Drs. WW Yan, Arthur CW Lau, Grace SM Lam and Kenny KC Chan, on behalf of the PYNEH ICU ECMO Team, Department of Intensive Care, Pamela Youde Nethersole Eastern Hospital, Hong Kong. Also published in the Hong Kong Thoracic Society Newsletter, vol 20, no 3, Sep/Oct 2010

Extracorporeal membrane oxygenation (ECMO) is a form of extracorporeal life support (ECLS) where an artificial circuit carries venous blood to a gas exchange device (the oxygenator) for oxygen enrichment and carbon dioxide removal. If the blood is returned to the venous system, it is called a veno-venous ECMO (VV-ECMO), whereas if blood is returned to the arterial system, it is called veno-arterial ECMO (VA-ECMO). VV-ECMO is considered for acute respiratory failure with good cardiac function, while VA-ECMO is used in cardiac failure with or without respiratory failure. The following discussion will focus more on the use of VV-ECMO for acute respiratory failure.

Indications and Contraindications
In general, VV-ECMO is indicated for severe but potentially reversible respiratory failure. From the literature, there were reports of its use in severe pneumonia, acute respiratory distress syndrome (ARDS, primary or secondary), acute graft failure following transplant, pulmonary contusion, alveolar proteinosis,1 smoke inhalation, status asthmaticus, airway obstruction and aspiration syndromes, etc. VA-ECMO is used for cardiac failure, with or without lung failure, e.g., post cardiac surgery, acute myocardial infarction, early graft failure post heart or heart/lung transplant, acute fulminant myocarditis, drug overdose with cardiac suppression, etc. Contraindications include conditions that are potentially irreversible or technically difficult, immunosuppression, multiple organ failure and having contraindications for systemic anticoagulation.

Early trials
ECMO has been in use since the 1970’s. However, initial experience was so unsatisfactory that it put out much ensuing enthusiasm. In a prospective, randomized, multi-centered study from nine medical centres in the United States published in 1979,2 VA-ECMO as a therapy for severe acute respiratory failure was used in 90 adult patients. Patients were treated either with conventional mechanical ventilation (48 patients) or mechanical ventilation supplemented with partial VA-ECMO (42 patients). Only four patients in each group survived and an overall benefit of ECMO could not be shown. Gattinoni et al3 later developed an alternative approach called extracorporeal CO2 removal (ECCO2R) by using the veno-venous perfusion route and a blood flow of only 20 to 30% of cardiac output combined with low frequency mechanical ventilation and additional oxygen insufflation. A randomized trial using this ECCO2R device on 40 patients with severe ARDS by Morris et al4 in 1994 was stopped early because of futility, when benefit in survival at 30 days could not be shown. Overall patient survival was 38% (15 of 40). Seven out of 19 cases on ECCO2R had bleeding resulting in premature discontinuation of treatment.

The above studies were done in the era of primitive ECMO designs. The lungs were either not put to rest or ventilated with high pressure despite low frequency and with 100% oxygen, a strategy that would cause lung injury with the present current concept of protective lung strategy since 1998 after the study of Amato et al.5 and ARDSNet trials.6 High bleeding complications were noted with the use of intensive heparinization. ECMO was many a time initiated only late in the course of illness.

Recent trials
Many advances have taken place after these earlier trials. Better membrane design was achieved with silicone and then microporous membranes compared with older bubble oxygenators. Newly developed fibres offer further improvements by combining a microporous texture with a thin closed layer on the surface to prevent plasma leakage and to extend the running time. Heparin-coated membrane and catheter surfaces allow lower levels of anticoagulation and increase haemocompatibility. Centrifugal pumps with impellers allow generation of high flow with lower likelihood of tube rupture caused by roller pumps. Studies on ECMO carried out in the modern era showed that ECMO could be applied with resulting survival rates exceeding 50%.

