardiovasc Surg 1983;85: Lamberti JJ, Carlisle J, Waldman JD, et al. Systemicpulmonary shunts in infants and children. Early and late results. J Thorac Cardiovasc Surg 1984;88: Gladman G, McCrindle BW, Williams WG, Freedom RM, Benson LN. The modified Blalock-Taussig shunt: clinical impact and morbidity in Fallot s tetralogy in the current era. J Thorac Cardiovasc Surg 1997;114: Williams JA, Bansal AK, Kim BJ, et al. Two thousand Blalock- Taussig shunts: a six-decade experience. Ann Thorac Surg 2007;84:2070 5; discussion Gold JP, Violaris K, Engle MA, et al. A five-year clinical experience with 112 Blalock-Taussig shunts. J Card Surg 1993;8: Curzon CL, Milford-Beland S, Li JS, et al. Cardiac surgery in infants with low birth weight is associated with increased mortality: analysis of the Society Of Thoracic Surgeons Congenital Heart Database. J Thorac Cardiovasc Surg 2008; 135: Ahmad U, Fatimi SH, Naqvi I, et al. Modified Blalock- Taussig shunt: immediate and short-term follow-up results in neonates. Heart Lung Circ 2008;17: Swain SK, Dharmapuram AK, Reddy P, Ramdoss N, Raghavan SS, Kona SM. Neonatal Blalock-Taussig shunt: technical aspects and postoperative management. Asian Cardiovasc Thorac Ann 2008;16: Soueidi A, El-Rassi I. Modified Blalock-Taussig shunt in neonates. Heart Lung Circ 2009;18: Sivakumar K, Shivaprakasha K, Rao SG, Kumar RK. Operative outcome and intermediate term follow-up of neonatal blalock-taussig shunts. Indian Heart J 2001;53:66 70.

10 Ann Thorac Surg PETRUCCI ET AL 2011;92: MORTALITY AND MORBIDITY AFTER NEONATAL BTS Erez E, Bush D, Tam VK, Doublin NA, Stakes J. Outcome in infants less than 3 kilograms for placement of saphenous venous homografts as systemic-to-pulmonary arterial shunts. Cardiol Young 2008;18: Fenton KN, Siewers RD, Rebovich B, Pigula FA. Interim mortality in infants with systemic-to-pulmonary artery shunts. Ann Thorac Surg 2003;76:152 6; discussion Alkhulaifi AM, Lacour-Gayet F, Serraf A, Belli E, Planche C. Systemic pulmonary shunts in neonates: early clinical outcome and choice of surgical approach. Ann Thorac Surg 2000;69: Welke KF, O Brien SM, Peterson ED, Ungerleider RM, Jacobs ML, Jacobs JP. The complex relationship between pediatric cardiac surgical case volumes and mortality rates in a national clinical database. J Thorac Cardiovasc Surg 2009;137: Hanley FL, Sade RM, Blackstone EH, Kirklin JW, Freedom RM, Nanda NC. Outcomes in neonatal pulmonary atresia with intact ventricular septum. A multiinstitutional study. J Thorac Cardiovasc Surg 1993;105:406 23, 424 7; discussion DISCUSSION DR JOHN W. BROWN (Indianapolis, IN): Dr Petrucci and his coauthors should be congratulated for this very important report. This is what we need to learn more about in our specialty, large volumes, multi-institutional, to find out really what the practice of congenital heart surgery is all about. I learned a lot. This represents 1273 neonates from 70 hospitals over a 7-year period, a large volume, and many institutions. So this gives you a pretty good snapshot of the practice of congenital heart surgery in a recent 7-year period. I thought the mortality of 7% was staggering, and I thought the 13% significant morbidity, including extracorporeal membrane oxygenation (ECMO), low cardiac output, and a 7% return to the operating room were also really quite high, but that is what it is. And when you reviewed the literature, which you did nicely in your manuscript, the literature points out that it is not too different than what is reported. The significance of this paper is that you have looked at the highest risk group, the neonates in the first month of life with a complex anatomy, and shown what the mortality and morbidity truly is. I found that 62% of the patients who went on ECMO went on to die, and that is just a surrogate for the sickest, worst patients, and we just prolong their death, not their lives, in many instances when their anatomy is very complex. I learned that 158 of your patients, or 10% of the patient population, went on bypass to get their Blalock-Taussig (BT) shunt, something except for hypoplastic left heart syndrome, which you excluded, I have never done. So it was just interesting how the practice of congenital heart surgery is different from one institution to the next. The risk factors for death and morbidity, weight under 2.5 kg, was 15%, again, another sort of staggering high figure, but that is what it is. This prompted a number of questions, but I will ask them one at a time. Do you have an opinion as to what we can do to change this high mortality? I realize the database doesn t tell you, but after reviewing all of this data, do you have an opinion? Are we using too big a shunt? Are we not anticoagulating the patients early enough after their shunt and that is why some of them have to go back to the operating room early because they thrombosed their shunt? Should we be using either a smaller shunt or a different shunt material to lessen the mortality that you have demonstrated? Again, wonderful report, something that I learned a great deal 21. Hanley FL, Sade RM, Freedom RM, Blackstone EH, Kirklin JW. Outcomes in critically ill neonates with pulmonary stenosis and intact ventricular septum: a multiinstitutional study. Congenital Heart Surgeons Society. J Am Coll Cardiol 1993;22: Hoskote A, Bohn D, Gruenwald