Abstract
Background This study investigated dyspnoea intensity and respiratory muscle ultrasound early after extubation to predict extubation failure.
Methods The study was conducted prospectively in two intensive care units in France and Canada. Patients intubated for at least 48 h were studied within 2 h after an extubation following a successful spontaneous breathing trial. Dyspnoea was evaluated by a dyspnoea visual analogue scale (Dyspnoea-VAS) ranging from 0 to 10 and the Intensive Care Respiratory Distress Observational Scale (IC-RDOS). The ultrasound thickening fraction of the parasternal intercostal and the diaphragm was measured; limb muscle strength was evaluated using the Medical Research Council (MRC) score (range 0–60).
Results Extubation failure occurred in 21 out of 122 enrolled patients (17%). The median (interquartile range (IQR)) Dyspnoea-VAS and IC-RDOS were higher in patients with extubation failure versus success: 7 (4–9) versus 3 (1–5) (p<0.001) and 3.7 (1.8–5.8) versus 1.7 (1.5–2.1) (p<0.001), respectively. The median (IQR) ratio of parasternal intercostal muscle to diaphragm thickening fraction was significantly higher and MRC was lower in patients with extubation failure compared with extubation success: 0.9 (0.4–2.1) versus 0.3 (0.2–0.5) (p<0.001) and 45 (36–50) versus 52 (44–60) (p=0.012), respectively. The thickening fraction of the parasternal intercostal and its ratio to diaphragm thickening showed the highest area under the receiver operating characteristic curve (AUC) for an early prediction of extubation failure (0.81). AUCs of Dyspnoea-VAS and IC-RDOS reached 0.78 and 0.74, respectively.
Conclusions Respiratory muscle ultrasound and dyspnoea measured within 2 h after extubation predict subsequent extubation failure.
Abstract
Respiratory muscle ultrasound and dyspnoea intensity early after extubation predict extubation failure https://bit.ly/39Uqo9o
Introduction
Extubation failure occurs in 10–15% of patients [1], and is associated with prolonged duration of mechanical ventilation and increased mortality [2]. The timing of re-intubation is likely to influence the outcome: delayed re-intubation is associated with a higher mortality rate [3]. Therefore, early identification of patients at high risk of re-intubation is of great importance. In clinical practice, after extubation, recognition of clinical worsening can be delayed because key elements of respiratory monitoring (e.g. tidal volume) are no longer available and respiratory rate alone is not a good indicator of inspiratory effort [4].
Dyspnoea, a key feature of acute respiratory failure, could be a warning sign of extubation failure. For instance, in patients admitted to the intensive care unit (ICU) for acute respiratory failure, moderate-to-severe dyspnoea is independently associated with noninvasive ventilation failure [5]. Therefore, assessing for dyspnoea after extubation could help predict the risk of extubation failure. The intensity of dyspnoea is strongly correlated with the activity of extra-diaphragmatic inspiratory muscles such as the parasternal intercostal and scalene muscles [6–8]. These muscles are commonly activated together with the diaphragm in acute respiratory failure and weaning failure [9, 10]. Accordingly, the activity of the inspiratory muscles relative to the diaphragm could provide another early predictor of extubation failure. The activity of the parasternal intercostal [11] and the diaphragm [12, 13] can be easily quantified by ultrasound, which could be useful during the weaning process. A recent study failed to identify differences in terms of diaphragm activity between patients who succeed and those who fail extubation in patients at high risk of extubation failure when ultrasound was performed during a spontaneous breathing trial [14]. Here, we tested the hypothesis that, early after extubation, dyspnoea and parasternal intercostal muscle/diaphragm ultrasound can predict subsequent extubation failure in nonselected patients separated from the ventilator after a successful spontaneous breathing trial.
Methods
This study complies with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement. It was conducted between August 2016 and May 2018 in two sites: the Medical and Surgical ICU, St Michael's Hospital, Toronto, ON, Canada (August 2016–November 2017) and the Pulmonology and Intensive Care Dept, Hôpital Pitié-Salpêtrière, Paris, France (August 2017–May 2018). All patients or their next of kin provided informed consent. The study was approved by the Institutional Review Boards at both participating institutions (Comité de Protection des Personnes Ouest 17/048-3; St Michael's Hospital REB 16-161) and was performed in accordance with the ethical standards laid down in the 2008 Declaration of Helsinki.
