Acute respiratory distress syndrome (ARDS) develops from acute pneumonia, nonpulmonary sepsis, aspiration of gastric content, and major trauma. All of these pathologies induce severe inflammation of the lung that becomes clinically apparent as systemic hypoxia due to impairment of pulmonary gas exchange. The clinical correlates of this inflammatory process have been defined using three categories of severity of ARDS—mild, moderate, and severe—depending on the oxygenation level [
In cases where the sensitive equilibrium between safe ventilation and oxygenation is imbalanced, extracorporeal membrane oxygenation (ECMO) can ensure oxygenation and decarboxylation until the lung heals. EMCO can lead to less invasive mechanical ventilation and reduces the risk of additional damage to the injured lungs. This risk reduction is achieved by establishing an extracorporeal circuit for venovenous ECMO (VV-ECMO). The use of EMCO has steadily increased recently [
After ECMO initiation, the value of the
All data were collected retrospectively from the medical records of the University Hospital of Tübingen. The study was approved by the ethics committee of the University Hospital Tübingen (768/2018BO2), which waived the need for informed consent, because patient anonymity was maintained. All the methods were approved by the local IRB and performed in accordance with the Declaration of Helsinki and the relevant guidelines. All patients treated with ECMO for ARDS at the Department for Anesthesiology and Intensive Care Medicine between December 2014 and May 2018 were retrospectively screened for inclusion into this observational cohort study.
Patients were admitted to our institution either directly or via secondary transfer from other hospitals for ARDS treatment. The standard of care at our institution recommends ECMO as a rescue therapy for ARDS when a patient requires extensive invasive mechanical ventilation that exceeds the recommendations of the respective clinical guidelines [
VV-ECMO placement was performed under echocardiographic guidance [
During the first 48 hours after VV-ECMO was initiated, all patients received sedation targeting a Richmond Agitation and Sedation Scale of −5. An ultra-low tidal volume ventilation was the main therapeutic target after VV-ECMO placement per institutional protocol. Using a pressure-controlled ventilation mode, we aimed for a tidal volume of 3.5 ml/kg ideal body weight in all patients. Ideally, this was achieved using an inspiratory plateau pressure ≤25 mbar and a positive end-expiratory pressure arbitrarily set to 15 mbar.
In a subset of patients, we analyzed the available echocardiographic images. For this, the picture archiving and communication system (PACS) of the University Hospital Tübingen was searched for echocardiographic images from both cohorts. The respective images were further analyzed, if recorded within 24 hours after ECMO implantation. All measurements were performed offline according to the current
For propensity score matching, we screened all patients admitted to our intensive care unit (ICU) from December 2014 to May 2018. Next, we selected patients who received the ICD-10 code J80 (
To compare survivors to nonsurvivors, a 1 : 1 matching was performed using the propensity score matching method based on the following variables that were expected to be associated with the ICU outcome: age, ARDS severity based on the Berlin definition [
To analyze the mechanics of the respiratory system, we extracted the ventilator measurements and settings from the
Continuous variables are expressed as the mean ± standard deviation and were compared using the Mann–Whitney
The matching process is depicted in Figure
Patient selection and matching strategy.
Demographic data and comorbidities.
Nonsurvivors ( |
Survivors ( |
| |
---|---|---|---|
|
|||
Age, yr (mean ± SD) | 54 ± 10 | 53 ± 15 | 0.7727 |
Male sex, no. (%) | 20 (71.4%) | 21 (75.0%) | >0.99 |
Height (cm) | 173 ± 10 | 175 ± 9 | 0.4469 |
Weight (kg) | 91 ± 30 | 91 ± 22 | 0.5050 |
Body Mass Index (kg/m2) | 30 ± 8 | 30 ± 6.0 | 0.6993 |
|
|||
|
|||
Diabetes mellitus | 6 (21.4%) | 7 (25.0%) | >0.99 |
Nicotine use | 10 (35.7%) | 10 (35.7%) | >0.99 |
Chronic obstructive pulmonary disease | 4 (14.3%) | 3 (10.7%) | >0.99 |
Chronic renal failure | 3 (10.7%) | 4 (14.3%) | >0.99 |
Autoimmune disease | 4 (14.3%) | 4 (14.3%) | >0.99 |
Arterial hypertension | 6 (21.4%) | 12 (43.0%) | 0.1516 |
History of malignancy | 3 (10.7%) | 6 (21.4%) | 0.4688 |
History of substance abuse (incl. alcohol) | 6 (21.4%) | 5 (17.9%) | >0.99 |
Peripheral atherosclerotic disease | 4 (14.3%) | 4 (14.3%) | >0.99 |
Coronary artery disease | 5 (17.8%) | 2 (7.1%) | 0.4216 |
Neurological disease | 6 (21.4%) | 7 (25.0%) | >0.99 |
The study groups did not differ in their major critical illness prediction scores, the duration of mechanical ventilation, length of ICU stay,
ICU patient variables.
