Associated Atrial Septal Defects Increase Perioperative Morbidity After Ventricular Septal Defect Repair in Infancy

Associated Atrial Septal Defects Increase Perioperative Morbidity After Ventricular Septal Defect Repair in Infancy Christopher J. Knott-Craig, MD, Ronald C. Elkins, MD, Kalyanakrishnan Ramakrishnan, MD, Debbie A. Hartnett, RN, Mary M. Lane, PhD, Edward D. Overholt, MD, Kent E. Ward, MD, and Jerry R. Razook, MD Sections of Thoracic Surgery and Pediatric Cardiology, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma Although closure of ventricular septal defects (VSDs) is currently associated with a relatively low risk, infants with associated atrial septal defects (ASDs) seem to have a higher perioperative morbidity. To clarify this impression, we reviewed our entire experience (since 1977) with closure of simple VSDs in 163 infants (age, _<12 months). Of these, 57 had significant ASDs (ASD-VSD subgroup). Hospital mortality was 3.7% (6/163) overall and 1.4% (2/145) since 1980. Actuarial survival at 10 years was 92% ± 5%. Significant morbidity occurred in 15.5% (16/103) of the VSD subgroup versus 48.1% (26/54) of the ASD-VSD subgroup (p _< 0.001). Multivariate analysis identified the presence of multiple VSDs and early date of operation as risk factors for hospital death, and younger age, an associated ASD, the size of the VSD, and use of hypothermic circulatory arrest as risk factors for significant perioperative morbidity. Compared with the VSD subgroup, the ASD-VSD subgroup had a higher hospital mortality (5.3% [3/57] versus 2.8% [3/106]), were younger (5.1 ± 2.9 versus 7.2 ± 2.9 months; p = 0.001), had a higher preoperative pulmonary artery pressure (70.2 ± 19.0 versus 62.7 ± 21.8 mm Hg; p = 0.08), needed more inotropic support (12.3% versus 3.7%; p = 0.07), needed more prolonged ventilation (3.3 versus 1.8 days; p = 0.02), and had longer postoperative hospital stays (11 versus 8 days; p = 0.005). The increased postoperative morbidity associated with infants who have a significant ASD in addition to a VSD is generally unappreciated, and may relate to the different hemodynamics associated with left-to-right shunting at both the atrial and ventricular levels. (Ann Thorac Surg 1995;59:573-8) A n isolated ventricular septal defect (VSD) is the most common congenital cardiac lesion, representing 20% to 30% of all congenital heart malformations, and has an incidence of about I per 1,000 live births [1-3]. The hospital mortality associated with surgical closure of a simple VSD in many series is currently less than 5% [4-9], and the associated morbidity is generally,low, though less well defined in the literature. It has been our impression, however, that infants with VSDs who also have significant shunting at the atrial level and who need surgical intervention within the first year of life, behave differently from those with isolated VSDs: They seem to suffer increased morbidity after repair, often require more inotropic support postoperatively, and take longer to recover from their surgical procedure. Because this observation is not well documented, we undertook this review to better define the relationship between associated atrial septal defects (ASDs) and the outcome from the repair of VSDs in infancy. Presented at the Forty-first Annual Meeting of the Southern Thoracic Surgical Association, Marco Island, FL, Nov 10-12, 1994. Address reprint requests to Dr Knott-Craig, Section of Thoracic and Cardiovascular Surgery, University of Oklahoma Health Sciences Center, PO Box 26901, Oklahoma City, OK 73190. Material and Methods Patient Population During March 1994, the hospital records and cardiothoracic surgery database were reviewed to identify all patients less than 12 months of age at the time of surgical closure of an isolated VSD at the Children's Hospital of Oklahoma. Included were patients (1) with a concomitant atrial septal defect (ASD) or patent ductus arteriosus, or both, (2) who had undergone repair of coarctation or pulmonary artery banding, or both, and (3) with other