Rationale: Excessive stress (distending pressure), strain (volume deformation), and drop in inspiratory alveolar pressure are proposed mechanisms for patient self-inflicted lung injury. Objectives: To dissect the influence of inspiratory effort, respiratory mechanics, and ventilation mode on lung stress, strain, and drop in inspiratory alveolar pressure; and explore their impact on oxygenation and lung compliance. Methods: International cohort study analyzing respiratory recordings (esophageal pressure) of patients with acute hypoxemic respiratory failure. Association between muscular pressure (Pmus), surrogates of stress (driving trans-alveolar pressure), strain (tidal volume), and inspiratory alveolar pressure relative to PEEP were explored with mixed-models, including interactions for ventilation mode, respiratory system elastance, and synchrony. Association between these and changes in oxygenation and lung compliance were explored. Measurements and main results: 60 patients from 15 centers represented 528 recordings (339,796 breaths). For each cmH(2)O Pmus increase there was an increase in driving trans-alveolar pressure (median[CI 95%] 0.28[0.27-0.29]cmH(2)O) and tidal volume (0.16[0.16-0.17]ml/kg of predicted body weight) and decrease in alveolar pressure (-0.25[0.24-0.6]cmH(2)O, p<0.001). Volume-control ventilation showed less increase in stress and strain surrogates than pressure-targeted modes, but more drop in alveolar pressure (p<0.001, Pmus:mode interaction). Breath-stacking was infrequent and associated with higher stress. Lower inspiratory alveolar pressure relative to PEEP was associated with subsequent worsening oxygenation (p=0.04) and higher stress with worsening lung compliance (p=0.023). Conclusion: Strong efforts are associated with high surrogates for lung stress, strain, and lower inspiratory alveolar pressure relative to PEEP, differently according to the mode of ventilation, being associated with subsequent worsening oxygenation and lung compliance.

The beneficial effects of prophylactic noninvasive ventilation (NIV) after extubation in patients without hypercapnia are uncertain. Our objective was to assess the effects of prophylactic NIV on reintubation among patients without hypercapnia at the time of extubation. Post hoc analysis of two multicenter clinical trials including high-risk patients (i.e., patients older than 65 years or with underlying cardiac/respiratory disease). Our analysis focused on the 829 patients without hypercapnia (PaCO2 ≤ 45 mmHg), the day of extubation who received NIV alternating high-flow nasal cannula (HFNC) oxygen or HFNC alone after extubation. The primary outcome was the proportion of patients who required reintubation within seven days following extubation. We used G-computation to robustly estimate the marginal causal effect of treatment on the risk of reintubation. After extubation, 540 patients (65

Background:In patients with septic shock as well as in the critically ill, the impact of hypothermia and core temperature changes during the first 24 h on mortality remains uncertain. In this post-hoc analysis of the SEPSISPAM trial, we investigated the association between hypothermia at inclusion, hypothermia trajectories and 90-day mortality in patients with septic shock. Methods:This post-hoc analysis of the SEPSISPAM trial included patients with septic shock enrolled within 6 h of vasopressors initiation. Core temperature was assessed every 2 h during the first 24 h. Hypothermia was defined by a temperature <36 °C. Mortality was assessed at day 90. Results:We included 691 patients from the SEPSISPAM trial, of whom 103 (14.9%) presented with hypothermia at inclusion. After adjustment for confounding factors, as compared with patients without hypothermia at inclusion, patients with hypothermia at inclusion had a higher mortality (HR 1.92, 95% CI [1.38-2.67], p < 0.001). Three groups of patients were identified according to the evolution of their core temperature: "without hypothermia" (86.6%), i.e., patients without any hypothermia during the first 24 h; "transient hypothermia" (10%), i.e., patients with hypothermia at inclusion and becoming normothermic during the first 24 h, and "persistent hypothermia" (3.4%), i.e., patients with sustained hypothermia both at inclusion and during the first 24 h. Compared with patients without hypothermia, the "persistent hypothermia" group had the highest mortality rate at day 90 (78.3%, HR 2.83 [1.62-4.95], p < 0.0001). Mortality at day 90 increased according to temperature trajectories, being highest in patients with persistent hypothermia (78%), followed by those with transient hypothermia (49%), and lowest in patients without hypothermia (40%). Conclusion:In patients with septic shock, hypothermia at inclusion and persistence of hypothermia during the first 24 h were associated with higher mortality at day 90. Mortality increased according to the course of hypothermia during the first 24 h, being highest in patients with persistent hypothermia, followed by those with transient hypothermia, and lowest in patients who never developed hypothermia.

