Further discussion in ECMO Neonatal

One of the most controversial topics related to neonatal ECMO treatment is the patient selection and qualifying criteria. Due to the invasive nature of ECMO and the severe risks associated with the treatment, ECMO has always been held in reserve to treat only those infants in whom other, less invasive and less dangerous therapies have failed. Initially this was determined subjectively by the physician. However, as ECMO expertise has improved and patient application broadened, objective criteria have been developed to define a population of patients in whom the benefits of ECMO would outweigh the risks.

ECMO Patient Selection Criteria

The standard accepted patient selection criteria are based on historic experience, patient safety and mechanical physics relative to the equipment (Table I).

Gestational Age > 34 weeks. The need for systemic heparinization of the ECMO patient limits the population in whom the benefits of ECMO out-weigh the risks. In the early experience with ECMO, premature infants with respiratory failure were offered ECMO. The result was significant morbidity and mortality related to intracranial hemorrhage (1). Refinement of ECMO techniques in the 1980s greatly improved safety, however, premature infants continue to be at particular risk for intracranial hemorrhage. A retrospective review of the early ECMO patients by Bui et al. (2) concluded that with improvements in diagnosis, patient care, and refinement of the ECMO technique, treatment of the premature infant may become possible. However, until the effects of unilateral jugular and/or carotid ligation on brain perfusion are better delineated, this area remains controversial. New technologic advances, resulting from better understanding of the clotting cascade and blood-surface interactions may improve the safety of ECMO in this population.

Birth Weight > 2,000 grams. The principle reason for this limitation in size is related to cannula diameter. The smallest ECMO cannula available from a manufacturer is an 8 French (Fr.). Flow through a tube is directly related to the radius of the tube to the

TABLE I. Neonatal ECMO Patient Selection Criteria

Gestational age > 34 weeks Birthweight > 2,000 grams No significant coagulopathy or uncontrollable bleeding No major intracranial hemorrhage Mechanical ventilation < 10-14 days Reversible lung injury No lethal malformations No major cardiac malformation

4th power. Therefore, as the internal diameter of the cannula is reduced, the ability to provide adequate ECMO flow is reduced by a power of 4. Cannulation of infants smaller than 2,000 grams with an 8 Fr. ECMO cannula depends on the size of the blood vessels and the skill of the surgeon. Some cannulae are manufactured to maximize internal diameter by reducing cannula wall thickness, thus optimizing flow. Using cannulae not specifically designed for ECMO, or a cannula smaller than 8 Fr. is discouraged. Patient size alone should not be an absolute contraindication to ECMO.

Lack of Major Coagulopathy or Active Bleeding. The need for ongoing systemic heparinization while on ECMO places the patient at risk for bleeding (3). Heparin acts at multiple sites in the normal coagulation cascade, inhibiting fibrin clots. In patients with coagulopathy, heparin may not be effective. Patients considered for ECMO should be screened for coagulopathy before the initiation of bypass. If a coagulopathy exists, all attempts to correct the deficits should be made prior to the initiation of ECMO. Patients with uncorrectable coagulopathy, ongoing uncontrolled bleeding, or sepsis are at increased risk for bleeding complications while supported with ECMO. In these patients, the risks associated with treatment may outweigh the benefits.

No Major Intracranial Hemorrhage. Heparin use during ECMO and altered cerebral blood flow increases the risk of extending a pre-existing intracranial bleed. Patients with small interventricular hemorrhages (Grade I-II) may be considered for ECMO on a case by case basis. Patients with intracranial bleeds, cerebral infarcts and other risk factors (such as prematurity, coagulopathy, ischemic central nervous system injury or sepsis) are particularly high risk patients (1,4). In these situations the risks of the procedure may outweigh the benefits.

Mechanical Ventilation Less Than 10-14 Days and Reversible Lung Disease. Prolonged exposure to high concentrations of oxygen and positive pressure ventilation is known to lead to the development of bronchopulmonary dysplasia (BPD) (5). Pulmonary recovery following this injury takes weeks to months to occur. The impact of ECMO treatment is dependent upon the ability of the patient to resolve the underlying pulmonary disease in a relatively short time. Even a lengthy ECMO course would not allow sufficient time to permit a recovery following significant pulmonary injury due to barotrauma or oxygen toxicity, while the risks associated with ECMO (thrombo-embolic phenomenon, infection, mechanical failures) would increase with the duration of the procedure.