simple cardiac lesions such as mild pulmonary stenosis and vascular rings. Patients with more complex associated cardiac defects were excluded. During the 17-year period between 1977 and March 1994, 163 infants fulfilling these criteria were identified, and they constitute the study group. One hundred fortyfive patients have undergone surgical treatment since January 1980. There were 92 female and 71 male infants. The mean age at operation was 6.5 + 3.0 months (range, 0.6 to 12 months), and the mean weight was 5.4 +-_ 1.6 kg (range, 2.2 to 9.8 kg) (Table 1). Forty-five infants (27.6%) had associated morphologic syndromes, including Down's syndrome (n = 33), fetal alcohol syndrome (n = 3), Vater's syndrome (n = 2), and Goldenhauer's syndrome (n = 2). Some infants had 1995 by The Society of Thoracic Surgeons 0003-4975/95/$9.50 0003-4975(94)01005-W

574 KNOTI'-CRAIG ET AL Ann Thorac Surg MORBIDITY IN INFANTS INITH ASD AND VSD 1995;59:573-8 Table 1. Characteristics in the 163 Patients Characteristic Age (too) Weight (kg) Associated atrial septal defect (No.) Previous cardiac operation (No.) Noncardiac syndrome (No.) Sex (No.) Ventricular septal defect closure (suture/patch) Ventriculotomy for ventricular septal defect (No.) Mean hypothermic circulatory arrest (n = 34) (min) Mean pulmonary artery pressure (mm Hg) Pulmonary-to-systemic resistance ratio Pulmonary arteriolar resistance index Follow-up (90.1% complete) (too) Value 6.5 + 3.0 5.4-+1.6 kg Yes, 57; no, 106 28 (17.1%) 45 (27.6 /0) Male, 71; female, 92 22/141 Right ventricle, 21; left ventricle, 4 33 (21-47) 41 -+ 15 0.23 -+ 0.17 3.1 -+ 2.9 50 + 47 congenital and acquired noncardiac pathologic conditions independently associated with significant mortality and morbidity; these included diaphragmatic hernia with pulmonary hypoplasia, tracheoesophageal fistula, and hydrocephalus. One 6-month-old ventilator-dependent pre-term infant was mentally retarded and suffered from cerebral apneic episodes, severe bronchopulmonary dysplasia and pulmonary hypertension, renal calcinosis and failure, severe gastroesophageal reflux, and little or no discernible shunt across a moderately sized VSD before operation. Relating perioperative morbidity to a relatively simple cardiac operation in patients such as this clearly poses intrinsic difficulties. To prevent bias, however, we have included all types of morbidity in this analysis. The VSD was perimembranous (conoventricular) in 116 patients, an inlet septal defect (atrioventricular canal type) in 19, an outlet septal defect (conal septal defect) in 5, and a muscular (trabecular) defect in 10; it constituted multiple VSDs in 13 [10, 11]. Of the 33 patients with Down's syndrome, 23 had perimembranous defects and 10 had inlet-type defects. The VSDs were arbitrarily classified as large if they were closed with a patch and small if they were suture obliterated. Forty patients underwent 45 surgical procedures before their VSD repair. These included aortic coarctation repair in 28, tracheoesophageal fistula repair in 2, a Ladd procedure for duodenal atresia in 2, and placement of a ventriculoperitoneal shunt for hydrocephalus in 1 (Table 2). Fifty-seven infants had ASDs closed at the time of their VSD repair; these constitute the ASD-VSD subgroup. In addition, a ductus arteriosus was obliterated in 50 patients, pulmonary commissurotomy or patch angioplasty was done in 7, debanding of the pulmonary artery was performed in 4, and division of a vascular ring was carried out in 1. Often the ductus arteriosus was suture obliterated at operation, irrespective of whether it was shown to be patent preoperatively, to prevent possible air embolization from occurring during circulatory arrest. The median postoperative hospital stay was 7 days (range, 5 to 68 days) and the median total hospital stay was 9 days (range, 6 to 93 days). A mean follow-up of 50 _+ 47 months (range, 0 to 170 months) was achieved in 147 patients (90.1%). Outcome Hospital mortality is defined as death occurring during the initial period of hospitalization