Introduction Acute respiratory distress syndrome (ARDS) is a major public health problem, accounting for 23% of intubated patients and associated with high mortality rates. Although lifesaving, invasive mechanical ventilation can worsen lung injury when ventilator settings are poorly adjusted to lung physiology. We hypothesise that individualising ventilator settings via (1) the bedside assessment of lung recruitability using a one-breath derecruitment manoeuvre and measurement of airway opening pressure to set positive end-expiratory pressure (PEEP), (2) controlling the distending pressure and (3) controlling respiratory drive improves ARDS outcomes.Methods and analysis The CAreful Ventilation In ARDS trial is an investigator-led multicentre (33 centres in eight countries), open-label, randomised controlled basket trial comparing two ventilation strategies in two subpopulations of moderate-to-severe ARDS: induced or not by COVID-19. A total of 740 patients will be randomised (370 in each substudy) in a 1:1 ratio to individualised ventilator settings or to using traditional PEEP to inspired fraction of oxygen tables for PEEP setting. Indications for proning and weaning strategies are similar in both arms. The primary outcome is all-cause mortality at day 60. Secondary outcomes include duration of mechanical ventilation, duration of intensive care unit (ICU) and hospital stay, organ dysfunction, barotrauma and mortality in ICU, at day 28 and in hospital.Ethics and dissemination Ethics approval has been obtained for all participating centres: Unity Health Toronto Research Ethics Board (for three centres: St Michael’s Hospital, Toronto General Hospital and Toronto Western Hospital); Comité de Ética de Investigación con Medicamentos del Hospital Universitari Vall d’Hebron; Comité de protection des personnes Ile de France III; Comité d'Ética de la Investigatción con Medicamentos de la Fundació de Gestió Sanitària del Hospital de la Santa Creu i Sant Pau; Comitato Etico—Fondazione Policlinico Gemelli; Comitato Etico di Area Vasta Emilia Centro; NYU Langone Health Institutional Review Board; Comité Ético Científico de Ciencias de la Salud; Il Comitato Etico Area 1 dell’Azienda Ospedaliero-Universitaria ‘Ospedali Riuniti’ di Foggia; HIGA ‘Eva Perón’ Comité de Bioética; Comité de Revisión Institucional del Hospital Británico Comité de Ética en Investigación; Complejo Médico Churruca-Visca Comité de Ética Biomédica; Comité de Ética SATI Comité de Ética en Investigación; Comité de Ética en Investigación del CEMIC; Comité de Ética SATI Comité de Ética en Investigación; Medical Research Ethics Committees United. Findings will be disseminated in peer review journals and conference presentations.Trial registration number NCT03963622.

Background:Whether or not a direct tracheostomy without any prior weaning attempt is beneficial in patients intubated for Guillain-Barré syndrome (GBS) remains to be determined. Our objective was to determine whether direct tracheostomy results in earlier or later ventilator weaning compared to standard weaning. Methods:Multicenter, retrospective cohort study including all patients intubated for GBS over a 10-year period (2014-2023). We compared patients having undergone direct tracheostomy without any prior weaning attempt with those having undergone standard weaning (spontaneous-breathing trial or extubation attempt). We also compared direct tracheostomy with tracheostomy after weaning failure. The primary outcome was the proportion of patients alive and free from mechanical ventilation 28 days after weaning initiation. Results:Overall, 221 patients (45%) underwent direct tracheostomy and 273 (55%) standard weaning. Compared to patients with standard weaning, those with direct tracheostomy were less likely to be alive and free from mechanical ventilation at day 28 (43% vs. 83%, p < 0.001). Using G-computation to estimate the marginal causal effect of the weaning strategy, standard weaning was more effective than direct tracheostomy in reducing the risk of remaining under mechanical ventilation at day 28: -39.5% [95% Confidence Interval, -47.3% to -31.5%]. Even when compared to patients having undergone tracheostomy after weaning failure, those with direct tracheostomy were less likely to be alive and free from mechanical ventilation 28 days after the tracheostomy procedure. Interpretation:A direct tracheostomy without any prior weaning attempt might significantly delay the ventilator weaning process in patients intubated for GBS. Clinical Trial Registration:NCT07022028.

