Diffuse pulmonary opacification in infants undergoing extracorporeal membrane oxygenation: clinical and pathologic correlation

Early continuous renal replacement therapy during infant extracorporeal life support is associated with decreased lung opacification

Lung opacification on chest radiography (CXR) is common during extracorporeal life support (ECLS), often resulting from pulmonary edema or inflammation. Concurrent use of continuous renal replacement therapy (CRRT) during ECLS is associated with improved fluid balance and cytokine filtration; through modification of these pathologic states, CRRT may modulate lung opacification observed on CXRs. We hypothesize that early CRRT use during infant ECLS decreases lung opacification on CXR. We conducted a retrospective cohort study comparing CXRs from infants receiving ECLS and early CRRT (n = 7) to matched infants who received ECLS alone (n = 7). The CXR obtained prior to ECLS, all CXRs obtained within the first 72 h of ECLS, and daily CXRs for the remainder of the ECLS course were analyzed. The outcome measure was the degree of opacification, determined by independent assessment of two, blinded pediatric radiologists using a modified Edwards et al.'s lung opacification scoring system (from Score 0: no opacification to Score 5: complete opacification). 220 CXRs were assessed (cases: 93, controls: 127). Inter-rater reliability was established (Cohen's weighted к = 0.74; p < 0.0001, good agreement). At baseline, the mean opacification score difference between cases and controls was 1 point (cases: 1.8, controls 2.8; p = 0.049). Using mixed modeling analysis for repeated measures accounting for differences at baseline, the average overall opacification score was 1.2 points lower in cases than controls (cases: 2.1, controls: 3.3; p < 0.0001). The overall distribution of scores was lower in cases than controls. Early CRRT utilization during infant ECLS was associated with decreased lung opacification on CXR.

The Role of Imaging During Extracorporeal Membrane Oxygenation in Pediatric Respiratory Failure

OBJECTIVE

Extracorporeal membrane oxygenation (ECMO) is increasingly widely used in pediatric respiratory failure. Despite playing a key part in patient management during ECMO, the role of radiology is not widely reported. We discuss the principles of ECMO support and the normal imaging appearances. Radiologic findings arising from the complications of ECMO are highlighted.

CONCLUSION

Radiology has a central role in establishing well-designed imaging protocols and vigilant reporting of ECMO apparatus positions and complications.

Dexamethasone therapy in neonates treated with extracorporeal membrane oxygenation

OBJECTIVE

To test the hypothesis that infants who received dexamethasone would have a shorter length of time on extracorporeal membrane oxygenation (ECMO). Study design Infants placed on ECMO for respiratory failure were randomly assigned to receive either dexamethasone for 3 days or placebo. Chest radiographs were scored through the use of a validated standard scoring system to assess lung injury.

RESULTS

Thirty infants received dexamethasone and 29 received placebo. The median (25th%, 75th%) duration of time on ECMO was 143.5 (100, 313) hours in the dexamethasone group and 160 (111, 303) hours in the placebo group (not significant). Survival was 80% in the dexamethasone group and 83% in the placebo group. Radiographic lung injury scores (mean+/-SEM) were significantly improved in the dexamethasone group (10.5+/-0.6) versus placebo (12.3+/-0.5) on day 3 of ECMO (P=.013). Hypertension developed in 27 of the 30 infants receiving dexamethasone and 13 of the 29 infants in the placebo group during ECMO (P<.01).

CONCLUSIONS

Dexamethasone given during the first 3 days of ECMO results in significant improvement in lung injury scores by day 3 of ECMO but does not significantly decrease the duration of ECMO or improve survival. The preponderance of evidence would not support the use of dexamethasone in this setting.

Measurement of the ductal L-R shunt during extracorporeal membrane oxygenation in the lamb

OBJECTIVE

In neonates, initially a ductal shunt is often observed during veno-arterial extracorporeal membrane oxygenation (ECMO). Depending on the degree of pulmonary hypertension in these patients, the ductal shunt will be right to left (R-L), left to right (L-R), or bidirectional. A ductal L-R shunt will possibly lead to pulmonary hyperperfusion and interact with ECMO weaning. The aim of this study was to give more insight in this ductal L-R shunt during ECMO by quantification of this shunt in relation to cardiac output and ECMO flow.