Nonsurvivors ( |
Survivors ( |
| |
---|---|---|---|
Median hours of mechanical ventilation (h) | 442 (188–800) | 523 (327–765) |
|
SAPS II (mean ± SD) | 44.8 ± 10.7 | 44.9 ± 16.0 |
|
APACHE II (mean ± SD) | 25.3 ± 7.7 | 26.8 ± 13.4 |
|
SOFA | 11.9 ± 2.8 | 10.7 ± 2.7 |
|
Renal replacement therapy in ICU | 17 (60.7%) | 15 (53.6%) |
|
Median days length of ICU stay (interquartile range) | 18.5 (9–36.75) | 24 (15.5–32.00) |
|
Median days duration of ECMO (interquartile range) | 16 (7–28.75) | 18 (8.5–24.0) |
|
|
|||
|
|||
Median days of invasive ventilation (interquartile range) | 1.0 (1–5) | 1 (0–3) |
|
Median |
62 (48–89) | 67 (61–90) |
|
Median |
51 (47–60) | 51 (47–57) |
|
|
|||
|
|||
Median fluid balance (interquartile range) (ml) | 2125 (0–4017) | 1375 (131–3334) |
|
|
|||
|
|||
Median norepinephrine (interquartile range) ( |
|||
Before ECMO | 0.14 (0.00–0.33) | 0.1 (0–0.33) |
|
12 h after ECMO | 0.08 (0.00–0.24) | 0.09 (0.02–0.20) |
|
Median serum lactate (interquartile range) (mmol/l) | |||
Before ECMO | 2.0 (1.2–3.6) | 1.4 (0.9–3.0) |
|
12 h after ECMO | 1.9 (1.2–3.4) | 1.5 (1.0–2.6) |
|
Heart rate (beats/minute) | |||
Before ECMO | 106 ± 21 | 98 ± 20 |
|
12 h after ECMO | 91 ± 21 | 88 ± 18 |
|
Mean arterial pressure (mmHg) | |||
Before ECMO | 78 ± 13 | 80 ± 15 |
|
12 h after ECMO | 74 ± 11 | 72 ± 9 |
|
Next, we investigated whether the nonsurvivors and survivors diverged in the etiologies of ARDS. We categorized the suspected etiologies of ARDS based on the patients’ medical records. Our study cohort included many patients who developed primary ARDS because of pneumonia (40 of 56 patients, 66%). A difference was found between the survivors and nonsurvivors regarding the microbiological profile causing pneumonia; the survivors of ECMO developed ARDS mainly because of a viral pneumonia (
Etiologies of acute respiratory distress syndrome.
All patients ( |
Nonsurvivors ( |
Survivors ( |
| |
---|---|---|---|---|
Pneumonia | 37 | 16 | 21 | 0.4219 |
Bacterial infection | 16 | 10 | 6 | 0.3753 |
Viral infection | 21 | 6 | 15 |
|
Extrapulmonary bacterial infection | 4 | 3 | 1 | 0.6110 |
Aspiration of gastric content | 7 | 5 | 2 | 0.4216 |
Thorax trauma | 2 | 1 | 1 | >0.9999 |
Morbus Wegener | 2 | 1 | 1 | >0.9999 |
Acute pancreatitis | 4 | 2 | 2 | >0.9999 |
As patients outcome could have been influenced by acute or chronic dysfunction in myocardial contractility, we analyzed echocardiographic images taken within 24 hours after VV-ECMO. Unfortunately, we could only retrieve sufficient cardiac images in 23 patients (12 survivors and 11 nonsurvivors). Regarding these patients, right and left ventricular function did not differ between the two study groups (Supplementary
We investigated whether mechanical ventilation as a surrogate for the extent of respiratory mechanics varied between the groups. For this analysis, we compared the ventilatory parameters within 30 min before ECMO initiation and at 12 h after the beginning of ECMO (Table
Ventilator variables.
