Univentricular Repair: Is Routine Fenestration Justified? Balram Airan, MCh, Rajesh Sharma, MCh, Shiv Kumar Choudhary, MCh, Smruti R. Mohanty, MCh, Anil Bhan, MCh, Ujjwal Kumar Chowdhari, MCh, Rajnish Juneja, DM, Shyam Sunder Kothari, DM, Anita Saxena, DM, and Panangipalli Venugopal, MCh Cardiothoracic Sciences Centre, All India Institute of Medical Sciences, New Delhi, India Background. A decade after the introduction of baffle fenestration, the outcome of Fontan-type repair for hearts with a functional single ventricle finally looks promising. Our study was designed to assess the impact of fenestration on the outcome of univentricular repairs. Methods. From January 1988 to December 1997, 348 patients (104 with tricuspid atresia and 244 with other morphological diagnoses) underwent univentricular repair at our institute. Since 1994, routine fenestration of the atrial baffle was performed in all patients (n 126). Results. The overall Fontan failure rate was 14% (50 of 348) and included 45 early deaths and five Fontan take downs. Absence of fenestration was the only and highly significant predictor of Fontan failure (risk ratio [RR] 3.3, 95% confidence interval [CI] 1.49 to 7.31, p 0.002). Significant pleural effusion was seen in 27% of patients. Absence of fenestration of the atrial baffle (RR 3.97, 95% CI 2.17 to 7.26, p < 0.001) and aortic cross-clamp time more than 60 minutes (RR 2.15, 95% CI 1.3 to 3.5, p 0.002) were found to be significant risk factors. The follow-up ranged from 6 to 120 months (mean 46.0 18.0 months). There were 12 late deaths and 5 patients were lost to follow-up. Actuarial survival (Kaplan Meier) at 90 months was 81% 4%. Two hundred and fifty-eight patients (90%) were in New York Heart Association class I at their last follow-up visit. Oxygen saturation in the fenestrated group ranged from 85% to 94% (mean 89%). Thirty patients (26%) had spontaneous closure of the fenestration over a mean period of 34 months, and there has been no incidence of late systemic thromboembolism. In no instance has there been a need to close the fenestration. Conclusions. Elective fenestration of the intraatrial baffle is associated with decreased Fontan failure rate and decreased occurrence of significant postoperative pleural effusions. Routine elective fenestration of the atrial baffle may, therefore, be justified in all univentricular repairs. (Ann Thorac Surg 2000;69:1900 6) 2000 by The Society of Thoracic Surgeons Over the last 27 years, since Fontan and Baudet [1] first described their technique for univentricular repair in 1971, various modifications have been introduced in the Fontan circuitry. Fenestration of the atrial baffle was proposed in 1988 [2] on the basis that it allows a protective right-to-left shunt at the atrial level to overcome the period of low cardiac output in the early postoperative period at the expense of mild to moderate desaturation [3 5]. In this center, we converted from a policy of no fenestration to routine fenestration in 1994 after results with fenestration demonstrated an improvement in outcome [6]. This study is an extension of the previous one [6], as we are now in a position to draw more definite conclusions based on a larger sample size and longer duration of follow-up. Patients and Methods Three hundred and forty-eight patients who have undergone univentricular repair for various morphological diagnoses (Table 1) at our institute from January 1988 to December 1997 form the study population. The group included 261 male and 87 female patients. The patients who had undergone extracardiac Fontan procedure were not included in this study, as those cases were performed on a beating heart without any cardiac arrest. The mean age was 7.4 6.0 years. The youngest patient was 8 months and the oldest was 52 years of age (median 5.6 years). The clinical profile of the patients is shown in Table 2. One hundred six patients (30.4%) had various palliative procedures before