Low cardiac output is defined as persistent hypotension, oliguria (urine output < 0.5 ml/kg/hr) with clinical signs of tissue hypoperfusion despite maximum inotropic support. (Combination of dopamine, adrenaline, milrinone or enoximone and noradrenaline or vasopressin). In this study the term cardiorespiratory failure refers to severe respiratory failure leading to refractory hypoxemia resulting in cardiac dysfunction. Mechanical complications were complications related to ECMO circuit such as clot in the bladder or oxygenator failure.

Bleeding complications is defined as uncontrolled mediastinal bleeding despite correction of coagulopathy requiring surgical intervention. Abnormal neurology is defined as clinical signs of neurological dysfunction with definite radiological findings like cerebral bleeding or infarct on computerized tomographic scan of brain. Arrhythmias after ECMO support refer to the occurrence of non-sinus rhythm anytime after withdrawal of ECMO support. The term sepsis was defined as clinical evidence of sepsis with positive microbiology.

The ECMO circuit consists of a venous cannula draining the blood by gravity to a compliant 60 ml silicon bladder, which is a servo regulating venous return monitor. A roller pump (Sorin Biomedical, New haven, West Sussex, UK) propels the blood from the bladder to an appropriately sized poly-methoxy pentane membrane oxygenator (Medos, Stolberg, Germany) with in-built heat exchanger (Betta-Tech controls, Milton Keynes, UK). Avecor silicone membrane oxygenators (Avecor Cardiovascular Products, Plymouth, Minnesota, USA) were used till 2000. The oxygenated blood from the oxygenator is returned to the patient through an arterial cannula. There is a connecting bridge between arterial and venous line of the ECMO circuit (Figure 1). All the patients in this study were supported by VA ECMO according to our published protocol 10.

Univariate and multivariate analysis was performed using peri ECMO variables with Genstat Release version-6.1 (VSN International, Oxford, UK) statistical package. Variables included in the analysis are explained in Table 2. Analysis of the simple association, 2 × 2 table of association of death/survival classified by all the other binary factors, has been done using a chi-squared test and was checked by using Fisher exact test. Continuos variables were examined by logistic regression.

The indications for ECMO support were a) low cardiac output (n = 16), b) failure to wean from cardiopulmonary bypass (n = 13), c) cardiac arrest in PICU (n = 10) and d) cardiorespiratory failure (n = 14).

Thirty-five patients (66%) were successfully weaned off from ECMO support. Twenty-nine patients (54%) survived to hospital discharge. Twenty patients (37.7%) were alive after a mean follow up period of 75 months. Survivals to hospital discharge based on their indication for ECMO support were as follows: 75%(12/16) in low cardiac output patients, 54% (7/13) in failure to come off CPB patients, 50% (5/10) in cardiac arrest patients and 36% (5/14) in cardiorespiratory failure patients.

Patient related complications such as bleeding and sepsis are charted in Table 4. The details of mechanical complications related to circuit are as follows: Thirteen patients (24%) had small blood clots identified in bladder box and bridging tube and 7 patients (13%) had clots in oxygenator. Three patients (6%)had oxygenator failure needed oxygenator changing and other two patients (4%)had ECMO circuit failure requiring new circuit. None of these mechanical complications were fatal.

Twenty-four patients (45%) died in hospital, eighteen patients (34%) died during ECMO support and six patients (11%) died in hospital after withdrawal of ECMO. About half of the patients died due to irreversible multiorgan failure (29%) and neurological damage (21%). Failed myocardial recovery (12%), hemorrhage (17%) including pulmonary and gastrointestinal hemorrhage and other causes (21%) including profound sepsis were responsible for the death of the remaining patients.

On univariate analysis the following factors were associated with increased hospital mortality: postoperative arrhythmias, duration of ECMO more than 168 hours, bleeding complications, requirement of renal replacement therapy (RRT) on ECMO, arrhythmias and cardiac arrest after ECMO support. On multivariate analysis, arrhythmias after ECMO, bleeding complications and abnormal neurology during ECMO were significantly associated with hospital mortality. Comparatively extra-thoracic cannulation had significantly less bleeding on ECMO (p = 0.031)

Rhythm disturbances can often result in hemodynamic instability during the immediate postoperative period and it is a major cause of morbidity and mortality. Arrhythmias during the crucial period of myocardial recovery should always raise the possibility of progressive cardiac dysfunction or acute respiratory decompensation. Recurrent episodes of postoperative arrhythmias before the initiation of ECMO were significantly associated with the hospital mortality in this study (p < 0.001). Most of the arrhythmias were resistant to anti-arrhythmatic medications. Hoskote, A and collegues also found that refractory tachydysrhythmia before initiation of ECMO support was a risk factor for mortality in functional single ventricle patients25. Extracorporeal life support has also been used successfully for the treatment of supraventricular tachycardia (SVT) as well 26.

Some of these patients required longer duration of mechanical support to get adequate myocardial recovery. Severe initial myocardial insult before the initiation of ECMO support could be the likely reason for prolonged myocardial recovery time. In this study it was found that one fourth of patients (26.4%) required ECMO support over a week and four patients (7.54%) needed more than two weeks. Statistically it was observed that duration of ECMO more than 168 hours is clearly associated with hospital mortality (p < 0.001). Previous studies have demonstrated the fact that if the patient needed a longer duration of ECMO support, their survival tends to be poor372224. In fact, Duncan et al observed poorer outcomes even in the children who were supported by VAD for more than 48 hours7.

