Further discussion in ECMO Pediatric

The survival of ECLS-treated children with ARDS is over 60%, so ECLS should be considered in any child with acute, severe, reversible respiratory failure. The problem is defining severe (very high mortality risk) and reversible. This uncertainty was reflected in the report of the 1990 NIH workshop1; which recommended that pediatric ECMO should be done only by experienced teams with careful data collection. Selection of pediatric ECMO patients is complicated by continuously changing conventional treatment and results 2,3. Although the Morn's study4 was conducted in adults, the 42% survival with conventional treatment may apply to children as well. Therefore, this discussion must begin with an analysis of the methods and results of conventional management of pediatric ARDS.

Pre-ECLS Respiratory Support

Developing selection criteria for ECLS of patients with ARDS is further complicated because no mode of mechanical ventilator support has proved to improve survival when tested with a prospective randomized trial 5. As with many therapies, initial enthusiasm from anecdotal experiences has withstood neither broad application nor randomized clinical trials. Anecdotal experiences for extraordinarily high PEEP (up to 40 or 45 cm H2O) waxed and waned 6, 7. Initial experience with high-frequency ventilation, even in children 8, also failed to decrease mortality in ARDS when subjected to trials 9, 10, although a recent pediatric experience was optimistic that intermediate outcome parameters are improved 11. So called "permissive hypercapnea" has been championed by Hickling 12-14 and is becoming more widely used before ECLS for ARDS is considered.

Permissive hypercapnea has been excellently reviewed 15, 16. In brief, paramount is limitation of inflation pressures below 40 cm H2O (and preferably 35 cm H20). Minute ventilation is tolerated so that arterial CO2 is controlled only to maintain the pH above 7.2 16. Oxygenation is maintained by the titration of oxygen, PEEP, and mean airway pressure. Some centers aggressively attempt to minimize the effects of pulmonary edema by achieving dry weight 17. An appreciation of these concepts is needed to understand current selection criteria for ECLS, as patients must be rigorously managed in a prospective manner before ECLS is instituted. For example, the University of Michigan approach to respiratory failure is summarized in Figure 1 18. Without such a consensus (plus rigorous data collection and analysis) on pre-ECLS care, the effect of introduction of ECLS technology will be difficult to assess. The data may still not be comparable to another institution's nor to national trial data. Without appropriate documentation of pre-ECLS treatment outcome, justification of the cost and effort will be difficult if not impossible.

Michigan Pediatric Respiratory Failure Algorithm


Ventilator Treatment Oxygenation

O2 Delivery

Systemic Treatment Fluids

TV 5cc/Kg

TV, rate to PaCO2 40

Limit: PIP 40

FIO2. 0.5

PEEP 5 Titrate O2 and PEEP to SvO2 > 70

Limit : FiO2.6

SaO2 > 90%

Hct > 40

increase CO to SvO2 > 70


> Dry weight Diurese Nutrition

decrease CO

(If no improvement)

decrease VCO2 Treat infection Sedation Paralysis Lipid feed Tolerate hypercardia ECMO decr easeVO 2 Treat infection Sedation Paralysis reduce Catechol drugs PCIRV Prone Position Tolerated hyplexia ECMO

Figure 1: Approach to pediatric respiratory failure at the University of Michigan

Patient Selection for ECLS Management

There are no broadly accepted indications for ECLS therapy for ARDS in children. Thus, indications remain institution-specific and are based on historical data. However, substantial efforts to analyze existing anecdotal and multi-institutional studies have been produced and allow some refinement and insight.

Suchyta et al, reported higher than predicted survival of conventionally managed adult ARDS patients with severe hypoxemia using the 1979 NIH ECMO Trial Criteria 2, 19. The authors prospectively evaluated morbidity in ARDS patients meeting the NIH Trial entry criteria. 178 ARDS patients were prospectively studied. Fifty-one of those patients met the older NIH ECMO trial criteria. A 45% survival rate was observed with conventional ICU management. In 1975-1979, patients with severe hypoxemia who met the same criteria had a survival of 11%. There were no obvious differences in etiology, Apache II scores, organ system failure, or the incidence of sepsis between the survivors and non-survivors. The authors have suggested that conventional ventilator management done in a protocolized manner may result in a much higher survival rate than that observed in 1979.