In a retrospective case-control study Kornhauser et al. (6) reviewed the records of ECMO patients treated over a 66 month period. Patients with known pulmonary hypoplasia were excluded, as were non-survivors. Patients with oxygen dependency at 1 month of age and x-ray findings consistent with BPD were compared to patients without these findings. Patients with BPD were placed on ECMO at an older mean age than non-BPD patients (135 versus 50 hours), and had longer mean courses of ECMO (203 versus 122 hours). Patients with RDS were at particular risk for the development of BPD. The authors suggest that the risk of BPD from high levels of ventilatory support may occur with as little as 4 days assisted ventilation.

No Lethal Congenital Anomalies. Every effort should be made to clearly establish the patient's diagnosis before the initiation of ECMO. If that is not possible, the diagnosis should be pursued during the course of ECMO. ECMO is not intended to be used to delay an inevitable death. However, non-survivable conditions sometimes present themselves in the same manner as reversible processes (congenital alveolar proteinosis, alveolar capillary misalignment, pulmonary hypoplasia, etcetera). Additionally, certain diagnoses that are potentially treatable such as total anomalous pulmonary venous return (TAPVR) may initially present with respiratory failure and be unrecognized. When giving consent for ECMO, the family must understand that not all patients are able to reverse their disease processes in a reasonable amount of time and that although we assume that the patient has a reversible disease process, that may not be the case.

No Major Uncorrectable Cardiac Lesion. Prior to the initiation of ECMO all patients should undergo a complete cardiologic evaluation, including cardiac ultrasound. TAPVR frequently presents with respiratory failure and/or persistent pulmonary hypertension of the newborn and may go unrecognized unless a detailed evaluation is performed. Additionally, cardiac function or carotid branching pattern off of the aorta may influence the approach to cannulation (V-A versus V-V). Coarctation of the aorta is a contraindication to V-A bypass. Patients with severe lung disease complicating congenital heart disease may be offered ECMO before the repair of their cardiac lesion assuming that the circulatory physiology is appropriate. Decisions regarding appropriate blood gases, timing and logistics of cardiac repair and approach to weaning from bypass should be discussed before ECMO is initiated in these special cases.

We have found several patients with severe hypoxic or diabetic cardiomyopathy who present with shock and poor cardiac output, are unresponsive to volume and inotropic support, but have normal pulmonary compliance and oxygenation. Use of ECMO in these patients has proven to be very successful.

Failure of Optimal Medical Management. One of the most difficult selection criteria to define, or agree upon from center to center, is failure of "optimal medical management". Optimal, " the most favorable condition" (7) is in contrast to maximal, which is "being the greatest or highest possible". Unlike the ECMO qualifying criteria, it is impossible to define optimal management in strictly numerical terms. Each neonatal center has a subtly different approach to the sick neonate. Each center also has specific areas of management strengths and weaknesses. For that reason, each ECMO center has strived to develop its own selection criteria and its own concept of optimal management.

Current optimal medical (non-ECMO) management may include pharmacologic support with vasodilator or vasoconstrictive agents, inotropic agents, sedatives, analgesic and paralytic agents. Ventilatory support usually begins with conventional physiologic strategies, but may change to include exogenous surfactant administration, achievement of respiratory alkalosis, hyperoxia, high PEEP, inverse I:E ratios, or high frequency ventilation. Early administration of steroids in the course of RDS has been proposed to decrease pulmonary inflammation (8), and achievement of metabolic alkalosis may be preferred to respiratory alkalosis.

Many reports of "successful management without ECMO" of patients who qualified for ECMO have been published. These studies of alternative management (high frequency ventilation, permissive hypercarpnea with normoxia, nitric oxide) lead us to question if our criteria are stringent enough for current management techniques. Others are looking at their current approach to management and questioning if their concept of maximum ventilator management may be too strict.