METHODS

In 7 lambs, closure of the duct was prevented by infiltration of the ductal wall with 10% formaline. This patent duct could be closed using a vesselloop around the duct. Ultrasound flowprobes were installed around the pulmonary artery, ascending aorta, and around the ECMO circulation tube. Right and left ventricular output and ECMO flow were measured. Ductus flow was defined as ductal left to right shunt (Qduct L-R) = flow in ascending aorta (Qao) - flow in central pulmonary artery (Qpa) and Qduct R-L = Qpa = Qao.

RESULTS

In 6 of 7 lambs a ductal L-R shunt was observed with a mean shunt of 44% (range, 11 to 79) of left ventricular output (Qduct L-R/Qao). Comparison with ECMO flow (Qduct L-R/Qecmo) showed a mean shunt of 76% (range 15 to 230). When compared with the total systemic circulating volume (Qpa + flow in the ECMO circuit [Qecmo]), the mean ductal L-R shunt showed a percentage of 51% (range, 7% to 142%).

CONCLUSIONS

During ECMO, mostly a ductal L-R shunt is observed in this lamb model. This ductal shunt is hemodynamically important. The percentages of this shunt in comparison with left ventricular output, and total circulating volume will support the idea that a ductal L-R shunt during ECMO could be another deteriorating factor in the often critical circulation of the neonate on veno-arterial ECMO.

The influence of ductal left-to-right shunting during extracorporeal membrane oxygenation

BACKGROUND/PURPOSE

The aim of this study was to analyze whether a ductal left-to-right (L-R) shunt will prolong extracorporeal membrane oxygenation (ECMO) in neonates with severe pulmonary hypertension. This report discusses the onset and termination of a ductal L-R shunt and its potential influences on ECMO when pulmonary hypertension decreases during venoarterial bypass.

METHODS

Twenty-nine neonates were monitored during veno-arterial ECMO, using bedside echocardiography with 12-hour interval observations.

RESULTS

Up to 43% of the patients showed this type of shunt already after 12 hours on bypass. In total, this type of ductal shunt was found between 12 and 72 hours on ECMO in 62% of the patients. After 72 hours, the ductal L-R shunt no longer was detected. In 38% of the patients, no ductal L-R shunt was found during ECMO. Comparisons between these 2 patient groups showed a significantly longer ECMO duration in patients with ductal L-R shunt (P <.007). The mean prolongation time was 46 hours. Also, a significant decrease of left atrium to aorta ratio (P <.01) was observed during ECMO in the ductus group after closure of the duct, illustrating the decrease in volume load for the left heart and lungs.

CONCLUSIONS

Ductal L-R shunting is related with a substantial prolongation of the ECMO course (mean prolongation of almost 2 days). The authors suggest that on one side, the ductal L-R shunt will lead to pulmonary hypercirculation and on the other side, postductal stealing from the descending aortic circulation will lead to prerenal failure. Possibly because of interactions with pulmonary and renal function, a ductal L-R shunt will, among other factors, interfere with weaning from ECMO, resulting in a prolonged bypass time.

Extracorporeal membrane oxygenation exposes infants to the plasticizer, di(2-ethylhexyl)phthalate

OBJECTIVES

To determine the exposure to, and evaluate the potential toxicity from, the plasticizer, di(2-ethylhexyl)phthalate (DEHP) during extracorporeal membrane oxygenation (ECMO) therapy.

DESIGN

Protocol 1 consisted of a prospective comparison of three ECMO circuit designs in vitro. Protocol 2 consisted of a prospective, comparative clinical study evaluating DEHP plasma concentrations in ECMO vs. non-ECMO patients with respiratory failure.

SETTING

Neonatal intensive care unit at The Children's National Medical Center, Washington, DC.

PATIENTS

In protocol 2, 28 consecutive term infants were referred for ECMO therapy. Eighteen infants required ECMO; ten control patients received conventional ventilation and improved without ECMO.

INTERVENTIONS

In protocol 1, three ECMO circuit designs were primed in vitro with normal saline, albumin, and human blood, which was maintained at 37 degrees C and recirculated at 400 mL/min for 48 hrs. Plasma samples were obtained at time 0, 1 hr, and every 6 hrs. In protocol 2, ventilatory and cardiovascular management of the patients in the study was conducted by the attending physician. Patients were placed on ECMO when they met the institutional criteria for ECMO therapy. Daily plasma concentrations for DEHP were collected until 3 days after decannulation from bypass in the ECMO group. Control patients were sampled daily until extubation. Evidence of cardiac, liver, or lung toxicity was evaluated by Chest Radiographic Scores, liver function studies, and echocardiograms obtained on day 1, day 3, and the day of decannulation in the ECMO group, or at the time of extubation in the control group. Sedation, blood product transfusions as indicated, antibiotics, and hyperalimentation were administered to all patients.