or within 30 days of repair. Significant perioperative morbidity was defined as (1) the need for more than 48 hours of ventilation postoperatively, or for reintubation after extubation; (2) the need for more than one intravenous inotropic agent for longer than 48 hours; (3) development of the postpericardiotomy syndrome; (4) the occurrence of wound or mediastinal infection; (5) the development of permanent or temporary arrhythmia that prolonged the hospital stay beyond 7 days; (6) the development of any infection, atelectasis, or congestive heart failure, or the need for oxygen resulting in a prolonged postoperative hospital stay beyond 7 days; and (7) the development of any seizure activity or neurologic deficit in the postoperative period. Technique of Operation All VSD repairs were performed using cardiopulmonary bypass and at least moderate hypothermia. Hypothermic circulatory arrest was used in 72 patients (37 of the 57 in the ASD-VSD subgroup and 35 of the 106 in the VSD subgroup), usually in the smaller ones. Cold crystalloid or blood cardioplegia was used to arrest the heart. The VSDs were repaired through the right atrium (n = 136; 83.4%), a right ventriculotomy (n = 21; 12.9%), a left ventriculotomy (n = 4; 2.5%), or the pulmonary artery (n = 2; 1.2%). Statistical Analysis The operation reports, discharge summaries, cardiac catheterization data, follow-up records, correspondence, and full hospital records were reviewed, and the collected data entered into a single database. Complete Table 2. Previous Operative Procedures Performed in 40 Patients Operation Aortic coarctation repair Repair of duodenal atresia Repair of tracheoesophageal fistula Pulmonary artery banding Nissen fundoplication and gastrostomy Ventriculoperitoneal shunt for hydrocephalus Repair of congenital diaphragmatic hernia Other miscellaneous procedures Number 28 2 2 4 3 1 1 4

Ann Thorac Surg KNOTI~-CRAIG ET AL 575 1995;59:573-8 MORBIDITY IN INFANTS WITH ASD AND VSD catheterization data were only available in 29 patients. Within this database, the subgroup of 57 infants with ASDs closed at operation was created and their data compared with those in the 106 patients with isolated VSD (VSD subgroup) (Table 3). Differences between these two subgroups were analyzed using the two outcome events of interest: hospital death and perioperative morbidity. Dichotomous variables were compared using X 2 tests, and continuous variables were compared with Wilcoxon rank-sum analysis. Using the study group as a whole, the following variables were analyzed univariately: age at operation; weight at operation; type of VSD; presence of associated ASD; associated patent ductus arteriosus, Down's syndrome, or a noncardiac abnormality; previous cardiac operation; transventricular closure of VSD; sex; preoperative pulmonary artery pressure; vascular resistance ratios; and use of hypothermic circulatory arrest. All variables approaching significance (p < 0.10) univariately were entered into a stepwise logistic-regression model and retained in this multivariate analysis for statistical significance at a p < 0.05 level (SAS system logistic procedure, Version 6.10; SAS Institutes, Cary, NC). Results Mortality Hospital mortality was 3.7% (6/163). This was 2.8% for the VSD subgroup (3/107) and 5.3% for the ASD-VSD sub- Table 3. Comparative Analysis of Perioperative Variables VSD ASD-VSD Variable Subgroup (n = 106) Subgroup (n = 57) p Value Age (mo) 7.2 + 2.9 5.1 + 2.9 0.0001 Weight (kg) 5.9-1.6 4.5 +- 1.2 0.0001 Down's syndrome (No.) 16 (15%) 17(29%) 0.02 Circulatory arrest used (No.) 35 (33%) 37 (64%) 0.0005 Circulatory arrest (min) 35 _+ 14 34 + 13 NS Pulmonary artery pressure (systolic) (mm Hg) 62.7 + 21.8 70.2 + 19.0 0.08 Pulmonary-to-systemic 0.22 + 0.17 0.26 + 0.16 NS resistance ratio Ventilation (days) 1.8 + 4 3.3 + 7 0.02 Arrhythmias (%) 3.7 14.0 0.03 Neurologic complications (%) 1.8 7.0 0.22 Postpericardiotomy syndrome (%) 3.7 5.2 NS Wound infection (%) 1.8 3.5 NS Pulmonary complications (%) 8.4 22.8 0.02 Low cardiac output (%) 3.7 