Non-invasive ventilation (NIV) is widely used in emergency settings for acute respiratory failure, with NIV failure, usually defined by the need for tracheal intubation, as its primary complication. In emergency settings where patients may not be intubated or or where NIV represents the ceiling of care, a pragmatic understanding of NIV failure requires a broader definition that incorporates early clinical deterioration, including presumptive intubation criteria. This study assessed the prevalence of early clinical deterioration under NIV initiated in emergency settings (emergency department [ED] or mobile emergency medical services [EMS]) and its associated variables.

A prospective multicentre study was conducted in 68 French EDs and EMS in the Initiative Recherche Urgences (IRU) network. Adult patients (≥ 18 years) requiring NIV in emergency settings were included, excluding those with a known do-not-resuscitate order or low autonomy. The primary endpoint was early clinical deterioration under NIV at 1 h. Early clinical deterioration under NIV was defined as either (1) the need for tracheal intubation or; (2) the presence of presumptive criteria for intubation. Secondary endpoints were baseline factors associated with failure, the need for tracheal intubation or death within 7 days among patients surviving without tracheal intubation at 1 h, and 7-day mortality.

A total of 198 patients were included over 5 days. Early clinical deterioration at 1 h was reported in 41% of the patients. Early clinical deterioration under NIV was associated with a Glasgow Coma Scale score < 14 (adjusted odds ratio [aOR] = 5.5, 95% confidence interval [CI] [1.8 –19.4]), heart rate > 115 beats per minute (aOR = 2.5, 95%CI [1.3–5.2]), and signs of increased work of breathing (aOR = 2.8, 95%CI [1.2–7.1]). Among the surviving patients not intubated at 1 h, 12% required intubation within 7 days in the Early Clinical Deterioration group and 3% in the No Early Clinical Deterioration group (p < 0.001). Within 7 days, 28% died in the Early Clinical Deterioration group and 10% in the No Early Clinical Deterioration group (p = 0.001). NIV failure was associated with increased 7-day mortality (aHR = 4.1, 95%CI [1.8–9.1]).

Early clinical deterioration under NIV is common in EDs, affecting nearly one out of two patients, and is associated with higher 7-day mortality.

Clinical trial registration Registered 2024 january, 23th. NCT06213623. Prior to the first inclusion.

Non-invasive ventilation (NIV) is a mechanical ventilation procedure that provides ventilatory support through a facemask, thereby avoiding tracheal intubation. NIV includes variable positive airway pressure devices (commonly referred to as “bilevel” or “BiPAP” [bilevel positive airway pressure]), which consist of higher inspiratory positive airway pressure and lower expiratory pressure, as well as continuous positive airway pressure (CPAP) [1]. In the emergency department (ED), NIV is widely used to manage patients with acute respiratory failure (ARF) resulting from acute exacerbation of chronic obstructive pulmonary disease (COPD) or cardiogenic pulmonary oedema (CPO). NIV improves clinical patterns and outcomes, including reduced tracheal intubation and mortality rates [1, 2].

Although NIV is initiated in approximately 15% of patients admitted to the ED for dyspnoea, the prevalence, associated risk factors, and consequences of NIV failure have only been investigated in retrospective or single-centre studies [3,4,5,6]. NIV failure, which often results in delayed intubation, is associated with poor outcomes and increased mortality rates. NIV failure has been well studied in the intensive care unit (ICU) but remains poorly explored in the ED. NIV failure is commonly defined as the need for tracheal intubation in patients receiving NIV. However, this definition is questionable in emergency settings, where some patients may have a do-not-intubate status or receive NIV as a ceiling of care. Consequently, broader definitions have been proposed that incorporate clinically worsening symptoms or presumptive criteria for intubation [3]. In addition, exploring “early clinical deterioration” under NIV appears to be more relevant in the ED setting.

This study aimed to assess the prevalence of early clinical deterioration under NIV initiated in emergency settings (ED or mobile intensive care unit [MICU]) and its associated variables.

We conducted a prospective, observational study in 68 EDs of the Emergency Research Initiative (Initiative Recherche Urgences, IRU) network from 15 January 2024 to 20 January 2024. The IRU is a research network based on the initiative of the French Society of Emergency Medicine (SFMU, Société Française de Médecine d’Urgence), with the aim of promoting and coordinating multicentre research projects over short inclusion periods (3–5 days) in the field of emergency medicine. The French medical system facilitates the deployment of ambulances staffed by an emergency physician, a nurse, and an ambulance driver. The MICU can initiate NIV during the prehospital period and provide early respiratory support before hospital admission.

