[Development and validation of a predictive model for high-flow nasal cannula oxygen therapy failure in patients with type I respiratory failure].

Authors

Fei Ma, Yali Chao, Linyan Xu, Ziwen Wang

Published in

Zhonghua wei zhong bing ji jiu yi xue. Volume 38. Issue 9. Pages 812-818.

Abstract

To explore the independent risk factors for failure of high-flow nasal cannula oxygen therapy (HFNC) in patients with type I respiratory failure, and to establish and validate an effective risk prediction model.
A retrospective observational study was conducted, Patients with type I respiratory failure who received HFNC treatment in the intensive care unit (ICU) of the Affiliated Hospital of Xuzhou Medical University from January 2019 to December 2024 were selected as the research subjects. Clinical data within 24 hours after admission to the ICU were collected, including demographic information and general conditions, underlying diseases, early acute organ dysfunction, vital signs, laboratory examinations, disease severity scores, and treatment measures. Patients were divided into the HFNC success group and the HFNC failure group based on whether HFNC failure (endotracheal intubation and/or death during treatment) occurred. The overall cohort was randomly divided at a ratio of 7 : 3 into a training set and a validation set. Multivariate logistic regression analysis was used to screen independent predictors of HFNC failure, and a nomogram prediction model was constructed accordingly. The receiver operator characteristic curve (ROC curve), Hosmer-Lemeshow goodness-of-fit test, calibration curve, and decision curve analysis (DCA) were used to evaluate the model's discrimination, calibration, and clinical utility.
A total of 700 patients with type I respiratory failure were included, of whom 601 experienced HFNC success and 99 experienced HFNC failure, with a failure rate of 14.1%. Compared with the HFNC success group, patients in the HFNC failure group had lower height, proportion of coronary heart disease, serum sodium, oxygenation index, Glasgow Coma Score (GCS), and proportion of patients receiving human albumin, while exhibiting significantly higher incidence of acute kidney injury (AKI), heart rate, respiratory rate, blood urea nitrogen level, serum creatinine level, blood lactate level, Sequential Organ Failure Assessment (SOFA) score, and Acute Physiology and Chronic Health Evaluation II(APACHE II) score, as well as longer prothrombin time (all P<0.05). Multivariate logistic regression analysis showed that the presence of AKI [odds ratio (OR)=2.113, 95% confidence interval (95%CI) was 1.159-3.854, P=0.015] and elevated blood lactate level (OR=1.148, 95%CI was 1.027-1.283, P=0.015) were independent risk factors for HFNC failure, while elevated oxygenation index (OR=0.973, 95%CI was 0.963-0.983, P<0.001) and higher GCS score (OR=0.815, 95%CI was 0.760-0.874, P<0.001) were independent protective factors. Based on these four independent predictors, a nomogram prediction model was established. The ROC curves showed that the area under the curve (AUC) of the training set and validation set were 0.844 (95%CI was 0.794-0.895) and 0.794 (95%CI was 0.706-0.881), respectively. At a cutoff value of 0.160, the sensitivities were 73.9% and 60.0%, the specificities were 81.0% and 82.8%, and the accuracies were 80.0% and 79.5% in the training set and validation set, respectively. The Hosmer-Lemeshow test showed good fit for the model in both the training set (χ ²=8.636, P=0.374) and validation set (χ ²=6.652, P=0.575). The calibration curves indicated good agreement between the model predicted probabilities and the actual observations. DCA demonstrated that the model exhibited good clinical net benefit over a wide range of threshold probabilities.
The nomogram model constructed based on AKI, blood lactate level, oxygenation index, and GCS score has good predictive performance for the risk of HFNC failure in patients with type I respiratory failure, providing a reference for early identification of high-risk patients and optimization of clinical decision-making.

PMID:
42875239
Bibliographic data and abstract were imported from PubMed on 15 Oct 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.