Patients
Consecutive adult patients who were mechanically ventilated for at least 48 h and who were extubated after a successful spontaneous breathing trial were eligible. Exclusion criteria were related to “do not re-intubate” orders, unplanned extubation, current use of extracorporeal membrane oxygenation, inability to perform ultrasound of the diaphragm and parasternal intercostal muscle (morbid obesity and thoracic dressings), tracheostomy, and pre-existing neuromuscular diseases or known diaphragm paralysis.
Patients were eligible for a spontaneous breathing trial as soon as they met predefined readiness-to-wean criteria: arterial oxygen saturation measured by pulse oximetry (SpO2) ≥90% on inspiratory oxygen fraction ≤0.4 and positive end-expiratory pressure ≤8 cmH2O, and low/no doses of vasopressors [15]. The spontaneous breathing trial was performed while patients were connected to the ventilator with no pressure assist of any kind (zero positive end-expiratory pressure and a pressure support level of zero) for 30 min. Success or failure of the spontaneous breathing trial and the decision to extubate were determined by the physician in charge who had no role in the study (usual criteria for spontaneous breathing trial success and for extubation are listed in the supplementary material). Prophylactic noninvasive ventilation or high-flow nasal cannula could be applied in patients with pre-identified risk factors for extubation failure (chronic respiratory disease, chronic cardiac disease and age >65 years) [16]. The decision to use noninvasive ventilation or high-flow nasal cannula was taken before the extubation by the physician in charge who had no role in the study.
Measurements
Dyspnoea evaluation
Presence and severity of dyspnoea were evaluated by using self-evaluation of dyspnoea and dyspnoea observation scales [17], depending on whether patients were communicative or noncommunicative. Patients were considered as communicative when the Richmond Agitation–Sedation Scale was between −1 and +1 and if they were able to consistently self-report dyspnoea, attested by a dyspnoea visual analogue scale (Dyspnoea-VAS) variation not exceeding 1 cm for three consecutive measures [18, 19]. In communicative patients only, self-evaluation was performed by means of the Dyspnoea-VAS that consisted of a 10-cm scale ranging from “no respiratory discomfort” to “intolerable respiratory discomfort”. Hetero-evaluation was performed in communicative and noncommunicative patients, by means of the Intensive Care Respiratory Distress Observational Scale (IC-RDOS) [17], which is based on respiratory, neurovegetative and behavioural signs, and includes five observable items (heart rate, use of neck muscles during inspiration, abdominal paradox during inspiration, facial expression of fear and supplemental oxygen). IC-RDOS correlates strongly with Dyspnoea-VAS [20] and has been previously validated in noncommunicating ICU patients [21]. Clinically significant dyspnoea was defined as Dyspnoea-VAS ≥4 [17] and IC-RDOS ≥2.4, since this latter predicts Dyspnoea-VAS ≥4 with equal sensitivity and specificity (72%) [20].
Respiratory muscle ultrasound
All investigators involved in ultrasound measurements received a training programme with a minimum of 15 ultrasounds supervised by an investigator experienced in respiratory muscle ultrasound imaging. Respiratory muscle measurements were performed by investigators who were independent of the clinical team. At the time of ultrasound, patients were breathing spontaneously without noninvasive ventilation or high-flow oxygen therapy. When indicated, noninvasive ventilation or high-flow oxygen therapy was applied after ultrasound measurements. Ultrasound was performed by using two machines: Sparq (Philips Healthcare, Bothell, WA, USA) and Sonosite (Fujifilm Sonosite, Bothell, WA, USA). The methods used to evaluate diaphragm and parasternal intercostal thickness and thickening have been extensively detailed and validated elsewhere (supplementary material) [11, 22]. Briefly, the parasternal intercostal muscle was evaluated by using a 10–15 MHz linear array transducer positioned at the level of the second right intercostal space. The second right parasternal intercostal muscle was identified as a three-layered biconcave structure: two linear hyperechoic membranes running, respectively, from the anterior and posterior aspects of the adjoining ribs, and a medial portion with muscle echotexture. Using M-mode, the thickness of the parasternal intercostal muscle was measured on frozen images at end-expiration and at peak inspiration. Change in thickness determined the thickening fraction of the parasternal intercostal muscle (TFic) as: TFic=(peak inspiration thickness–end-expiratory thickness)/end-expiratory thickness. Only the right parasternal intercostal muscle was evaluated for simplicity.
