Palliative operations and interventional cardiac catheterization have several applications in patients with congenital heart disease. They are used to permit survival of critically ill neonates, to prepare the circulation for later more curative surgery, and to permit an acceptable quality of life in patients with complex malformations not amenable to corrective surgery. A low-risk palliative operation, which results in a good haemodynamic state, can sometimes be a preferable option to a high-risk ‘corrective’ operation. The later development of heart surgery for complex congenital heart disease was only made possible by two pioneering palliative procedures: the Blalock-Taussig shunt to increase pulmonary blood flow, and the Rashkind balloon atrial septostomy to improve mixing at the atrial level. More recently, survival of infants with cardiac malformations in which either the pulmonary or systemic circulation is dependent on a patent ductus arteriosus has been dramatically improved by the use of prostaglandin treatment to maintain duct patency.

In order to have a satisfactory quality of life certain haemodynamic conditions must be met: systemic venous and pulmonary venous return must be unobstructed; systemic arterial outflow from the heart must be unobstructed; pulmonary blood flow must be sufficient to give acceptable arterial oxygen saturations, but at a relatively normal pressure, and not so high as to cause congestive heart failure; and if the great vessels are transposed there must be adequate mixing of systemic and pulmonary venous return.

Some patients with ‘univentricular hearts’ and pulmonary stenosis fulfil these criteria and may remain completely asymptomatic into adulthood; however, in many patients with complex lesions palliative procedures are used to create these conditions.

Transposition of the great vessels is typical of a condition in which severe arterial desaturation results from inadequate mixing of systemic and pulmonary venous return. There are, however, many other conditions where poor mixing or unfavourable streaming of systemic and pulmonary venous return result in severe hypoxia in the systemic circulation. The initial palliative treatment should be an infusion of prostaglandin E&sub1; or E&sub2;. This usually improves arterial oxygenation significantly by allowing mixing at ductal level, and by increasing pulmonary blood flow. However, this solution is only temporary and further management depends on the type of eventual surgical correction that is planned. If an arterial switch procedure is planned in the next few days prostaglandin treatment alone may be sufficient; if an atrial redirection procedure or a later arterial switch is planned it is usually necessary to improve the arterial oxygenation by an atrial septostomy. In the first days or weeks of life this can be achieved by the use of a Rashkind septostomy catheter, introduced either via the sapheno-femoral venous junction or via the umbilical vein. This procedure used to be undertaken only in the cardiac catheterization laboratory, but can be performed in the paediatric intensive care unit under the guidance of ultrasound scanning (Fig. 1) 1676.

In later life, or if a larger opening in the atrial septum is required, a surgical atrial septostomy is generally the procedure of choice. This septostomy can be performed as a closed procedure (Blalock-Hanlon septostomy) or, more commonly, as an open atrial septostomy using cardiopulmonary bypass. Rarely, where there are relative contraindications for surgery, a cardiac catheter blade septostomy, using a percutaneously introduced catheter with a knife blade, can be used to enlarge existing atrial septal defects. However, the opening achieved is not as large as a surgically created defect, and neighbouring structures may accidentally be damaged.

A group of cardiac malformations presents with cyanosis secondary to reduced pulmonary blood flow. The most common is tetralogy of Fallot; others include pulmonary atresia with or without a ventricular septal defect, tricuspid atresia with aorta arising from the main chamber and the pulmonary artery from the outlet chamber, and a double outlet right ventricle with valvar or infundibular pulmonary stenosis. If hypoxia is severe in the neonate prostaglandin E&sub1; or E&sub2; infusion temporarily reopens a closing ductus arteriosus. Oral prostaglandin E&sub2; administration may be used to maintain an open ductus for a period of weeks or even months, but the need for frequent administration makes it a difficult home-based treatment.

In many patients it is necessary to consider creating a surgical shunt to increase pulmonary blood flow. In the neonate the procedure of choice is the modified Blalock-Taussig shunt (Fig. 2) 1677. This involves a prosthetic tube (usually Gore-Tex or Impra) being anastomosed end-to-side with either the right or the left subclavian artery and connected end-to-side with the ipsilateral pulmonary artery. If the subclavian artery is extremely small the innominate artery is occasionally used. This shunt has a good patency rate, particularly if a tube of 5 mm, rather than 4 mm, diameter can be used. As long as the shunt is unilateral and connected to confluent central pulmonary arteries it rarely causes excessive pulmonary blood flow, and it is comparatively easy to take down at the time of further surgery. Many surgeons feel that a 2- to 3-month period of treatment with low-dose aspirin and dipyridamole reduces the risk of thrombotic occlusion.