In a retrospective case review by Brogan et al7 of the Extracorporeal Life Support Organization (ELSO) registry of respiratory failure adult patients for the entire period of 1986 – 2006 and the more recent years (2002 – 2006), 50% of the 1473 patients survived to discharge. Median age was 34 years. 78% were supported by VV-ECMO. In a multi-variate logistic regression model, the following pre-ECMO factors were associated with increased odds of death: increasing age, decreasing weight, longer days on mechanical ventilation before ECMO, arterial blood pH ≤ 7.18, and being Hispanic and Asian (vs. white population). In the analysis of patients of the more recent years (n = 600), the following factors were associated with increased odds of death: increasing age, PaCO2 ≥ 70 mmHg (vs. PaCO2 ≤ 44), ARDS (vs. diagnostic categories of acute respiratory failure and asthma), VA-ECMO (vs. VV-ECMO), need for CPR and presence of complications while on ECMO, and arterial blood pH < 7.2 or >7.6.

Table 1.  Murray Score (average score of all 4 parameters) - calculator here
Parameter / Score
A.           PaO2/FIO2
(On 100% Oxygen)
B.           CXR
1 point per quadrant infiltrated
C.           PEEP
D.           Compliance (ml/cmH2O)
The CESAR trial (Conventional Ventilation or ECMO for Severe Adult Respiratory Failure)8 published in the Lancet in 2009 was a single-centre, randomized, controlled trial by intention-to-treat analysis performed at the Glenfield Hospital, UK. Inclusion criteria were patients with severe respiratory failure, being defined as having Murray score ≥3 (Table 1) or uncompensated hyper-capnea with pH <7.20. Recruitment was conducted from July 2001 to August 2006. The primary outcome was survival without severe disability, defined as being confined to bed, or unable to dress/wash oneself, by six months. Ninety patients were randomized into receiving ECMO and 90 to conventional therapy. In the ECMO group, 68 (75%) actually received ECMO, compared with the 90 patients in the conventional group. The relative risk reduction was in favour of ECMO (0.69, 95% CI 0.05 to 0.97, p = 0.03). The number-needed-to-treat (NNT) to achieve one more survivor without severe disability was only six. Though being a positive study, it has been criticized that it was only a single-centre study, and the control patients remained at the referring centre where the standard of care was unknown and protocol of management not standardized.

ECMO in the H1N1 pandemic: the worldwide situation
It happened that in the same year of 2009, the world witnessed the pandemic of influenza A (H1N1), and Hong Kong also identified our first case on 1st May 2009. Recent publications showed that ECMO had been adopted for the management of H1N1-related respiratory failure in all major countries and cities, including Hong Kong, for both paediatric and adult patients, including obstetric patients.9 The largest published series so far was from the Australia and New Zealand Extracorporeal Membrane Oxygenation (ANZ ECMO) Influenza Investigators.10 Patients with the 2009 Influenza A (H1N1) Acute Respiratory Distress Syndrome from 1 June 2009 to 31 August 2009 who had been on ECMO were reviewed. ECMO had been used in 68 patients (34%) out of a total of 133 patients on intermittent positive pressure ventilation. For patients given ECMO, at the time of writing, 48/68 (71%) survived ICU admission, 6/68 (9%) still in ICU, and 14/68 (21%) died.

The Extracorporeal Support Organisation (ELSO), an international consortium of health care professionals and scientists dedicated to the development and evaluation of novel therapies for support of failing organ systems, has set up a H1N1 Registry for H1N1 cases that have been placed on ECLS.11 By voluntary reporting (and hence reporting bias), 248 patients from 73 centres have used ECMO for H1N1-related respiratory failure at May 28, 2010, from the statistics published on its website. There were 151 adult, 90 paediatric and 7 neonatal patients. The average hours on ECMO were 264. As for the outcome, 137 (55.2%) were alive, 83 (33.5%) died, 14 (5.6%) still on ECMO, 14 (5.6%) off ECMO but were still in hospital. Both VV or VA modes had been adopted. Age-standardized mortality rates of various ages were around 35%, but for those aged 50 and above, the rate rose to 48.1%. A trend of having higher mortality rates was also noted with more ventilator days before ECMO: 0-2 days: 34/117 (29.1%), 3-6 days: 18/57(31.6%), >= 7 days: 30/73(41.1%).

Local Experience of ECMO for influenza A (H1N1) and other conditions
The first trial of using VV-ECMO for the management of influenza A (H1N1) was carried out in the ICU of United Christian Hospital in July 2009,12 followed by Pamela Youde Nethersole Eastern Hospital and Prince of Wales Hospital. During the period from 1 May 2009 to 28 Feb 2010, 120 patients had been mechanically ventilated in Hong Kong ICUs, and seven cases had used ECMO. A detailed report of these cases is now in press.13 One out of the seven (14.3%) ECMO patients died. Nosocomial infection was the commonest complication. There was otherwise no significant complication directly attributable to ECMO.