Our reporting is in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. The study was registered at ClinicalTrials.gov with the identifier 06213623.

Patients were enrolled in the study if they met all the following criteria: (1) age ≥ 18 years; (2) admitted to a participating ED or cared for by an MICU outside the hospital; and (3) requiring ventilator support with NIV. Exclusion criteria were as follows: (1) known do-not-resuscitate (DNR) order; (2) contraindications to NIV use; (3) low autonomy defined by the World Health Organization (WHO) score equal to or greater than 3; and (4) patient’s opposition to the use of their data.

NIV was initiated at the discretion of the attending physician and local protocols were followed. In addition, the research protocol did not interfere with patient management throughout the study.

The primary endpoint was the proportion of patients with early clinical deterioration under NIV 1 h after NIV initiation. Given the emergency setting, clinical deterioration was defined as either (1) the need for tracheal intubation or; (2) the presence of presumptive criteria for intubation in accordance with the SRLF-SFMU consensus on oxygen therapy in ARF at 1 h. These criteria were defined as the persistence or worsening of specific signs at 1 h compared to baseline and included at least one of the following: (a) tachypnoea with a respiratory rate > 30 breaths/min; (b) new or worsening vigilance disorders; (c) clinical signs of respiratory distress (accessory muscle use, abdominal respiration); (d) haemodynamic failure requiring vasopressors; or (e) premature NIV interruption due to patient agitation, intolerance, or opposition [7]. This broader definition was chosen to better reflect real-life ED practice, where early clinical deterioration often necessitates rapid reassessment of ventilatory support. To limit interobserver variability across centres, the assessment of the primary outcome was centralised and validated by the coordinating investigator team based on standardised clinical documentation collected at the bedside. This methodological choice aimed to ensure the consistent and reproducible application of the criteria across all participating sites.

Secondary endpoints were (1) identification of patient characteristics at baseline independently associated with clinical deterioration at 1 h; (2) need for tracheal intubation or death within 7 days of inclusion in patients surviving without tracheal intubation at 1 h; (3) death within 7 days; (4) survival without intubation within 7 days; and (5) association between clinical deterioration and death within 7 days of inclusion in patients surviving at 1 h.

Demographic characteristics (age, sex, comorbidities, and autonomy using the WHO score), clinical patterns (heart rate, arterial pressure, respiratory rate, signs of increased work of breathing, SpO2 value, and Glasgow Coma Scale) at baseline and 1 h after starting NIV, biological patterns at baseline (blood gas analysis, blood chemistry panel, arterial lactate, and blood count), final diagnosis upon leaving the ED, ventilation support settings, and the place where NIV was started (mobile emergency medical services [EMS] or ED) were collected for each patient.

To minimise selection bias, all consecutive patients who met the inclusion criteria during the study period were enrolled, regardless of their prognosis or treatment response. Data collection was conducted prospectively by IRU network investigators at each participating centre to reduce information bias. Standardised definitions and pre-specified criteria were presented to the investigators prior to the study and were used to ensure consistency in data interpretation.

The study size was determined based on feasibility, considering the short inclusion period and the observational nature of the study. All eligible patients who met the inclusion criteria within the predefined time frame were enrolled to maximise their representativeness. No formal sample size calculation was performed beforehand, because the primary objective was to describe the prevalence and risk factors for early clinical deterioration in a real-life emergency setting. Based on the estimated prevalence of NIV use in the ED, we anticipated enrolling 400–500 patients across 100 centres within the IRU network, aiming to capture at least 60 (15%) events and include six covariates in the multivariate analysis.

Qualitative data are reported as numbers and proportions and compared between groups using the chi-square test or Fisher’s exact test, as appropriate. Quantitative data are reported as median and interquartile range (IQR) and compared between groups using the Mann–Whitney U test.