In designing prospective trials to assess the effectiveness of ECMO as it influences survival in ARDS, these authors suggest that a survival rate on conventional management in 1991 may be closer to 45% than the previously reported 15%. The authors cite three explanations for the increased survival in their institution: 1) patient selection, 2) changes in medical technology in patient management, and 3) hospital-specific reasons. In this study, the patients were older than in the previous studies reported by Zapol et al in the 1979 trial, and in the extracorporeal CO2 removal trial of Gattinoni, et al 20. However, there was a common predominant risk factor for ARDS, i.e. pneumonia, in all of the studies. The authors speculated that advances in medical technology, including antimicrobial agents, cardiac assist devices, organ transplantation, and nutritional support all contribute to an increased survival in the modern era. They point out that evolution in the use of such modalities as steroids and changes in fluid management are different now than 10 or 15 years ago. They suggest that permissive hypoventilation allowing hypercapnia by the use of smaller tidal volumes may lead to less diffusely damaged lungs in ARDS patients. Could the observations of this study have been hospital-specific? The authors' institution was involved at this time in a randomized trial aimed at the 'Development of Detailed Computerized Protocols for the Management of Oxygenation in ARDS Patients'. The authors speculate that interest and commitment to the ARDS patient may have increased significantly and, therefore, effected increased survival.

This study has been used by critics of ECLS to support the argument that ECLS technology is not needed. However, one must be careful to draw sweeping conclusions from single institution data. Knaus has pointed out that definitions based on degrees or measures of hypoxemia include a broad array of individual patient risk. 21 They report a cohort study of 423 adult ICU admissions with a PaO2/FiO2 of < 300 mm Hg and a diagnosis of ARDS. A variety of nonpulmponary risk factors including bilirubin levels and previous hospital stay significantly predicted mortality. However, indices of hyopoxemia alone were not predictive. When a PaO2/FiO2 of < 150 mm Hg was used to stratify patients, mortality rates varied from < 10% to > 90%. Thus other indicators such as Apache III must be included to accurately produce a group of homogeneous patients of similar high risk of mortality. Small studies using only pulmonary criteria may thus include patients at the extremes of high and low risk of death. Unless possible confounding variables are accounted for , strategies designed to prove superiority of a modality may in fact be comparing groups of low or high mortality, thus inappropriately biasing the results and conclusions.

Three papers have studied the prediction of mortality in pediatric respiratory failure. Rivera reported predictors of mortality in children with respiratory failure aimed at developing objective predictors of death with conventional management 22. To determine the predictors of death, the authors retrospectively evaluated the charts of 42 children, aged one month to 18 years, admitted to the Intensive Care Unit at Royal Children's Hospital in Parkfield, Victoria, Australia. The authors defined ARDS as a syndrome of low lung compliance with hypoxia and diffuse alveolar and interstitial lung disease radiographically. All children were ventilated for more than 12 hours, received greater than 90% oxygen, had a peak inspiratory pressure greater than 25 cm H2O, and had no pre-existing neurodevelopmental handicap. The study excluded children who died from neurologic sequelae after hypoxic-ischemic or septic insult. The authors developed a combination variable reflecting ventilation and another variable reflecting oxygenation which reliably predicted death. A ventilation index (respiratory rate times the PCO2, times the peak inspiratory pressure, divided by 1000) greater than 40 and an oxygenation index (mean airway pressure times the FiO2, divided by the PaO2) greater than 0.4 was associated with a 77% chance of mortality. This produced a sensitivity of 65% and a specificity of 74%. A combination of peak inspiratory pressures greater than 40 cm H2O and an A-aDO2 greater than 580 mm Hg had a specificity of 79%.