MEASUREMENTS AND MAIN RESULTS

All DEHP plasma concentrations were measured by gas chromatography. In protocol 1, three circuits were studied: circuit A (small surface area); circuit B (larger surface area); and circuit C (surface area of A but with heparin-bonded tubing in the circuit). DEHP leached from circuit A at 0.32 +/- 0.12 microgram/ mL/hr, compared with 0.57 +/- 0.14 microgram/mL/hr from circuit B (p < .05). This amount of DEHP extrapolates in the ECMO patient to a potential exposure of 20 to 70 times that exposure from other medical devices or procedures, such as transfusions, dialysis, or short-term cardiopulmonary bypass. Circuit C showed almost no leaching from the circuit; DEHP concentrations decreased at a rate of 0.2 +/- 0.04 microgram/mL/ hr. In protocol 2, DEHP was undetected in the control patients. DEHP concentrations in ECMO patients were greater in the early course of ECMO. However, most patients cleared this compound from the plasma before decannulation. In contrast to the in vitro results in protocol 1, the average highest concentration at any time on bypass was 8.3 +/- 5.7 micrograms/mL or 2 mg/kg.

CONCLUSIONS

DEHP leaches from ECMO circuits, with potential exposure concentrations related to the surface area of the tubing in the ECMO circuit. Heparin bonding of the tubing eliminates this risk. Although significant concentrations of DEHP leach from the nonheparin-bonded circuits over time, our in vivo studies showed that the DEHP plasma concentrations were less than the previously reported values and do not correlate with any observable short-term toxicity. This compound may be either efficiently metabolized by the newborn, or redistributed into various tissues. Although signs of toxicity were not found in this study, long-term complications from chronic exposure to DEHP have not been determined.

Neutrophil and cytokine activation with neonatal extracorporeal membrane oxygenation

OBJECTIVE

To determine whether extracorporeal membrane oxygenation (ECMO), like cardiopulmonary bypass, produces systemic inflammatory responses that could potentiate organ injury in infants with respiratory failure.

STUDY DESIGN

We evaluated the effects of neonatal ECMO on neutrophil surface adherence proteins, elastase release, and cytokine levels in blood samples from 15 patients before and during ECMO, and from banked blood and ECMO circuit blood before cannulation. Neutrophil elastase, tumor necrosis factor alpha, and interleukin types 1 beta, 6, and 8 were measured. Chest radiographs were evaluated by a radiologist using a lung injury score in blinded fashion.

RESULTS

Primed ECMO circuit blood, in comparison with patient pre-ECMO blood, demonstrated marked up-regulation of CD11b (mean fluorescence intensity 1660 +/- 109 vs 361 +/- 81; p < 0.001 (mean +/- SEM)), shedding of L-selectin (mean fluorescence intensity 10 +/- 2 vs 89 +/- 38; p < 0.01), and elevated elastase levels (349 +/- 76 vs 154 ng/ml +/- 38; p < 0.001), consistent with neutrophil activation. During ECMO, neutrophil CD11b levels increased but L-selectin was not significantly shed. Concentrations of circulating neutrophil elastase increase significantly during ECMO. Corrected circulating quantities of interleukin-8 also rose significantly, but the responses of tumor necrosis factor alpha and interleukin-1 beta were minimal. Radiographic lung injury scores worsened with the initiation of ECMO (median score: 6 before ECMO vs 11 in first hour of ECMO; p = 0.012), in conjunction with indicators of neutrophil activation.

CONCLUSION

Neonates with respiratory failure have activation of the inflammatory cascade. ECMO incites additional neutrophil and cytokine activation in association with early pulmonary deterioration. Routine leukodepletion of blood for circuit priming to remove activated neutrophils may be beneficial.