In the bivariate analysis, qualitative data were compared between groups using the chi-square test or Fisher’s exact test, as appropriate, and quantitative data using the Mann–Whitney U test. Independent covariates associated with early clinical deterioration under NIV were identified using multivariable analysis by logistic regression, including relevant covariates (p < 0.20 on bivariate analyses), using the optimal cut-point for quantitative data, if relevant, with cutpointr 1.1.2 package, and excluding covariates with missing data > 20%. Model quality was assessed using the Akaike information criterion (AIC), and nested models were compared using likelihood ratio tests. In the presence of mild overdispersion, a quasi-binomial model was used to correct the standard errors. The final model selection was based on statistical performance and clinical relevance. The results were reported as adjusted odds ratios (aORs) and 95% confidence intervals (95%CIs). The association between the primary outcome and tracheal intubation within 7 days and 7-day mortality was assessed using time-to-event analyses with Kaplan–Meier estimates and Cox proportional hazards models. For the analysis of intubation within 7-days, only patients who were not intubated at 1 h were included. All patients were included in the mortality analysis. The results were reported as adjusted hazard ratios (aHRs) with 95%CIs. Statistical analyses were performed using the R 4.0.2 software (R-project, Vienna, Austria).

This prospective, observational study was conducted in accordance with the ethical principles of the Declaration of Helsinki. It complies with French regulations on data protection (General Data Protection Regulation [GDPR]) and the amended French Data Protection Act n°78–17). In accordance with the French data protection authority’s reference methodology MR-004 and in the absence of any intervention or data collection specific to research, this study qualified as a non-interventional study outside the scope of the Loi Jardé and did not require approval from an ethics committee. The study was declared to the French Commission Nationale Informatique et Liberté (CNIL), and was approved by the Research Directorate of the Poitiers University Hospital and by the research committee of the SFMU. All patients were informed about the study, and non-opposition was obtained in compliance with French regulations. The data were fully anonymised prior to the analysis.

From 15 January 2024 to 20 January 2024, 198 patients (115 men [58%], median age 77 years [IQR 68–86]) were included (Table 1). Most of the patients had several comorbidities.

At baseline, 101 patients (51%) had ARF and 147 (74%) showed signs of increased work of breathing; the respiratory rate was 30 [25–35] breaths/min, SpO2:FiO2 ratio was 208 [147–297], and 105 of the 151 patients (70%) with blood gas analysis had acute hypercapnic acidosis (pH < 7.35 and PaCO2 > 45 mmHg).

The main indications for NIV were CPO in 96 patients (49%), acute exacerbation of COPD in 76 (38%), and de novo hypoxaemic respiratory failure in 26 (13%). NIV was initiated by mobile EMS in one of five cases and was mostly managed with BiPAP (Table 2).

NIV was delivered using BiPAP in 184 of 198 (94%) patients, CPAP in 12 (6%), and unknown in 2 patients. The NIV interface was an orofacial mask for all the patients. The median PEEP was 6 cmH₂O [IQR 5–7] with CPAP and BiPAP, and the median pressure support was 9 cmH₂O [IQR 8–10] with BiPAP. Pressure support and PEEP values did not significantly differ between the groups. FiO₂ values were 50% [35–60]. FiO₂ was higher in patients with early clinical deterioration than in those without (50% [40–80] vs 40% [35–55]; p = 0.009); NIV was initiated by mobile EMS in 21% of cases and in the ED in 79%. NIV was performed with a turbine-based ventilator in 182 (92%) patients, a pneumatic-based ventilator in 6 (3%), and a specific CPAP device in 10 (5%).

Overall, 81 patients (41%) met the criteria for early deterioration at 1 h (Table 3). An early deterioration was linked to tracheal intubation in 6 (7%) patients and the presence of presumptive criteria for intubation for 75 (93%), including ARF symptoms in 51 patients (63%), new or worsening vigilance disorders in 11 patients (14%), haemodynamic impairments requiring vasopressors in 4 patients (5%), and premature interruption in nine patients (11%) due to patient agitation (n = 4) or opposition (n = 5).

Bivariate analysis showed that the factors associated with early deterioration 1 h after NIV implementation were Glasgow coma scale, respiratory rate, heart rate, HACOR (heart rate, acidosis, consciousness, oxygenation and respiratory rate) score, and arterial lactate. On multivariate analysis including relevant and statistically related covariates, early deterioration was associated with a Glasgow Coma Scale score lower than 14 (aOR = 5.3 [95%CI, 1.8 to 19.1]), heart rate higher than 115 beats/min (aOR = 2.5 [1.2 to 5.2]) and signs of increased work of breathing (aOR = 2.9 [1.2 to 8.1]). No significant interaction was found between the main predictors identified in the multivariable analysis (Supplemental material). After adjusting for covariates, the diagnostic category was not significantly associated with early deterioration. Compared with patients with acute heart failure, those with exacerbation of COPD had an aOR of 0.91 (95%CI [0.42–4.38], p = 0.80) and those with de novo hypoxaemic respiratory failure an aOR of 1.51 (95%CI [0.52–4.38], p = 0.44) (Fig. 1).