In older children, particularly those with a non-correctible cardiac malformation who require permanent palliation, the classical Blalock-Taussig shunt (Fig. 3) 1678 can produce an excellent long-term result. This shunt is created by anastomosing the right or the left subclavian artery end-to-side to the ipsilateral pulmonary artery and ligating the distal subclavian artery. The arm still receives adequate blood supply from collaterals around the shoulder. The particular advantage with this shunt is the fact that no prosthetic material is used and that the subclavian artery will grow with the patient. Some of Helen Taussig's original patients are still alive, and reasonably well on their initial shunt.

The Waterston shunt (a fenestration between ascending aorta and right pulmonary artery) and the Potts shunt (a fenestration between descending aorta and left pulmonary artery) are rarely used nowadays. The former has a relatively high incidence of pulmonary vascular disease and of severe distortion of the right pulmonary artery, and the latter is extremely difficult to take down at the time of later surgery. Some surgeons still use the Waterston shunt when the pulmonary arteries are extremely small, but a central shunt using a tube prosthesis from the aortic arch to the main pulmonary artery is probably preferable (Fig. 4) 1679. The central shunt is also useful in patients with pulmonary atresia, ventricular septal defect, and multifocal pulmonary blood supply, in whom it helps to enlarge a hypoplastic central pulmonary artery confluence when used in conjunction with attempts at unifocalization of pulmonary blood supply.

The Glenn shunt is a venous shunt which, in its classical form, constructed by anastomosing the right superior vena cava end-to-end with the divided right pulmonary artery. For a period it went out of favour because some patients developed late pulmonary arteriovenous fistulae with increasing cyanosis. This complication was thought to have been a reaction of the pulmonary vasculature to the absence of pulsatile flow. There has been a great resurgence in the use of this shunt recently, as in many ways it is ideal for preparing the circulation for a later Fontan-procedure (Fig. 5) 1680, since it provides good pulmonary blood flow at very low pressure, thus avoiding risk of precipitating a rise of pulmonary vascular resistance. Furthermore it does not cause a volume overload of the systemic ventricle, which may lead to dysfunction, particularly diastolic, of this chamber. Nowadays this shunt is usually constructed as a bidirectional Glenn shunt by anastomosing the right superior vena cava end-to-side to the right pulmonary artery. This has the advantage of distributing pulmonary blood flow equally to the two lungs and preserving some pulsatility in the pulmonary vascular bed. Used in this way it is sometimes referred to as a ‘hemi-Fontan’. In patients with laevoisomerism (‘bilateral left-sidedness’) systemic venous drainage may be via a left, or bilateral venae cavae, and under these circumstances a left-sided, or bilateral Glenn shunt may be constructed. The only drawback of this shunt is that it cannot be used in neonates as their pulmonary vascular resistance is too high. It is mostly used in children over 12 months of age.

Where reduced pulmonary blood flow is mainly due to valvular obstruction an obvious approach is to enlarge the pulmonary valve orifice; however, in all patients with a large coexisting ventricular septal defect there is the risk of flooding the pulmonary circulation at too high a pressure. The most controlled method is probably via catheter balloon dilatation with a carefully sized balloon, but a surgical outflow–tract enlargement is occasionally used as a first preparatory stage in the treatment of patients with tetralogy of Fallot and a severely hypoplastic main pulmonary artery.

In pulmonary atresia with intact ventricular septum the right ventricle is usually severely hypoplastic and unable to support the pulmonary circulation, even if continuity between right ventricle and pulmonary artery can be re-established with a valvotomy. However, if the management is by a systemic-to-pulmonary shunt only, the right ventricle will not grow. Optimal management therefore consists of securing pulmonary blood flow, usually with a modified Blalock shunt but occasionally with long-term prostaglandin treatment, followed by a valvotomy or right ventricular outflow reconstruction to establish forward flow through the right ventricle. It is important that the outflow obstruction is relieved as completely as possible: there is usually substantial tricuspid valve incompetence in this condition, and this severely compromises forward flow if there is a large outflow gradient remaining. If there is only a patent foramen ovale, or if the atrial septal defect is small, it is important also to secure unobstructed inflow of systemic venous blood across the atrial septum into the left heart chambers by carrying out a septostomy, usually a Rashkind septostomy.