12.2 0.07 Hospital stay (total days) 11 19 0.0001 Postoperative hospital stay 8 11 0.005 (days) Hospital mortality (%) 2.8 5.3 NS Total morbidity (%) 15.5 48.1 0.001 ASD = atrial septal defect; NS = not significant; VSD = ventricular septal defect. Percent 100 ~- I I" 80 60 40 20 161 80 67 47 31 15 4 0 0 2 4 6 8 10 12 14 Follow-up in Years Fig 1. Actuarial survival with 70% confidence limits for 163 infants undergoing repair of a ventricular septal defect. group (3/57) (p = not significant). Three of the 6 patients who suffered hospital deaths had multiple VSDs. There were no deaths among the 33 infants with Down's syndrome. Since 1980, the hospital mortality was 1.4% (2/145), which is significantly less than that for the earlier era (p = 0.001). Multivariate analysis identified the presence of multiple VSDs and earlier date of operation as risk factors for hospital mortality (both p = 0.001). Four additional patients died within 6 months of their operation, for a 6-month survival of 93.9% (1531163). Three of the four late deaths were related to noncardiac abnormalities. The actuarial survival at the 10-year follow-up was 92% + 5% (Fig 1). Perioperative Morbidity Forty-two of the 157 hospital survivors (26.7%) experienced perioperative morbidity. This was 15.5% (161103) for the VSD subgroup and 48.1% (26154) for the ASD- VSD subgroup (p --- 0.001). Multivariate analysis identified younger age at operation, use of hypothermic circulatory arrest, the presence of an associated ASD (ASD- VSD subgroup), and size of the VSD as risk factors for perioperative morbidity (Table 4). For the final model the area under the receiver operating characteristic curve (ROC curve) was 0.864. The most common cause of morbidity was related to significant pulmonary complications, such as atelectasis, consolidation, and infection, and resulted in a prolonged hospital stay in 22 patients (13.5%). Low cardiac output requiring intravenous inotropic support for longer than 48 hours was needed in 11 infants (6.7%). Significant perioperative arrhythmias requiring temporary pacing or medication occurred in 11 of the patients (6.7%). However, only 1 patient required a permanent pacemaker postoperatively. Neurologic complications occurred in 6 patients (3.7%), and constituted only perioperative seizure activity in 4; all 6 of these patients had hypothermic circulatory arrest during their repair. Two patients required reoperation for the control of postoperative bleeding. The causes of significant morbidity are given in Table 5. One patient with Vater's syndrome and multiple other noncardiac abnormalities underwent suture obliteration

576 KNOTT-CRAIG ET AL Ann Thorac Surg MORBIDITY IN INFANTS WITH ASD AND VSD 1995;59:573-8 Table 4. Risk Factors for Perioperative Morbidity After Repair of Ventricular Septal Defect Univariate Multivariate Variable (p value) (p value) /3-Coefficient Odds Ratio Circulatory arrest 0.0001 0.0001 Age at operation 0.0001 0.0005 Associated atrial 0.0001 0.02 septal defect Size of 0.01 0.04 ventricular septal defect Previous coarctation 0.04 NS Noncardiac syndrome 0.03 NS Previous operation 0.03 NS NS = not significant. +1.60 4.9 (1.7-14.2) -0.29 1.3 (1.1-1.6) +1.00 2.7 (1.1-6.7) +2.23 9.3 (1.0-88.4) of a small residual VSD 6 months after his initial repair. Another was found to have a partly dehisced VSD patch at postmortem examination 6 months after repair. A hemodynamically insignificant VSD was noted by colorflow Doppler imaging at follow-up in an additional 17 patients (11.0%). Comparison between the ASD-VSD subgroup and the VSD subgroup revealed that the ASD-VSD patients were younger and smaller at operation, more often had circulatory arrest used in the repair, and tended to have higher preoperative pulmonary artery pressures. In addition, they more often needed substantial inotropic support postoperatively, had more pulmonary complications, and were ventilated longer postoperatively. Both their total hospital stay and postoperative hospital stays were also longer (see Table 3). Comment Hospital