Diaphragm ultrasound was conducted using a 10–15 MHz linear array transducer. As for the parasternal intercostal muscle, diaphragm thickness (including pleural and peritoneal membranes) was measured at end-expiration and at peak inspiration, and the thickening fraction (TFdi) was calculated offline as: TFdi=(peak inspiration thickness–end-expiratory thickness)/end-expiratory thickness.
All ultrasound recordings were analysed offline by one single investigator (M. Dres) who was blinded to the clinical outcomes. All ultrasound measurements were repeated on at least three separate breaths and their average was reported. Reproducibility of diaphragm and parasternal intercostal muscle ultrasound has been reported elsewhere and was not tested in the present study [11, 22].
Clinical data collection
Demographic data were collected from the electronic medical charts of the patients: age, gender, comorbidities (chronic hypertension, chronic pulmonary disease, diabetes, chronic renal failure and left heart dysfunction), Sequential Organ Failure Assessment and Acute Physiology and Chronic Health Evaluation III scores, date of ICU admission, date of intubation, main reason for intubation, weight and height at admission, number of spontaneous breathing trials before extubation, and ventilation mode before extubation. At the time of dyspnoea evaluation and respiratory muscle ultrasound, the following variables were collected: arterial blood pressure, heart rate, SpO2 and respiratory rate. We also evaluated limb muscle strength by the Medical Research Council (MRC) score in communicative patients [23]. ICU-acquired weakness was defined by an MRC score <48 [24]. Finally, cough strength was assessed by using a semiquantitative categorical scale that classified cough strength as “weak”, “moderate” or “strong” [14].
Study protocol
Patients were enrolled in the study following extubation, and ultrasound of the diaphragm and the parasternal intercostal muscle was performed on the right side within 2 h after extubation. Immediately before ultrasound measurements, dyspnoea was assessed with Dyspnoea-VAS in communicative patients and the items of IC-RDOS were collected in all patients and the sum was computed offline.
The primary end-point was extubation failure defined as re-intubation or death within the 7 days following planned extubation.
Statistical analyses
Variables are presented as median (interquartile range (IQR)) or number (percentage). As, to the best of our knowledge, no previous publications have evaluated the level of dyspnoea or respiratory muscle ultrasound after extubation, a formal sample size calculation was deemed not possible, and we estimated that a sample of 15–20 patients with extubation failure would be appropriate to make a relevant comparison between groups regarding their dyspnoea scales and ultrasound respiratory muscle indices. Based on an estimate rate of extubation failure of 15% [25], we planned to enrol 120 patients.
Differences between patients with and without extubation failure were assessed using the Mann–Whitney or Chi-squared test, where appropriate. Potential risk factors for extubation failure were assessed by univariate analysis and their predictive performances were computed with receiver operating characteristic (ROC) curves. Sensitivities, specificities, positive and negative predictive values, positive and negative likelihood ratios, and areas under the ROC curves (AUCs) were calculated. AUCs were determined to identify optimal cut-off values in predicting failure and these estimates were obtained using bootstrapping with 1000 replications. The best threshold value for each index was determined as the value associated with the best Youden index for the prediction of failure.
The Spearman correlation was used to evaluate the relationship between ultrasound indices (TFic, TFdi and TFic/TFdi) and dyspnoea scales (Dyspnoea-VAS and IC-RDOS). In all cases, a p-value <0.05 was considered statistically significant and two-tailed testing was used to test hypotheses. All analyses were performed using Prism version 8 (GraphPad, San Diego, CA, USA) and MedCalc (MedCalc Software, Ostend, Belgium).