In the Intensive Care Unit of Pamela Youde Nethersole Hospital at the time of writing (September 2010), we have used VV-ECMO in 12 patients (5 males and 7 females): 10 influenza A pneumonia (9 pandemic H1N1, 1 pending subtyping by specific serology), 1 human metapneumovirus pneumonia and 1 mycoplasma pneumonia. Age ranged from 19 to 56. Four patients have enjoyed good past health, while comorbidities were noted in others including hypertension (4 patients), diabetes mellitus (4 patients), morbid obesity with BMI  >= 35 kg/m2 (3 patients), hyperlipidemia (1 patient), hepatitis B carrier (1 patient), schizophrenia (1 patient), obstructive sleep apnea (1 patient) and non-toxic nodular goiter (1 patient). Duration of ECMO ranged from 4 days (2 patients) to 24 days (1 patient). All patients survived: eleven have already been discharged home and the one with the longest duration has also been weaned off ventilator and started rehabilitation.

A point of note is that all patients we have so far treated were in extremis when ECMO was started; they fulfilled the same entry criteria as for the CESAR study, and all of them had relentless desaturation while on FiO2 of 1.0 and high ventilatory support, or considered almost total treatment failure when ECMO was used as a desperate last resort. In other words, we believe they would not have survived had ECMO not been adopted.

Technical aspects
The VV-ECMO setup (Figure 1) used in our hospital involves a console (Maquet PLS set BE-PLS 2050, Fig 2), a Centrifugal Pump (Rotaflow RF 32, Maquet, Rastatt, Germany, Fig. 3), a hollow fiber membrane oxygenator (Quadrox-D Oxygenator, Maquet, Rastatt, Germany, Fig. 4) connected with biocoated tubes, an air-oxygen blender (Model 20090, Sechrist, California, USA, Fig. 5), and a heated water bath (Heater Unit HU 35, Maquet, Rastatt, Germany) for blood temperature control.

FIg 1. VV-ECMO setup

Fig 2. Control over the speed of revolution of centrifugal pump and hence the flow rate

Fig 3. The Rotaflow centrifugal pump driven by magnetic field to drive blood forward

Before use, the whole circuit needs priming with normal saline and any bubbles to be driven out and possible leakage excluded. Cannulae from 15 to 23 French (HLS, Maquet, Rastatt, Germany) were employed for femoral and/or jugular vein cannulation. Vascular ultrasound is used to confirm position, size and patency of the veins before insertion by the Seldinger technique. Appropriately sized cannulae (Figure 6a and b) are chosen to provide the desired extracorporeal blood flow. Major ECMO centres use different perfusion routes (jugular-femoral, femoral-jugular and femoral-femoral). Drainage of blood can be facilitated by fluid administration, using a catheter of a large size and with side holes to facilitate outflow by lessening sucking onto the inferior vena cava, or inserting one more cannula. The access and return cannulae should be placed at some distance apart while also taking into account the direction of blood flow to minimize circuit recirculation. A sub-circuit for continuous renal replacement therapy (CRRT) is possible to be set up within the ECMO circuit (Figure 7). Surgical repair of vein after decannulation is usually not necessary for percutaneously inserted cannula.

Fig 4. Gas of adjustable FiO2 controlled by an air-oxygen blender will go through this oxygenation membrane to allow gaseous diffusion

Fig 5. The air-oxygen blender. FiO2 and gas flow rate can be adjusted

Fig 6a. Drainage cannulae

Fig 6b. Return cannulae

Fig 7. Sub-circuit for continuous renal replacement therapy is set up within the ECMO circuit

Another extracorporeal device which has also been used in Hong Kong is the arterio-venous pumpless Novalung® iLA Membrane Ventilator. It obtains blood from the femoral artery driven by blood pressure, and returns to the femoral vein. Because it is a low flow device, it only allows for efficient CO2 removal, but the ability to oxygenate is not great. This device will not be further discussed here.