Tamburro, from the University of Tennessee, evaluated the use of the alveolar-arterial oxygen gradient as a predictor of outcome in patients with non-neonatal pediatric respiratory failure 23. The authors performed a four-year observational descriptive study of all patients admitted to the LeBonheur Children's Medical Center Pediatric Intensive Care Unit with respiratory failure. During that time, 4800 patients were admitted to the unit. The first two and one-half years of this study were retrospective, while the last year and one-half was prospective. Patients with acute lung disease who required mechanical ventilation with a PaO2 of less than 60 mm Hg, while receiving an FiO2 of greater than 0.5, were eligible. Patients with any pathologic process other than acute lung disease, such as chronic lung disease, congenital heart disease, neuromuscular disease, and severe head injury were excluded. Also excluded were patients with severe hypoxic-ischemic encephalopathy or terminal disease. The study did not include patients less than one month of age. The authors identified 37 patients that met entrance criteria. Five of these were excluded because they died before completion of the 48-hour study period. Of the remaining 32 patients, 18 survived. Thirteen survivors and 10 non-survivors were studied retrospectively; five survivors and four non-survivors were studied prospectively. Of the 14 deaths, 93% (13) resulted from cardiorespiratory failure as a result of lung injury, while one death was caused by a hemothorax. An A-aDO2 of greater than 450 mm Hg for 16 hours predicted death with 86% sensitivity, 100% specificity, and a p-value less than .01, and was associated with a 100% mortality rate.

Timmons reported a retrospective chart review of mortality rates and prognostic variables in children with adult respiratory distress syndrome 24. In this review of cases from 1987 to 1990, the authors attempted to identify physiologic variables predictive of death. They identified 44 children, with a mortality rate of 75%. Significant differences between survivors and non-survivors were found in intrapulmonary venous admixture, mean airway pressure, A-aDO2 gradient, oxygenation index, and peak inspiratory pressure. Shunt fraction greater than 0.5, peak airway pressure greater than 23 cm H2O, and alveolar arterial oxygen difference of more than 470 mm Hg were 93%, 90%, and 81% predictors of death respectively. Sensitivity and specificity were enhanced when multiple predictors were linked. Notably, data included patients with near-drowning, asphyxia, and sudden infant death syndrome. Thus, patients with a high mortality rate from another related disease were included. For example, septic patients were included (mortality rate of 70%), and drowning patients (90% mortality). There were three patients with sudden infant death syndrome who had a 100% mortality. One would assume the mortality rate in these patients would be extraordinarily high with or without respiratory failure.

Moler et al have recently analyzed available data in the ELSO registry in an attempt to discover predictors of survival from hypoxic respiratory failure for children with ECMO therapy. Data from that study are seen in Tables I and II. Stepwise multivariate logistic regression modeling found patient age and days of mechanical ventilation prior to the initiation of ECLS to correlate with survival 25.

The same investigators have suggested that predictors of outcome for a single disease entity can be drawn from existing data, such as in the ELSO Registry. The authors examined all cases of respiratory syncytial virus-induced respiratory failure treated with ECLS. Fifty-three patients reported to the ELSO Registry through June of 1992 were evaluated. The average patient age was five months; duration of mechanical ventilation before ECLS was eight days. Forty-nine percent survived to discharge. Multivariate logistic regression analysis revealed four variables to be predictive of death: male gender, duration of ventilation in days before bypass, peak inspiratory pressure, and lower PaO2/FiO2 ratio. Receiver operator characteristic curve analysis at a cutoff point of r = 0.5 revealed 92% sensitivity and 81% specificity. Thus, appropriate statistical analysis of historical data provide some insight into the development of criteria for ECLS therapy 26.

