Abstract
Rationale: Many patients with severe acute respiratory distress syndrome (ARDS) caused by influenza A(H1N1) infection receive extracorporeal membrane oxygenation (ECMO) as a rescue therapy.
Objectives: To analyze factors associated with death in ECMO-treated patients and the influence of ECMO on intensive care unit (ICU) mortality.
Methods: Data from patients admitted for H1N1-associated ARDS to French ICUs were prospectively collected from 2009 to 2011 through the national REVA registry. We analyzed factors associated with in-ICU death in ECMO recipients, and the potential benefit of ECMO using a propensity score–matched (1:1) cohort analysis.
Measurements and Main Results: A total of 123 patients received ECMO. By multivariate analysis, increasing values of age, lactate, and plateau pressure under ECMO were associated with death. Of 103 patients receiving ECMO during the first week of mechanical ventilation, 52 could be matched to non-ECMO patients of comparable severity, using a one-to-one matching and using control subjects only once. Mortality did not differ between the two matched cohorts (odds ratio, 1.48; 95% confidence interval, 0.68–3.23; P = 0.32). Interestingly, the 51 ECMO patients who could not be matched were younger, had lower Pao2/Fio2 ratio, had higher plateau pressure, but also had a lower ICU mortality rate than the 52 matched ECMO patients (22% vs. 50%; P < 0.01).
Conclusions: Under ECMO, an ultraprotective ventilation strategy minimizing plateau pressure may be required to improve outcome. When patients with severe influenza A(H1N1)–related ARDS treated with ECMO were compared with conventionally treated patients, no difference in mortality rates existed. The unmatched, severely hypoxemic, and younger ECMO-treated patients had, however, a lower mortality.
Extracorporeal membrane oxygenation (ECMO) benefit for H1N1 severe acute respiratory distress syndrome is still debated because different countries have heterogeneous results.
We identified new factors associated with survival in ECMO patients. In a propensity score–matched analysis, ECMO-treated patients and conventionally treated patients have similar mortality rates.
Extracorporeal membrane oxygenation (ECMO) has been proposed early after its first description as a possible management of the most hypoxemic cases of acute respiratory distress syndrome (ARDS), but two randomized controlled trials (RCT) could not confirm the superiority of the technique over more conventional management (1, 2), and its use was long restricted to a few selected centers. Renewed interest in this therapy has been stimulated by technical improvements and by the positive results of a recent RCT (3). This technique was thus more widely used during the recent influenza A(H1N1) pandemic (pdm) as a rescue therapy for the most severe ARDS cases who could hardly be managed with conventional therapy (4, 5).
Analysis of patients with influenza A(H1N1)pdm pneumonia provides a unique opportunity to describe the results of ECMO in patients suffering from a relatively homogeneous form of ARDS. In a recent study from the United Kingdom, patients referred to ECMO centers had lower mortality than propensity score (PS)–matched nonreferred patients (6). Soon after the first cases occurred in France, the REVA/French Society of Intensive Care (SRLF) H1N1 registry was set up to prospectively collect the characteristics of patients infected with influenza A(H1N1)pdm hospitalized in participating intensive care units (ICUs), and included 123 patients treated with ECMO between July 2009 and May 2011 (7, 8). In this study, we describe the baseline characteristics, ICU management, and variables associated with ICU mortality in the largest cohort to date of patients with severe influenza A(H1N1)pdm–induced ARDS receiving ECMO; we also estimated the potential benefit of ECMO on survival of patients after PS matching of ECMO recipients with other patients from the same registry having severe ARDS managed without this technique. Some of the results of these studies have been previously reported in the form of abstracts in May 2012 during the American Thoracic Society International Conference (9, 10).
From July 2009 to March 2010, the baseline characteristics, ICU management, and outcome of all patients infected with influenza A(H1N1)pdm admitted to 114 participating French ICUs were prospectively collected in a web-based registry (REVA-SRLF H1N1 registry). During the subsequent wave of the epidemic (September 2010 to March 2011), only patients treated with ECMO were recorded in the database. The database thus contained patients infected with influenza A(H1N1)pdm with ARDS treated or not treated with ECMO.
All adult patients hospitalized with influenza A(H1N1)pdm–related ARDS and treated with ECMO were included in this ECMO cohort (Figure 1). Patients were followed-up until ICU discharge or death, or up to June 1, 2011.
Figure 1. Flow chart of patients included in the severe acute respiratory distress syndrome (ARDS) and extracorporeal membrane oxygenation (ECMO) cohort. From July 2009 to March 2010, all patients infected with A(H1N1)pdm admitted to participating French intensive care units (ICUs) were prospectively recorded. During the subsequent wave of the epidemic (September 2010 to March 2011), only patients infected with H1N1 treated with ECMO were recorded in the database. Therefore, the non-ECMO cohort is composed of patients admitted in ICUs during the 2009–2010 pandemic and the ECMO cohort includes patients admitted between 2009 and 2011 and treated with ECMO. *Severe ARDS was defined by Pao2/Fio2 < 100 mm Hg, Sao2 < 90%, pH < 7.21, or LIS > 3. LIS = modified lung injury score; MV = mechanical ventilation.
[More] [Minimize]Patients treated with ECMO in the first week of ARDS were matched to patients from the whole database having influenza A/H1N1–related severe ARDS and treated without ECMO using a PS exposed–nonexposed one-to-one matching approach. Severe ARDS was defined by the classical definition of ARDS (11) plus a modified Lung Injury Score (12) greater than 3 or any of the following three criteria: (1) an arterial pH less than 7.21, (2) a ratio of arterial partial pressure of oxygen to the fraction of inspired oxygen (Pao2/Fio2) less than 100 mm Hg, or (3) an arterial oxygen saturation (Sao2) less than 90%.
A confirmed case of influenza A(H1N1)pdm–associated ARDS was considered when a polymerase chain reaction was positive.
Baseline characteristics were collected on admission; ICU evolution and management were recorded daily (variables are detailed in the online supplement).
Descriptive statistics included frequency (percentages) for categorical variables, mean and standard deviation, or median and interquartile ranges for continuous variables. Comparisons of proportions were made using chi-square test, Fisher exact test, or McNemar test for matched patients; continuous variables were compared using Student t test or Wilcoxon rank sum test and the corresponding paired tests for matched patients.
To determine variables associated with ICU mortality in the whole cohort of ECMO-treated patients, covariates assumed to be associated with death and those associated with ICU mortality at P less than 0.10 in univariate analyses were entered in a multivariate stepwise backward logistic regression model.