Results
Population
Among the 470 patients eligible during the study period, 122 patients were enrolled in the study (figure 1). The main characteristics of the patients are shown in table 1. There were significant differences between both centres regarding some characteristics of the patients (supplementary table S1). The main reasons for intubation were hypoxaemic acute respiratory failure and coma, and the median (IQR) duration of mechanical ventilation at the time of enrolment was 5 (3–8) days. Extubation failure occurred in 21 (17%) of the 122 patients. The median (IQR) duration between extubation and extubation failure was 1 (0–3) days (supplementary table S2). Of these 21 patients, one patient was not re-intubated because of a decision of “do not re-intubate” taken after extubation (see supplementary table S2 for the detailed causes of extubation failure). Respiratory distress was the reason for extubation failure in 18 patients; three patients failed extubation for other reasons (one for coma, one for seizure and one for sepsis). No stridor occurred after extubation.
Flowchart of the study. ICU: intensive care unit; SBT: spontaneous breathing trial.
Characteristics of the patients
The 101 (83%) remaining patients were classified as extubation success. Of these 101 patients, two received rescue noninvasive ventilation not followed by a re-intubation. Post-extubation management and outcomes of the patients are reported in supplementary table S3.
Post-extubation assessment
Post-extubation assessment was performed 23 (0–60) min after extubation; there was no difference in timing assessment between patients with extubation success and those with extubation failure (30 (5–59) versus 22 (0–61) min, respectively; p=0.45). Median (IQR) Dyspnoea-VAS was 4 (2–5) and was ≥4 (i.e. clinically important dyspnoea) in 49 (52%) of the 95 patients with Dyspnoea-VAS measurements. Median IC-RDOS was 1.9 (1.6–3.5) and was ≥2.4 (i.e. clinically important) in 68 out of 122 (56%) patients. 34 patients (36%) had ICU-acquired weakness, defined by MRC <48. Cough strength was estimated as weak in 14 patients (13%), moderate in 72 patients (65%) and strong in 24 patients (22%).
Factors associated with extubation failure
Clinical features and arterial blood gases before and after the spontaneous breathing trial were not different in patients depending of the extubation outcome (supplementary table S4). Except for the respiratory rate that was higher in patients with extubation failure, the rest of the clinical examination was not different between groups at the time of ultrasound and dyspnoea evaluation (table 2). The proportion of patients with weak, moderate and strong cough was not different between groups (table 2). The MRC score was higher in patients with extubation success compared with patients with extubation failure (table 2). Clinically relevant dyspnoea (Dyspnoea-VAS ≥4) was more frequent and dyspnoea was more intense in patients who failed extubation (table 2 and figure 2). TFdi was significantly higher in patients who succeeded extubation, whereas TFic was significantly higher in patients who failed extubation (figure 2 and table 2). TFic/TFdi was higher in patients with extubation failure (table 2). There was a moderate but significant correlation between TFic and IC-RDOS and between TFic and Dyspnoea-VAS, but there was no correlation between TFdi and IC-RDOS or between TFdi and Dyspnoea-VAS (supplementary table S5).
Clinical evaluation and respiratory muscle ultrasound indices on enrolment
a) Dyspnoea visual analogue scale (Dyspnoea-VAS), b) Intensive Care Respiratory Distress Observation Scale (IC-RDOS), c) Medical Research Council (MRC) score, d) thickening fraction of the parasternal intercostal muscle (TFic), e) thickening fraction of the diaphragm (TFdi) and f) TFic/TFdi in patients with extubation success and extubation failure. Box plots represent the median and interquartile range (with minimum–maximum) of the distributions.
The performance of IC-RDOS, Dyspnoea-VAS, MRC score, TFic, TFdi and TFic/TFdi to predict extubation failure was evaluated by computing their AUCs (figure 3 and supplementary table S6). The best thresholds to predict extubation failure were IC-RDOS >3.3, Dyspnoea-VAS >4.0, MRC score <50, TFic >8.6%, TFdi <15.6% and TFic/TFdi >0.44 (supplementary table S5). IC-RDOS was <3.3 in eight out of 21 patients who failed extubation. Among these eight patients, five had TFic/TFdi >0.44. Therefore, combining IC-RDOS and TFic/TFdi predicted extubation failure in 18 out of 21 patients. A sensitivity analysis comparing patients with extubation failure (re-intubation and rescue noninvasive ventilation) and extubation success reached similar findings (supplementary table S7).