Table I: Survival Trend by Age Groups in 220 Pediatric Respiratory Failure Patients Receiving ECLS (25)

Age Group



% Survived

1 month to 1 year




1 to 2 years




2 to 8 years




greater than 8 years




Table II: Survival Trend of Duration of Mechanical Ventilation Prior to ECLS in 220 Pediatric Respiratory Failure Patients Receiving ECLS (25)

Mechanical ventilation duration before ECLS


Survival %

0 to 3 days



4 to 6 days



7 to 10 days



greater than 10 days




To further the single institution study of Timmons' 24, the Pediatric Critical Care Study Group sponsored an ambitious effort to define predictors of mortality for children with ARDS. The data collection was driven by the hypothesis that physiologic data could be used to develop a multivariate prediction tool to estimate mortality within 28 days of the diagnosis (with a positive predictive value of 0.8 or greater) 27.

Forty-one institutions screened approximately 8,000 charts for patients with ARDS during calendar 1991. Each institution determined the most effective way to identify patients; most sites used respiratory therapy or ICU logs. Further, CPT codes 31500, 94656, and 94657 and ICD-9 codes 93.92, 96.04, 96.05, 96.70, 96.71, and 96.72 were searched at each site. Patients included in the study were between two weeks and 18 years old, who had 12 continuous hours of EEP of 6 or greater and an FiO2 of 0.5. Excluded were patients with intracardiac shunts or operations on the cardiovascular system during the screened admission. Demographic data was recorded on each patient. Physiologic data (FiO2, respiratory rate, peak inspiratory pressure, PEEP, exhaled tidal volume, mean airway pressure, PaO2, PaCO2, and mechanism of death) obtained closest to 6:00 a.m. and 6:00 p.m. were recorded for the first 14 days after entry criteria were met. In patients continuing on mechanical ventilation at 14 days, an additional worst-value-data point was collected once during week three and again during week four. PRISM scores were calculated within 24 hours of meeting the entry criteria. Patients validly exited that data collection at the time of death, hospital discharge, or on day 28.

Data was collected on 744 patients. Sixty-five patients were excluded (four with a cardiac diagnosis, 11 with ineligible ages, 33 cases did not meet EEP and FiO2 entry criteria, six were not admitted in calendar 1991, ten because they were transferred between data collection sites and were thus counted twice, one patient met criteria twice, but only the first episode was considered) leaving 679 cases for analysis.

The median age was 1.88 years. Three hundred three of the patients were girls. Overall mortality was 52%. Median length of stay was 26 days for survivors, and 13 days for non-survivors. Three hundred twelve of the 352 deaths occurred within the first 28 days.

The population was further restricted to those patients who had not received ECMO, high frequency ventilation, surfactant therapy, or inter-hospital transfer (112 patients). Further, 96 cases were judged unlikely to benefit from heroic measures as evidenced by fixed, dilated pupil, progression to brain death, or withdrawal of life support because of neurologic futility. The remaining 471 patients had an overall mortality of 43%. Because mean airway pressure and exhaled tidal volume data were often missing, these parameters were not utilized.

A Pediatric Respiratory Failure score was generated with the age, operative status, PRISM score 28, FIO2, respiratory rate, peak inspiratory pressure, EEP, PaO2, and PaCO2. 14 separate logistic regression were generated which predict mortality at 12 hour intervals for 7 days. Area under the receiver operating characteristic curve was 0.769 at entry and was greater than 0.8 after 36 hours. When the score was applied to the validation subset of patients, goodness-of-fit chi-square showed no significant difference between estimated and actual mortality between entry and 96 hours. These data provide time sensitive predictors of mortality from pediatric respiratory failure from a multi-institutional data base and thus are the only existing estimate of "state of the art" results using conventional management techniques. It must be remember, however, that therapies for ARDS have changed ("evolved" implies a logical progression) since 1991. At best, it should be expected that mortality be no worse than, that predicted with historical controls.

ECMO in the ELSO-PCCSG Study

A more focused analysis of the same data set was performed by Green, Timmons, Fackler, et al to specifically address the issue of the impact of ECMO on survival in children with ARDS. Excluded from the 679 patients described above were 143 patients who did not have complete data (as above, usually mean airway pressure and exhaled tidal volume). Patients were more broadly excluded for possible ECMO contraindications (including those with ICD-9 codes 286 and 287 of coagulopathy) and, therefore, 255 patients were also excluded. Consequently, 331 patients from 32 hospitals had complete data for further analysis and had no diagnostic exclusions for ECMO. Thirty-eight patients were treated with ECMO.