Covariates presumed to be associated with the decision of ECMO treatment or with ICU death were included in a multivariable logistic regression analysis with ECMO treatment in the first week as the dependent variable to determine the PS of ECMO treatment for each patient (i.e., the probability of receiving ECMO treatment conditionally on the observed covariates). The independent covariates selected were age; sex; pregnancy status; body mass index (BMI); risk factors for influenza-related complications as defined by the Centers for Disease Control and Prevention (13) (including immunosuppression, chronic lung disease, diabetes, chronic renal failure, chronic heart failure, obesity, and pregnancy); SAPS3 score (14, 15), SOFA score (16), bacterial coinfection, and shock at the time of admission; and the use of corticosteroids or of a rescue therapy, worst arterial pH, Pao2/Fio2 ratio, and Lung Injury Score (12) before ECMO implantation or during the first week for non-ECMO patients with ARDS. Missing data were imputed to complete the PS. Therefore, we selected the model leading to a reduction of the imbalances (17) between the two matched cohorts for a maximal number of variables (plotted standardized differences before and after matching) (see Figure E1 in the online supplement).
For the matched cohort analysis, patients treated with ECMO were matched with patients with severe ARDS managed without ECMO according to the PS, using a 1:1 matching procedure without replacement (a matched control patient cannot be reused for matching another case) and a caliper width of 0.2 (i.e., 0.2 × standard deviation of the logit of the PS) resulting in a relatively narrow difference between matched variables, as recommended (18, 19). This PS matching method was rather conservative and these points as well as the results obtained with less stringent methods are explained in more details in the online supplement.
Multivariate analyses of variables associated with ICU death in the ECMO cohort were performed with and without data imputation. Several models with covariates usually related with mortality were performed to limit confounding. To compare our results with those of the analysis conducted on the UK cohort (6), the PS-matched analysis was repeated allowing replacement for matching ECMO with non-ECMO patients (i.e., reusing control patients more than once when no other control subject could be found). We then performed the three analyses done by the UK group on our dataset (manual matching with replacement, PS matching analysis with replacement, GenMatch matching with replacement [20–22]) (see online supplement).
All statistical tests were two-sided and were not adjusted for multiple comparisons. P values of 0.05 or less were considered statistically significant. Statistical analyses were performed with R 2.10.1 (http://www.R-project.org) software packages.
Data collected were directly downloaded as electronic files from the REVA web registry. The database was completed when needed after direct contact with the ICU physicians. Duplicate notifications were systematically checked, and patients transferred from one ICU to another were counted as a single admission. We followed the Strengthening the Reporting of Observational Studies in Epidemiology statement guidelines for observational cohort studies (23).
The national committee for protection of patient’s rights and electronic data recording approved data collection in the national H1N1 registry. The SRLF Ethics Committee gave approval for the registry and this analysis. The authors had full access to the data and take responsibility for their integrity.
Between July 29, 2009 and March 26, 2011, 127 patients infected by influenza A(H1N1)pdm and treated with ECMO were recorded into the REVA/SRLF H1N1 registry (Figure 1, Table 1). The inclusiveness for ECMO patients treated in French ICUs was cross-checked with mandatory notifications to the National Institute for Public Health Surveillance (InVS, St Maurice, France) and only two patients were missing from the REVA registry. Four of these 127 patients were excluded for not fulfilling the H1N1pdm-related ARDS criteria (Figure 1). Despite a lack of virologic confirmation, 16 patients considered as having a high probability of influenza A(H1N1)pdm–induced ARDS because of a typical clinical presentation and no other cause for ARDS were kept in the analysis (2 in the ECMO group, 14 in the ARDS group).
| Mean (SD), Median (IQR), or N (%) | ||
| Baseline characteristics | ||
| Age, yr | 42 (13) | |
| Male sex | 61 (50%) | |
| MacCabe 1* | 108 (88%) | |
| Risk factor for influenza complication | 93 (76%) | |
| Pregnancy or postpartum | 18 (15%) | |
| BMI, kg/m2 | 30.5 (8.5) | |
| On admission | ||
| SAPS3 | 58 (14) | |
| SOFA | 9.5 (4) | |
| Bacterial coinfection | 28 (23%) | |
| Before ECMO | ||
| Shock | 60 (49%) | |
| Corticosteroid therapy | 42 (34%) | |
| Rescue therapy | 91 (74%) | |
| Inhaled nitric oxide | 83 (72%) | |
| Prone positioning | 51 (45%) | |
| Almitrine | 7 (7%) | |
| HFOV | 3 (2%) | |
| Time from MV to ECMO, d | 2 (1–5) | |
| V-V ECMO | 107 (87%) | |
| Pre-ECMO | First Day on ECMO | |
| Tidal volume, ml/kg PBW | 6.7 (1.6) | 3.9 (1.4) |
| Respiratory rate, min−1 | 27 (6) | 19 (8) |
| PEEP, cm H2O | 13 (4) | 13 (4) |
| Plateau pressure, cm H2O | 32 (5) | 26 (4) |
| Pao2/Fio2 ratio, mm Hg | 63 (21) | 109 (74) |
| Sao2, % | 84 (11) | 95 (5) |
| Arterial pH | 7.26 (0.12) | 7.39 (0.12) |
| Paco2, mm Hg | 57 (18) | 38 (9) |
| Arterial lactate, mM | 2.9 (3) | 4.2 (5.6) |
| Driving pressure, cm H2O† | 19 (6) | 14 (5) |
| LIS | 3.4 (0.6) | Not collected |
| Complications and outcome | ||
| Nosocomial pneumonia | 68 (62%) | |
| Length of ECMO, d | 11 (8–22) | |
| Length of MV, d | 28 (15–44) | |
| Length of ICU stay, d | 33 (17–59) | |
| Mortality | 44 (36%) | |
The mean age of the 123 remaining patients was 42 years and 50% were men; 76% had one or more risk factors for influenza-related complications (13), mostly obesity with a BMI higher than 30 kg/m2 (40%), immunosuppression (19%), and pregnancy or postpartum status (15%). On admission, 28 (23%) patients had concomitant bacterial infection, and 49% had septic shock.
All ECMO patients received antiviral therapy. ECMO was initiated a median of 2 days after mechanical ventilation, and 103 (84%) patients had been receiving mechanical ventilation for less than 8 days before ECMO (Table 1); only 16 (13%) received venoarterial ECMO and 107 received venovenous ECMO. Before initiation of ECMO, the mean modified Lung Injury Score was 3.4. Rescue therapies were used in 74% of patients, with 72% of patients receiving inhaled nitric oxide, and 45% prone positioning. A total of 42 patients (34%) had received steroids before ECMO initiation.
Before initiating ECMO, ventilation parameters included a tidal volume of 6.7 ml/kg predicted body weight, a high respiratory rate, and high positive end-expiratory pressure (PEEP) resulting in a mean plateau pressure (Pplat) of 32 cm H2O. Patients remained severely hypoxemic (Pao2/Fio2 ratio, 63 mm Hg; Sao2, 84%), hypercapnic (Paco2, 57 mm Hg), and acidotic (pH = 7.26), with a mean arterial lactate of 2.9 mmol/L.