univentricular repair. One hundred and four patients (30%) had tricuspid atresia as the underlying pathological diagnosis; the Accepted for publication Dec 15, 1999. Address reprint requests to Dr Airan, Department of Cardiothoracic & Vascular Surgery, All India Institute of Medical Sciences, Ansari Nagar, New Delhi-110029, India; e-mail: bairan@medinst.ernet.in. This article has been selected for the discussion forum on the STS Web site: http://www.sts.org/section/atsdiscussion/ 2000 by The Society of Thoracic Surgeons 0003-4975/00/$20.00 Published by Elsevier Science Inc PII S0003-4975(00)01247-9
Ann Thorac Surg AIRAN ET AL 2000;69:1900 6 UNIVENTRICULAR REPAIR 1901 Table 1. Cardiac Morphology in Patients Undergoing Univentricular Repair (N 348) Morphology n % Tricuspid atresia 104 30 DORV, VSD, PS, with Hypo RV/LV 97 28 SV with PS/PA 57 16 dtga, VSD PS with Hypo RV/LV 27 8 LTGA, VSD PS with Hypo RV/LV 29 8 PA with IVS 01 0.3 Tetralogy of Fallot with Hypo RV/LV 08 2.3 Hypoplastic right heart syndrome 11 3.2 TAPVC with functional SV with PS 09 2.6 DILV with PS 02 0.6 DOLV, VSD, PS 02 0.6 RV tumor 01 0.3 Total 348 100 DILV double inlet left ventricle; DOLV double outlet left ventricle; DORV double-outlet right ventricle; dtga dextrotransposition of the great arteries; Hypo hypoplastic; IVS intact ventricular septum; LTGA congenitally corrected transposition of the great arteries; LV left ventricle; PA pulmonary atresia; PS pulmonic stenosis; RV right ventricle; SV single ventricle; TAPVC total anomalous pulmonary venous connection; VSD ventricular septal defect. remaining patients had complex congenital heart diseases with a functional single ventricle. Nine patients had presented with single ventricle and total anomalous pulmonary venous drainage (TAPVC). All patients underwent preoperative echocardiography and cardiac catheterization. Pulmonary artery pressures were measured either by direct entry into the pulmonary artery (n 54) or by measuring pulmonary venous wedge pressure (n 105). In the remaining patients (n 189), pulmonary artery pressure was measured intraoperatively before cardiopulmonary bypass by direct needle insertion into the most accessible part of the pulmonary arterial tree. A mean pulmonary artery pressure 20 mm Hg or greater was generally considered a Table 2. Preoperative Status of Patients Undergoing Univentricular Repair McGoon ratio 2.7 0.6 Mean pulmonary artery pressure (mm Hg 5.5 5.9 measured in operating room) Bifurcation/pulmonary artery stenosis 22 (6.3%) Significant systemic atrioventricular valve 7 (2%) regurgitation Small pulmonary artery (as measured in 23 (16.6%) b operating room) a Previous systemic pulmonary artery shunt 85 (24%) Previous pulmonary artery band 7 (2%) Previous bidirectional Glenn 14 (4%) Coexisting total anomalous pulmonary venous 9 (2.6%) connection Coexisting subaortic stenosis 1 (0.3%) Ventricular dysfunction 3 (0.9%) a Measured in operating room and compared with normogram (7.8); small means value less than predicted in normogram. b Out of 138. contraindication to a one-stage complete univentricular repair. Similarly, patients with systemic ventricular end diastolic pressure 12 mm Hg or greater were not considered suitable for univentricular repair. Pulmonary vascular resistance was not measured routinely. To decide about performing a single-stage procedure, pulmonary artery size (McGoon ratio) was estimated angiographically in all the patients. A localized pulmonary artery narrowing producing a reduction in diameter equal to or greater than 50% of the diameter of the adjacent pulmonary artery before its branching was an indication for a pulmonary arterioplasty. This finding generally was against a one-stage Fontan type repair. Since 1994 (n 138), we also estimate pulmonary artery size in the operating room by sizing the pulmonary artery with Hegar s dilators and comparing it with the expected value for that age and body surface area [7, 8]. The final decision