More than half of patients (57%) required CVVH in this study, the need of RRT on ECLS in these patients implies the severity of secondary organ dysfunction. The majority of these patients were treated by continuous venovenous hemofiltration while few of them had peritoneal dialysis. In this cohort, requirement of renal replacement therapy during ECMO support was identified as a significant risk factor for hospital mortality (p < 0.001). Kolovos et al also demonstrated that the post cardiotomy ECMO supported children requiring hemofiltration were five times more likely die than other patients 9.

The incidence of neurological complications during VA-ECMO support varies from 22% to 31% 71617. In our study total of ten patients (21%) had abnormal neurology; three of them had cardiopulmonary resuscitation (CPR) for more than 45 minutes before initiation of ECMO. It is difficult to attribute ECMO support was the reason for neurological complications in these 3 patients. Kulik et al mentioned about thirty percent of patients with neurological complication who had previous CPR before ECMO, moreover neurological complications was one of the reasons for withdrawal of ECMO support in their study 27. In our study Four patients had intracranial bleeding (8%), Two patients had localized cerebral infracts (4%), Three patients had hypoxic ischemic encephalopathy (5%) and one patient had cerebral atrophy (2%). we found that abnormal neurology is one of the variables strongly associated with hospital mortality (p < 0.001). In this cohort only two of long-term survivors had functional neurological deficit (Hemiperesis) where as 39% of long-term survivors had neurological deficits in Chow et al study 17. They also found that CPR prior to ECLS was a predictor of hospital death but whether there is a relation between the CPR prior to ECLS and neurological deficit is not clear in their study 17. Where as Chaturvedi and colleagues have proved that brain death was very strongly associated with CPR before initiation of ECMO 24.

Bleeding complications are an inevitable problem during this life saving therapy. Heparin is not the only factor responsible for bleeding in these patients since most of them were bound to have altered homeostasis resulting in deranged hemostasis. Mediastinal bleeding requiring surgical reexploration was another risk factor significantly affecting hospital survival in this group of patients (p < 0.001). In this study 36% of patients had intrathoracic bleeding within 48 hours of initiation of ECMO. Once the abnormal coagulation was corrected with Fresh frozen plasma, Cryoprecipitate and platelet transfusion. Heparin was withheld temporarily in these patients to control the bleeding. Aprotinin and AMICAR were also used. Failure of these conservative measures resulted surgical reexploration. Upon reopening the chest almost more than 90% of patient no surgical source of bleeding was found. Their sternum left open with skin closed for 24 to 48 hours. Majority of them had more than one surgical reexploration. Meliones JN and collegues were also identified mediastinal bleeding as risk factor for mortality in their review of extracorporeal life support organization registry 28. The initiation of ECMO leads to an initial worsening of coagulopathy in these sick patients29 Whether using heparin coated extra-corporeal circuit 30 or using recombinant Factor VIIa would make any difference in these patients needs further evaluation31. In fact, Ahoran et al reported all the patients who had transthoracic ECMO cannulation needed surgical reexploration for bleeding but the actual incidence surgical bleeding in that group was not mentioned 3. The relationship between type of cannulation and bleeding complications in our study will be discussed below.

Rhythm disturbances after withdrawal of ECMO are potentially serious complications adversely affecting the outcome. Occurrence of arrhythmias in these patients may be due to electrolyte abnormalities, metabolic disturbances (acidosis) and poor myocardial recovery despite maximal ECMO support. Nine patients (29%) developed recurrent arrhythmias after withdrawal of ECMO. Six patients had SVT (Junctional ectopic tachycardia in five, atrial flutter in one patient) and three patients had ventricular tachycardia. These patients were treated with appropriate anti-arrhythmatic medications and electro-cardioversion. In this study arrhythmias after withdrawal of ECMO support was identified as a risk factor for hospital mortality (p < 0.001). Becker et al found that cardiovascular complications including arrhythmias have significant negative impact on the survival of the ECMO patients 32. Cardiac complications were also noted to be one of the major causes for mortality in the ELSO registry data 28. Identifying these high-risk patients with poor myocardial recovery and address them with bridge to transplant or recovery might improve the survival 33.

Appropriate size of the cannula is the crucial factor in providing the support by extracorporeal circuit. In these patients even if cannulation were done in the operating theatre, the preferred type of cannulation was cervical. The cannulation technique has been described previously 34. Seventy nine percent of patients were cannulated via the internal jugular vein and carotid artery while only twenty one percent had transthoracic cannulation. In this study we identified that the extrathoracic cannulation was associated with significantly less bleeding complications (p = 0.031). Duncan and colleagues also found this fact in their study involving both surgical and non-surgical patients 7. Moreover the relationship between cannulation type and the abnormal neurology was assessed using Fisher's exact test and was not found to be significant (p = 0.67).

This is a retrospective study from a single centre. Sample size is relatively small and only three patients with univentricular hearts were involved. Post cardiotomy paediatric patients following surgical correction of congenital cardiac defect who required VA-ECMO only were included in this study. Some laboratory variables like lactate could not be included in the analysis because of incomplete data. In spite of the limitations, our study is one of the few reported a long-term survival of pediatric patients following VA-ECMO support.