After starting ECMO, physicians reduced tidal volume by 2.8 ml/kg predicted body weight and respiratory rate by 8 breaths/min, but did not change the PEEP level. Both Pplat and driving pressure (difference between Pplat and PEEP) decreased by 6 cm H2O and arterial blood gas improved (increase in Pao2/Fio2 by 42 mm Hg, decrease in Paco2 by 18 mm Hg).
The median duration of ECMO, intubation, and ICU stay were 11, 28, and 33 days, respectively. Forty-four patients (36%; 95% confidence interval [CI], 27–44%) died in ICU: 22 of multiorgan failure, 8 of refractory hypoxemia, 6 of refractory shock (two hemorrhagic, one septic, and three unspecified shock), 5 of intracranial hemorrhage, and 3 patients had unspecified cause of death. Overall, 65 (53%) patients had one or more ECMO-related complication (see Table E2).
Except for a higher rate of early administration of steroids in nonsurvivors (P = 0.03), there was no difference between survivors and nonsurvivors in baseline characteristics and arterial blood gas results before and after ECMO in the two groups (Table 2). Nevertheless, nonsurvivors had higher plateau (P < 0.01) and driving (P = 0.03) pressures and arterial lactate levels (P = 0.004) and a smaller reduction in Pplat (P < 0.01) on the first day under ECMO (Figure 2). After multivariate analysis, age (odds ratio [OR], 1.09; 95% CI, 1.04–1.15; P < 0.01), higher Pplat (OR, 1.33; 95% CI, 1.14–1.59; P < 0.01), and lactate under ECMO (OR, 1.42; 95% CI, 1.18–1.82; P < 0.01) were the three variables that remained significantly associated with ICU death. Driving pressure was not retained in the model, but was highly correlated with Pplat (see Figure E4).
| Mean (SD), Median (IQR), or N (%) | |||
| Survivors (n = 79) | Nonsurvivors (n = 44) | P Value | |
| Baseline characteristics | |||
| Age, yr | 40 (13) | 45 (13) | 0.06 |
| Male sex | 37 (47%) | 24 (55%) | 0.53 |
| MacCabe 1* | 71 (90%) | 37 (84%) | 0.51 |
| Risk factor for influenza complication | 62 (78%) | 31 (70%) | 0.44 |
| Pregnancy or postpartum | 14 (18%) | 4 (9%) | 0.30 |
| BMI, kg/m2 | 31.5 (8.9) | 28.6 (7.9) | 0.09 |
| On admission | |||
| SAPS3 score | 57 (13) | 61 (17) | 0.22 |
| SOFA score | 9.5 (3.9) | 9.4 (4.6) | 0.95 |
| Bacterial coinfection | 17 (22%) | 11 (26%) | 0.80 |
| Before ECMO | |||
| Shock | 35 (44%) | 25 (57%) | 0.25 |
| Steroids | 21 (27%) | 21 (48%) | 0.03 |
| Rescue therapy | 57 (72%) | 34 (77%) | 0.68 |
| Time from MV to ECMO, d | 2 (1–5) | 2 (1–6) | 0.51 |
| V-V ECMO | 71 (90%) | 36 (82%) | 0.32 |
| Pre-ECMO | |||
| Tidal volume, ml/kg PBW | 6.4 (5.8–7.4) | 6.7 (5.8–7.2) | 0.96 |
| Respiratory rate, min−1 | 27 (5) | 28 (7) | 0.33 |
| PEEP, cm H2O | 13 (4) | 12 (4) | 0.13 |
| Plateau pressure, cm H2O | 32 (5) | 32 (5) | 0.86 |
| Pao2/Fio2 ratio | 62 (17) | 65 (27) | 0.58 |
| Sao2, % | 84 (11) | 84 (11) | 0.80 |
| Arterial pH | 7.27 (0.12) | 7.24 (0.14) | 0.32 |
| Paco2, mm Hg | 60 (14) | 58 (20) | 0.75 |
| Arterial lactate, mmol/L | 2.4 (2) | 4.1 (4.4) | 0.06 |
| Driving pressure† | 19 (7) | 20 (6) | 0.33 |
| LIS | 3.5 (0.5) | 3.3 (0.5) | 0.27 |
| First day on ECMO | |||
| Tidal volume, ml/kg PBW | 3.6 (2.7–4.5) | 4.1 (3.5–5.4) | 0.07 |
| Respiratory rate, min−1 | 19 (8) | 20 (8) | 0.51 |
| PEEP, cm H2O | 13 (4) | 12 (4) | 0.93 |
| Plateau pressure, cm H2O | 25 (3) | 29 (5) | <0.01 |
| Pao2/Fio2 ratio | 105 (66) | 116 (88) | 0.53 |
| Sao2, % | 95 (4) | 94 (6) | 0.36 |
| Arterial pH | 7.40 (0.09) | 7.37 (0.16) | 0.20 |
| Paco2, mm Hg | 37 (9) | 39 (9) | 0.26 |
| Arterial lactate, mmol/L | 2.7 (2.1) | 7 (8.5) | <0.01 |
| Driving pressure† | 13 (4) | 16 (7) | 0.03 |
| Delta plateau pressure | −7 (6) | −3 (5) | <0.01 |
| Complications and outcome | |||
| Nosocomial pneumonia | 45 (64%) | 24 (58%) | 0.69 |
| Length of ECMO, d | 12 (8–23) | 10 (4–21) | 0.1 |
| Length of MV, d | 32 (23–53) | 17.5 (9.5–32) | <0.01 |
| Length of ICU stay, d | 44 (29–67) | 16.5 (9.5–34) | <0.01 |
Figure 2. Boxplots of arterial lactate and respiratory parameters according to intensive care unit outcome for the 123 patients who received extracorporeal membrane oxygenation (ECMO) support. Top left: Plateau pressure (Pplat) on the first day under ECMO. Top right: Pplat under ECMO − Pplat before ECMO. Bottom left: Driving pressure on the first day under ECMO. Bottom right: Arterial lactate on the first day under ECMO.