Studies of hypoxemic/reoxygenation injury with aortic clamping: XI. Cardiac advantages of normoxemic versus hyperoxemic management during qardiopulmonary bypass

The conventional way to start cardiopulmonary bypass is to prime the cardiopulmonary bypass circuit with hyperoxemic blood (oxygen tension about 400 mm Hg) and deliver cardioplegic solutions at similar oxygen tension levels. This study tests the hypothesis that an initial normoxemic oxygen tension strategy to decrease the oxygen tension-dependent rate of oxygen free radical production will, in concert with normoxemic blood cardioplegia, limit reoxygenation damage and make subsequent hyperoxemia (oxygen tension about 400 mm Hg) safer. Thirty-five immature (3 to 5 kg, 2 to 3 week old) piglets underwent 60 minutes of cardiopulmonary bypass. Eleven control studies at conventional hyperoxemic oxygen tension (about 400 mm Hg) included six piglets that also underwent 30 minutes of blood cardioplegic arrest. Of 25 studies in which piglets were subjected to up to 120 minutes of ventilator hypoxemia (reducing fraction of inspired oxygen to 5% to 7%; oxygen tension about 25 mm Hg), 11 underwent either abrupt (oxygen tension about 400 mm Hg, n = 6) or gradual (increasing oxygen tension from 100 to 400 mm Hg over a 1-hour period, n = 5) reoxygenation without blood cardioplegia. Fourteen others underwent 30 minutes of blood cardioplegic arrest during cardiopulmonary bypass. Of these, nine were reoxygenated at oxygen tension about 400 mm Hg, and five others underwent normoxemic cardiopulmonary bypass and blood cardioplegia (oxygen tension about 100 mm Hg) with systemic oxygen tension raised to 400 mm Hg after aortic unclamping. Measurements of lipid peroxidation (conjugated dienes and antioxidant reserve capacity) and contractile function (pressure-volume loops, conductance catheter, end-systolic elastance) were made before and during hypoxemia and 30 minutes after reoxygenation. Hyperoxemic cardiopulmonary bypass did not produce oxidant damage or reduce functional recovery after cardiopulmonary bypass in nonhypoxemic controls. In contrast, abrupt and gradual reoxygenation without blood cardioplegia produced significant lipid peroxidation (84% increase in conjungated dienes), lowered antioxidant reserve capacity 68% +/- 5%, 44% +/- 8%, respectively, and decreased functional recovery 75% +/- 6% (p < 0.05), 66% +/- 4% (p < 0.05). Similar impairment followed abrupt reoxygenation before blood cardioplegic myocardial management, because conjungated diene production increased 13-fold, antioxidant reserve capacity fell 40%, and contractility recovered only 21% +/- 2% (p < 0.05). Conversely, normoxemic induction of cardiopulmonary bypass and blood cardioplegic myocardial management reduced conjungated diene production 73%, avoided impairment of antioxidant reserve capacity, and resulted in 58% +/- 11% recovery of contractile function.(ABSTRACT TRUNCATED AT 400 WORDS)

Measurement of lung volumes and pulmonary mechanics during weaning of newborn infants with intractable respiratory failure from extracorporeal membrane oxygenation

Newborn infants with intractable respiratory failure who require extracorporeal membrane oxygenation (ECMO) experience diffuse pulmonary atelectasis shortly after initiation of ECMO. Atelectasis is likely due to the primary lung injury and the reduction of applied inspiratory ventilator pressure when the respirator settings are changed to the "rest settings." These pathophysiologic changes result in a decrease in lung compliance and lung volumes. We hypothesized that improving lung functions observed during ECMO and indicated by an increase in lung volumes will predict successful weaning from ECMO. Sixteen infants (mean +/- SEM: gestational age, 40.3 +/- 0.3 weeks; birth weight, 3.5 +/- 0.1 kg) with meconium aspiration syndrome (n = 13), sepsis (n = 2), and persistent pulmonary hypertension (n = 1) were studied. We measured passive respiratory system mechanics and lung volumes initially during full ECMO support (115 +/- 18 h on ECMO, Study I), and then within 24 h prior to weaning from ECMO (Study II). Respiratory system compliance (Crs), respiratory system resistance (Rrs), functional residual capacity (FRC), and tidal volume (VT) were measured. Prior to Study I lung volumes were too small to be detected. Crs increased between Study I and Study II (0.41 +/- 0.05 to 0.63 +/- 0.05 mL/cmH2O/kg, P < 0.05), and VT increased between Study I and Study II (5.6 +/- 0.6 to 10.4 +/- 0.8 mL/kg, P = 0.0005). FRC increased from 3.6 +/- 1.0 to 7.9 +/- 0.9 mL/kg (P = 0.0001). There was no change in Rrs (88 +/- 8 to 89 +/- 6 cm H2O/L/s, P = 0.9). The combination of Crs > 0.5 mL/cmH2O/kg and FRC > 5 mL/kg was a better predictor (P = 0.0002) of readiness to wean from ECMO than either Crs (> 0.5 mL/cmH2O/kg, P = 0.057) or FRC (> 5 mL/kg, P = 0.007) alone. The combination of FRC and Crs had a sensitivity of 73.3% and specificity of 100% for successful decannulation. We conclude that repeated measurements of FRC and Crs can assess lung recovery and may assist in establishing criteria for successful weaning from ECMO.