was based on the intraoperative measurement of pulmonary artery size. Pulmonary arteries with Z value less than zero were considered as small, and if Z value was less than 2, a univentricular repair was not performed. Atrioventricular valve regurgitation was judged as none, mild, moderate, or severe by echocardiographic or cineangiographic evaluation. Moderate and severe regurgitation was considered significant. Similarly, ventricular function was also assessed by echocardiography/ cineangiography. An ejection fraction less than 45% was designated as suboptimal. The type of surgical modification used was directed by the underlying cardiac anatomy before 1994 and included atriopulmonary connection (n 113) or total cavopulmonary connection (n 104). Subsequently, it was entirely the total cavopulmonary connection (n 131). All patients operated upon after January 1994 had elective fenestration of the intraatrial baffle. Fenestration was done using a 4- or 5-mm coronary punch in the baffle. Cardiopulmonary bypass and myocardial preservation techniques were essentially the same over the entire study period. Assesment of Operative Outcome Outcome variables assessed were Fontan failure and postoperative pleural effusions. Fontan failure included all patients who died within 30 days of operation as well as those who survived Fontan takedown. Hemodynamic Fontan failure was defined as persistent low cardiac output after a univentricular repair with need to maintain central venous pressure higher than 18 mm Hg or inordinately high fluid requirement to maintain systemic arterial pressure [6]. Any pleural effusion that lasted for more than 10 days after operation came under the category significant. If there was any recurrence of effusion after the removal of chest tubes, it was also included in the list of significant pleural effusions. Since 1994, low-dose oral anticoagulation was given for a period of 6 months in all the patients undergoing univentricular repair. This was followed by antiplatelet therapy. The patient s clinical course (survival, symptom-
1902 AIRAN ET AL Ann Thorac Surg UNIVENTRICULAR REPAIR 2000;69:1900 6 Table 3. Type of Modification of Fontan Operation Used (N 348) Procedure n % Atriopulmonary connection 113 33 TCPC A (lateral tunnel using PTFE 190 55 patch) TCPC B (lateral tunnel using atrial 40 11 septum) TCPC C (bilateral superior cavopulmonary connection inferior venacava to left pulmonary artery through coronary sinus) 5 1 PTFE polytetrafluoroethylene; connection. atic class, need for cardiac medications, occurrence of late complications) after hospital discharge were monitored. Echocardiographic studies were performed as routine follow-up protocol. A group of patients, presenting for follow-up between July 1997 and September 1998, underwent 24-hour ambulatory electrocardiographic monitoring (Holter study) using an H-12 Holter unit (Mortara Instruments, Inc, Milwaukee, WI). Recording was analyzed on a standard Mortara Holter analyzer by a senior consultant in cardiac electrophysiology. Statistical Analysis All data were analyzed with BMDP statistical software (Biomedical Data Processing Statistical Software, Inc, Los Angeles, CA). Continuous and interval-related data were expressed as the mean standard deviation. Numeric data were analyzed with unpaired Student s t test. Categorical variables were analyzed with 2 analysis. A multiple logistic regression model was used to identify independent risk factors (Appendix) for Fontan failure and occurrence of significant pleural effusion. Selection of independent variables was a forward stepwise method with a critical probability value of 0.15 for variable inclusion and exclusion. A p value of less than 0.05 was considered significant in the final model. Analysis of time-related survival was performed using the Kaplan- Meier method. Results TCPC total cavopulmonary A total of 348 patients were operated on. Various procedures performed are shown in Table 