[More] [Minimize]The 103 patients who received ECMO within the first week of mechanical ventilation were compared with 157 patients who had severe ARDS criteria but did not receive ECMO (non-ECMO group). These two groups differed in many respects (Table 3 “before matching”). Patients receiving ECMO were younger, included a higher proportion of pregnant women and obese patients, had less comorbidities and immune suppression and less bacterial infection on admission, and a lower proportion received early steroid treatment. They had more organ failures and a more severe pulmonary failure (with a higher Lung Injury Score, a lower Pao2/Fio2 ratio, and higher Pplat). In-ICU mortality rate was similar between the ECMO and non-ECMO groups (36% vs. 34%; P = 0.90). Matching (without replacement) on the PS allowed selecting only 52 unique pairs of patients with similar medical history and initial severity, minimizing all baseline differences (Table 3,“after matching”). Using this well-matched subgroup of patients, we found no significant difference in ICU mortality (40% in the conventional treatment group vs. 50% in the ECMO group; OR for death of ECMO patients, 1.48; 95% CI, 0.68–3.23; P = 0.32).
| Before Matching | After Matching | |||||
| Non-ECMO Patients (n = 157) | ECMO Patients (n = 103) | P Value | Non-ECMO Patients (n = 52) | ECMO Patients (n = 52) | P Value | |
| Baseline characteristics | ||||||
| Age, yr | 48 (15) | 42 (13) | <0.01 | 45 (15) | 45 (13) | 0.86 |
| Male sex | 83 (53%) | 49 (48%) | 0.41 | 29 (56%) | 30 (58%) | 0.83 |
| MacCabe 1† | 112 (71%) | 92 (89%) | <0.01 | 39 (75%) | 45 (87%) | 0.24 |
| Risk factor for influenza complication | 119 (76%) | 81 (79%) | 0.70 | 39 (75%) | 37 (71%) | 0.80 |
| Pregnancy or postpartum | 3 (2%) | 16 (16%) | <0.01 | 2 (4%) | 3 (6%) | 1.00 |
| BMI, kg/m2 | 29 (9) | 32 (9) | 0.02 | 31 (11) | 30 (7) | 0.44 |
| Immunosuppression | 49 (31%) | 18 (17%) | 0.02 | 16 (31%) | 12 (23%) | 0.48 |
| On admission | ||||||
| SAPS3 score | 58 (17) | 60 (14) | 0.31 | 58 (18) | 60 (14) | 0.29 |
| SOFA score | 8.6 (4) | 9.8 (4.2) | 0.05 | 9.5 (4) | 9.6 (4.6) | 0.98 |
| Bacterial coinfection | 55 (35%) | 22 (21%) | 0.02 | 10 (19%) | 13 (25%) | 0.81 |
| Shock | 113 (72%) | 77 (75%) | 0.72 | 35 (67%) | 40 (77%) | 0.54 |
| Before Day 7 or before ECMO | ||||||
| Steroids | 74 (47%) | 32 (31%) | 0.01 | 23 (44%) | 24 (46%) | 1.00 |
| Rescue therapy | 67 (42%) | 84 (82%) | <0.01 | 36 (69%) | 40 (77%) | 0.54 |
| TV, ml/kg PBW | 7 (1.2) | 6.7 (1.6) | 0.18 | 6.8 (1.1) | 6.6 (1.4) | 0.40 |
| Respiratory rate, min−1 | 28 (5) | 27 (6) | 0.13 | 28 (5) | 28 (6) | 0.80 |
| PEEP, cm H2O | 13 (3) | 13 (4) | 0.65 | 13 (3) | 13 (4) | 0.49 |
| Plateau pressure, cm H2O | 29 (4) | 32 (6) | <0.01 | 31 (5) | 31 (5) | 0.24 |
| Pao2/Fio2 ratio | 83 (27) | 63 (22) | <0.01 | 68 (20) | 70 (26) | 0.76 |
| Sao2, % | 90 (7) | 83 (11) | <0.01 | 88 (8) | 87 (9) | 0.99 |
| Arterial pH | 7.27 (0.12) | 7.26 (0.12) | 0.39 | 7.25 (0.16) | 7.24 (0.13) | 0.75 |
| Paco2, mm Hg | 55 (14) | 54 (16) | 0.8 | 55 (16) | 56 (15) | 0.97 |
| Lactate, mM | 3.7 (5.4) | 3.1 (3.1) | 0.36 | 4.6 (6) | 3.4 (2.9) | 0.63 |
| LIS | 3.2 (0.6) | 3.6 (0.4) | <0.01 | 3.3 (0.7) | 3.3 (0.7) | 1.00 |
| Complications and outcome | ||||||
| Length of MV, d | 17 (9–25) | 25 (14–40) | <0.01 | 13.5 (7–21) | 22 (11.7–35) | <0.01 |
| Length of ICU stay, d | 21 (12–32) | 32 (15.25–56) | <0.01 | 19.5 (9–26) | 27 (12–52) | 0.04 |
| Mortality | 54 (34%) | 37 (36%) | 0.90 | 21 (40%) | 26 (50%) | 0.44 |
Of note, the 51 ECMO patients excluded from this analysis because of the absence of adequately matched non-ECMO patients (Table 4) were younger (P < 0.01), had a higher BMI (P = 0.03), included a higher proportion of pregnant women (P = 0.01), and less frequently received early steroids (P < 0.01) than those retained in the matched cohort; they also had lower Pao2/Fio2 (P < 0.01) and Sao2 (P < 0.01), and a higher Pplat before ECMO (P < 0.01), but had a twice lower ICU mortality (50% vs. 22%; P < 0.01).