Whereas at birth the pulmonary vascular resistance is approximately equal to systemic vascular resistance, postnatally it should normally drop rapidly, so that at 3 months of age it is normally a small fraction of that of the systemic circulation (resistance ratio about 0.02–0.1). Thus, if there are large defects in the interventricular septum, if both arteries arise from the same ventricle, or if a functionally common ventricle exists, the pulmonary circulation will be flooded by high flow at systemic pressure. This often precipitates congestive heart failure and carries the risk of causing irreversible pulmonary vascular disease. If it is caused by a single ventricular defect alone immediate surgical closure of the defect is usually appropriate. On the other hand, if there are multiple ventricular septal defects, particularly if they are of the ‘Swiss cheese’ variety, or if there is a common atrioventricular septal defect, or associated lesions such as coarctation of the aorta, the pulmonary circulation can be protected by pulmonary artery banding (Fig. 6) 1681. In this operation a silk tie is applied around the main pulmonary artery and tightened until a desired ratio between distal pulmonary artery pressure and systemic artery pressure is achieved. Pulmonary artery banding is also used to protect the pulmonary circulation in patients with complex heart disease, but the actual tightness of the band may vary depending on the timing of future planned procedures. Since the band does not grow with the patient, the average pulmonary artery band will initially still admit moderately excessive flow. It usually produces an ideal balance between pulmonary and systemic flow at around 12 months of age, and then gradually restricts pulmonary blood flow so that some right to left shunting starts occurring usually between 2 and 4 years of age. In some patients with complex heart disease it may be appropriate to detach the pulmonary arteries from the heart and to feed the pulmonary circulation through a shunt; this applies, for example, in the Norwood stage 1 procedure for hypoplastic left heart syndrome, where the pulmonary artery is used to reconstruct an aortic arch.

The most common form of inflow obstruction in congenital heart disease consists of atresia of one of the atrioventricular valves. In patients with atresia of the right atrioventricular valve (usually morphologically a tricuspid valve) the systemic venous return has to cross a patent foramen ovale or an atrial septal defect to reach the ventricles. If this opening is restrictive in a neonate a Rashkind balloon atrial septostomy is performed; in the older infant or child an open atrial septostomy is preferable. Atresia of the left atrioventricular valve (usually morphologically a mitral valve) is associated with pulmonary venous congestion unless a good size atrial septal defect is present to allow unrestricted flow into the right atrium, and a Rashkind atrial septostomy is often required in the neonate. In later life the interatrial communication may often need to be enlarged by an open atrial septectomy. Unless it is anticipated that atrial septal structures will be used in subsequent surgery, the excision of the atrial septum in this situation should be wide, rather than including only the fossa ovalis membrane, so that functionally a single atrium is achieved. Where there is congenital severe stenosis, as opposed to atresia, of an atrioventricular valve balloon dilatation of the valve, at least of the right atrioventricular valve, may find a role in management but this still remains to be evaluated.

Correctable causes of outflow obstruction, such as valvular stenosis or coarctation, should be corrected in patients with complex congenital heart disease, since if they coexist with an unprotected pulmonary circulation the effect of pulmonary artery banding will be severely diminished by the systemic outflow gradient, and irreversible pulmonary vascular disease will occur early. An example of a difficult situation is the child with either tricuspid atresia or double-inlet left ventricle where the great arteries are ‘transposed’ so that the pulmonary artery comes off the main chamber (the left ventricle), and the outflow chamber is filled via a restrictive ventricular septal defect and gives rise to the aorta, often with muscular subaortic stenosis and/or coarctation of the aorta. In the past these children were treated with pulmonary artery banding with or without coarctation repair, but the mortality rate was high and the few survivors developed pulmonary vascular disease. In recent years more radical palliative procedures are therefore becoming adopted. One approach is a variant of the Damus-Kaye-Stansel procedure, in which the pulmonary artery is transected near the bifurcation and the proximal end is anastomosed end-to-side to the aorta. The pulmonary circulation is then fed via a shunt. The mortality rate associated with this procedure is substantial, often related to the difficulty in the immediate postbypass period in establishing sufficient pulmonary blood flow through a shunt into a pulmonary vascular bed that is adapted to flow at high pressures. There have also been some long-term problems with incompetence of the previously pulmonary, now systemic, valve, but generally the haemodynamic state of the survivors is good. Lincoln has advocated using surgical enlargement of the ventricular septal defect (Fig. 7) 1682 (with or without a subaortic gusset) in these patients, and the surgical mortality of this procedure appears lower. However, it is difficult to achieve a sufficiently large ventricular septal defect in a small neonate, and restriction of flow between main chamber and outflow chamber may recur. This procedure needs to be combined with pulmonary artery banding if carried out as the first palliation. Again, the symptomatic state of survivors is good. One advantage with this operation is that it can be used to improve the haemodynamics even in patients with pulmonary vascular disease.