In 1985, Wung et al. (9) presented a non-traditional approach to the management of patients with persistent pulmonary hypertension. Hyperventilation and hyperoxia were not encouraged, and muscle relaxants were not used. Permissive hypercapnea was allowed with peak inflation pressures determined by clinical assessment of chest excursion. The authors presented 15 patients who met institutional ECMO criteria and survived with this approach to care. We and others have had successful experiences with this approach. The authors remind us that the primary goals of ventilatory management are tissue perfusion, oxygenation, and reduction of barotrauma.


Ventilator Settings and Arterial Blood Gases in Neonatal Patients
Immediately Prior to the Initiation of ECMO












PIP (cm H2O)




PEEP (cm H2O)




MAP (cm H2O)








PaCO2 (torr)




PaO2 (torr)




* p<0.05 survivors versus non-survivors

High frequency oscillatory ventilation (HFOV) has recently become a standard component of optimal management which may significantly improve respiratory status in certain disease processes and decrease the need for ECMO (10). Several retrospective reports indicate that many infants who met ECMO qualifying criteria when supported with "maximal" conventional mechanical ventilation (CMV), no longer met ECMO criteria after being placed on HFOV or jet ventilation, and survived without ECMO (11,12,13,14,15). Clark et al. (16) also explored the effectiveness of HFOV in a prospective trial of CMV versus HFOV for "near-ECMO" candidates. In this multicenter trial, HFOV was successful in 51% of the infants who met traditional ECMO criteria. Predicting which infant is more likely to fail HFOV and need ECMO is somewhat dependent on diagnosis. A large retrospective report from deLemos et al. (15) indicated that RDS and pneumonia respond well to HFOV (80%), while MAS is successfully treated in less than half of cases (35%). Congenital diaphragmatic hernia and pulmonary hypoplasia are rarely responsive to HFOV (25%). HFOV use in neonatal patients is currently common enough that it should be considered as part of standard medical care. Therefore, the benefits of HFOV on ECMO eligible infants should promote earlier transfer of sick neonates to an institution that can provide both HFOV and ECMO.

It is important for each center to know not only what optimal medical therapy is, but also how long to employ it before declaring a failure. Some investigators are looking at reducing morbidity for those infants who ultimately require ECMO by examining the timing of the decision to cannulate. Kanto (17) supervised a review of patients referred for ECMO with meconium aspiration syndrome. They noted that of the infants who had a peak inflation pressure > 35 cm H2O, 45 of 58 (78%) eventually required ECMO. They concluded that manipulating the peak inflation pressure higher than 35 cm H2O may simply be an exercise that postpones the eventual need for ECMO, while contributing to the morbidity and mortality in the ECMO population. They recommended that an upper limit be placed on the use of peak inflation pressure in the conventional ventilation of infants with meconium aspiration syndrome.

Using a logistic regression model, Kornhauser et al. (6) found that the risk for developing BPD was increased 11.4-fold if ECMO was initiated after 96 hours of age. They concluded that in patients with continuing high levels of ventilatory support who are not meeting traditional criteria for acute respiratory failure, ECMO intervention should be considered if no improvement has occurred by 96 hours of life.

Also complicating the management of many infants is the potential need to transfer the patient to an ECMO center. Boedy et al. (18) found a hidden mortality associated with the ECMO eligible population specifically related to the timing of transfer. When all patients referred for ECMO consideration were reviewed, a mortality of 27.5% was observed (including deaths before and during transport). They recommended that patients be transferred to the ECMO center before beginning to meet ECMO qualifying criteria.

The most recent ELSO National Registry information on patient condition and ventilator management is presented in Table II. There are many ECMO centers (including our own) that question the diminishing return of increasing the peak inflation pressure in any patient beyond 40 cm H2O. Qualifying for ECMO from high frequency ventilation also varies from center to center. However, most authors report that if there is no dramatic improvement noted in 1-6 hours (13) as evidenced by the ability to reduce the FiO2 and mean airway pressure, then the patient has failed to respond and will ultimately die unless placed on ECMO.

In this current era of multiple management strategies, including surfactant replacement, high frequency ventilation, nitric oxide and potentially, liquid ventilation, the patient may be nearly dead when other management is exhausted and the decision to initiate ECMO is made. This approach could significantly impact the ultimate outcome of the patient by subjecting him to increased morbidity and mortality risks.