Mortality Since the first repairs of a VSD performed by Lillehei and colleagues in 1954 using controlled cross-circulation [12] and by Kirklin in 1956 using the early mechanical pump- Table 5. Causes of Perioperative Morbidity in 42 Patients Cause No. of Cases (%) Pulmonary complications 22 (13.5) Significant inotropic support 11 (6.7) Arrhythmias 11 (6.7) Postpericardiotomy syndrome 7 (4.3) Neurologic complications 6 (3.7) Wound infection 4 (2.5) Bleeding 2 (1.2)...... oxygenator [13], the early mortality related to the surgical repair of VSDs in infancy has steadily decreased, from around 20% in the 1970s [14-16] to less than 5% in many recent series [4-6, 8, 9, 17, 18]. Improvements in preoperative diagnosis, surgical skills, operating room technology, and perioperative care made during the 1970s and early 1980s led to the widespread adoption of primary repair rather than preliminary pulmonary artery banding as the treatment of choice for symptomatic infants with isolated VSDs, except for those with multiple VSDs of the swiss cheese variety [4, 7, 19-21]. Our hospital mortality of 3.7% since 1977 and 1.4% since 1980 is further evidence that such a policy is well-founded. We, too, have found that the presence of multiple VSDs is an incremental risk factor for death (three of the six early deaths in our series occurred in patients with multiple VSDs; p = 0.001), and that this subset of patients is best managed by preliminary pulmonary artery banding and repair at about 2 years of age. This is particularly applicable to infants who are very small or in whom the defects are predominantly in the apical muscular septum. Many of these muscular defects may undergo spontaneous closure in the interim after pulmonary artery banding, facilitating the definitive repair [22, 23]. Prior repair of coarctation was not found to be a significant risk factor for subsequent VSD repair in this analysis; we continue to recommend initial coarctation repair only for the overwhelming majority of these patients, and reserve concomitant pulmonary artery banding for those with either large outlet-type VSDs (conal septal defects) or multiple large muscular VSDs. If spontaneous closure does not occur, or the infants continue to have congestive cardiac failure after their coarctation repair, we repair the VSD during the same hospitalization or soon thereafter. This is consistent with the practice described in other recent reports [24, 25]. The actuarial survival in our patients at the 10-year follow-up was 92% + 5%, with no appreciable mortality after I year following repair, indicating a good long-term prognosis for this patient cohort (see Fig 1). Perioperative Morbidity We defined perioperative morbidity rather liberally in this analysis. By these standards, 26.7% (42/157) of our patients experienced significant morbidity. Although the incidence of complete heart block, mediasfinitis, reoperation, and neurologic deficits was low, pulmonary complications such as infections, considerable atelectasis, and serous effusions were a common source of morbidity and prolonged hospital stay (see Table 5). This may reflect the degree of congestive heart failure and malnutrition commonly present preoperatively in these chronically ill small infants. Despite the limitations of a study such as this in which it is difficult to differentiate cardiac-related from noncardiac-related morbidity, multivariate analysis identified younger age at operation, the use of hypothermic circulatory arrest, the presence of an associated ASD, and the size of the VSD as independent risk factors for