| Selected for Matching (52) | Not Selected for Matching (51) | P Value | |
| Baseline characteristics | |||
| Age, yr | 45 (13) | 38 (13) | <0.01 |
| Male sex | 30 (58%) | 19 (37%) | 0.06 |
| MacCabe 1* | 45 (87%) | 47 (92%) | 0.55 |
| Risk factor for flu complication | 37 (71%) | 44 (86%) | 0.10 |
| Pregnancy or postpartum | 3 (6%) | 13 (25%) | 0.01 |
| BMI, kg/m2 | 30 (8) | 33 (10) | 0.03 |
| Obesity | 17 (33%) | 29 (57%) | 0.04 |
| Immunosuppression | 12 (23%) | 6 (12%) | 0.21 |
| On admission | |||
| SAPS3 score | 61 (14) | 58 (14) | 0.28 |
| SOFA score | 9.6 (4.8) | 10 (3.7) | 0.70 |
| Bacterial coinfection | 13 (25%) | 9 (18%) | 0.50 |
| Shock | 40 (77%) | 37 (73%) | 0.78 |
| Before ECMO | |||
| Steroids | 24 (46%) | 8 (16%) | <0.01 |
| Rescue therapy | 40 (77%) | 44 (86%) | 0.33 |
| Time from MV to ECMO | 2 (1–4) | 1 (0–3.5) | 0.47 |
| V-V ECMO | 44 (85%) | 44 (86%) | 0.97 |
| Pre-ECMO | |||
| Tidal volume, ml/kg PBW | 6.6 (1.4) | 6.8 (1.8) | 0.52 |
| Respiratory rate, min−1 | 28 (6) | 26 (5) | 0.22 |
| PEEP, cm H2O | 13 (4) | 13 (3) | 0.86 |
| Plateau pressure, cm H2O | 31 (5) | 33 (6) | 0.03 |
| Pao2/Fio2 ratio | 70 (26) | 54 (13) | <0.01 |
| Sao2, % | 87 (9) | 80 (11) | <0.01 |
| pH | 7.24 (0.13) | 7.26 (0.13) | 0.40 |
| Paco2, mm Hg | 56 (15) | 52 (17) | 0.22 |
| Lactate, mM | 3.4 (2.9) | 2.9 (3.3) | 0.42 |
| Driving pressure | 21 (7) | 18 (6) | 0.03 |
| LIS | 3.4 (0.6) | 3.6 (0.4) | 0.19 |
| First day under ECMO | |||
| Tidal volume, ml/kg PBW | 4 (1.5) | 3.7 (1.3) | 0.17 |
| Respiratory rate, min−1 | 19 (8) | 20 (7) | 0.84 |
| PEEP, cm H2O | 13 (5) | 12 (4) | 0.22 |
| Plateau pressure, cm H2O | 26 (4) | 27 (5) | 0.52 |
| Pao2/Fio2 ratio, mm Hg | 108 (54) | 109 (76) | 0.97 |
| Sao2, % | 95 (5) | 95 (4) | 0.95 |
| Arterial pH | 7.36 (0.13) | 7.38 (0.10) | 0.46 |
| Paco2, mm Hg | 38 (10) | 38 (8) | 0.91 |
| Lactate, mmol/L | 5 (5.1) | 4.27 (6.8) | 0.54 |
| Driving pressure, cm H2O† | 13.5 (5) | 14 (5) | 0.46 |
| Complications and outcome | |||
| Nosocomial pneumonia, n | 32 (61%) | 22 (43%) | 0.09 |
| Length of ECMO, d | 9 (7–18) | 13 (9–23) | 0.31 |
| Length of MV, d | 22 (12–35) | 30 (15–42) | 0.11 |
| Length of ICU stay, d | 27 (11–52) | 34.5 (21–58) | 0.72 |
| In-ICU mortality, n | 26 (50%) | 11 (22%) | <0.01 |
To explain the differences between our results and those of a recent similar analysis (6), we repeated the matching using different techniques (see online supplement). In a complementary PS analysis using a matching procedure with replacement similar to that used in the UK study (6), 102 of the 103 ECMO-treated patients could be retained in the analysis; because of repeated allocations, however, only 58 unique non–ECMO-treated patients were then used for matching (see Table E3 and Figure E4). In this analysis, treatment with ECMO seemed to be associated with a significantly lower risk of death (OR, 0.45; 95% CI, 0.25–0.78; P < 0.01).
To our knowledge, this is the largest cohort of influenza A(H1N1)pdm–related ARDS treated with ECMO. We found several new and relevant findings: (1) in addition to age, higher lactate levels on the first day of ECMO treatment was associated with increased mortality, whereas a reduced Pplat the first day under ECMO was significantly associated with survival; (2) in a one-to-one matched paired propensity adjusted analysis, overall survival did not differ between ECMO-treated patients and those managed conventionally; and (3) the ECMO-treated patients who could not be matched to non–ECMO-treated patients were younger, had more severe respiratory failure, and had a markedly lower mortality.
Similarly to previous cohorts of patients infected with influenza A(H1N1)pdm treated with ECMO (4–6), patients were young, obese, previously healthy, and had a rapidly extensive viral pneumonia with extremely severe initial pulmonary injury. Nevertheless, some differences in management deserve to be highlighted: more French patients received inhaled nitric oxide or prone positioning (72% and 45%, respectively; 43% received both) before ECMO than Australian (4) (32% and 20%), UK (6) (19% and 34%), or Italian (5) patients (15% and 28%), suggesting that French intensivists considered ECMO as a last-line rescue therapy. Whereas other countries relied on 4 (UK) or 15 (Australia and New Zealand) referral centers, more than 30 different centers provided ECMO therapy in France, which was possibly associated with a higher heterogeneity in patients’ management; however, overall mortality did not differ between the most experienced French centers and those that cared for fewer ECMO patients (data not shown).
Pre-ECMO ventilator settings in our cohort (Table 1) show that physicians generally followed recommendations for maximal protective ventilation and ventilator settings algorithms issued by French scientific societies (24) soon after the pandemic started (including a low tidal volume, a high respiratory rate to maintain pH between 7.30 and 7.45, and high PEEP levels to allow for a maximal Pplat of 28–30 cm H2O). Ventilation settings under ECMO were also consistent with recommendations (Table 1): tidal volume and respiratory rate were lowered, resulting in a lower Pplat while PEEP level was maintained.
More than half ECMO patients experienced at least one device-related complication (see Table E1). The rate of intracranial bleeding (4%) in our patients is higher than the 1% rate recorded in the Italian series (5), but lower than the 11% rate reported in the Australian or UK cohorts (4, 6). Comparison of these bleeding rates is difficult because the anticoagulation strategy and monitoring were not reported in any of these cohorts. Experts now recommend targeting a relatively low level of anticoagulation, but no guideline is available to date (25).
We identified older age, higher lactate, and Pplat under ECMO as all associated with increased odds of ICU death in ECMO recipients. Age was consistently found predictive of a poor outcome in several studies (26, 27), leading most teams to restrict ECMO treatment to the youngest patients. In one study of 42 patients receiving ECMO for cardiac failure, lactate measured after 48 hours of ECMO therapy was associated with mortality (28); persistently severe hypoxia or circulatory failure on the first ECMO day may account for this finding in our patients. Importantly, Pplat under ECMO was the only ventilator parameter associated with ICU death. Optimal ECMO settings with large cannulas and high circuit blood flows permitting full respiratory support might have allowed physicians to apply a maximal protective ventilation hence lowering Pplat to below 25 cm H2O, which in itself was beneficial on outcome of patients; indeed, it has been suggested that reducing Pplat below the accepted safe range of 28–30 cm H2O might enhance lung protection and reduce ARDS-associated mortality (29, 30).
Our major objective was to estimate the effect of ECMO treatment on patients’ outcome. Although a RCT to study the potential benefit of ECMO for severe ARDS caused by influenza A(H1N1)pdm did not seem feasible for logistical and ethical reasons, a PS-matched analysis appeared to be a suitable alternative. This well-validated method allows minimizing the biases and replicates a RCT by creating two groups of patients (treated or not) as similar as possible on measured confounders (19, 31). The matching procedure used in this study (matching without replacement using a caliper of width of 0.2 of the standard deviation of the logit of the PS) is considered to minimize differences between treatment groups and thus selection bias in estimating treatment effect (18, 19). Interestingly, this procedure allowed matching only 52 of our ECMO patients (50% of those receiving ECMO during the first week) to a unique similarly severe non-ECMO patient. Patients not selected for the matched analysis were younger and had more severe respiratory failure (Table 4). Despite their extremely severe respiratory failure, this subgroup had a twice lower ICU mortality than the ECMO subgroup selected for the matched analysis (22% vs. 50%; P < 0.01). Thus, ECMO patients not selected by our matching procedure seem to define a subgroup that may have actually benefited most from ECMO, although this could not be formally tested.