Some centers, including our own, advocate early recognition of the potential ECMO patient, and early aggressive medical management. We limit the use of peak inflation pressures beyond that which produces adequate chest excursion and breath sounds, confirmed when feasible by pulmonary function testing. We rarely increase the ventilatory rate beyond 60 breaths per minute and synchronize the timing of breaths with the patient. If these strategies are ineffective in improving ventilation, oxygenation, and perfusion, then we attempt high frequency ventilation. Pharmacologic management includes early use of surfactant, correction of metabolic acidosis, but rarely induced alkalosis, inotropic support and volume recussitation to promote normotension, and narcotic sedation (preferably with morphine). Several of us have had considerable success with the early use of tolazoline by continuous infusion (0.5-2 mg/Kg/hr). The goals of this management strategy are to provide good tissue perfusion and reasonable oxygenation. If this is unsuccessful in a short period of time, the patient is considered for ECMO.

ECMO Qualifying Criteria (Table III)

The first approach to objective recognition of overwhelming neonatal respiratory failure was described by Bartlett et al. (19,20). The newborn pulmonary insufficiency index (NPII) identified infants in the first 24 hours of life with 100% risk of death. The method of plotting serial values of inspired oxygen (FiO2) and arterial pH against time on a graph was further studied by Cimma et al. (21). They evaluated the NPII prospectively and retrospectively and found a score of 40 units to be predictive of ultimate outcome in 95% of the cases with 0.71 sensitivity (predicts non-survivors) and 0.988 specificity (excludes survivors).

In 1984 Krummel et al. (22) compared the use of the NPII with serial alveolar-arterial oxygenation gradient (A-aDO2) measurements. The authors retrospectively studied 50 consecutive severely hypoxic newborns to assess the predictive value of the two methods. The study population included patients with meconium aspiration, transient tachypnea of the newborn, congenital diaphragmatic hernia, neonatal asphyxia, pneumonia and primary pulmonary hypertension of the newborn. The study time encompassed the advent of ECMO availability in the nursery. Of the 50 patients studied, there were 14 survivors of the 38 conventionally treated infants, and 7 survivors of the 12 ECMO treated infants. The authors found that due to the medical manipulation of the pH through hyperventilation to a PaCO2 < 30 mm Hg, the NPII at either 12 or 24 hours was not predictive of mortality in their patients. However, an A-aDO2 > 600 for 12 hours was found to predict mortality with a sensitivity of 0.888 and a specificity of 0.933. The authors concluded that the NPII was not well suited to prediction of outcome in patients with induced alkalosis because the area encompassed between the simultaneous plots of pH and FiO2 became small. In their patients acidosis developed only as a pre-terminal event.

TABLE III Neonatal ECMO Patient Qualifying Criteria

AaDO2 = [atmospheric pressure -47 -(PaCO2 + PaO2)]/FiO2 (47 is the partial pressure of water vapor) 600-624 torr for 4 to 12 hours at sea level Oxygenation Index (OI) = [MAP x FiO2 x 100]/PaO2 (MAP is mean airway pressure) 25-60 for 1/2 to 6 hours PaO2 35-50mmHg for 2-12 hours Acute Deterioration PaO2 < 30-40 torr pH < 7.25 for 2 hours intractable hypotension

Beck et al. (23) emphasized the need for a consistent approach to the medical management in patients with persistent pulmonary hypertension of the newborn. They felt that only after "standard maximal medical management" has been defined, and applied consistently, can a discussion of ECMO entry criteria be undertaken. For their population, an A-aDO2 of 610 torr for 8 hours gave a sensitivity of 0.938 and a specificity of 0.714. Their standard maximal medical management included hyperventilation to achieve a PaCO2 <25 mmHg and a pH > 7.5 with FiO2 of 1.0, paralysis, hyperventilation, tolazoline, inotropic support, and sodium bicarbonate infusion to achieve pH >7.55. Patients with congenital diaphragmatic hernia were considered only after repair of the defect.