Lung management with perfluorocarbon liquid ventilation improves pulmonary function and gas exchange during extracorporeal membrane oxygenation (ECMO)

We investigated whether pulmonary function and gas exchange would improve with liquid perfluorocarbon ventilation (LV) during ECMO for severe respiratory failure. Lung injury was induced in 11 young sheep 15.1 +/- 3.7 kg in weight utilizing right atrial injection of 0.07 cc/kg oleic acid followed by saline pulmonary lavage. When (A-a)DO2 > or = 600 mmHg and PaO2 < or = 50 mmHg with FiO2 = 1.0, ECMO was instituted. Animals were then ventilated with either standard ECMO "lung rest" gas ventilator settings (ECMO, n = 5) or with "total" liquid ventilation at standard ventilator device settings (LIQ-ECMO, n = 6) utilizing perflubron (perfluooctyl bromide, Liquivent; Alliance Pharmaceutical Corp.). After 3 hours on ECMO, pulmonary physiologic shunt decreased (ECMO = 88 +/- 11% vs LIQ-ECMO = 31 +/- 1%; p < .001) and pulmonary compliance increased (ECMO = 0.50 +/- 0.06 cc/cmH2O/kg vs. LIQ-ECMO = 1.04 +/- 0.19 cc/cmH2O/kg; p < .001). The ECMO flow rate required to maintain the PaO2 in the 50-80 mmHg range was decreased significantly (ECMO = 116 +/- 14 ml/kg/min vs. LIQ-ECMO = 14 +/- 5 ml/kg/min; p < .001). In this model requiring extracorporeal support for severe respiratory failure, lung management with liquid ventilation improves pulmonary function and gas exchange.

Pulmonary changes following extracorporeal membrane oxygenation: autopsy study of 23 cases

Extracorporeal membrane oxygenation (ECMO) has become an established mode of therapy in many centers for potentially fatal neonatal respiratory failure refractory to conventional therapy. We reviewed the findings of 23 autopsies of patients placed on ECMO therapy during the period from 1988 to 1992 at our institution in order to document the pulmonary histopathologic changes and to correlate such changes with the duration of treatment. Interstitial and intra-alveolar hemorrhages, as well as hyaline membrane formation, were the most common findings during the first few days of therapy. Reactive epithelial hyperplasia (bronchial and type II pneumocytes), squamous metaplasia, and smooth muscle hyperplasia were observed as early as 2 to 3 days after initiation of ECMO therapy. Interstitial fibrosis was noted only after 7 days of ECMO therapy. In three patients treated for 15, 19, and 21 days there was replacement of the terminal airways and alveoli by tall columnar and mucin-producing epithelium. Alveolar and bronchiolar calcifications were noted in seven of the 23 cases in this series. Pulmonary vascular changes were seen in association with persistent fetal circulation, meconium aspiration, and respiratory distress syndrome. These changes are most likely due to the compounded effect of ECMO and the underlying pulmonary insult.

Improved pulmonary outcome after exogenous surfactant therapy for respiratory failure in term infants requiring extracorporeal membrane oxygenation

A blinded, randomized, controlled study was designed to test whether multiple-dose surfactant therapy would improve pulmonary outcome in term infants with respiratory failure, resulting in a shortened period of extracorporeal membrane oxygenation (ECMO). Infants > or = 34 weeks of gestational age in severe respiratory failure and receiving ECMO were stratified by diagnosis and then randomly assigned to the treatment or the control group. Four doses of modified bovine lung surfactant extract (beractant) were administered to the surfactant group (n = 28), and an equal volume of air was administered to the control group (n = 28). Lung compliance was initially low in both groups; after treatment, values were higher with time in the surfactant group (F = 5.40, p = 0.026). The ECMO treatment period was significantly shorter in the surfactant group (mean +/- SD: 107 +/- 33 hours vs 139 +/- 54 hours for the control group; U = 232, p = 0.023). Tracheal aspirate concentrations of surfactant protein A were low in both groups, and then increased steadily to a higher level in the surfactant group (F = 2.58, p = 0.04). The overall incidence of complications after ECMO was decreased in the surfactant group (18% vs 46% for the control group; chi-square value = 5.004, p = 0.025). Radiographic scores, echocardiographic findings, incidence of intracranial or pulmonary hemorrhage and bronchopulmonary dysplasia, time to extubation, duration of oxygen therapy, and duration of hospitalization did not differ between the two groups. Beractant in this population improved pulmonary mechanics, increased surfactant protein A content in tracheal aspirate, decreased time on ECMO duration, and reduced disease complications.