3. A 4- or 5-mm fenestration was performed in 126 patients. Associated TAPVC repair was performed in 9 patients and subaortic resection was done in 1. Fontan Failure Outcome of the patients undergoing univentricular repair is shown in Table 4. The overall Fontan failure rate was 14%, which included 45 early deaths and five Fontan takedowns. Various causes of early mortality are shown in Table 5. By univariate analysis, date of operation Table 4. Operative Outcome in Univentricular Repair (N 348) Outcome n % Early mortality 45 13 Fontan failure a 50* 14 Pleural effusion 94 27 Late mortality 12 3 Ventricular arrhythmias 14 4 Supraventricular arrhythmias 44 13 Median hospital stay (in days) 12 (range) (8 34) a Fontan failure (n 50) includes early deaths (n 45) and Fontan takedown (n 5). (before 1994), absence of fenestration, and systemic ventricular dysfunction were significant predictors of Fontan failure (Table 6). However, by multivariate logistic analysis, only the absence of fenestration was the significant predictor for Fontan failure (risk ratio [RR] 3.3, 95% confidence interval [CI] 1.49 to 7.31, p 0.002). Effusions Significant pleural effusion was seen in 27% of the patients. By univariate analysis, date of operation, absence of fenestration, McGoon ratio less than 2.0, cardiopulmonary bypass time more than 120 minutes, and aortic cross-clamp time more than 60 minutes were risk factors for postoperative significant pleural effusion (Table 7), but by multivariate analysis, only absence of fenestration (RR 3.97, 95% CI 2.17 to 7.26, p 0.001) and aortic cross-clamp time more than 60 minutes (RR 2.15, 95% CI 1.30 to 3.55, p 0.002) were found to be significant risk factors. Fourteen patients had undergone staged repair (10 nonfenestrated, 4 fenestrated). These patients were highrisk candidates for Fontan failure. Average number of Choussat criteria [9] violated was 2.5 as compared with 0.7 in the whole series. There was no Fontan failure in this subgroup. Significant pleural effusion was present in 3 patients (21%), which was not much different from the overall incidence of effusions (27%). Table 5. Causes of Early Mortality in Univentriucular Repair a Cause of death n % Hemodynamic Fontan failure 25 7.2 Sepsis b 10 2.9 Acute renal failure 2 0.6 Respiratory failure 2 0.6 Neurologic complications 2 0.6 Pulmonary embolism 2 0.6 Intractable ventricular tachycardia 1 0.3 Unexplained sudden death 1 0.3 Total 45 13.0 a Early mortality: death within 30 days or before discharge. b Mediastinitis and wound related sepsis in 6 patients and septicemia in 4 patients.
Ann Thorac Surg AIRAN ET AL 2000;69:1900 6 UNIVENTRICULAR REPAIR 1903 Table 6. Risk Factors for Fontan Failure in Univentricular Repair (Univariate Analysis) Variable n Fontan Failure Odds Ratio 95% Confidence Interval p Age 4 years 139 20 0.9 0.5 2.0 1.0 4 15 years 168 23 15 years 41 7 0.8 0.3 2.0 0.7 Date of Operation Before 1994 217 40 2.7 1.3 6.1 0.01 After 1994 131 10 Morphology Tricuspid atresia 104 13 0.8 0.4 1.6 0.6 Others 244 37 Fenestration Yes 126 9 0.3 0.1 0.8 0.006 No 222 41 Operation Total cavopulmonary connection 235 30 0.7 0.4 1.3 0.3 Atriopulmonary connection 113 20 Previous systemic to pulmonary artery shunt Yes 85 13 1.1 0.5 2.3 0.9 No 263 37 Previous bidirectional Glenn Yes 14 1 0.4 0.01 3.4 0.7 No 334 49 McGoon ratio 2.0 69 9 0.9 0.4 2.0 0.8 2.0 279 41 Atrioventricular valve regurgitation Yes 7 1 0.9 0.9 8.6 0.6 No 341 49 Pulmonary artery stenosis Yes 22 3 0.9 0.5 3.5 0.8 No 326 47 Systemic ventricular dysfunction Yes 3 2 12 0.8 201 0.07 No 345 48 Cardiopulmonary bypass time 120 minutes 160 18 0.6 0.3 1.2 0.2 120 minutes 188 32 Aortic cross-clamp time 60 minutes 183 25 0.9 0.5 1.7 0.8 60 minutes 165 25 Late Outcome There were 12 (3.4%) late deaths. One of the patients died of severe hemoptysis in spite of repeated coil embolization of aortopulmonary collaterals, 2 died of sepsis after recurrent pleural effusion, 1 died of infected thrombus