Marsh et al. (24) undertook a similar review of patients, comparing the PaO2 with the A-aDO2. The purpose of their study was to determine the institutional mortality prediction criteria and then maximize and compare the criteria with regard to specificity (exclusion of the patient who would survive without ECMO). Previous studies had focused on identification of neonatal mortality (sensitivity) (23). The authors concluded that for their population, a PaO2 < 50 mmHg for four hours resulted in the best sensitivity/specificity combination (0.86/ 0.96). They emphasized several important considerations in their discussion of the difficulties encountered in determining appropriate criteria. The first is that there are patients who are not predictable, most notably, the patients with sepsis. Secondly, the consistency of A-aDO2 as a mortality predictor is dependent on consistency in PaCO2 management. Patients who are hyperventilated, will meet A-aDO2 criteria with a higher PaO2, than non-hyperventilated patients. Conversely, patients with high PaCO2 (such as with congenital diaphragmatic hernia) may not qualify, even with a very low PaO2. Finally, using PaO2 as an ECMO qualifying criteria permits fewer potential calculation errors, and allows centers at varying altitudes to use the same criteria.

Most recently, the oxygenation index (OI) has been applied in the ECMO population as a means of predicting mortality. The OI has the advantage of being simple to calculate, is not sensitive to ventilatory approach (pH or PaCO2), and includes a measure of maximal ventilator settings in the mean airway pressure (25). The authors describe a predicted mortality risk of 80-90% in patients with an OI of > 40. Patients with OI 25-40 are predicted to have 50-80% mortality. The authors recommended that each neonatal ECMO center derive its own predictors of mortality through review of previous experience.

Schumacher et al. (26) hypothesized that initiation of ECMO earlier in the acute disease process would reduce hospital days and cost. They also measured pulmonary and neurologic outcome at discharge and at 1 year of age. Forty-one patients met study criteria (OI=25-40; 50% mortality). Twenty two patients randomized to early ECMO with twenty survivors. Of the remaining nineteen patients, five improved without ECMO, one died due to neurologic abnormalities after randomization, and thirteen met late ECMO criteria (OI > 40) with 12 survivors. Although there was a trend toward lower mean hospital costs ($49,500 versus $53,700) and fewer mean intensive care days (14 versus 19) in patients where ECMO was initiated early, statistical significance when compared with late ECMO was not found, due to wide variability. Patients who were randomized to late ECMO but never met criteria had the greatest mean hospital charges ($63,800) and longest mean intensive care stay (21 days). At age one year, patients treated with early ECMO trended toward higher mean scores on the Bayley Scales of Infant Development MDI (115 versus 103) when compared with late ECMO treated patients. Patients randomized to late ECMO who never met criteria had mean one year MDI scores of 86+25. Three of the five patients in this group had neurologic sequelae. Four patients had pulmonary sequelae. The study lacked sufficient power to conclude that early ECMO was of benefit. However, the authors cautioned against waiting too long to initiate ECMO, as increased morbidity and mortality could result.

Retrospective review of potential ECMO patient populations has been criticized for inconsistent approaches to respiratory management when compared with currently managed populations. The only prospective randomized application of retrospectively acquired ECMO criteria to a single patient population was published by O'Rourke et al. (27). Retrospective analysis predicted 85% mortality using a ratio of the PaO2 to the PAO2 (where PAO2 is atmospheric pressure - water vapor pressure - alveolar partial pressure of CO2) of < 0.15 at 12 to 72 hours of life. Prospective application of these criteria resulted in only a 40% mortality. The authors, as with previous studies (28) expressed concern with the prospect of randomly assigning infants to two therapies with potentially different survival rates. Therefore, they applied an adaptive study design with both a random and non-random phase. The randomized portion of the study was halted after a significant difference in survival was reached.

Another criticism of the current approach to ECMO qualification is the lack of disease specific criteria (29). The uniform application of ECMO criteria to multiple disease processes assumes that the pathophysiology and clinical course of these diseases are similar. The "unpredictability" of pulmonary hypertension when coupled with congenital diaphragmatic hernia or sepsis has long been recognized, yet specific management strategies and ECMO qualifying criteria in these patient populations have not been developed.