Ann Thorac Surg KNOTt-CRAIG ET AL 577 1995;59:573-8 MORBIDITY IN INFANTS WITH ASD AND VSD perioperative morbidity (see Table 4). When weight at operation was substituted for age at operation, the results of the analysis were essentially the same. This is not surprising, as the technical aspects of repairing a VSD in a 2.5-kg infant are little different from those used in a 6.0-kg infant; however, the effects of cardiopulmonary bypass and hypothermic circulatory arrest, and the "third-spacing" of fluid often occurring postoperatively in this subgroup of patients are associated with increased morbidity and a longer time to extubation. These, as well as other perioperative factors peculiar to major surgical procedures in very small infants (eg, airway management, fluid and electrolyte control, and feeding), result in their slower initial recovery compared with that in their older, larger infant counterparts. Hence, the finding of younger age at operation, use of circulatory arrest, and size of VSD as risk factors for morbidity is not unexpected. Our analysis confirmed our clinical impression that the presence of a concomitant ASD per se increases the risk of perioperative morbidity. This is not difficult to conceptualize: additional left-to-right shunting at the atrial level increases the total left-to-right shunt; this correspondingly diminishes the cardiac output and aggravates the cardiac failure and failure to thrive. In addition, the hemodynamic characteristics of both ventricles in the setting of a large ASD in terms of preload and compliances may differ from that encountered in the setting of a simple VSD, a situation analogous to that occurring in the small subset of infants with a large isolated ASD who present with severe congestive heart failure. Unfortunately, in a historical analysis such as this, there are not sufficient appropriate data (eg, ventricular mass and volume indices) to permit these issues to be addressed. However, these infants are usually in severe congestive cardiac failure preoperatively, and have often been hospitalized numerous times for tube feeding, and for the control of cardiac failure and respiratory infections by the time they are referred for surgical treatment. Earlier operation, performed within the first I to 2 months of life, may lead to a lower perioperative morbidity by minimizing both the pulmonary effects of chronic congestive failure and the negative impact of poor nutritional status at the time of operation. In our series, the ASD-VSD patients were smaller and younger at operation than were the VSD patients, and both characteristics have been associated with a poorer prognosis in terms of recovering and "'catching up" after repair [5, 8, 18]. Our ASD-VSD population also tended to have a higher preoperative pulmonary artery pressure, suggesting higher shunt ratios or higher pulmonary vascular resistance, or both. Because increasingly fewer patients presenting for VSD repair in infancy undergo preoperative cardiac catheterization studies, we do not have sufficient data to analyze these variables in a meaningful manner. The ASD-VSD subgroup had a higher percentage of Down's patients than did the VSD group (29% versus 15%; p = 0.02); however, the presence of Down's syndrome did not increase the risk of either morbidity or mortality after operation. The findings from our analysis corroborate a recent observation that the muscular septurn is usually free of defects in Down's syndrome patients [26]: of 33 such patients in our analysis, 23 had perimembranous defects and 10 had inlet-type defects. The ASD-VSD subgroup in this analysis tended to have more postoperative arrhythmias (most of them temporary), more pulmonary complications, a higher incidence of low cardiac output, and needed a longer period of postoperative ventilation than did the VSD group; they also had a longer postoperative hospital stay (see Table 3). Although they more often required hypothermic circulatory arrest for their repair, the period of arrest was within the safe range and was no different from that of the other patients (34 + 13 versus 35 -+ 14 minutes; p = not significant). From the findings yielded by this study, it is apparent that currently most VSDs can be repaired in infancy with a low risk of death; those infants who are very small or who have multiple muscular VSDs would probably still benefit from a two-stage approach to their repair. However, those with concomitant large intraatrial shunts are at increased risk for