on the right and left side of the intraatrial baffle, 3 died of pneumonia, 1 of protein-losing enteropathy, 1 of anorexia nervosa, 1 of acute pancreatitis after exploratory laparotomy, and 2 during reoperation for baffle leak. Of the survivors, 287 patients have been followed up for periods ranging from 6 to 120 months (mean 46 18.0 months). Actuarial survival was 81% 4% at 90 months (Fig 1). Two hundred and fifty-eight patients (90%) were in New York Heart Association class I at their last follow-up visit. Twenty-four hours Holter monitoring was performed in 136 patients, after a mean interval of 49.0 21 months. Ventricular arrhythmias were present in 4% of patients and supraventricular arrhythmias in 15% of patients. Though there was no significant correlation between preoperative variables (Appendix) and the occurrence of arrhythmias, supraventricular arrhythmias were signifi-
1904 AIRAN ET AL Ann Thorac Surg UNIVENTRICULAR REPAIR 2000;69:1900 6 Table 7. Risk Factors for Pleural Effusion in Univentricular Repair (Univariate Analysis) Variable n Pleural Effusion Odds Ratio 95% Confidence Interval p Age 4 years 139 33 1.4 0.8 2.3 0.3 4 15 years 168 50 15 years 41 11 1.2 0.5 2.7 0.8 Date of Operation Before 1994 217 75 3 1.7 5.7 0.001 After 1994 131 19 Morphology Tricuspid atresia 104 25 0.8 0.5 1.5 0.5 Others 244 69 Fenestration Yes 126 16 0.3 0.1 0.5 0.001 No 222 78 Operation Total cavopulmonary connection 235 60 0.9 0.5 1.5 0.8 Atriopulmonary connection 113 34 Previous systemic to pulmonary artery shunt Yes 85 24 1.1 0.7 2.0 0.9 No 263 70 Previous bidirectional Glenn Yes 14 3 0.7 0.02 3.0 0.8 No 334 91 McGoon ratio 2.0 69 27 2.0 1.1 3.7 0.02 2.0 279 67 Atrioventricular valve regurgitation Yes 7 1 0.4 0.01 3.8 0.7 No 341 93 Pulmonary artery stenosis Yes 22 6 1.0 0.03 2.8 0.8 No 326 88 Cardiopulmonary bypass time 120 minutes 160 33 0.5 0.3 0.9 0.01 120 minutes 188 61 Aortic cross-clamp time 60 minutes 183 38 0.5 0.3 0.8 0.01 60 minutes 165 56 cantly less in patients with fenestration (4 of 50) as compared with those without fenestration (17 of 86). However, the mean follow-up duration in the fenestrated group (26 12 months) was much less than that in the nonfenestrated group (63 19 months). Oxygen saturations ranged from 85% to 94% (mean 89%) in patients having functioning fenestrations at a mean follow-up period of 34 months. Thirty of these (26%) had spontaneous closure of their fenestration as demonstrated by lack of any right-to-left shunt by contrast echocardiography, and in no patient has there been any incidence of cerebrovascular accidents. None of the patients with fenestration has been taken up for closure of the fenestration, either surgically or by catheter. Comment Improvement in results of univentricular repair at this center dates to the introduction of routine fenestration of the baffle in 1994. Another factor was abandonment of atriopulmonary connection. Hence, the bias of the date of operation cannot be removed from the analysis of outcome. However, the univentricular repairs have been performed long before 1994 at our center, and thus, the learning curve does not seem to be responsible for the improved results. This study included patients who underwent the Fontan operation according to the modified criteria of selection (discussion of Sharma and associates [6]), and our
Ann Thorac Surg AIRAN ET AL 2000;69:1900 6 UNIVENTRICULAR REPAIR 1905 Fig 1. Actuarial survival curve of patients undergoing univentricular repair. policy was strict with reference to pulmonary artery size, pulmonary artery distortion, and ventricular function. Pulmonary artery entry was the exception at preoperative cardiac catheterization because most of the patients had severe pulmonary stenosis and were deeply cyanosed. Preoperative pulmonary artery pressure was not available in most patients and was measured directly during operation in these. Calculation of pulmonary vascular resistance was routinely not available