Even as the ECMO qualifying criteria are criticized as not being sufficiently specific or stringent, many ECMO clinicians are questioning the rational of using high mortality criteria rather than high morbidity risk. Schumacher et al. (26) attempted to show that delaying ECMO until the patient was moribund would result in increased morbidity and hospital costs. Although his results failed to reach significance, the finding of lower MDI scores and 80% ongoing morbidity at 1 year of age in control patients not qualifying for ECMO is quite chilling. These patients had the highest utilization of resources in the study. He recommended considering the use of ECMO when continuing current management poses a significant risk for morbidity, and definitely using it in situations of high mortality risk. These conditions corresponded to an OI of 25-40. Kornhauser et al. (6) raised similar questions regarding the delayed initiation of ECMO. They found the duration of mechanical ventilation prior to ECMO correlated with the development BPD when it extended beyond 96 hours. Patients placed on ECMO after 96 hours of age had significantly longer ECMO courses (203 + 73 versus 122 + 51 hours). This implies increased costs in these patients, as well as increased late morbidity related to BPD. They recommend considering earlier initiation of ECMO in patients with continuing high levels of ventilatory support, even if traditional qualifying criteria are not met.

These recommendations were further supported when Kornhauser et al. (30) followed their population of ECMO patients with BPD for at least four years. When compared to ECMO patients of the same age without BPD, the BPD patients were found to have significantly lower scores on neurodevelopmental testing. Sixty-five percent of the BPD patients had mild neurologic disabilities and 18% had severe disabilities. Thirty-four percent of the non-BPD patients had mild disabilities and only 4% had severe neurologic disabilities. These significant findings further uphold the growing concern for increased morbidity associated with late initiation of ECMO.

In a separate study, Katikaneni et al. (31) followed their neonatal ECMO population for at least 2.5 years. They evaluated the degree of neurodevelopmental disability and compared their findings to a population of patients who were very ill but did not qualify for ECMO. There were more ECMO treated patients with severe neurodevelopmental disability (36%) compared to non-ECMO patients(19%) examined at greater than two years of age. They found that the ECMO treated patients with severe disability were placed on ECMO later that the non-disabled ECMO patients(168 hrs versus 46 hrs). The authors suggest that delaying the initiation of ECMO may increase the risk of adverse neurodevelopmental outcome. There is an increasing body of information supporting the decision to initiate ECMO earlier, before permanent morbidity results.

Neonatal ECLS Flow Chart

Assessment Parameters Assisted Ventilation Manipulations

Blood Gas Goals:

pH 7.35 - 7.45 PaCO2 35 - 45 PaO2 50 - 80
Pulmonary Perfusion Manipulations

Pulmonary Perfusion Goals:

Assessed by improved PaO2 Lower pulmonary artery pressure assessed by Cardiac U/S

Systemic Perfusion Manipulations

Systemic Perfusion Goals:

Mean arterial >50th percentile SvO2 > 70% Urine output > 1 cc/kg/hr Capi1lary refill Lactic acid < 2 mmoles/ml


Ventilation OI < 25 PaO2 > 60 MAP < 15 Perfusion BP<50th%tile urine output <2cc/kg/hr No airleak Conventional vent Optimize PIP/PEEP Consider surfactant Airleak Consider HFV Consider surfactant If diagnosis is PPHN Consider tolazoline 0.5-2 mg/kg/min Consider Morphine as sedative Volume support Inotropic support Consider Head U/S Consider Heart U/S Consider transfer to ECMO Center
Ventilation OI 25 - 40 MAP 15 - 20 Perfusion BP<25th%tile urine output <1.5cc/kg/hr ABG Q 30 - 90/ min Optimize PIP/PEEP Consider surfactant Consider SIMV Sedation Trial of HFV Consider ECMO Consider Tolazoline NO Optimize heart function Consider vasodilation (if lactic acidosis with good BP) If unable to achieve this, go to ECMO CXR Head U/S Heart U/S Consider PFTs to optimize assisted ventilation Transfer to ECMOCenter
Ventilation OI > 40 PaO2 < 50 x 4 hrs Perfusion BP<5th%tile urine output <1cc/kg/hr If no improvement in 4 - 6 hrs, place patient on ECMO If no improvement in 4 - 6 hrs, place patient on ECMO If no immediate improvement, place patient on ECMO Review transfer information Repeat if needed