Our finding of a lack of survival difference in the matched treated and untreated groups differs from the results of the UK study based on a similar approach (6). First, this points out that PS analysis can be done in several ways (32) (see online supplement), and that clearly reporting the whole process of the PS approach in published studies should be done. There are several important methodologic differences between the two analyses: we included more covariates in the PS and used more stringent matching parameters (narrower caliper and matching without replacement). To better understand the differences between these two studies, we repeated the analysis with replacement (i.e., reusing the same controls several times), and this procedure provided very similar results to the UK study, with an odds ratio of survival under ECMO of 0.45 (95% CI, 0.25–0.78; P < 0.01) (see online supplement). By using more ECMO-treated patients, and the same severe non-ECMO patients, this latter procedure may artificially favor the ECMO group.
In this study describing the largest cohort of ECMO-treated patients for influenza A(H1N1)pdm–related ARDS, we included almost all French patients treated with this procedure. Some of these patients have already been included in previous publications but none of these describe the whole ECMO population treated in France (7, 8, 33–36). Data were collected prospectively and pre-ECMO characteristics were consistent with previous published data (4, 5). Although every effort was made to minimize missing data by contacting all ICUs, some remained missing (see Table E1), but multiple imputation of missing data was used to determine the propensity PS and for the sensitivity analysis. The results of our multivariate and the matched analyses seemed robust, however, because all the different models performed with different sets of covariates concluded the same results. Our analysis was limited to in-ICU mortality because hospital mortality was not available for every patient after ICU discharge, but the two main published cohorts on influenza A(H1N1)pdm ECMO patients from Australia and Italy have shown identical mortality at ICU and hospital discharge (4, 5).
Data from non-ECMO patients were not collected in the 2010–2011 winter to minimize physicians’ workload. Nevertheless, ICU patients with severe influenza had similar characteristics during the 2 consecutive years of the study (37).
The PS-matched analysis decreased our sample size from 103 to 52 ECMO patients comparable with 52 control patients. As other countries reported their experience on ECMO for H1N1-related ARDS (4–6), an international study linking available datasets could increase the capability to better define the potential benefit of this technique in a larger population.
Our results may not be generalizable to all patients with ARDS from various causes. Indeed, the effectiveness and safety of venovenous ECMO in patients with ARDS remains an open question (38). Whether these patients should be treated by such an invasive technique still remains debatable and more information will be brought in the near future by ongoing studies (EOLIA; ClinicalTrial.gov Identifier NCT 01470703).
In summary, our results indicate that an ultraprotective ventilation strategy with reduction of Pplat to around 25 cm H2O from the first day under ECMO may be required to improve survival. We could not demonstrate a benefit of ECMO on ICU survival when compared with conventional management in matched patients of similar severity; however, only 50% of ECMO patients were successfully matched. A specific subgroup of young patients with severe respiratory failure receiving ECMO remained unmatched and had a more favorable outcome. Whether such a better outcome may be related to ECMO is still an open issue because it may also be caused solely by patient selection.
The authors are grateful to Dr Isabelle Bonmarin (Institut de Veille Sanitaire, St Maurice, France) for her help in checking the exhaustiveness of ECMO cases. They thank all the medical and nursing staff of participating centers.
Collaborators (REVA Research Network) are as follows:
ECMO centers (CHU = university hospital; CH = non-university hospital):
CHU d’Amiens: E. Zogheib, H. Dupont
CHU d’Angers: A. Mercat, M. Pierrot
CHU J. Minjoz, Besançon: G. Capellier
CHU L. Pradel, Bron: B. Verdiere, O. Bastien, J.-J. Lehot
CHU de Brest: J. M. Tonnelier
CHU Clermont-Ferrand: D. Guelon
CHU Côte de Nacre, Caen: D. du Cheyron
CHU Henri Mondor, Créteil: A. Thille, T. Pham, C. Brun-Buisson
CHU Dijon: P. Quenot
CHU A. Michallon, Grenoble (Medical ICU): C. Arasomoano, C. Minet, J.-F. Timsit
CHU A. Michallon, Grenoble (Surgical ICU): G. Dessertaine
CHU de Lille: L. Robriquet, E. Jaillette
CHU Marseille: M. Castanier, A. Roch, L. Papazian
CHU Arnaud de Villeneuve, Montpellier: J. Eliet, P. Gaudard, S. Machado
CHG Emile Muller, Mulhouse: K. Kuteifan
CHU Nancy: A. Kimmoun, B. Levy
CHU Nantes: P. Bizouarn, D. Villiers, C. Guitton
CHU Archet 1, Nice: J. Dellamonica
CHR Orléans: T. Boulain
CHU Bichat Claude Bernard, Paris: B. Mourvillier, J. Bailly Salin, M. Wolff
CHU Cochin, Paris: J. Charpentier, J.-D. Chiche, S. Ricome
CHU Lariboisière, Paris: B. Megarbane
CHU Pitié Salpetrière, Paris: A. Combes, M. Schmidt, N. Bréchot, J. Chastre
CHU Haut Lévêque, Pessac: H. Rozé, A. Ouattara
CHU Poitiers: R. Robert, E. Carise
CHD Les Oudaries, La Roche s/Yon: J. Reignier
CHU Charles Nicolle, Rouen: G. Beduneau, J.-C.M. Richard
CHU Pontchaillou, Rennes: Y. Le Tulzo
CHU St. Etienne: F. Zeni, R. Jospe
CHU de Strasbourg: C. Kummerlen, J. P. Gouello
Hôpital Foch, Suresnes: C. Cerf, J. Devaquet
CHU Purpan, Toulouse: B. Riu-Poulenc, A. Luzi