Phospholipid and surfactant protein A concentrations in tracheal aspirates from infants requiring extracorporeal membrane oxygenation

To test the hypothesis that infants with severe respiratory failure and the need for extracorporeal membrane oxygenation (ECMO) are surfactant deficient, we measured the amount of surfactant phospholipids, disaturated phosphatidylcholine, surfactant protein A, and protein in tracheal aspirates from 22 infants, who received ECMO therapy for respiratory failure with meconium aspiration syndrome (n = 18) or pneumonia (n = 4). Tracheal suction material was obtained in a standardized way every 4 hours during the period of ECMO treatment and pooled for 24-hour periods. During ECMO, mean total phospholipid, disaturated phosphatidylcholine, and surfactant protein A values in tracheal aspirates increased and protein values decreased significantly, predominantly during the 72-hour period before infants were weaned from ECMO. Of the 22 infants, 14 had an increase in tracheal aspirate phospholipid values of more than 200% and were found to need a shorter period of ECMO support (p less than 0.005) and post-ECMO ventilatory support (p less than 0.025) than did the eight infants with stationary or only moderate increases in tracheal aspirate phospholipid values, three of whom had pneumonia. We conclude that infants with respiratory failure who require ECMO treatment often have surfactant deficiency. We speculate that surfactant treatment might decrease the need for or the duration of ECMO support.

A prospective, multicenter, randomized study of high versus low positive end-expiratory pressure during extracorporeal membrane oxygenation

To test the hypothesis that increased positive end-expiratory pressure (PEEP) could prevent deterioration of pulmonary function and lead to more rapid recovery of lung function, we randomly assigned 74 patients undergoing extracorporeal membrane oxygenation (ECMO) at four centers to receive either high (12 to 14 cm H2O) or low (3 to 5 cm H2O) PEEP. The two groups were similar in terms of weight, gestational age, diagnosis, and pre-ECMO course. All other aspects of care were identical. Dynamic lung compliance was measured at baseline and every 12 hours. Radiographs of the chest were obtained daily. Survival rates were similar in the two groups: 36 of 40 for low PEEP and 34 of 34 for high PEEP. The duration of ECMO therapy was 97.4 +/- 36.3 hours in the high-PEEP group and 131.8 +/- 54.5 hours in the low-PEEP group (p less than 0.01). Dynamic lung compliance throughout the first 72 hours of ECMO was significantly higher in patients receiving high PEEP. Radiographic appearance of the lungs correlated well with lung compliance: patients receiving high PEEP had significant deterioration of the radiographic score less frequently than those receiving low PEEP. High PEEP also was associated with significantly fewer complications. We conclude that PEEP of 12 to 14 cm H2O safely prevents deterioration of pulmonary function during ECMO and results in more rapid lung recovery than traditional lung management with low PEEP.

Ultrasonographic findings (CNS, thorax, abdomen) in infants undergoing extracorporeal oxygenation therapy

Extracorporeal membrane oxygenation (ECMO) has been performed on 45 neonates at the Children's Hospital of Michigan in a 39-month period. Ultrasound evaluation of these patients prior to and during ECMO therapy has demonstrated abnormalities in the central nervous system including intracranial hemorrhage (21), extra-axial fluid collections (5), and ventricular enlargement (2). Ultrasonic evaluation of the thoracic cavity in 12 infants revealed pleural fluid in 8. There were seven children with varying types of peritoneal fluid. Two children had visceral abnormalities - 1 with liver hemorrhage and 1 with hydronephrosis found prior to ECMO. Most of these findings could not have been diagnosed without ultrasound and may lead directly to alterations in clinical management. Ultrasound is an extension of physical examination which is important in hour-by-hour clinical care of patients on ECMO.