Receiver operating characteristic (ROC) curves (with 95% CIs) of a) dyspnoea visual analogue scale (Dyspnoea-VAS), b) Intensive Care Respiratory Distress Observation Scale (IC-RDOS), c) Medical Research Council (MRC) score, d) thickening fraction of the parasternal intercostal muscle (TFic), e) thickening fraction of the diaphragm (TFdi) and f) TFic/TFdi. AUC: area under the ROC curve.
Discussion
This study investigated the performance of dyspnoea scales and respiratory muscle ultrasound following extubation after a successful spontaneous breathing trial to predict the risk of extubation failure. The main findings are: 1) patients who eventually failed extubation experienced higher self-reported dyspnoea intensity and had a higher observational dyspnoea scale after extubation, 2) there was a moderate but significant correlation between dyspnoea and TFic and TFic/TFdi as evaluated by ultrasound, and 3) TFic/TFdi predicted extubation outcome with a good performance (the higher the ratio, the higher the risk of subsequent re-intubation).
Weaning failure occurs when respiratory loading (pulmonary oedema, secretions, atelectasis and hyperinflation) exceeds the compensatory capacity of the respiratory muscles [26]. This load capacity imbalance stimulates the respiratory drive [27], which in turn induces the recruitment of accessory inspiratory muscles [9, 11], both well-established determinants of dyspnoea. Dyspnoea is a frequent but underestimated symptom that is encountered in ∼50% of ICU patients [5, 20, 28, 29]. Dyspnoea is strongly associated with the recruitment of extra-diaphragmatic respiratory muscles [6, 8, 30] and behaves as a reliable surrogate of the increase in respiratory drive that is subsequent to the load capacity imbalance. The correlation between dyspnoea and extra-diaphragmatic respiratory muscle activity has already been established with electromyography [6, 30]. Our study confirms this physiological relationship with ultrasound. In contrast, there was no correlation between dyspnoea and TFdi, a finding already highlighted in a previous study [30]. Therefore, the recruitment of extra-diaphragmatic inspiratory muscles seems to be a better indicator of the respiratory load capacity imbalance than diaphragm activity. This may explain why dyspnoea, which is closely linked to respiratory drive, is better correlated with parasternal intercostal activity than with diaphragm activity alone.
While our patients had successfully passed a spontaneous breathing trial, suggesting a relatively adequate load capacity balance, the prevalence of dyspnoea was as high as 52% after extubation. Since dyspnoea was not evaluated before extubation, it cannot be ruled out that some patients were dyspnoeic while they did not develop classical criteria of spontaneous breathing trial failure. Interestingly, the intensity of dyspnoea predicted extubation outcome. Several studies have already reported on the relationship between dyspnoea (as a warning sign) and clinical outcomes. Dyspnoea seems to be a proxy for the severity of a respiratory or cardiac disease [31]. For instance, there is an association between poor respiratory comfort and hospital mortality in patients with suspected acute myocardial infarction [32, 33], in those admitted for acute chronic obstructive pulmonary disease (COPD) exacerbation [34], and even in patients without previously diagnosed cardiopulmonary diseases [35]. Regarding the prediction of intubation, a recent study suggested that in patients receiving noninvasive ventilation for acute respiratory failure, moderate-to-severe dyspnoea can predict noninvasive ventilation failure and subsequent intubation [5]. These findings and ours suggest that evaluation of dyspnoea in nonintubated patients (before or after extubation) may be of clinical interest. Eventually, further studies will have to investigate whether preventive strategies for extubation failure could rely on dyspnoea scales and timely application of noninvasive ventilatory supports.
It is notable that IC-RDOS, an observational scale that can evaluate dyspnoea in noncommunicative patients as opposed to Dyspnoea-VAS that requires patient cooperation, performed as well as Dyspnoea-VAS to predict extubation failure. This is of great help since freshly extubated patients may show delirium and be unable to self-report dyspnoea. Finally, it is important to note that, with a median Dyspnoea-VAS of 4, dyspnoea was present in a substantial proportion of patients. Similar pain intensity would have immediately triggered a medical response involving the administration of analgesic medication. The suffering that dyspnoea induces in extubated patients should not be neglected as dyspnoea contributes to the dark recollections following an ICU stay [36]. For some authors, failing to address dyspnoea breaches human rights [37, 38].