Typical settings and targets
ECMO blood flow is set to 50 to 80 ml/kg/min, gas flow at 50 to 80 ml/kg/min to achieve ventilation to perfusion matching of 1:1. FiO2 (sweep gas) starts with 1.0. Once the ECMO starts operating, patient’s oxygenation typically improves promptly within one to two minutes. Oxygenated blood inside the return catheter can be appreciated to appear obviously redder than deoxygenated blood inside the drainage catheter. The target is to achieve an arterial oxygen saturation of 85 to 95%, PaCO2 4.7 to 6.0 kPa, pH 7.35 to 7.45, mean arterial pressure > 65 mmHg, haemoglobin level around 10 g/dl, aPTT 50 to 60s, and platelet count >100,000. A detailed review of the technical aspects of ECMO in critically ill patients has been published.14

For the ventilator, the goals are to maximize lung rest and minimize lung injury to avoid baro-/volu-traumas. If the ECMO is running well without too much recirculation, ventilatory support can be set very low till recovery. Usually we will continue to adopt an open and protective lung strategy (e.g. FiO2 0.30 to 0.40, tidal volume 3 to 4 ml ml/kg, PEEP 10 to 20 cmH2O, low respiratory rate of 8 to 10 ml/min). We have also used oesophageal catheters to guide PEEP setting and measure of transpulmonary pressure.15

Patient monitoring and possible complications
Important monitoring parameters are oxygenation, carbon dioxide clearance, anticoagulation status and bleeding complication, ECMO circuit status including cannula patency due to direct mechanical problem or volume status, oxygenator status, ventilator management, sedation, as well as the overall patient condition. Possible problems and complications can include bleeding, haemolysis, DIC, circuit rupture, pump failure, air and clot in circuit and accidental decannulation. Table 2 shows common complications or conditions and their trouble-shooting.

Table 2. Monitoring, complications and trouble-shooting
Oxygen saturation of arterial blood in VV-ECMO depends on the ECMO blood flow, patient’s systemic venous return, oxygen saturation of systemic venous blood, patient’s residual lung function, and the degree of circuit recirculation. If oxygenation cannot be maintained with persistent metabolic acidosis. Make sure gas flow to the oxygenator is not interrupted.  Increase the extracorporeal blood flow and treat any sepsis that causes an abnormal increase in cardiac output. In access insufficiency which may manifest as chattering, increase intravascular volume, or insert a second access cannula. Adjust position of the cannulae to reduce the degree of recirculation. Transfuse blood transfusion to maintain a haematocrit level between 40-45%. In cardiac failure, use volume, inotropes, or conversion to VA-ECMO for cardiac support.
Carbon dioxide clearance
Carbon dioxide removal is related to the gas flow and the function of the oxygenator. If pCO2 is high, check oxygenator function, use 100% oxygen as sweep gas. Begin with a sweep gas flow rate of 6L/min. After the extracorporeal blood flow has been adjusted, set the sweep gas to extracorporeal blood flow ratio to 1:1. Titrate sweep gas flow rate according to carbon dioxide partial pressure, increase sweep gas flow rate to increase carbon dioxide clearance. Monitor end-tidal CO2 which should be > 20 mmHg to avoid lung tissue alkalosis.
Monitor activated partial thromboplastin time (aPTT) every 6 hours and titrate heparin to achieve a target of aPTT of 50 to 60s. Apply direct pressure to accessible bleeding sites. In case of bleeding at the cannulation site, rule out decannulation, decrease heparin. Stop heparinization for severe bleeding or bleeding at critical sites, replace deficient clotting elements, transfuse platelets, cryoprecipitate until fibrinogen >1.5, fresh frozen plasma until an INR >1.3, give tranexamic acid for their antifibrinolytic effect and to improve platelet function.
Haemolysis, thrombocytopenia and DIC
It is caused by mechanical damage. Monitor Hb, plasma free haemoglobin and platelet concentration regularly. Increase intravascular volume to increase venous return, insert a second access cannula, change out clotted parts of the circuit, reset anticoagulation targets, review pump speed settings.
Circuit rupture
Increase ventilator setting and replace volume, stop pump and clamp circuit on either side of rupture, change damaged circuit.
Pump failure
Turn off pump then re-engage if dislodged, or replace pump head. If battery power also fails, crank the pump by hand.
Air in circuit
To prevent air embolism, it is necessary to maintain the pressure at the blood side higher than that at the gas side: Keep the oxygenator below the level of the patient. Stop the pump, clamp the circuit near the patient, increase ventilator support, position the patient head down and consider aspiration of the right heart for air. Identify and treat source of air-leak, remove air from circuit by aspiration.
Clotting in circuit
Clotting can occur at the pump, oxygenator, or cannulae. The risk of clotting can be decreased by minimizing the number of circuit access sites. Clots larger than 5mm or enlarging clots on the return side of the circuit should be removed. Clotted oxygenator should be replaced.
Decannulation by accident
Turn off pump and clamp the circuit close to the patient. Increase ventilator settings and replace volume, control bleeding by direct pressure, reinsert cannula and restart circuit if necessary