BACKGROUND:Data are needed on the effect of oxygen delivered through a high-flow nasal cannula, as compared with standard oxygen therapy, on intubation and mortality in patients with acute hypoxemic respiratory failure. METHODS:In this multicenter, open-label trial, we randomly assigned patients who had acute hypoxemic respiratory failure to receive high-flow-oxygen or standard-oxygen therapy. All the patients had a ratio of the partial pressure of arterial oxygen to the fraction of inspired oxygen of 200 or less, a respiratory rate of more than 25 breaths per minute, and pulmonary infiltrate on chest imaging. The primary outcome was death by day 28. RESULTS:A total of 1116 patients underwent randomization. Of these patients, 1110 (556 in the high-flow-oxygen group and 554 in the standard-oxygen group) were included in the analysis. Mortality at day 28 was 14.6% (in 81 of 556 patients) in the high-flow-oxygen group and 14.6% (in 81 of 554 patients) in the standard-oxygen group (difference, -0.05 percentage points; 95% confidence interval [CI], -4.21 to 4.10; P = 0.98). The incidence of intubation by day 28 was 42.4% (in 236 of 556 patients) in the high-flow-oxygen group and 48.4% (in 268 of 554 patients) in the standard-oxygen group (difference, -5.93 percentage points; 95% CI, -11.78 to -0.08). Serious adverse events (cardiac arrest or pneumothorax) occurred during spontaneous breathing in 13 patients (2.3%) in the high-flow-oxygen group and in 6 patients (1.1%) in the standard-oxygen group. CONCLUSIONS:Among patients with acute hypoxemic respiratory failure, the use of oxygen delivered through a high-flow nasal cannula did not significantly reduce mortality at day 28. (Funded by the French Ministry of Health and Fisher and Paykel Healthcare; SOHO ClinicalTrials.gov number, NCT04468126.).

Positive end-expiratory pressure (PEEP) remains a cornerstone of acute respiratory distress syndrome management. However, randomized trials of PEEP strategies have yielded conflicting results, reflecting inter-individual heterogeneity. PEEP-induced alveolar recruitment enhances lung protection by redistributing tidal volume over a larger functional lung and reducing atelectrauma and bronchiolotrauma. Conversely, in poorly recruitable lungs, PEEP predominantly increases stress in already aerated regions, potentially exacerbating ventilator-induced lung injury and cardiovascular compromise. The net effect of PEEP reflects a patient-specific balance between recruitment and overdistension, shaped by hemodynamic tolerance. Clinical markers traditionally used for assessing PEEP response (gas exchange, compliance, and driving pressure), while informative at a population level, remain unreliable for clinical decision-making at the individual level. With the exception of specific subgroups (e.g., obesity), patients with PaO2/FiO2 > 200 mmHg are unlikely to have substantial alveolar collapse amenable to recruitment and can be managed with lower PEEP (e.g., 5–8 cmH2O) facilitating transition to assisted ventilation. In contrast, higher PEEP is more likely to benefit patients with PaO2/FiO2 ≤ 200 mmHg, in whom a physiologically grounded framework should begin with assessment of lung recruitability (e.g., computed tomography, gas recruitment indices). This should account for airway closure, since PEEP below the airway opening pressure does not modify lung volume and may confound respiratory mechanics interpretation. In patients with significant recruitability, PEEP titration should integrate global measures (plateau pressure, stress index) with regional monitoring (electrical impedance tomography, transpulmonary pressure) when available. Conversely, poorly recruitable patients are unlikely to benefit from higher PEEP and can be managed with lower levels.