The present study shows that ultrasound of respiratory muscles could identify patients who further presented extubation failure. Indeed, patients who developed extubation failure had a lower TFdi and a higher TFic. The association between either increased TFic or decreased TFdi and weaning failure has already been well established [11, 39]. It is noticeable that, at the time of respiratory muscle ultrasound, patients who further developed extubation failure had similar clinical presentation compared with their counterparts. Furthermore, respiratory rate was associated with a poor ROC curve, suggesting that ultrasound was able to detect a subclinical impairment of the respiratory load/capacity balance.
Because diaphragm dysfunction is compensated for by an increase in extra-diaphragmatic inspiratory muscle activity [40–42], the relationship between TFic and TFdi is inverse, i.e. the lower the TFdi, the higher the TFic [11]. TFic/TFdi combines the evaluation of the diaphragm and the extra-diaphragmatic inspiratory muscle. When the TFic/TFdi ratio increases, it indicates a recruitment of parasternal intercostal muscles (high TFic) because of a weak diaphragm (low TFdi). This ratio may appropriately reflect the respiratory load/capacity balance. Indeed, in the case of increased respiratory drive, extra-diaphragmatic respiratory muscles are activated more prematurely and more vigorously than the diaphragm [43]. Recruitment of scalene muscles predicts weaning failure in ICU patients [9] and recruitment of the parasternal intercostal muscle predicts poor outcome in patients with COPD exacerbation [44]. These data and ours raise the relevance of monitoring extra-diaphragmatic inspiratory muscles, which seem to be a better indicator of the respiratory load capacity imbalance than diaphragm activity alone. However, combining TFdi with TFic did not offer any predictive advantage for extubation failure, probably because TFdi provides a lower performance than TFic to predict extubation failure, as already reported by Vivier et al. [14] who observed similar values of TFdi in patients who succeeded and in those who failed extubation. In Vivier et al. [14], diaphragm ultrasound was performed when patients were still under mechanical ventilation while on a T-piece and not after extubation (such as in our study). In addition, patients in Vivier et al. [14] were selected as being at high risk of extubation failure, which was not particularly the case of our population. Our approach was to evaluate a strategy aiming to predict extubation outcome after and not before extubation, which could be questionable. We reasoned that despite a successful spontaneous breathing trial, the risk of extubation failure still ranges between 10% and 20% [25, 45]. This is explained by the fact that the spontaneous breathing trial behaves like any other diagnostic test and will never reach a perfect specificity (i.e. 100%) [46]. Accordingly, we believe that it is still relevant to evaluate the risk of extubation failure after extubation. The implications (not investigated in the present work) would be to personalise the management of these patients in order to prevent post-extubation acute respiratory failure and subsequent re-intubation (noninvasive ventilation, high-flow nasal oxygen, physiotherapy, mobilisation and delayed ICU discharge).
Strengths and limitations of the study
The study was conducted in two centres in two different countries and used a standardised weaning protocol. In addition, the inclusion criteria were broad, which might help generalise our findings. However, this study has limitations. First, the number of patients is limited due the weak occurrence of extubation failure and there were significant differences between both centres regarding some characteristics of the patients; further studies will have to evaluate the extrinsic validity of our calculated cut-offs and to confirm the predictive thresholds identified in the present work in a validation cohort. Second, the reproducibility of Dyspnoea-VAS and IC-RDOS was not evaluated in our study, although previous studies reported good reproducibility [20, 47]. Third, in order to standardise measurements, ultrasound was performed while patients were breathing spontaneously. Accordingly, ultrasound measurements may not reflect the patients’ conditions under prophylactic measures (noninvasive ventilation or high-flow oxygen therapy). In addition, the use high-flow oxygen therapy may have influenced the outcome of some patients. At the time of study, no evidence-based guidelines were available on the use of high-flow oxygen therapy, which was driven by the experience of the physician in charge.