Three separate analyses were performed on the data from the 331 patients. First, a logistic regression analysis was performed (separately from the analysis described above) to identify factors associated with survival. Second, each of the 38 ECMO-treated patients was matched with up to two non-ECMO-treated patients with the same respiratory diagnosis and with a mortality risk within 15% of each other (with the mortality risk that did not use ECMO use as a regressor). Third, the study population was stratified into risk quartiles, based on this latter logistic regression, and comparisons of ECMO and non-ECMO-treated patients were performed.

Fifteen hospitals performed ECMO on one or more patients. Seventeen hospitals did not perform ECMO on a patient in this data set. The most common diagnoses associated with respiratory failure were bacterial infections, viral infections, trauma (including burns and foreign bodies, and status post surgical procedures). Patients treated with ECMO had higher FIO2 values (p < .01), higher mean airway pressures (p < .02), and lower PaO2 (p < .001) and AaDO2 (p < .01). Twenty-six percent of patients who received ECMO had also received high frequency ventilation, compared with 11% of patients who did not receive ECMO (p < .02).

Termination of the forward logistic regression analysis before inclusion of ECMO yielded the weighted regressor equation:

R = (0.0491 * oxygenation index) + (0.0893 * PRISM -3.3217)

For 24 ECMO-treated patients, both of the desired two non-ECMO-treated case matches were identified. Five ECMO only had one non-ECMO-treated case match identified. The remaining nine ECMO-treated patients had no acceptable case match and were, therefore excluded from further analysis. Overall, 52 pairs of ECMO and non-ECMO-treated patients matched for respiratory diagnosis and mortality risk were identified.

Mortality in the ECMO-treated group was 26.3% compared with 49.1% in the non-ECMO-treated controls (p < .01). When stratified by risk quartile, ECMO-treated patients in the 50 to 75% mortality risk group had a 28.6% mortality compared to 71.4% for the case-matched controls. There were no differences between the ECMO-treated patients and controls in the other three quartiles. However, small sample sizes, particularly in the quartile with greater than 75% predicted mortality, may have limited the power of the comparisons (see Figure 2).

Current ARDS Studies

A randomized clinical trial of ECLS for ARDS is underway and should be complete in 1997 29. All patients aged two weeks to eighteen years are eligible for study. Patients will be enrolled into a rigorous data collection when a patient is receiving 50% oxygen and an end expiratory pressure of 6 cm H2O or greater for more than 12 hours. Patients are eligible for randomization when the predicted mortality reaches 65%. At this point, informed consent is obtained and eligible patients will be randomized to either protocol driven mechanical ventilation (i.e. control) or ECMO (i.e. experimental) therapies. Sixty five percent mortality is chosen because analysis of the retrospective data mentioned above suggested that in the 50 to 75% predicted mortality quartile, ECMO for pediatric ARDS was shown effective . The preliminary data did not suggest an effect of ECMO in the lowest two, or highest, risk quartiles.

The prediction of mortality will be from OI-PRISM score generated from the work of Green et al 30. Specifically, the predicted mortality is 65% when the OI and PRISM97 are paired as below. As the PRISM score rises, the necessary OI to predict 65% morality diminishes. The minimum OI for which randomization is appropriate is 25. Patients who have PRISM scores greater than 30 are eligible for randomization if their OI reaches 25 (as long as with their high admission severity index they have no exclusion criteria).