for all patients. Age less than 4 years and more than 15 years, as reported by Fontan and others [9 11] as a significant risk factor for early mortality, was not in accordance with our observation, which showed no significant impact of age on operative outcome. The same finding has also been noted by others [12 14], and patients under 4 years of age can be safely offered the fenestrated Fontan option when they do not violate other selection criteria. While analyzing the risk factors for Fontan failure, it was observed that absence of fenestration was the only significant factor, and introduction of elective routine fenestration reduced the Fontan failure rate significantly. Aortic cross-clamp time more than 60 minutes and absence of fenestration were risk factor for postoperative pleural effusion. Elective fenestration, though it reduces the pleural effusion rate significantly, is not protective enough if aortic clamp time exceeds 60 minutes. Successful outcome of Fontan procedure is dependent on the pulmonary vascular resistance and pulmonary artery size [15]. Though sizing the pulmonary artery by calculating the McGoon ratio from preoperative angiography had no significant impact on either early mortality or effusions, we believe that the sizing of pulmonary arteries by Hegar s dilator intraoperatively is more informative about the decision making for a one-stage Fontan. This may be explained by the observation that the pulmonary artery size determined by angiography in patients with decreased pulmonary blood flow is an underestimate of true pulmonary artery size (ie, measured intraoperatively). Arrhythmias have been reported as a sequel to Fontan operation due to the distention of the atrium, and secondary to multiple suture lines [16]. Supraventricular arrhythmias reduced from 20% to 8% with introduction of fenestration, which may signify reduced distension of exposed atrial wall. However, the mean follow-up duration in the fenestrated group (26 12 months) was much less than that in the nonfenestrated group (63 19 months), and thus, the difference in incidence of arrhythmias may also be attributed to the duration of follow-up. Higher incidence of supraventricular tachyarrhythmias in patients with nonfenestrated Fontan may also be attributed to a greater proportion of patients having atriopulmonary connection.
1906 AIRAN ET AL Ann Thorac Surg UNIVENTRICULAR REPAIR 2000;69:1900 6 Atrial thrombosis resulting in thromboembolism has been reported after Fontan operation [17]. Atrial thrombosis occured in 1 of our patients who had undergone fenestrated repair and contributed to his demise. Despite its presence, fenestration did not contribute to occurrence of systemic thromboembolism. Staging the Fontan procedure has been established to have a good outcome in spite of the presence of a number of risk factors [5, 18]. There was no Fontan failure or early mortality in this group, while the effusion rate was similar (21% vs 27% of the series) to the single-stage univentricular repairs. This is a significant observation because these patients had more risk factors for Fontan operation as compared with the group undergoing onestage repair, but still had a favorable outcome. The small number, however, makes any derivations highly speculative. Follow-up of patients has shown excellent results with respect to New York Heart Association functional class (90% in class I). Thirty patients (26%) had spontaneous closure of a 4-mm fenestration over a mean period of 34 months, and oxygen saturation of 85% to 94% postoperatively in these patients raises a question regarding the need for closure of the fenestration. The fact that in none of our patients did we think it necessary to close the fenestration subsequently infers the recommendation of routine fenestration of atrial baffle with no need for subsequent closure. Conclusions Fenestration of the atrial baffle is associated with reduced Fontan failure rate and decreased occurrence of pleural effusion. Subsequent closure of the fenestration is usually not required. Ischemic time more than 60 minutes continues to be a risk factor for postoperative effusion despite fenestration. We thank Rajvir Singh, MSc(Stat) for statistical analysis of the work. References 1. Fontan F, Baudet E. Surgical repair of tricuspid atresia. Thorax 1971;26:240 8. 