BP = blood pressure CXR = chest x-ray HFV = high frequency ventilation MAP = mean airway pressure NO = nitric oxide OI = oxygenation index PFTs = pulmonary function tests PPHN = persistent pulmonary hypertension of the newborn SIMV = synchronized intermittent mandatory ventilation U/S = ultrasound

Escalation of ventilator settings, simply to attain an arbitrary "maximal" value, may delay prompt initiation of ECMO and may contribute to morbidity and mortality (17). In many ECMO centers, including our own, this practice has been abandoned. The focus of patient management and qualifying for ECMO should include a thoughtful and prompt attempt to optimize tissue perfusion and oxygenation, minimizing the risks of barotrauma, while optimizing pulmonary management. If this cannot be accomplished the patient should be supported with ECMO. At our institution the most commonly used qualifying criteria for ECMO is PaO2 < 50 for 4 hours. A recent review of our patient population showed that only approximately 20% of our patients had an OI > 40. The remainder had OI values of 25-40.

Application of "ECMO qualifying criteria" to newer approaches in ventilatory management such as permissive hypercapnea, and high frequency ventilation, without the benefit of a control group, leaves many of us in a quandary. How does one objectively predict mortality or morbidity accurately, yet with enough time to permit a good quality "rescue", when the approach to care is changing, the patient population is changing and the medical equipment and safety features are changing. Our understanding of the disease processes and our ability to manipulate the patient physiology is rapidly changing as well. Through all this, we must maintain a commitment to a critically objective approach to identification and selection of neonatal ECMO patients. This can only be achieved through pooled, rather than individualized experiences, and respect for each other's skill and knowledge.


(1) Cilley RE, Zwischenberger JB, Andrews AF, Bowerman RA, Roloff DW, Bartlett RH. Intracranial hemorrhage during extracorporeal membrane oxygenation in neonates. Pediatrics1986;78(4):699-704.

(2) Bui KC, LaClair P, Vanderkerhove J, Bartlett RH. ECMO in premature infants: Review of factors associated with mortality. ASAIO Trans 1991;37:54-59.

(3) Sell LL, Cullen ML, Whittlesey GC, Yedlin ST, Philippart AI, Bedard MP, Klein MD. Hemorrhagic complications during extracorporeal membrane oxygenation: Prevention and treatment. J Pediatr Surg 1986;21(12):1087-1091.

(4) Allmen D, Babcock D, Matsumoto J, Flake A, Warner BW, Stevenson RJ, Ryckman FC. The predictive value of head ultrasound in the ECMO candidate. J Pediatr Surg 1992;27(1):36-39.

(5) Northway WH, Rosan RC, Porter DY. Pulmonary disease following respirator therapy of hyaline-membrane disease. N Engl J Med 1967;276(7):357-368.

(6) Kornhauser MS, Cullen JA, Baumgart S, McKee LJ, Gross GW, Spitzer AR. Risk factors for bronchopulmonary dysplasia after extracorporeal membrane oxygenation. Arch Pediatr Adolesc Med 1994;148:820-825.

(7) The American Heritage Dictionary, 2nd College Edition, 1985, Houghton Mifflin Co., Boston.

(8) Sanders RJ, Cox C, Phelps DL, Sinkin RA. Two doses of early intravenous dexamethasone for the prevention of bronchopulmonary dysplasia in babies with respiratory distress syndrome. Pediatr Res 1994;36(1):122-128.

(9) Wung JT, James LS, Kilchevsky E, James E. Management of infants with severe respiratory failure and persistence of the fetal circulation, without hyperventilation. Pediatrics 1985;76(4):488-494.

(10) Clark RH. High-frequency ventilation. J Ped 1994;124(5,1):661-670.

(11) Kohelet D, Perlman M, Kirpalani H, Hanna G, Koren G. High-frequency oscillation in the rescue of infants with persistent pulmonary hypertension. Crit Care Med 1988;16:510-516.