CHU Rangueil, Toulouse: B. Georges, N. Mayeur
CHU Bretonneau, Tours: E. Mercier, A. Guillon
CH Bretagne Atlantique, Vannes: M. L. Eustache
Non-ECMO centers (CHU = university hospital; CH = non-university hospital):
Hôpital de Sallanches: L. Monon, S. Cagnin, J. Ronda
CH d’Agen: A. Khechache, B. Morteau
CH d’Ajaccio: B. Lecomte
CH d’Alençon: K. Merouani
CH d’Annecy: M. Siradot
CH d’Antibes Juan les Pins: F. Tiger
Hôpital Privé d’Antony: M. Benhamou
CH d’Armantières: C. Canevet
CH R. Ballanger, Aulnay sous bois: F. Santoli
CH d’Avignon: K. Debbat
CHI Oise, Beauvais: A.-M. Guerin
CH de Bézier: L. Favier
CHU Pellegrin, Bordeaux: G. Hilbert, Y. Castaing
CHU Saint André, Bordeaux: J. Youssef
CHU A. Paré, Boulogne Billancourt: C. Charron
Clinique J. Villar, Bruges: P. Ragot
CH de Cannes: A. Freche
CH de Carpentras: B. Colin
CHU L. Mourier, Colombes: J.-D. Ricard
CH de Douai: C. Boulle Geronimi
CHSF, Evry: Y. Abadie
CH Val d’Ariège, Foix: J. Desbordes
CHU de Garches: M. Antona
CH de Hyere: C. Cheval
CH de la Dracerie, Draguignan: R. Farhat
CH du Havre: M. Bousta
CH de Laon: J.-C. Guilbaud
CHU R. Salengros, Lille: F. Fourrier
CHU de Limoges: H. Gastinne, M. Clavel, P. Vignon
CH de Lons le Saulnier: A. Gill
CH de Bretagne Sud, Lorient: P. Quiniot
CHU Croix Rousse, Lyon: C. Guerin
CH Saint Luc, Lyon: S. Rosselli
CHU Lyon Sud: V. Piriou
CHU E. Herriot, Lyon: L. Argaud
CH du Mans: F. Grelon
CH F. Quesnay, Mantes
CH de Martigues: P. Courtin
CH de Montauban: S. Vimieux
CH de Mont de Marsan: M. Pascal, A. Karcenty
CHI A. Grégoire, Montreuil: M. Daumal
CHU de Nancy: P.-E. Bollaert
CH de Nîmes: C. Bengler
CHG de Niort: J. Voultoury
CH d’Orléans: T. Boulain
CH d’Orsay: N. Rezgui
CHU Hotel Dieu, Paris: A. Rabat
CHU Pitié Salpétrière: A. Duguet
CHU G. Pompidou, Paris: C. Faizy
CHU St. Antoine, Paris: B. Guidet
CHU St. Louis, Paris: V. Lemiale, E. Azoulay
CHU Tenon, Paris: M. Fartoukh
CHG F. Mitterand, Pau: P. Badia
CH de Périgueux: Y. Monseau
CH de Pontoise: J. Richecoeur
CH L. Binet, Provins: T.-H. Chau
CH de Roanne: P. Beuret
CH de Rodez: A. Delahaye
CH de Saint Bieux: G. Guivarch
CH de Saint Germain: J.-C. Lachérade
CH de Salon de Provence: J. Theodore
CH G. Ramon, Sens: D. Tonduangu
CHU de Strasbourg: P. Lutun
CH Bigorre, Tarbes: T. Dulac, C. Boubien
CH Font pré, Toulon: J.-M. Arnal
CH de Clocheville, Tours: S. Cantagrel
CH de Trappes: B. Granclerc
CH de Vesoul: C. Floriot
CH Mignot Versailles: M. Henry-Lagarrigue
| 1. | Zapol WM, Snider MT, Hill JD, Fallat RJ, Bartlett RH, Edmunds LH, Morris AH, Peirce EC, Thomas AN, Proctor HJ, et al.. Extracorporeal membrane oxygenation in severe acute respiratory failure: a randomized prospective study. JAMA 1979;242:2193–2196. |
| 2. | Morris AH, Wallace CJ, Menlove RL, Clemmer TP, Orme JF, Weaver LK, Dean NC, Thomas F, East TD, Pace NL, 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. |
| 3. | Peek GJ, Mugford M, Tiruvoipati R, Wilson A, Allen E, Thalanany MM, Hibbert CL, Truesdale A, Clemens F, Cooper N, et al.. 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;374:1351–1363. |
| 4. | Davies A, Jones D, Bailey M, Beca J, Bellomo R, Blackwell N, Forrest P, Gattas D, Granger E, Herkes R, et al.. Extracorporeal membrane oxygenation for 2009 influenza A(H1N1) acute respiratory distress syndrome. JAMA 2009;302:1888–1895. |
| 5. | Patroniti N, Zangrillo A, Pappalardo F, Peris A, Cianchi G, Braschi A, Iotti GA, Arcadipane A, Panarello G, Ranieri VM, et al.. The Italian ECMO network experience during the 2009 influenza A(H1N1) pandemic: preparation for severe respiratory emergency outbreaks. Intensive Care Med 2011;37:1447–1457. |
| 6. | Noah MA, Peek GJ, Finney SJ, Griffiths MJ, Harrison DA, Grieve R, Sadique MZ, Sekhon JS, McAuley DF, Firmin RK, et al.. Referral to an extracorporeal membrane oxygenation center and mortality among patients with severe 2009 influenza A(H1N1). JAMA 2011;306:1659–1668. |
| 7. | Fuhrman C, Bonmarin I, Bitar D, Cardoso T, Duport N, Herida M, Isnard H, Guidet B, Mimoz O, Richard JCM, et al.. Adult intensive-care patients with 2009 pandemic influenza A(H1N1) infection. Epidemiol Infect 2011;139:1202–1209. |
| 8. | Brun-Buisson C, Richard J-CM, Mercat A, Thiébaut ACM, Brochard L. Early corticosteroids in severe influenza A/H1N1 pneumonia and acute respiratory distress syndrome. Am J Respir Crit Care Med 2011;183:1200–1206. |
| 9. | Pham T, Combes A, Rozé H, Richard J-CM, Chevret S, Mercat A, Bastien O, Roch A, Brun-Buisson C, Brochard L. Extra-corporeal membrane oxygenation (ECMO) for influenza A(H1N1) induced acute respiratory distress: preliminary results of a pairwise-matched propensity based analysis. Am J Respir Crit Care Med 2012;185:A6013. |
| 10. | Pham T, Combes A, Richard J-CM, Rozé H, Chevret S, Roch A, Bastien O, Mercat A, Brun-Buisson C, Brochard L. Extra-corporeal membrane oxygenation (ECMO) for influenza A(H1N1) induced acute respiratory distress syndrome (ARDS): analysis of the factors associated with death in 122 French patients. Am J Respir Crit Care Med 2012;185:A6019. |
| 11. | Bernard GR, Artigas A, Brigham KL, Carlet J, Falke K, Hudson L, Lamy M, Legall JR, Morris A, Spragg R. The American-European Consensus Conference on ARDS. Definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994;149:818–824. |
| 12. | Murray JF, Matthay MA, Luce JM, Flick MR. An expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988;138:720–723. |
| 13. | CDC. Seasonal influenza (flu): people at high risk of developing flu-related complications [accessed 2011 Aug 21]. Available from: http://www.cdc.gov/flu/about/disease/high_risk.htm |