2. Laks H, Haas GS, Pearl JM, et al. The use of an adjustable interatrial communication in patients undergoing the Fontan and definitive heart procedures [Abstract]. Circulation 1988;78(Suppl II):357. 3. Bridges ND, Lock JE, Castaneda AR. Baffle fenestration with subsequent transcatheter closure. Modification of the Fontan operation at increased risk. Circulation 1990;82:1681 9. 4. Bridges ND, Mayer JE, Lock JE, et al. Effect of fenestration on outcome of modified Fontan repair. Circulation 1992;86: 1762 9. 5. Jacobs ML, Norwood WI. Fontan operation: influence of modifications on morbidity and mortality. Ann Thorac Surg 1994;58:945 52. 6. Sharma R, Iyer KS, Airan B, et al. Univentricular repair, early and midterm results. J Thorac Cardiovasc Surg 1995;110: 1692 701. 7. Rowlat UF, Rimaldi MJA, Lev J. The quantitative anatomy of the normal child s heart. Pediatric Clin North Am 1963;10: 499 588. 8. Kirklin JW, Barratt-Boyes BG. Cardiac surgery. Morphology, diagnostic criteria, natural history, techniques, results and indications, 2nd ed. New York: Churchil Livingstone, 1993: 33 40. 9. Choussat A, Fontan F, Besse P, Vallot F, Chauve A, Bricand H. Selection criteria for Fontan s procedure. In: Anderson R, Shineborne E, eds. Paediatric cardiology. New York: Churchill Livingstone, 1977:559 66. 10. Mayer JE, Bridges ND, Lock JE, Hanley FL, Jonas RA, Castaneda AR. Factors associated with marked reduction in mortality for Fontan operations in patients with single ventricle. J Thorac Cardiovasc Surg 1992;103:444 52. 11. Mair DD, Hagler DJ, Puga FJ, Schaff HV, Danielson GK. Fontan operation in 176 patients with tricuspid atresia. Circulation 1990;82(Suppl):164 9. 12. Mayer JE, Helgason H, Jonas RA, et al. Extending the limits of modified Fontan procedures. Thorac Cardiovasc Surg 1986;92:1021 8. 13. Weber HS, Gleason MM, Myer JL, Waldhausen JA, Cyran SC, Baylean BG. The Fontan operation in infants less than 2 years of age. J Am Coll Card 1992;19:828 33. 14. Uemura H, Yagihara T, Kawashima Y, et al. What factors affect ventricular performance after a Fontan-type operation? J Thorac Cardiovasc Surg 1995;110:405 15. 15. Fontan F, Fernandez G, Costa F, et al. The size of the pulmonary arteries and the results of the Fontan operation. J Thorac Cardiovasc Surg 1989;98:711 24. 16. Cromme-Dijkhuis AH, Hess J, Hählen K, et al. Specific sequelae after Fontan operation at mid- and long-term follow-up. Arrhythmia, liver dysfunction, and coagulation disorders. J Thorac Cardiovasc Surg 1993;106:1126 32. 17. Jahangir M, Ross DB, Redington AN, Lincoln C, Shinebourne EA. Thromboembolism after the Fontan procedure and its modification. Ann Thorac Surg 1994;58:1409 14. 18. Kaulitiz R, Ziemer G, Luhmer I, Kallfelz H-C. Modified Fontan operation in functionally univentricular hearts: preoperative risk factors and intermediate results. J Thorac Cardiovasc Surg 1996;112:658 64. Appendix. List of Variables Preoperative variables Age Morphology Previous systemic pulmonary artery shunt Previous pulmonary artery band Previous bidirectional Glenn Pulmonary artery size (McGoon ratio) Pulmonary artery stenosis Systemic ventricular dysfunction Systemic atrioventricular valve dysfunction Subaortic stenosis Total anomalous pulmonary venous drainage Pulmonary artery pressure Left ventricular end diastolic pressure Operative variables Year of operation (1993 and before vs 1994 and after) CPB time Aortic cross-clamp time Simultaneous pulmonary artery reconstruction Additional procedure Type of connection (total cavopulmonary connection vs others) Fenestration of atrial partition Surgeon identity