Logistics of service provision in Hong Kong
In Hong Kong, three hospitals (Queen Mary Hospital, Queen Elizabeth Hospital and Prince of Wales Hospital) would be the initial ECMO referral centres for H1N1-related respiratory failure. The service might be extended eventually to other hospitals. Exact referral criteria and indications are being worked out by the HA COC (ICU) ECMO Working Group.

As PYNEH has accumulated experience in ECMO, besides treating our own patients, we welcome referrals of acute respiratory failure of various causes from other HA or private hospitals if considered necessary. Acceptance of cases will be decided on a case-by-case basis according to our availability of beds, equipment and staff.

Patients with acute respiratory failure can be too sick for a safe transport. The ICU of PYNEH has successfully performed on-site VV-ECMO setup in another HA hospital for the management of a young patient with severe mycoplasma pneumonia. We sent out a team of four doctors and a nursing specialist to perform the on site VV-ECMO setup, with subsequent escort of the patient back to our ICU for continuation of treatment. The whole process of catheter insertion and transport was smooth and took three hours to complete.

As technology improves, use of ECMO appears to revive recently. Interest was rekindled by the positive CESAR trial and experience of use during the H1N1 pandemic. In many situations, but not all, ECMO was definitely life-saving. Our observation also agrees with previous report that earlier ECMO application before lung injury as a result of baro-/volutrauma and oxygen toxicity sets in may provide even greater benefit. With further research, ECMO could become a revolution of our paradigm of acute respiratory failure management. Further trials should be focused to find out any subgroups of patients that will particularly benefit from this treatment, and some viral or atypical pneumonia can be one of these subgroups compared with bacterial pneumonia. From our experience, ECMO can be labour-intensive at the moment of setting up, but by using equipment of modern technology, subsequent care is less so. While it is also considered expensive, its cost effectiveness should be more thoroughly explored to see if shortened duration of mechanical ventilation with improved morbidity and mortality actually saves cost. Major Hong Kong ICUs are preparing to incorporate this technology into their armamentarium.

If you would like to obtain further local information on ECMO, please visit the Hong Kong Society of Critical Care Medicine website at www.hksccm.org (search for “ECMO”) where you could find articles, references, local experience, photos and videos about ECMO.

The successful introduction of ECMO in our ICU as well as the smooth on-site ECMO setup with interhospital transport was only possible with the close cooperation within the PYNEH ICU ECMO team comprising of doctors (WW Yan, Arthur CW Lau, Kenny KC Chan, Grace SM Lam, HP Shum, Helen HL Wu, Arthur MC Kwan and Jackie OY Tam) and all nurses.