| 14. | Metnitz PGH, Moreno RP, Almeida E, Jordan B, Bauer P, Campos RA, Iapichino G, Edbrooke D, Capuzzo M, Le Gall J-R. SAPS 3–From evaluation of the patient to evaluation of the intensive care unit. Part 1: Objectives, methods and cohort description. Intensive Care Med 2005;31:1336–1344. |
| 15. | Moreno RP, Metnitz PGH, Almeida E, Jordan B, Bauer P, Campos RA, Iapichino G, Edbrooke D, Capuzzo M, Le Gall J-R. SAPS 3–From evaluation of the patient to evaluation of the intensive care unit. Part 2: Development of a prognostic model for hospital mortality at ICU admission. Intensive Care Med 2005;31:1345–1355. |
| 16. | Vincent JL, Moreno R, Takala J, Willatts S, De Mendonça A, Bruining H, Reinhart CK, Suter PM, Thijs LG. The SOFA (sepsis-related organ failure assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996;22:707–710. |
| 17. | Ho DE, Imai K, King G, Stuart EA. Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Political Analyses 2007;15:199–236. |
| 18. | Austin PC. Some methods of propensity-score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations. Biom J 2009;51:171–184. |
| 19. | Gayat E, Pirracchio R, Resche-Rigon M, Mebazaa A, Mary J-Y, Porcher R. Propensity scores in intensive care and anaesthesiology literature: a systematic review. Intensive Care Med 2010;36:1993–2003. |
| 20. | Mebane W, Sekhon JS. Genetic optimization using derivatives: the rgenoud package for R. J Stat Softw 2011;42:1–26. |
| 21. | Sekhon JS, Grieve RD. A matching method for improving covariate balance in cost-effectiveness analyses. Health Econ 2012;21:695–714. |
| 22. | Sekhon JS. Multivariate and propensity score matching software with automated balance optimization: the matching package for R. J Stat Softw 2011;42:1–52. |
| 23. | von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med 2007;4:e296. |
| 24. | Mercat A, Richard J-CM, Combes A, Chastre J, Ricard JD, Dreyfuss D, Brochard L. SDRA lié à la grippe A (H1N1)-2009 recommandations pour l'assistance respiratoire Available from: http://www.srlf.org/rc/org/srlf/htm/Article/2011/20110903-134333-752/src/htm_fullText/fr/20090927_REVAGRIP_SDRA_grippe_A_H1N1_Recommandations_assistance_respiratoire.pdf |
| 25. | Combes A, Bacchetta M, Brodie D, Müller T, Pellegrino V. Extracorporeal membrane oxygenation for respiratory failure in adults. Curr Opin Crit Care 2012;18:99–104. |
| 26. | 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;35:2105–2114. |
| 27. | Hemmila MR, Rowe SA, Boules TN, Miskulin J, McGillicuddy JW, Schuerer DJ, Haft JW, Swaniker F, Arbabi S, Hirschl RB, et al.. Extracorporeal life support for severe acute respiratory distress syndrome in adults. Ann Surg 2004;240:595–605, discussion 605–607. |
| 28. | Formica F, Avalli L, Colagrande L, Ferro O, Greco G, Maggioni E, Paolini G. Extracorporeal membrane oxygenation to support adult patients with cardiac failure: predictive factors of 30-day mortality. Interact Cardiovasc Thorac Surg 2010;10:721–726. |
| 29. | Hager DN, Krishnan JA, Hayden DL, Brower RG. Tidal volume reduction in patients with acute lung injury when plateau pressures are not high. Am J Respir Crit Care Med 2005;172:1241–1245. |
| 30. | Terragni PP, Del Sorbo L, Mascia L, Urbino R, Martin EL, Birocco A, Faggiano C, Quintel M, Gattinoni L, Ranieri VM. Tidal volume lower than 6 ml/kg enhances lung protection: role of extracorporeal carbon dioxide removal. Anesthesiology 2009;111:826–835. |
| 31. | Stuart EA, Marcus SM, Horvitz-Lennon MV, Gibbons RD, Normand S-LT. Using non-experimental data to estimate treatment effects. Psychiatr Ann 2009;39:41451. |
| 32. | Austin PC. A critical appraisal of propensity-score matching in the medical literature between 1996 and 2003. Stat Med 2008;27:2037–2049. |
| 33. | Dubar G, Azria E, Tesnière A, Dupont H, Le Ray C, Baugnon T, Matheron S, Luton D, Richard J-C, Launay O, et al.. French experience of 2009 A/H1N1v influenza in pregnant women. PLoS ONE 2010;5:e13112. |
| 34. | Roch A, Lepaul-Ercole R, Grisoli D, Bessereau J, Brissy O, Castanier M, Dizier S, Forel J-M, Guervilly C, Gariboldi V, et al.. Extracorporeal membrane oxygenation for severe influenza A (H1N1) acute respiratory distress syndrome: a prospective observational comparative study. Intensive Care Med 2010;36:1899–1905. |
| 35. | Mercat A, Pham T, Rozé H, Cuquemelle E, Brun-Buisson C, Brochard L, Richard J-CM. Severe H1N1 2009 influenza infection in adults: the French experience. Reanimation 2011;20:162–168. |
| 36. | Richard J-CM, Pham T, Brun-Buisson C, Reignier J, Mercat A, Beduneau G, Régnier B, Mourvillier B, Guitton C, Castanier M, et al.. Interest of a simple on-line screening registry for measuring ICU burden related to an influenza pandemic. Crit Care 2012;16:R118. |
| 37. | Bonmarin I, Belchior E, Haeghebaert S, Servas V, Watrin M, Levy-Bruhl D. Severe cases of influenza admitted in intensive care units in France, 2010–2011. Bulletin Epidémiologique Hebdomadaire 2011;37–38:399. |
| 38. | Brodie D, Bacchetta M. Extracorporeal membrane oxygenation for ARDS in adults. N Engl J Med 2011;365:1905–1914. |
*A complete list of members may be found before the beginning of the References.
Author Contributions: Study concept and design, T.P., L.B., A. Combes, S.C., C.B.-B., A.M., and J.-C.M.R. Acquisition of data, T.P., C.B.-B., J.-C.M.R., and A. Constan. Analysis and interpretation of the data, T.P., S.C., H.R., L.B., and A. Combes. Drafting of the manuscript, T.P., A. Combes, H.R., S.C., J.-C.M.R., C.B.-B., and L.B. Revision of the manuscript: all authors. Statistical analysis, T.P. and S.C. Obtained funding, L.B. and A.M.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.201205-0815OC on November 15, 2012
Author disclosures are available with the text of this article at www.atsjournals.org.