Nonsurvivors | Survivors |
| |
---|---|---|---|
|
( |
( |
|
Blood flow (l/min) | 4.0 | 4.6 |
|
Rotation per minute (rpm) | 4160 | 3427 |
|
Sweep gas flow (l/min) | 3.967 | 3.882 |
|
|
|||
|
( |
( |
|
Before ECMO | 32.2 ± 4.8 | 35.0 ± 6.7 |
|
12 h after ECMO | 27.1 ± 4.1 | 25.6 ± 3.2 |
|
|
|||
|
( |
( |
|
Before ECMO | 15.0 ± 3.4 | 15.4 ± 3.1 |
|
12 h after ECMO | 14.0 ± 2.8 | 15.3 ± 2.7 |
|
|
|||
|
( |
( |
|
Before ECMO | 22.7 ± 4.8 | 22.0 ± 4.1 |
|
12 h after ECMO | 20.2 ± 2.5 | 19.2 ± 3.3 |
|
|
|||
|
( |
( |
|
Before ECMO | 21 ± 5 | 23 ± 8 |
|
12 h after ECMO | 15 ± 6 | 14 ± 5 |
|
|
|||
|
( |
( |
|
Before ECMO | 5.7 ± 2.3 | 6.3 ± 1.6 |
|
12 h after ECMO | 3.3 ± 4.8 | 3.7 ± 1.8 |
|
|
|||
|
( |
( |
|
Before ECMO | 28 ± 11 | 25 ± 10 |
|
12 h after ECMO | 22 ± 13 | 31 ± 18 |
|
|
|||
|
( |
( |
|
Before ECMO | 16.9 ± 3.8 | 18.4 ± 6.3 |
|
12 h after ECMO | 13.0 ± 4.3 | 10.3 ± 3.0 |
|
Since ΔP has been reported recently as an outcome-predicting variable in ARDS [
Driving pressures in the survivors and nonsurvivors. (a) Driving pressures (Δ
Logistic regression analysis was performed to assess the impact of known markers of ICU survival on mortality (Table
Factors associated with mortality.
Univariate logistic regression | Multivariate logistic regression | |||
---|---|---|---|---|
Odds ratio (95% CI) |
|
Odds ratio (95% CI) |
| |
|
0.988 (0.97 to 1.01) | 0.2079 | ||
|
||||
|
||||
Before ECMO | 1.12 (0.89 to 1.41) | 0.3463 | ||
12 h after ECMO | 1.12 (0.92 to 1.37) | 0.2529 | ||
|
||||
|
||||
Bacterial infection versus noninfectious | 0.19 (0.04 to 0.74) | 0.0338 | 0.09 (0.01 to 0.97) |
|
Viral infection versus noninfectious | 0.19 (0.04 to 0.74) |
|
0.08 (0.01 to 0.42) |
|
Bacterial versus viral infection | 1.02 (0.24 to 4.22) | 0.9830 | ||
|
||||
|
||||
Before ECMO | 0.94 (0.84 to 1.05) | 0.2770 | ||
12 h after ECMO | 1.25 (1.05 to 1.48) |
|
||
Absolute change | 0.82 (0.71 to 0.94) |
|
||
Relative change (per 1% decrease) | 0.96 (0.93 to 0.99) |
|
|
|
Receiver-operating characteristic curve (ROC) for relative driving pressure change. ROC analysis for the relative change in driving pressure and survival of ECMO therapy; area under the curve (AUC) 0.75265 (chi2 = 11.27336;
As the different Δ
Using propensity score matching, we identified 28 pairs of surviving and nonsurviving ARDS patients who underwent VV-ECMO. To identify outcome-predicting parameters, we focused on mechanical ventilation parameters collected before and 12 hours after VV-ECMO initiation. Our main finding was that despite the use of low tidal volume ventilation in both groups, the survivors exhibited a change in Δ
Δ
To determine Δ
In our cohort, the predominant underlying cause of ARDS was pneumonia. In terms of the microbiological agent, we found a trend towards more bacterial pneumonia in the nonsurvivors. In contrast, the surviving patients developed significantly more ARDS based on viral pneumonia. This result is in line with that of a retrospective analysis by Schmidt et al. who found that ECMO for severe asthma and viral pneumonia was independently associated with hospital survival [
The current clinical guidelines for treatment and mechanical ventilation of ARDS patients focus on plateau pressure targets and tidal volumes corrected to the ideal body weight [
To enhance the comparability between the two groups, we used a propensity score matching approach similar to that of other studies [
In summary, we report that patients surviving ECMO undergo a drastic decrease in driving pressure within the first 12 hours of ECMO, whereas in nonsurvivors, driving pressure changes are strongly attenuated. Future research needs to clarify whether strategies to adjust ventilator settings in ECMO patients based on driving pressure calculations may improve patient outcomes.
All data included in this study are available from the corresponding author upon reasonable request.
Approval for this study was obtained from the local institutional review board and the ethics committee of the University Hospital Tübingen (768/2018BO2).
The authors declare that they have no conflicts of interest regarding the publication of this paper.
HM, PR, and MK designed the study and performed data analysis and interpretation and statistical data analysis. HAH, MM, PH, and POV contributed to the acquisition of the data. VM contributed to the manuscript preparation, drafting, critique, and review. All authors approved the final version of the manuscript submitted.
Echocardiographic parameters of right and left ventricular function.