perioperative morbidity due largely to their more serious preoperative status and more severe congestive heart failure. Rather than attempting to buy time for the patient to grow, this risk may be negated by an earlier operation. Although these observations are intuitively known to most pediatric cardiologists and surgeons, this study was conducted to document them objectively. With the accent in medicine changing to cost containment and shortening of hospital and intensive care unit stays, it is important to recognize that this subgroup of patients does experience increased perioperative morbidity. The fact that they often need to stay in the hospital longer than would otherwise be anticipated becomes more relevant in terms of planning and the counsel given to parents, health care providers, and health care insurers. We thank Karen Dale for her assistance in preparation of the manuscript. References 1. Hoffman JIE, Rudolph AM. The natural history of ventricular septal defects in infancy. Am J Cardiol 1965;16:634-53. 2. Fyler DC. Report of the New England regional infant cardiac program. Pediatrics 1980;65:376-461. 3. Grabitz RG, Joffres MR, Collins-Nakai RL. Congenital heart disease: incidence in the first year of life. The Alberta heritage pediatric cardiology program. Am J Epidemiol 1988;128:381-8. 4. Rizzoli G, Blackstone EH, Kirldin JW, Pacifico AD, Bargeron LM Jr. Incremental risk factors in hospital mortality rate after repair of ventricular septal defect. J Thorac Cardiovasc Surg 1980;80:494-505. 5. Hardin JT, Muskett AD, Canter CE, Martin TC, Spray TL. Primary surgical closure of large ventricular septal defects in small infants. Ann Thorac Surg 1992;53:397-401. 6. Rizzoli G, Rubino M, Mazzucco A, et al. Progress in the surgical treatment of ventricular septal defect: an analysis of twelve years' experience. Thorac Cardiovasc Surg 1983;31: 382-8.

578 KNOTT-CRAIG ET AL Ann Thorac Surg MORBIDITY IN INFANTS WITH ASD AND VSD 1995;59:573-8 7. Serraf A, Lacour-Gayet F, Bruniaux J, et al. Surgical management of isolated multiple ventricular septal defects. Logical approach in 130 cases. J Thorac Cardiovasc Surg 1992; 103:437-43. 8. Weintraub RG, Menahem S. Early surgical closure of a large ventricular septal defect: influence on long-term growth. J Am Coll Cardiol 1991;18:552-8. 9. Lincoln C, Jamieson S, Joseph M, Shinebourne E, Anderson RH. Transatrial repair of ventricular septal defects with reference to their anatomic classification. J Thorac Cardiovasc Surg 1977;74:183-90. 10. Van Praagh R, Geva T, Kreutzer J. Ventricular septal defects: how shall we describe, name and classify them? J Am Coll Cardiol 1989;14:1298-9. 11. Soto B, Ceballos R, Kirklin J. Ventricular septal defects: a surgical viewpoint. J Am Coil Cardiol 1989;14:1291-7. 12. Warden HE, Cohen M, Read RC, Lillehei CW. Controlled cross circulation for open intracardiac surgery. Physiologic studies and results of creation and closure of ventricular septal defects. J Thorac Surg 1954;28:331-44. 13. DuShane JW, Kirklin JW, Patrick RT, et al. Ventricular septal defects with pulmonary hypertension. Surgical treatment by means of a mechanical pump-oxygenator. JAMA 1956;160: 950-3. 14. Blackstone EH, Kirklin JW, Bradley EL, DuShane JW, Appelbaum A. Optimal age and results in repair of large ventricular septal defects. J Thorac Cardiovasc Surg 1976;72: 661-79. 15. Barratt-Boyes BG, Simpson M, Neutze JM. Intracardiac surgery in neonates and infants using deep hypothermia with surface cooling and limited cardiopulmonary bypass. Circulation 1971;43:I25-30. 16. Cooley DA, Garrett HE, Howard HS. The surgical treatment of ventricular septal defect: an analysis of 300 consecutive surgical cases. Prog Cardiovasc Dis 1962;4:312-23. 17. Waldman JD. Why not close a small ventricular septal defect? Ann Thorac Surg 1993;56:1011-2. 18. Yeager SB, Freed MD, Keane JF, Norwood WI, Castaneda AR. Primary surgical closure of ventricular septal defect in the first year of life: results in 128 infants. J Am Coll Cardiol 1984;3:1269-76. 