C, et al. Extracorporeal life support after staged palliation of a functional single ventricle: subsequent morbidity and survival. J Thorac Cardiovasc Surg 2006;131: Holzer RJ, Sisk M, Chisolm JL, et al. Completion angiography after cardiac surgery for congenital heart disease: Complementing the intraoperative imaging modalities. Pediatr Cardiol 2009;30: Rao MS, Bhan A, Talwar S, et al. Modified Blalock-Taussing shunt in neonates: determinants of immediate outcome. Asian Cardiovasc Thorac Ann 2000;8: Li JS, Yow E, Berezny KY, et al. Clinical outcomes of palliative surgery including a systemic-to-pulmonary artery shunt in infants with cyanotic congenital heart disease: Does aspirin make a difference? Circulation 2007;116: from, and I congratulate you and your group for a very nice presentation. DR PETRUCCI: Thank you, Dr Brown. Definitely our data doesn t allow us to dissect these questions. We have no shunt size from this database, we have no anticoagulation regimen, but I believe the surgery seems too simple and the mortality too high. So our final message is, that it is not so simple. A conversation I have had with Dr Eghtesady is that maybe we have to find out some extra protection for the vascular pulmonary tree, because we are dispatching some extra blood that is passing through a polytetrafluroethylene (PTFE) tube, and blood is being activated in there. The activated blood comes into the pulmonary vasculature tree, and maybe triggers events and vasoconstriction or vasodilation of the pulmonary artery bed. So definitely, at this time, we are unable to answer what we can improve. About the shunt size, we are awaiting the data. When we asked for the data, we were anxious to see whether or not the shunt size was associated with mortality, but unfortunately, the database has no information about the shunt size. DR BROWN: Just to make a comment, the shunt size sure made a difference in the hypoplastic left heart syndrome group, and I am wondering whether some of this is attributed to the large shunt size; but again, the database doesn t give us that kind of information. But questions are just a stimulus for more studies. That is how we learn. Thank you. DR MICHAEL HINES (Winston-Salem, NC): One thing this study may do is to help us eliminate the phrase that our cardiologists and intensivists commonly use that it is just a shunt. We know it is not just a shunt. I think we may need to expand the database and start collecting more data to find out why. Is it an overcirculation problem and there is unrecognized ischemia and arrest, or is it because these shunts are clotting and how do we fix that? Over a period of about 5 years, I had three shunts clot acutely in the first 24 hours, and in hindsight, I got hypercoagulable studies on them and all 3 patients had factor V Leiden deficiency, which in our part of the country is about 5% to 7% of the population. So currently, I get factor V Leiden and factor II 20210

11 652 PETRUCCI ET AL Ann Thorac Surg MORTALITY AND MORBIDITY AFTER NEONATAL BTS 2011;92: genetic profiles before doing any elective shunt. In those patients who are positive, we treat aggressively with Warfarin (Bristol-Myers Squibb Co, Princeton, NJ) and haven t had any acute shunt thrombosis since. Now, it is very anecdotal, but that may be something to either start checking or to start putting in the database to try and find out why some of these kids are having problems. DR ROSS M. UNGERLEIDER (Winston-Salem, NC): I think both Mike, you and John, sort of emphasize that there may be not an issue of what you have been able to analyze but what is in the database and is it possible to add these fields to the database, that is, shunt size I know is there. So, you can get those data, but whether or not there is post-op anticoagulation, and the question I have, and maybe this needs to be in the database, is, the patients who are on mechanical circulatory support like ECMO, was the shunt left open or occluded? I think the studies we did in the 1990s at Duke demonstrated that if you occlude the shunt, the hemodynamics improve quickly but the lungs die and the patient dies with that. So that leaving the shunt open on ECMO was a turnaround in outcomes, and I don t know if that is in the database with the ECMO patients but may be another thing that would need to be there. I take it those data were not available to you? DR PETRUCCI: No, no. DR CHRISTOPHER KNOTT-CRAIG (Memphis, TN): Congratulations on a very nice presentation. Were you able to differentiate whether the shunts were placed on the left side or the right side in your database, and if so, did that influence the outcome at all? DR PETRUCCI: No, unfortunately, also, there is no data about which side the shunt was performed. No, we can t answer this question. DR JEFFREY JACOBS (St. Petersburg, FL): I would just like to take this opportunity to congratulate both Orlando and Pirooz for putting a substantial amount of time and effort into working with the Society of Thoracic Surgeons (STS) database and getting these data together. They put together some very important data, and I think that studies like this must make Gus Mavroudis smile when he sat down and set up the STS database back in the 1990s that here we are in 2010 and we are able to get data like this. So congratulations, guys. DR PETRUCCI: Thank you.