Forrest-plot of risk factors for Early clinical deterioration at 1-h. Optimal cutpoint were chosen according to cutpointr 1.1.2 package. NIV denotes non-invasive ventilation. De novo hypoxemic respiratory failure referred to an acute respiratory failure occurring in patients without chronic respiratory disease or clinical suspicion of acute heart failure, typically caused by pneumonia or ARDS

Full size image

Overall, 189 patients (95%) were hospitalised, including 87 (44%) admitted to the ICU, with no significant intergroup differences. In total, 4 patients (2%) died in the ED before hospitalisation, and 5 (3%) were discharged to home.

Of the 192 patients not requiring tracheal intubation at 1 h, 12 (6%) required it over the following 7 days, including 3 out of 117 patients (3%) in the No Early Deterioration group and 9 out of 75 (12%) in the Early Deterioration group (hazard ratio [HR] = 5; 95%CI [1.3–18.5]; p = 0.016). After multivariable analysis, early deterioration at 1 h remained significantly associated with an increased risk of tracheal intubation within the first 7 days (aHR = 5.1, 95%CI [1.2–20.1], p = 0.025) (Fig. 2). No other baseline covariates, including age, heart rate, respiratory distress signs, or oxygenation indices, were significantly associated with this outcome.

Cumulative incidence for intubation within day-7. aHR denotes adjusted Hazar ratio. aHR was calculated on patients without intubation within 1-h following NIV initiation (n = 192)

Full size image

Within 7 days, 36 patients (19%) died (12 [10%] in the No Early Deterioration group and 23 [28%] in the Early Deterioration group; p < 0.001). Early deterioration at 1 h was associated with a higher mortality rate within 7 days (HR = 3.1; 95%CI [1.5–6.3]; p = 0.001) (Table 4). After multivariable analysis, early deterioration at 1 h remained associated with 7-day mortality (aHR = 4.1, 95%CI [1.8–9.1], p = 0.001) (Fig. 3). Age > 79 years (aHR = 3.35, 95%CI [1.6–6.9], p = 0.001) and signs of increased work of breathing (aHR = 2.5 95%CI [1.2–5.3]; p = 0.002) were also independently associated with an increased mortality rate within 7 days. The other baseline clinical parameters were not significantly associated with mortality. Finally, in the Early Clinical Deterioration group, intubation at 1 h was not associated with a decrease in the mortality rate within 7 days (aHR = 0.57 95%CI [0.10–7.20]). The main causes of mortality within 7 days were acute respiratory distress syndrome in 10/36 (30%), end-stage heart failure in 10/36 (30%), multiple organ failure in 3 (8%), treatment-refractory exacerbation of COPD in 4 (11%), and unknown in 9 (25%) patients.

Kaplan–Meier Curve for survival probability within 7-day. aHR denotes adjusted Hazard ratio

Full size image

Finally, survival without intubation within 7 days concerned 147 (74%) patients who required NIV in an ED setting. Survival without intubation was independently associated with NIV performed for de novo hypoxaemic respiratory failure (aHR = 0.42 95%CI [0.18–0.97]; p = 0.042).

In this multicentre observational study conducted in emergency settings, early clinical deterioration under NIV at 1 h occurred in 41% of the patients. It was associated with a fourfold increase in 7-day mortality and was predicted by clinical signs at baseline, including increased work of breathing, elevated heart rate, and lower Glasgow Coma Scale scores.