Because congenital heart disease represents a confounding variable(s) that is beyond the scope of the study patients will be completely excluded they have either undergone cardiac surgery during the hospital admission when criteria are met or if they have congenital heart disease with right-to-left or left-to-right shunting. Further, patients will be excluded from randomization if patients meet the original entry criteria for more than 7 days. Further, situations were ECMO is unlikely to benefit will also be excluded from randomization. These latter conditions are (a) pre-existing lung disease (defined as a chronic oxygen requirement or chronic diuretic therapy); (b) immune compromise (defined as having a previous solid organ transplant, a previous bone marrow transplant, on going cancer chemotherapy, severe combined immune deficiency, or human immunodeficiency virus infection); (c) cardiac disease with left ventricular failure (defined as a PCWP greater than 18 mmHg); (d) profound acute central nervous system damage (defined as a Glasgow coma scale equal to or less than 7, fixed and dilated pupils, or pupil asymmetry of 2mm or greater). Other reasons for exclusion from randomization must be well documented.

However, patients who are excluded randomization are not excluded from the data collection. Hence, the overall study population will contain both randomized and non-randomized groups. To reiterate, the entry criteria were chosen to allow capture of all children with ARDS. Better definition of the ECLS criteria for ARDS in children will then be available.

Table III: Eligibility Criteria for ECMO

Duration of Ventilation: < 2 years <10 days 2 - 8 years < 8 days > 8 years < 6 days Respiratory Failure: PEEP > 8 cm H2O x 12 hours FiO2 > .8 x 12 hours PaO2/FiO2 < 150 P(A-a) O2 > 450 mm Hg Respiratory Acidosis: pH < 7.28 with Peak Inspiratory Pressure 40 cm H2O or Airleak Syndrome Maximal minute ventilation without air trapping Reasonable Medical Certainty of Quality of Life

Table IV: Exclusion Criteria for ECMO

Age Days of Ventilation < 2 years > 10 days 2 - 8 years > 8 days > 8 years > 6 days Major hemorrhage Immunosuppression Cardiac arrest with neurologic impairment Recent cerebral-vascular accident Low certainty of quality of life.


This review helps to define the incidence and nature of acute respiratory failure in pediatric patients, and identifies an overall mortality risk of 40-50%. Patients with high mortality risk (>80%) can be identifiend, and these are the patients for whom ECMO should be considered. The selection criteria currently used at the University of Michigan are shown in Table 3 and 4. Other scoring systems such as OI and PRISM, and new prospective studies of pediatric ARDS will help to define indications more precisely.


1. Wright LL. Report of the workshop on diffusion of ECMO technology. Rockville, MD: NIH Publication No.93-3399, 1993:7-13.

2. Suchyta MR, Clemmer TP, Orme JF, Jr., Morris AH, Elliott CG. Increased survival of ARDS patients with severe hypoxemia (ECMO criteria). Chest 1991; 99:951-955.

3. Milberg JA, Davis DR, Steinberg KP, Hudson LD. Improved survival of patients with acute respiratory distress syndrome(ARDS): 1983-1993. JAMA 1995; 273:306-9.

4. Morris AH, Wallace CJ, Menlove RL, et al. Randomized clinical trial of pressure-controlled inverse ratio ventilation and extracorporeal CO2 removal for adult respiratory distress syndrome. Am J Respir Crit Care Med 1994; 149:295-305.

5. Marini JJ. Ventilation of the acute respiratory distress syndrome: Looking for Mr. Goodmode. Anesthesiology 1994; 80:972-975.

6. Kirby RR, Downs JB, Civetta JM, et al. High level positive end expiratory pressure (PEEP) in acute respiratory insufficiency. Chest 1975; 67:156-163.

7. Gallagher TJ, Civetta JM, Kirby RR. Terminology update: optimal PEEP. Crit.Care Med. 1978; 6:323-326.

8. Arnold JH, Truog RD, Thompson JE, Fackler JC. High-frequency oscillatory ventilation in pediatric respiratory failure. Critical Care Medicine 1993; 21:272-278.

9. Carlon GC, Howland WS, Ray C, Miodownik S, Griffin JP, Groeger JS. High-frequency jet ventilation. A prospective randomized evaluation. Chest 1983; 84:551-559.

10. Hurst JM, Branson RD, Davis K, Jr., Barrette RR, Adams KS. Comparison of conventional mechanical ventilation and high- frequency ventilation. A prospective, randomized trial in patients with respiratory failure. Ann.Surg. 1990; 211:486-491.