Importance: Recent reports have highlighted an intense influenza activity related to the circulation of the influenza A(H3N2) subclade k variant. There is no data available on the impact of the emergence of H3N2 subclade k on the severity of the 2025/2026 epidemic or on the clinical phenotype of patients requiring admission to the intensive care unit (ICU). Objective: To compare the clinical presentation, hospital mortality and virological characteristics of patients with laboratory-confirmed influenza infection included in French intensive care units during the 2025/2026 epidemic season with those of patients admitted during the 2024/2025 season. We also aimed at measuring the impact of the A(H3N2) subtype on hospital mortality during the 2025/2026 season. Design: Prospective, multicenter, observational SEVARVIR cohort study including patients admitted during the 2024/2025 and 2025/2026 influenza seasons. Setting: Forty-two French ICUs Participants: Adult patients with laboratory-confirmed influenza infection Interventions: none Main Outcomes and Measures: The primary outcome measure was in-hospital mortality. Results: Patients admitted in intensive care units for influenza in 2024/2025 (n=360) and 2025/2026 (n=325) were included in the French nationwide prospective multicentre SEVARVIR study. There was no significant difference in day28 mortality between the seasons (12.7%, n=45/355 vs 16.5% n=28/170; p=0.28). In the 2025-26 season, 49% had the A(H1N1) subtype and 51% the A(H3N2) subtype (k subclade: 77%). The univariable Cox analysis revealed that patients infected with A(H3N2) viruses were at greater risk of death over time. Multivariable Cox analysis revealed that during the 2025-2026 season, age (adjusted hazard ratio, aHR=1.05 [1.00;1.11]; p=0.046) and the clinical frailty scale (aHR=1.82 [1.26;2.72]; p=0.001) were associated with an increased risk of death. The A(H3N2) subtype was not associated with an increased risk of death (aHR=1.13 [0.32;4.51]; p=0.85). Phylogenetic analyses from our ICU cohort together with 300 contextual sequences from community-acquired influenza cases collected during the same period showed no clustering according to severity. Conclusions and Relevance: This French national prospective observational study, found that the emergence of the influenza A(H3N2) subclade K was associated with an increased risk of death in univariable but not multivariable analysis, adjusting for host-related factors. ### Competing Interest Statement Slim Fourati has served as a speaker for GlaxoSmithKline, AstraZeneca, MSD, Pfeizer, Cepheid and Moderna, SANOFI. Jean-Michel Pawlotsky has served as an advisor or speaker for Abbvie, Arbutus, Assembly Biosciences, Gilead and Merck. Etienne Audureau has received fees for lectures from Alexion, Sanofi, Gilead and Pfizer. His hospital has received research grant from Pfizer, MSD and Alexion. Remi COUDROY received fees and reimbursement for travel expenses from Fisher and Paykel Healthcare. Sylvie LARRAT received reimbursement for travel expenses from GILEAD and Biomerieux. All others authors have nothing to disclose. ### Clinical Protocols ### Funding Statement The SEVARVIR study has been funded by the EMERGEN consortium ANRS Maladies Infectieuses Emergentes (ANRS0153). ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The study was approved by the Comite de Protection des Personnes Sud-Mediterranee I (Number EudraCT/ID-RCB: 2021-A02914-37). Informed consent was obtained from all patients or their relatives. The study was conducted in accordance with the 1964 Helsinki Declaration and subsequent amendments. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes the clinical datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request

Clinical practice guidelines recommend prophylactic noninvasive ventilation (NIV) after extubation in high-risk patients. The effects of high-flow nasal cannula (HFNC) oxygen during breaks from NIV after extubation have been poorly explored. Our objective was to assess whether HFNC rather than standard oxygen during breaks from NIV may prevent extubation failure. Observational study based on two multicenter clinical trials including patients at high-risk of extubation failure (> 65 years or with underlying cardiac or respiratory disease). We included the 1077 patients who received prophylactic NIV after extubation, alternating with HFNC (NIV/HFNC) or standard oxygen (NIV/O2). The primary outcome was the proportion of patients who failed extubation (reintubation or death within the seven days following extubation). We used G-computation to estimate the causal effect of HFNC on the risk of extubation failure, while accounting for confounding factors. After extubation, 655 patients (61

BACKGROUND:Invasive meningococcal disease (IMD) is traditionally associated with younger populations, but its impact on older adults is rising. We aimed to describe the clinical characteristics and outcomes of critically ill older patients with IMD and identify risk factors for in-hospital mortality. METHODS:We conducted an ancillary analysis of the French nationwide multicenter RETRO-MENINGO cohort (2016-2024). All adults admitted to 102 ICUs with microbiologically confirmed IMD were included and stratified by age (≥65 vs. <65 years). The primary outcome was all-cause in-hospital mortality. Multivariable logistic regression identified factors independently associated with death. RESULTS:Among 654 patients, 114 (17%) were aged ≥65 years, of whom only 0.9% were reported as vaccinated. Compared to younger adults, older patients presented more frequently with hemodynamic failure and less often with classic purpura or meningeal symptoms. Serogroup W predominated in the older group, while serogroup B was more common in younger patients. Older adults required more organ support and had significantly higher in-hospital mortality (28.1%, n=32/114 vs. 9.6%, n=52/540 p<0.001). After adjustment, age ≥65 years remained independently associated with in-hospital mortality (aOR 2.99; 95% CI 1.57-5.72; p<0.001), while administration of a third-generation cephalosporin before ICU admission was protective (aOR 0.45; 95% CI, 0.24-0.84; p=0.01). CONCLUSIONS:Critically ill older adults with IMD exhibit atypical clinical features, a high prevalence of serogroup W, and nearly triple the in-hospital mortality of younger patients. These findings emphasize the need for high clinical suspicion, rapid antibiotic therapy, and potential expansion of vaccination strategies to include older populations.

Background:Evidence comparing high-flow nasal cannula (HFNC) and non-invasive ventilation (NIV) in acute hypercapnic respiratory failure remains controversial. We compared their short-term effects on breathing effort, ventilation, CO2 clearance, and preference. Methods:A randomized, crossover, non-inferiority trial was conducted in patients with stabilized hypercapnic exacerbation requiring NIV or HFNC. Baseline oxygen therapy was followed by a randomized sequence of NIV and HFNC at 30 and 50 L.min-1. The primary endpoint was to assess non-inferiority of HFNC 50 L.min-1 compared to NIV. Diaphragm, parasternal intercostal, and transversus abdominis muscle activity were assessed using thickening fraction (TF) and the product of TF and respiratory rate (TF•RR). Ventilation was evaluated using electrical impedance tomography and transcutaneous partial pressure of carbon dioxide (tcCO2). Results:21 patients (mean ± SD age 69 ± 11 years, 82% COPD) were enrolled. In 17, diaphragm thickening fraction (TFdi) was available: HFNC 50 L.min-1 was non-inferior to NIV in reducing TFdi (p = 0.122, 95% CI: -19.1-3.4), as was HFNC at 30 L.min-1 (p = 0.413, 95% CI: -17.0-5.7). Only HFNC 50 L.min-1 reduced TFdi•RR (p = 0.036) and respiratory rate compared to baseline (p = 0.001). HFNC at 50 L.min-1 decreased the baseline TFdi by 18% ± 36% (p = 0.033), whereas NIV did not decrease it. HFNC and NIV reduced tcCO₂ compared to baseline. Minute ventilation and the estimated ventilatory ratio were lower with HFNC than NIV (p < 0.01). HFNC was the preferred strategy by the patients. Conclusions:In stabilized hypercapnic exacerbation, HFNC and NIV reduced tcCO₂, but only HFNC lowered ventilatory ratio and minute ventilation. HFNC at 50 L.min-1 reduced diaphragm activity and was non-inferior to NIV in this regard, while being preferred by patients.

Background:Sleep disturbances are associated with a poor prognosis in intensive care units (ICUs). Nursing care during the night can further disturb patient sleep. We hypothesized that nursing rounds and care would be less harmful to sleep if guided by a real-time sleep monitoring system. Methods:This was a quasi-experimental, prospective, before-after study including non-sedated ICU patients. In the usual care group (first group recruited), nursing rounds and care were performed every four hours as daily practice, regardless of the patient's sleep/wake status. In the sleep-guided care group (second group recruited), nursing rounds and care were guided by a sleep-EEG monitoring system indicating the patient's sleep status in real time on a tablet positioned at the entrance of the room. When patients fell asleep, the tablet displayed a symbol asking caregivers to postpone non-urgent care. Otherwise (patient awake), nursing care was encouraged. The number of room entries was measured using an entry-exit counter, synchronized with sleep-EEG recorder. The system stored EEG in both groups. Two experts, blind to the patients' groups, provided consensual sleep scoring. Primary outcome was continuous sleep (i.e. time spent in long sleep episodes). Results:Forty-six patients mainly admitted for acute respiratory failure (72% of cases) were analysed. Patients' characteristics did not significantly differ between groups. The proportion of room entries while patients were asleep decreased from 22% [4-32] in usual care group to 6% [0-13] in sleep-guided care group (p = 0.015). Continuous sleep was longer in the sleep-guided care group (20 patients) compared to the usual care group (26 patients): 170 min [75-240] vs. 80 min [53-128] (p = 0.03). Deep sleep was likewise longer in the sleep-guided care group. Conclusion:Our study reports for the first time that real-time sleep monitoring can guide nursing care and improve sleep quality in ICUs.

Invasive meningococcal disease (IMD) is a rare but potentially fatal infection caused by Neisseria meningitidis. In adults requiring admission to intensive care unit (ICU), IMD typically presents with two distinct clinical presentations: neurological (meningitis) and hemodynamic (sepsis). These presentations are often conflated, despite the differences in pathophysiology and outcome. RETRO-MENINGO is a nationwide, multicentre, retrospective cohort study conducted in 102 French ICUs between January 1, 2016, and December 31, 2024. Adults (≥ 18 years) admitted to the ICU with a microbiologically confirmed IMD were categorised as having either a neurological or a hemodynamic presentation according to the main reason for ICU admission. The primary outcome was day-60 mortality in each presentation. Of 654 patients (median age 33 years [IQR 21–56]; 63.5

Extubation failure leading to reintubation is associated with high mortality. In patients at high-risk of extubation failure, clinical practice guidelines recommend prophylactic non-invasive ventilation (NIV) over high-flow nasal oxygen (HFNO) immediately after extubation. However, the physiological effects supporting the beneficial effect of NIV have been poorly explored. We hypothesized that NIV may reduce patient inspiratory efforts to a greater extent than HFNO after extubation. In a prospective physiological study, patients at high-risk of extubation failure (> 65 years old or underlying cardiac or respiratory disease) were included to receive after planned extubation prophylactic NIV and HFNO in a randomized crossover order, followed by standard oxygen. Inspiratory efforts were assessed by calculation of the simplified esophageal pressure–time-product per minute (sPTPes in cmH2O s/min). Tidal volumes, distribution and homogeneity of ventilation were estimated using electrical impedance tomography. Twenty patients were retained in the analysis. Inspiratory efforts were lower with NIV than with HFNO (sPTPes 196 cm H2O s/min [116–234] vs. 220 [178–327], p < 0.001) whereas tidal volumes were larger with NIV than with HFNO (8.4 mL/kg of predicted body weight [6.7–9.9] vs. 6.9 [5.3–8.6], p = 0.005). There was a non-significant increase in dorsal region ventilation under NIV compared to HFNO. In patients at high-risk of extubation failure, prophylactic NIV significantly decreased inspiratory efforts with increased tidal volumes compared to HFNO. The clinical benefits of NIV to prevent reintubation in patients at high-risk may be mediated by these physiological effects. Trial registration Clinicaltrials.gov: ID NCT04036175), retrospectively registered 17 June 2019.

Severe forms of Guillain–Barré syndrome (GBS) and myasthenia gravis (MG) may result in prolonged mechanical ventilation in ICUs. Data on weaning from mechanical ventilation are scarce, and our objective was to assess ventilator weaning characteristics in a large-scale multicenter study. Multicenter retrospective cohort study including all patients intubated for at least 48 h for GBS or MG over a 10-year period (from January 2014 to December 2023). The primary outcome was the proportion of patients having experienced prolonged weaning (at least 7 days). Some 886 patients admitted to 47 ICUs in France were retained in the analysis, including 513 (58

Background Mortality of immunocompromised patients is particularly high in intensive care units (ICUs) and mainly depends on severity at admission. Moreover, mortality is also high during the months following ICU discharge. The reasons for these poor outcomes after ICU discharge have not been adequately studied. Research question We hypothesized that the factors associated with poor outcomes after ICU discharge of immunocompromised patients would be different from those associated with in-ICU mortality. Study design and methods This is a post-hoc analysis of a multicenter clinical trial comparing two noninvasive oxygenation strategies in immunocompromised patients admitted to ICU for acute hypoxemic respiratory failure. Multivariable analyses were performed to determine early factors (i.e within 6 h of admission) associated with in-ICU mortality, as well as factors associated with poor functional outcomes (i.e death or survival with poor performance status) at 6 months, only in ICU survivors. Results Among the 299 patients analyzed, the mortality rate was 31% (94 patients) in the ICU and 49% at 6 months (146 patients). Solid cancer (adjusted odds ratio 2.92 [95% confidence interval, 1.22–7.28]), severity SOFA score at admission (aOR 1.29 [1.14–1.48]), the extent of pulmonary infiltrates on chest X-ray (aOR 1.57 [1.17–2.15]) and increased discomfort one hour after initiation of noninvasive respiratory support (aOR 2.08 [1.12–3.85]) were independently associated with in-ICU mortality. Out of the 202 ICU survivors whose performance status was reported, solid cancer (aOR 3.03 [1.33–9.09]) and poor performance status before ICU admission (aOR 2.43 [1.03–5.88]) were both associated with poor outcome at 6 months, independently from the decision to forgo life-sustaining therapies (aOR 5.88 [2.17–20.00]). Interpretation Whereas in-ICU mortality of immunocompromised patients with acute respiratory failure was mainly driven by severity, poor outcomes at 6 months were mainly driven by performance status before ICU admission. Solid cancer was independently associated with both poor short as well as longer-term outcomes.Trial registration Clinical trial registration: NCT04227639