Conclusions
In patients who successfully passed a spontaneous breathing trial and were subsequently extubated, dyspnoea as assessed by self-report or observational scales and respiratory muscle activity assessed with ultrasound predicted extubation failure. Although these results need to be confirm by a larger study, they suggest that monitoring dyspnoea in every newly extubated patient may be useful to predict a future re-intubation. Our results pave the way for future studies evaluating whether monitoring dyspnoea following extubation may enable tailoring the oxygenation and ventilation strategy in the post-extubation period.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-00002-2021.Supplement
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-00002-2021.Shareable
Footnotes
Author contributions: M. Dres and L. Brochard designed the study. M. Dres coordinated the study. M. Dres, T. Pham, L. Sergenyuk, W. Ouechani, I. Telias, D.L. Grieco, M.C. Sklar, D. Junhasavasdikul, L. Melo, C. Santis, L.F. Damiani, L. Degravi, M. Decavèle and L. Brochard were responsible for patient screening, enrolment and follow-up. M. Dres, T. Similowski, L. Brochard, A. Demoule and E.C. Goligher analysed the data. M. Dres, A. Demoule and T. Similowski wrote the manuscript. All authors had full access to all of the study data, contributed to drafting the manuscript or critically revised it for important intellectual content, approved the final version of the manuscript, and take responsibility for the integrity of the data and the accuracy of the data analysis.
This article has supplementary material available from erj.ersjournals.com
Conflict of interest: M. Dres reports personal fees from Lungpacer SA, outside the submitted work.
Conflict of interest: T. Similowski reports personal fees from AstraZeneca France, Boehringer Ingelheim France, GSK France, Teva France, Chiesi France, Lungpacer Inc. and ADEP Assistance, personal fees and nonfinancial support from Novartis France, grants from Air Liquide Medical Systems, outside the submitted work.
Conflict of interest: E.C. Goligher reports personal fees and nonfinancial support (research equipment) from Getinge, nonfinancial support (research equipment) from Timpel, outside the submitted work.
Conflict of interest: T. Pham has nothing to disclose.
Conflict of interest: L. Sergenyuk has nothing to disclose.
Conflict of interest: I. Telias reports grants from CIHR, personal fees for lectures from MBMed SA, Argentina and Covidien, Argentina, outside the submitted work.
Conflict of interest: D.L. Grieco reports grants from GE, nonfinancial support from Dimar, Maquet and Fisher & Paykel, outside the submitted work.
Conflict of interest: W. Ouechani has nothing to disclose.
Conflict of interest: D. Junhasavasdikul has nothing to disclose.
Conflict of interest: M.C. Sklar has nothing to disclose.
Conflict of interest: L.F. Damiani has nothing to disclose.
Conflict of interest: L. Melo has nothing to disclose.
Conflict of interest: C. Santis has nothing to disclose.
Conflict of interest: L. Degravi has nothing to disclose.
Conflict of interest: M. Decavèle has nothing to disclose.
Conflict of interest: L. Brochard reports grants from Medtronic Covidien, grants and nonfinancial support from Fisher & Paykel, nonfinancial support from Sentec and Philips, other (patent) from General Electric, outside the submitted work.
Conflict of interest: A. Demoule reports personal fees from Medtronic, Baxter, Hamilton, Getinge and Lowenstein, grants, personal fees and nonfinancial support from Philips, Respinor and Lungpacer, personal fees and nonfinancial support from Fisher & Paykel, grants from French Ministry of Health, outside the submitted work.
Support Statement: M. Dres was supported by the French Intensive Care Society (SRLF bourse de mobilité 2015), the 2015 Short Term Fellowship programme of the European Respiratory Society, the 2015 Bernhard Dräger Award for Advanced Treatment of ARF of the European Society of Intensive Care Medicine, Assistance Publique Hôpitaux de Paris, Fondation pour la Recherche Médicale (FDM 20150734498) and Mitacs Globalink Sorbonne Universités. Funding information for this article has been deposited with the Crossref Funder Registry.
- Received January 3, 2021.
- Accepted March 30, 2021.
- Copyright ©The authors 2021. For reproduction rights and permissions contact permissions{at}ersnet.org