11. Arnold JH, Hanson JH, Toro-Figuero LO, GutiƩrrez J, Berens RJ, Anglin DL. Prospective, randomized comparison of high-frequency oscillatory ventilation and conventional mechanical ventilation in pediatric respiratory failure. Crit Care Med 1994; 22:1530-1539.

12. Hickling KG. Ventilatory management of ARDS: can it affect the outcome? Intensive.Care Med. 1990; 16:219-226.

13. Hickling KG, Henderson SJ, Jackson R. Low mortality associated with low volume pressure limited ventilation with permissive hypercapnia in severe adult respiratory distress syndrome. Intensive.Care Med. 1990; 16:372-377.

14. Hickling KG, Walsh J, Henderson S, Jackson R. Low mortality rate in adult respiratory distress syndrome using low-volume, pressure-limited ventilation with permissive hypercapnia: A prospective study. Crit Care Med 1994; 22:1568-1578.

15. Bidani A, Tzouanakis AE, Cardenas VJ, Zwischenberger JB. Permissive hypercapnia in acute respiratory failure. Jama 1994; 272:957-962.

16. Tuxen DV. Permissive hypercapnic ventilation. Am J Respir Crit Care Med 1994; 150:870-874.

17. Simmons RS, Berdine CG, Seidenfeld JJ. Fluid balance and the adult respiratory distress syndrome. Am Rev Resp Dis 1987; 135:924-932.

18. Bartlett RH. Use of the mechanical ventilator. In: Holcroft T, ed. Care of the Surgical Patient. Vol. 1. New York: Scientific American, 1993.

19. Zapol WM, Snider MT, Hill JD, et al. Extracorporeal membrane oxygenation in severe acute respiratory failure. A randomized prospective study. JAMA 1979; 242:2193-2196.

20. Gattinoni L, Pesenti A, Mascheroni D, et al. Low-frequency positive-pressure ventilation with extracorporeal CO2 removal in severe acute respiratory failure. JAMA 1986; 256:881-886.

21. Knaus WA, Sun X, Hakim RB, Wagner DP. Evaluation of definitions for adult respiratory distress syndrome. Am J Respir Crit Care Med 1994; 150:311-7.

22. Rivera RA, Butt W, Shann F. Predictors of mortality in children with respiratory failure: possible indications for ECMO. Anaesth.Intensive.Care 1990; 18:385-389.

23. Tamburro RF, Bugnitz MC, Stidham GL. Alveolar-arterial oxygen gradient as a predictor of outcome in patients with nonneonatal pediatric respiratory failure. J.Pediatr. 1991; 119:935-938.

24. Timmons OD, Dean JM, Vernon DD. Mortality rates and prognostic variables in children with adult respiratory distress syndrome. J.Pediatr. 1991; 119:896-899.

25. Moler FW, Palmisano JM, Custer JR. Extracorporeal life support for pediatric respiratory failure: Predictors of survival from 220 patients. Crit Care Med 1993; 21:1604-1611.

26. Moler FW, Palmisano JM, Green TP, Custer JR. Predictors of outcome of severe respiratory syncytial virus-associated respiratory failure treated with extracorporeal membrane oxygenation. J Pediatr 1993; 123:46-52.

27. Timmons OD, Havens PL, Fackler JC. Predicting death in pediatric patients with acute respiratory failure. Chest In Press.

28. Pollack MM, Ruttimann UE, Getson PR. Pediatric risk of mortality (PRISM) score. Crit Care Med 1988; 16:1110-6.

29. Fackler JC. A randomized clinical trial of ECMO for ARDS, Annual ELSO Symposium, Ann Arbor, MI, 1994.

30. Green TP, Timmons OD, Fackler JC, Moler FW, Thompson AE, Sweeney MF. The impact of extracorporeal membrane oxygenation on survival in pediatric patients with acute respiratory failure. Crit Care Med In Press.

back to top