To address the challenges in emergency settings, we assessed early clinical deterioration rather than conventional NIV failure. In the ICU, NIV failure is traditionally characterised by the requirement for tracheal intubation and invasive mechanical ventilation[8,9,10]. However, for several reasons, we considered this definition inappropriate or too narrow for patients managed in the ED. First, EDs often manage older patients with more comorbidities than those treated in ICUs. In such cases, tracheal intubation and consequent admission to the ICU are often considered appropriate. Second, the absence of clinical improvement after 1 h, indicating NIV ineffectiveness, is not systematically associated with patient death, as some patients may spontaneously improve after premature NIV discontinuation. Third, NIV sessions in EDs typically last 1–2 h, whereas in ICUs, NIV failure is generally assessed after 48 h or even later, corresponding to significantly longer treatment sessions [11]. Finally, while tracheal intubation was not the sole criterion for defining NIV ineffectiveness, other parameters included in our definition—persistence of respiratory failure, neurological impairment, or haemodynamic impairment—are factors that should prompt the consideration of tracheal intubation in patients undergoing NIV in the ED [7].

Early clinical deterioration under NIV at 1 h occurred in 41% of patients and was associated with a fourfold increase in 7-day mortality. This underscores the importance of the early identification of patients for whom NIV is unlikely to result in clinical improvement to promptly reassess the therapeutic strategy. Although our criteria differ from the classical definition of NIV failure, our results are consistent with previously reported data in emergency settings. In three small-scale retrospective studies conducted in EDs, NIV failure rates ranged from 30% in patients with undetermined ARF[3, 10] to 50% in patients admitted with community-acquired pneumonia [5]. Similarly, two single-centre prospective ED studies reported failure rates of 24% and 35% for acute CPO and[6] de novo hypoxaemic respiratory failure, respectively [4]. This association with mortality is well documented across various aetiologies, including de novo acute hypoxaemic respiratory failure [12,13,14], acute exacerbation of COPD,[15] and acute CPO[6]. Delayed tracheal intubation likely contributes to poorer outcomes, including prolonged hospital stay [3], higher risk of complications [16], and consequently, increased mortality rates [6, 9, 12,13,14,15,16]. However, previous studies have assessed this association using a definition of NIV failure based solely on the need for tracheal intubation. Notably, more patients died within 7 days than were intubated. This likely reflects evolving goals of care, including limitations on life-sustaining treatments or patient refusal of intubation. Our findings demonstrate that even with a broader definition underscoring NIV efficacy, including clinical signs of treatment inefficacy or presumptive to intubation, the negative impact on prognosis remains unchanged.

Early deterioration under NIV at 1 h was independently associated with a lower Glasgow Coma Scale score, elevated heart rate, or signs of increased work of breathing at baseline [17]. In the ICU, parameters such as heart rate, level of consciousness, oxygenation, and acidosis, which are components of the HACOR score, are associated with NIV failure [14, 15]. We observed similar results, although acidosis and the PaO2:FiO2 ratio could not be included because of missing data. We hypothesised that these clinical signs reflected a more severe presentation upon admission, possibly related to diaphragmatic fatigue or reduced haemodynamic and neurological tolerance to respiratory distress. In the context of ARF, tachycardia may reflect the cardiovascular consequences of hypoxaemia, further exacerbating the imbalance between oxygen supply and demand, and ultimately compromising prognosis, particularly in older patients with cardiovascular comorbidities [17]. Similarly, signs of increased work of breathing reflect excessive spontaneous respiratory effort, which is often driven by impaired gas exchange and elevated respiratory drive. Available data suggest that sustained inspiratory effort, particularly during pressure support ventilation, may predispose patients to patient self-inflicted lung injury, a mechanism associated with worse clinical outcomes [18]. Moreover, a decrease in Glasgow Coma Scale score may reflect neurological impairment, reduced alertness due to fatigue, or an increased risk of vomiting and aspiration, particularly in the early phase of ARF. Rather than being interpreted in isolation, they should be collectively considered as part of a broader clinical picture that justifies close monitoring and may prompt early modification or escalation of treatment. In light of these findings, patients identified as being at a high risk for early deterioration under NIV, particularly those with de novo hypoxaemic respiratory failure, may benefit from alternative strategies such as high-flow nasal cannula therapy, which has been shown to be more effective than NIV in reducing the risk of intubation and mortality.