(12) Carlo W, Beoglos A, Chatburn RL, Walsh M, Martin R. High-frequency jet ventilation in neonatal pulmonary hypertension. Am J Dis Child 1989;143:233-238.

(13) Baumgart S, Hirschl RB, Butler SZ, Coburn CE, Spitzer AR. Diagnosis-related criteria in the consideration of extracorporeal membrane oxygenation in neonates previously treated with high-frequency jet ventilation. Pediatrics 1992;89:491-494.

(14) Varnholt V, Lasch P, Suske G, Kachel W, Brands W. High-frequency oscillatory ventilation and extracorporeal membrane oxygenation in severe persistent pulmonary hypertension of the newborn. Eur J Pediatr 1992;151:769-774.

(15) deLemos R, Yoder B, McCurnin D, Kinsella J, Clark R, Null D. The use of high-frequency oscillatory ventilation (HFOV) and extracorporeal membrane oxygenation (ECMO) in the management of the term/near term infant with respiratory failure. Early Hum Develop 1992;29:299-303.

(16) Clark RH, Yoder BA, Sell MS. Prospective, randomized comparison of high-frequency oscillation and conventional ventilation in candidates for extracorporeal membrane oxygenation. J Peds 1994;124(3):447-454.

(17) Edwards G, Karp W, Davis H, Boedy R, Hatley R, Howell C, Kanto W. Defining the point of diminishing return in the use of mechanical ventilation in the treatment of meconium aspiration. ELSO IV, 1992; page 10.

(18) Boedy RF, Howell CG, Kanto WP. Hidden mortality rate associated with extracorporeal membrane oxygenation. J Ped 1990;117:462-464.

(19) Bartlett RH, Gazzaniga AB, Fong SW, Jefferies MR, Roohk HV, Haiduc N. Extracorporeal membrane oxygenator support for cardiopulmonary failure. J Thorac Cardiovas Surg 1977;73(3):375-386.

(20) Bartlett RH, Gazzaniga AB, Huxtable RF, Schippers HC, O'Connor MJ, Jefferies MR. Extracorporeal circulation (ECMO) in neonatal respiratory failure. J Thorac Cardiovasc Surg 1977;74(6):826-833.

(21) Cimma R, Risemberg H, White JJ. A simple objective system for early recognition of overwhelming neonatal respiratory distress. J Pediatr Surg 1980;15(4):581-585.

(22) Krummel TM, Greenfield LJ, Kirkpatrick BV, Mueller DG, Kerkering KW, Ormazabal M, Napolitano A, Salzberg AM. Alveolar-arterial oxygen gradients versus the neonatal pulmonary insufficiency index for prediction of mortality in ECMO candidates. J Pediatr Surg 1984;19(4):380-384.

(23) Beck R, Anderson KD, Pearson GD, Cronin J, Miller MK, Short BL. Criteria for extracorporeal membrane oxygenation in a population of infants with persistent pulmonary hypertension of the newborn. J Pediatr Surg 1986;21(4):297-302.

(24) Marsh TD, Wilkerson SA, Cook LN. Extracorporeal membrane oxygenation selection criteria: Partial pressure of arterial oxygen versus alveolar-arterial oxygen gradient. Pediatrics 1988;82(2):162-166.

(25) Ortiz RM, Cilley RE, Bartlett RH. Extracorporeal membrane oxygenation in pediatric respiratory failure. Pediatr Clin N Am 1987;34(1):39-46.

(26) Schumacher RE, Roloff DW, Chapman R, Snedecor S, Bartlett RH. Extracorporeal membrane oxygenation in term newborns. A prospective cost-benefit analysis. ASAIO J 1993;39:873-879.

(27) O'Rouke PP, Crone RK, Vacanti JP, Ware JH, Lillehei CW, Parad RB, Epstein MF. Extracorporeal membrane oxygenation and conventional medical therapy in neonates with persistent pulmonary hypertension of the newborn: A prospective randomized study. Pediatrics 1989;84(6):957-963.

(28) Bartlett RH, Roloff DW, Cornell RG, Andrews AF, Dillon PW, Zwischenberger JB. Extracorporeal circulation in neonatal respiratory failure: A prospective randomized study. Pediatrics 1985;76(4):479-487.

back to top