The most complete form of systemic outflow obstruction is aortic atresia, usually combined with left ventricular hypoplasia to create the hypoplastic left heart syndrome, which is the most common cause of death from congenital heart disease. As mentioned above, Norwood has described an initial palliative procedure, which by using the pulmonary artery to reconstruct the aortic arch, transforms the heart to a ‘univentricular heart of right ventricular morphology’, with the pulmonary circulation fed from a systemic-pulmonary shunt into a reconstructed pulmonary confluence. If pulmonary vascular resistance remains low in later life a ‘correction’ is attempted, using a Fontan procedure. The combined surgical mortality rate of this series of operations is very high, and there are doubts about the long-term ability of the right ventricle and half a right atrium to support the systemic circulation. Furthermore the few survivors usually require prolonged intensive care. For that reason many large centres in the United Kingdom do not feel that they can justify using scant resources in carrying out this procedure. The other possible active treatment in this condition is a neonatal heart transplant, and the Norwood stage 1 operation may serve as a holding procedure, with the aim of later transplantation. However, the supply of potential human neonatal heart donors will never match the need if all children with the hypoplastic left heart syndrome receive active treatment, and it is important to discuss fully with the parents whether they really wish for active treatment to be carried out.

There are many conditions in which right-to-left shunting across an atrial septal defect or a patent foramen ovale aggravates arterial hypoxia in obstructive right-sided heart lesions. This situation is usually best dealt with by correction of the underlying anomaly. One special case in which a ‘palliative’ closure of the atrial septal defect may be helpful is Ebstein's anomaly of the tricuspid valve. Because the tricuspid valve is usually severely incompetent, and sometimes mildly stenotic, in severe cases of the anomaly, right-to-left shunting at the atrial level is often substantial if there is an atrial septal defect present. If hypoxia is the major symptom, and the right ventricle is judged to be able to accommodate pulmonary blood flow, the mortality associated with closure of the atrial septal defect alone is much lower than if the procedure is combined with valve replacement or tricuspid reconstructive surgery.

It is important that the choice of palliative procedure is based on the fullest diagnostic information possible. This can often be obtained non-invasively by cardiac ultrasound and Doppler studies. Considerable thought has to be given to the likely final correction or palliation required. For instance, in a child with transposition of the great arteries, severe pulmonary stenosis and a ventricular septal defect with a straddling atrioventricular valve corrective surgery is not possible. Initially, a Rashkind balloon atrial septostomy should be considered, even if the child is not critically hypoxic, because long-term mixing at atrial level will become important. Later the child is likely to need a procedure to increase pulmonary blood flow. As the final palliation is likely to be a Glenn shunt combined with an open atrial septectomy, it is important that if any systemic-to-pulmonary shunt has to be performed before the child is large enough for a Glenn shunt, it should be a left-sided shunt, usually a modified Blalock, as a right-sided Blalock shunt would make a later Glenn shunt very difficult. Thus, if palliative procedures are chosen correctly, and carried out at the optimal time, even a child with uncorrectable congenital heart disease can have a good future. Good results require regular monitoring of the child and close liaison between the paediatric cardiologist and the cardiac surgeon.

Anderson RH, Macartney FJ, Shinebourne EA, Tynan M, eds. Paediatric cardiology. Vols. 1, 2. London: Churchill-Livingstone, 1987.

Anderson RH, Becker AE, Lucchese FA, Meier MA, Rigby ML, Soto B. Morphology of congenital heart disease. Angiographic, echocardiographic and surgical correlates. Tunbridge Wells, Kent: Castle House, 1983.

Ebert PA. Atlas of congenital cardiac surgery. New York: Churchill-Livingstone Inc, 1989.

Garson E Jr, Bricker JT, McNamara DG, eds. The science and practice of pediatric cardiology. Philadelphia: Lea & Febiger, 1990.

Kirklin JW, Barratt-Boyes BG. Cardiac Surgery. Vols. 1, 2. 2nd edn. New York: Churchill-Livingstone Inc, 1993.

Perloff JK, Child JS. Congenital heart disease in adults. Philadelphia: WB Saunders, 1991.

Stark J, de Leval M, eds. Surgery for congenital heart defects. New York: Grune and Stratton, 1983.