Our study has several limitations. First, although we initially anticipated enrolling 400–500 patients, only 198 were included, thereby limiting the robustness of our analyses. Our final sample size may have limited the statistical power and ability to identify other significant risk factors for early deterioration under NIV. Notably, although the diagnostic categories were not significantly associated with early deterioration after adjustment, we cannot exclude a higher risk of failure and poorer prognosis among patients with de novo hypoxaemic respiratory failure, as previously described in studies comparing oxygen therapy strategies. Second, arterial blood gas data were missing in 23% of the patients, which limited our ability to analyse this important clinical parameter. Although blood gas analysis is not routinely performed in all emergency settings [7], acidosis and elevated PaCO₂ are well-established predictors of NIV failure and have been incorporated into several risk stratification tools, including the HACOR score [14, 19]. Their absence prevented us from including these variables in our model, and this limitation should be considered when interpreting our findings. Although most patients were managed with turbine-driven transport ventilators, a subset received NIV via pneumatic transport devices. Experimental studies suggest that these systems offer reduced performance in terms of trigger sensitivity, pressure delivery, and leak compensation, potentially increasing the risk of patient–ventilator asynchrony and NIV failure [20]. Some measured factors may have influenced these results. Notably, variations in clinician experience and the absence of standardised local protocols may have affected the likelihood of NIV success, and patient prognosis in cases of intubation [21]. We limited our assessment to very early NIV failure. However, the literature describes different failure timeframes. Therefore, we cannot confirm that the predictors identified for early failure also apply to later forms of NIV failure [11]. In addition, we did not collect data on the time elapsed between the onset of clinical symptoms and NIV initiation, a parameter that is likely to be associated with NIV failure. Future studies should explore this aspect to gain a deeper understanding of its effect on patient outcomes. The short NIV trial duration (1–2 h) may be considered a limitation. However, in the ED, therapeutic decisions are often made rapidly to prevent further deterioration, and prolonged NIV trials may not always be feasible. Our study excluded patients with known DNR status, which may have limited the generalisability of our findings to this specific population. In clinical practice, NIV failure in these patients is often defined solely by patient intolerance or mortality, as tracheal intubation is not an option. Finally, the short follow-up period is also a limitation. Owing to the IRU network methodology (short inclusion and follow-up period across nearly 100 centres), the follow-up was limited to 7 days. As a result, we were unable to assess the impact of early clinical deterioration on ICU or hospital length of stay, or on 30-day mortality. These additional data could provide a more comprehensive understanding of patient outcomes and should be explored in future studies.

However, our study had several strengths. To our knowledge, this is the first multicentre prospective study to assess the NIV failure rate in the ED. Many mobile EMS and EDs participated in this study, ensuring the reproducibility and validity of the results. Although the definition of our primary outcome is debatable, we argue for its relevance in emergency settings. As mentioned previously, we assumed that the classical definition of NIV failure was not fully applicable in the ED because of the specific characteristics of ED patients.

In conclusion, early clinical deterioration under NIV is common in EDs, affecting nearly one out of two patients, and is associated with higher 7-day mortality.

aOR:

Adjusted Odds Ratio

aHR:

Adjusted Hazard Ratio

ARF:

Acute Respiratory Failure

BiPAP:

Bilevel Positive Airway Pressure

CI:

Confidence Interval

COPD:

Chronic Obstructive Pulmonary Disease

CPE:

Cardiogenic Pulmonary Edema

CPAP:

Continuous Positive Airway Pressure

ED:

Emergency Department

EMS:

Emergency Medical Services

FiO2:

Fraction of Inspired Oxygen

HACOR:

Heart rate, Acidosis, Consciousness, Oxygenation, and Respiratory rate

IQR:

Interquartile Range

IRU:

Initiative Recherche Urgences (Emergency Research Initiative)

NIV:

Non-invasive ventilation

PaCO2:

Partial Pressure of Carbon Dioxide

SFMU:

Société Française de Médecine d’Urgence (French Society of Emergency Medicine)

SpO2:

Peripheral Capillary Oxygen Saturation

STROBE:

Strengthening the Reporting of Observational Studies in Epidemiology

WHO:

World Health Organization

Download references

Conceptualization: NM, JL. Methodology: NM, JL, FB, CG. Analysis: NM, FB, JL. Writing – Original Draft: NM, JL. Resources: JL, FB, CG, FC. Review and editing: NM, JL, FB, CG, FC, JG, OM. Supervision: NM, OM. All authors approved the final version of the manuscript.

Correspondence to Nicolas Marjanovic.

All patients provided consent to participate in the study. In accordance with French law governing prospective non-interventional studies involving the collection of routine data, ethical committee approval was not required for the conduct of this study.

Not applicable.

The authors declare no competing interests.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions