Acute Respiratory Failure, Electrical Impedance Tomography (EIT), High-Flow Nasal Oxygen Therapy
Conditions
Keywords
Acute hypoxemic respiratory failure, High-flow nasal oxygen, Electrical impedance tomography, EIT-derived flow index, Prediction model, Multicenter prospective validation
Brief summary
This is a single-center, ambispective observational cohort study of adults with acute hypoxemic respiratory failure (AHRF) receiving high-flow nasal oxygen (HFNO) as initial respiratory support. The study includes three stages. Stage 1 (retrospective derivation): the first 57 consecutive patients (May-September 2025) were retrospectively included to develop and lock a dual-domain logistic model using baseline EFI, baseline PaCO2, and 0-30 minute changes in respiratory rate (ΔRR) and SpO2 (ΔSpO2). Stage 2 (retrospective temporal validation): the later 58 patients (October 2025-March 2026) were retrospectively included, and the frozen model was directly applied without refitting; results of Stages 1 and 2 have been reported. Stage 3 (prospective validation): an independent prospective validation cohort is currently being enrolled, and the frozen model will be applied without refitting. The primary outcome is endotracheal intubation within 72 hours after HFNO initiation. This registration covers the complete study; the retrospective stages have been reported, and the prospective stage is ongoing.
Detailed description
This was a single-center, ambispective observational cohort study conducted at Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, between May 2025 and March 2026, with an ongoing prospective validation stage. Adults with AHRF receiving HFNO as initial respiratory support were included. Patients with an immediate indication for endotracheal intubation, severe hemodynamic instability, marked consciousness disturbance or inability to cooperate with monitoring, or missing/inadequate EIT data were excluded. Clinical, physiological, blood gas, and EIT data were collected at baseline before HFNO and at 30 minutes after HFNO initiation. EFI was calculated from the global EIT impedance curve. Stage 1 (retrospective derivation): the first 57 consecutive patients were included retrospectively; baseline EFI, baseline PaCO2, ΔRR, and ΔSpO2 were used to develop and lock a dual-domain logistic model. Stage 2 (retrospective temporal validation): the later 58 patients were included retrospectively; the frozen model coefficients were directly applied without refitting. Results of Stages 1 and 2 have been reported. Stage 3 (prospective validation): an independent multicenter prospective validation cohort is currently being enrolled. The same inclusion/exclusion criteria, 30-minute reassessment protocol, and frozen model will be applied without refitting. The primary endpoint is endotracheal intubation within 72 hours after HFNO initiation. The retrospective stages were approved by the Ethics Committee of Ruijin Hospital (Approval No. 2025232), and informed consent was waived. The prospective validation stage was approved by the same committee (Approval No. (2026592), and written informed consent is obtained from all prospectively enrolled participants.
Interventions
Patients received HFNO as part of standard clinical care for acute hypoxemic respiratory failure. The treatment was titrated by the clinical team based on physiological parameters (SpO2, RR, ABG), aiming to maintain SpO2 ≥ 92% and reduce signs of respiratory distress.
PulmoVista 500 EIT device (Dräger Medical, Lübeck, Germany) was used for continuous real-time monitoring of regional lung ventilation and to derive the EIT-derived flow index (EFI) at baseline and 30 minutes.
Sponsors
Ruijin Hospital
Study design
Observational model
COHORT
Time perspective
OTHER
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
(must meet all): * Age ≥18 years * Acute hypoxemic respiratory failure (AHRF) requiring HFNO initiation * At least one of the following high-risk criteria: a) PaO₂/FiO₂ ≤200 mmHg or FiO₂ ≥0.40 to maintain SpO₂ ≥92%; b) Respiratory rate ≥25/min; c) APACHE II score ≥12; d) Bilateral infiltrates on chest imaging
Exclusion criteria
* Neuromuscular disease affecting spontaneous breathing * Pregnancy * Contraindication to EIT (e.g., chest wall wounds, pacemaker) * Do-not-intubate order * Hypercapnic respiratory failure as primary diagnosis (PaCO₂ \>50 mmHg with pH \<7.30 at baseline, unless mixed mechanism)
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| HFNO Failure (Endotracheal Intubation Within 72 Hours) | Within 72 hours after HFNO initiation. | HFNO failure was defined as endotracheal intubation within 72 hours after HFNO initiation due to refractory hypoxemia, progressive respiratory acidosis, severe respiratory distress, or hemodynamic instability. Escalation to noninvasive ventilation (NIV) without endotracheal intubation and within-tier adjustments (increasing flow or FiO2 without changing support modality) were NOT considered the primary failure event. The model is intended for reassessment at 30 minutes, but the outcome is assessed within 72 hours.Unit of Measure: Proportion of participants. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Model Discrimination in Retrospective Temporal Validation | Within 72 hours after HFNO initiation. | Area under the receiver operating characteristic curve (AUC) of the frozen dual-domain model for predicting endotracheal intubation within 72 hours in the retrospective temporal validation cohort.Unit of Measure: AUC (dimensionless, 0-1). |
| Model Calibration in Retrospective Temporal Validation | Within 72 hours after HFNO initiation. | Calibration intercept and slope of the frozen dual-domain model in the retrospective temporal validation cohort. Unit of Measure: Dimensionless. |
| Model Discrimination in Prospective Validation | Within 72 hours after HFNO initiation. | AUC of the frozen dual-domain model for predicting endotracheal intubation within 72 hours in the ongoing prospective validation cohort.Unit of Measure: AUC (dimensionless, 0-1). |
| Model Calibration in Prospective Validation | Within 72 hours after HFNO initiation. | Calibration intercept and slope of the frozen dual-domain model in the ongoing prospective validation cohort. Unit of Measure: Dimensionless. |
Countries
China
Contacts
CONTACTrui zhang, Dr
CONTACTHongping Qu
STUDY_CHAIRHongping Qu
Department of Critical Care Medicine，Ruijin Hospital，Shanghai Jiao Tong University School of Medicine
Outcome results
None listed