19. Stark J, Sethia B. Closure of ventricular septal defect in infancy. J Cardiac Surg 1986;1:135-50. 20. Kirklin JK, Castaneda AR, Keane JF, Fellows KE, Norwood WI. Surgical management of multiple ventricular septal defects. J Thorac Cardiovasc Surg 1980;80:485-93. 21. Danilowicz D, Presti S, Colvin S, Galloway A, Langsner A, Doyle EF. Results of urgent or emergency repair of symptomatic infants under one year of age with single or multiple ventricular septal defect. Am J Cardiol 1992;69:699-701. 22. Frontera-Izquierdo P, Cabezuelo-Huerta G. Natural and modified history of isolated ventricular septal defect: a 17-year study. Pediatr Cardiol 1992;13:193-97. 23. Kidd L, Driscoll DJ, Gersony WM, et al. Second natural history study of congenital heart defects. Results of treatment of patients with ventricular septal defects. Circulation 1993;87:I38-51. 24. Park JK, Dell RB, Ellis K, Gersony WM. Surgical management of the infant with coarctation of the aorta and ventricular septal defect. J Am Coil Cardiol 1992;20:176-80. 25. Quaegebeur JM, Jonas RA, Weinberg AD, Blackstone EH, Kirklin JW. Outcomes in seriously ill neonates with coarctation of the aorta: a multiinstitutional study. J Thorac Cardiovasc Surg 1994;108:841-54. 26. Marino B, Corno A, Guccione P, Marcelletti C. Ventricular septal defect and Down's syndrome. Lancet 1991;337:245-6. DISCUSSION DR ERIC L. CEITHAML (Jacksonville, FL): Doctor Knott-Craig, I congratulate you on a nice presentation. I noticed that in your comparative analysis of variables you failed to include multiple VSDs. It is our impression that this is a more difficult group with a higher surgical mortality. How many patients in your series had multiple VSDs, and were they equally matched between the two subgroups? DR KNOTT-CRAIG: Our data showed that multiple VSDs are both univariately and multivariately associated with increased mortality in this cohort of patients. There were too few patients with multiple VSDs repaired in infancy who survived for them to have an impact on the perioperative morbidity. DR THOMAS L. SPRAY (St. Louis, MO): This is a very large series and an interesting analysis. 1 wonder whether your patients with ASD and VSD come to operation earlier and therefore may have increased perioperative morbidity because in fact they are younger. Another issue is whether these patients simply have more congestive heart failure and protracted attempts to achieve weight gain and growth to a certain age and size have been made, at which point they are repaired. We presented some data a few years ago at this meeting that showed that early repair of VSDs, even in the first month or two of life, was preferable to trying to get patients to an older age and greater weight. Prolonged congestive heart failure while attempting to achieve weight gain by medical means may result in more morbidity than if you just perform repair earlier. Your data would tend to support that concept in the sense that the average age at repair, in the ASD and VSD group, was around 3 to 5 months, and more than 6 months in the general group of VSD patients. I wonder if you review the patients with ASD and VSD over the past 5 years, where a more aggressive approach might have been taken, whether you would still note an increased perioperative morbidity. DR KNOTT-CRAIG: I agree with you completely. I think the patients with the ASD-VSD pathology are younger and have more severe congestive heart failure, which develops earlier. The tendency, among cardiologists at least, is still to try and tide them over, tube feed them, or somehow get them to a bigger size or an older age in order that the operative risk can be minimized. I think, however, that the risk would be minimized if those patients came to operation earlier before the severe congestive heart failure and pulmonary infections developed, which are a significant cause of perioperative morbidity. With the low risk of complex intracardiac repairs that are performed in the neonatal period, there really is no good reason to delay operation in this group of patients.