1. Sihoe AD, Ng VM, Liu RW, Cheng LC. Pulmonary alveolar proteinosis in extremis: the case for aggressive whole lung lavage with extracorporeal membrane oxygenation support. Heart Lung Circ. 2008 Feb;17(1):69-72. Epub 2007 Mar 6.
2. Zapol WM, Snider MT, Hill JD, Fallat RJ, Bartlett RH, Edmunds LH, Morris AH, Peirce EC 2nd, Thomas AN, Proctor HJ, Drinker PA, Pratt PC, Bagniewski A, Miller RG Jr. Extracorporeal membrane oxygenation in severe acute respiratory failure. A randomized prospective study. JAMA. 1979 Nov 16;242(20):2193-6.
3. Gattinoni L, Pesenti A, Mascheroni D, Marcolin R, Fumagalli R, Rossi F, Iapichino G, Romagnoli G, Uziel L, Agostoni A, et al. Low-frequency positive-pressure ventilation with extracorporeal CO2 removal in severe acute respiratory failure. JAMA. 1986 Aug 15;256(7):881-6.
4. Morris AH, Wallace CJ, Menlove RL, Clemmer TP, Orme JF Jr, Weaver LK, Dean NC, Thomas F, East TD, Pace NL, Suchyta MR, Beck E, Bombino M, Sittig DF, Böhm S, Hoffmann B, Becks H, Butler S, Pearl J, Rasmusson B. 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 Feb;149(2 Pt 1):295-305.
5. Amato MB, Barbas CS, Medeiros DM, Magaldi RB, Schettino GP, Lorenzi-Filho G, Kairalla RA, Deheinzelin D, Munoz C, Oliveira R, Takagaki TY, Carvalho CR. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med. 1998 Feb 5;338(6):347-54.
6. The Acute Respiratory Distress Syndrome Network. Ventilation with Lower Tidal Volumes as Compared with Traditional Tidal Volumes for Acute Lung Injury and the Acute Respiratory Distress Syndrome. N Engl J Med 2000; 342:1301-1308May 4, 2000
7. Brogan TV, Thiagarajan RR, Rycus PT, Bartlett RH, Bratton SL. Extracorporeal membrane oxygenation in adults with severe respiratory failure: a multi-center database. Intensive Care Med. 2009 Dec;35(12):2105-14. Epub 2009 Sep 22.
8. Peek GJ, Mugford M, Tiruvoipati R, Wilson A, Allen E, Thalanany MM, Hibbert CL, Truesdale A, Clemens F, Cooper N, Firmin RK, Elbourne D; CESAR trial collaboration. Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial. Lancet. 2009 Oct 17;374(9698):1351-63. Epub 2009 Sep 15.
9. Robertson LC, Allen SH, Konamme SP, Chestnut J, Wilson P. The successful use of extra-corporeal membrane oxygenation in the management of a pregnant woman with severe H1N1 2009 influenza complicated by pneumonitis and adult respiratory distress syndrome. Int J Obstet Anesth. 2010 Aug 10. [Epub ahead of print]
10. Extracorporeal Membrane Oxygenation for 2009 Influenza A(H1N1) Acute Respiratory Distress Syndrome. Australia and New Zealand Extracorporeal Membrane Oxygenation (ANZ ECMO) Influenza Investigators, Davies A, Jones D, Bailey M, Beca J, Bellomo R, Blackwell N, Forrest P, Gattas D, Granger E, Herkes R, Jackson A, McGuinness S, Nair P, Pellegrino V, Pettilä V, Plunkett B, Pye R, Torzillo P, Webb S, Wilson M, Ziegenfuss M. JAMA. 2009 Nov 4;302(17):1888-95. Epub 2009 Oct 12.
11. H1N1 ECLS Registry, Statistics from the H1N1 Registry (as of May 28, 2010), Extracorporeal Life Support Organization. http://www.elso.med.umich.edu/H1N1Registry.html (Accessed on 15 September 2010)
12. Liong T, Lee KL, Poon YS, Lam SY, Chan CP, Yue CS, Chu CM, Yuen KY, Law KI. The first novel influenza A (H1N1) fatality despite antiviral treatment and extracorporeal membrane oxygenation in Hong Kong. Hong Kong Med J. 2009 Oct;15(5):381-4.
13. Chan KKC, Lee KL, Lam PKN, Law KI, Joynt GM, Yan WW. Hong Kong’s experience on the use of extracorporeal membrane oxygenation for the treatment of influenza A (H1N1). HK Med J (in press)
14. Marasco SF, Lukas G, McDonald M, McMillan J, Ihle B. Review of ECMO (extra corporeal membrane oxygenation) support in critically ill adult patients. Heart Lung Circ. 2008;17 Suppl 4:S41-7. Epub 2008 Oct 29.
15. Talmor D, Sarge T, Malhotra A, O'Donnell CR, Ritz R, Lisbon A, Novack V, Loring SH. Mechanical ventilation guided by esophageal pressure in acute lung injury. N Engl J Med. 2008 Nov 13;359(20):2095-104. Epub 2008 Nov 11.