Development and Validation of a Nomogram Model for Predicting the Risk of Weaning Failure in Prolonged Mechanical Ventilation Patients

Abstract

Background:

In Taiwan, the Respiratory Care Center (RCC) acts as a transitional unit for patients who are difficult to wean from mechanical ventilation in the ICU and thus require prolonged mechanical ventilation (PMV). This phase is critical for determining whether patients can be weaned from invasive mechanical ventilation. The main purpose of this study is to develop and validate an easy-to-use nomogram for predicting the risk of weaning failure and to compare its accuracy with commonly used weaning indicators, RSBI and MIP, to assist physicians in clinical decision-making.

Methods:

This retrospective study spanned from January 2010 to December 2021. We selected 1,060 patients who were difficult to wean due to respiratory failure from January 1, 2010, to December 31, 2015, and who were transferred from the ICU to the RCC. To test the accuracy of the model, another 612 patients with the same criteria were transferred to the RCC from January 1, 2016, to December 31, 2021, for internal validation. Clinical data collected on the day of transfer to RCC included basic demographics, etiology of mechanical ventilation, airway interfaces, blood test data, presence of cancer, GCS, APACHE II, Provent score, etc. Statistical analysis was conducted using SAS 9.4 and R software. (IRB number: 202201806B0).

Results:

Four variables were identified through multivariate analysis: etiology of mechanical ventilation, NLR, APACHE II, and cancer, and a nomogram was developed to predict the probability of weaning failure. The predictive capability of the model was assessed by the area under the ROC curve (AUC); the nomogram’s accuracy in the development group was AUC = 0.688 and in the validation group AUC = 0.626. Additionally, to assess the clinical applicability of the results from the development group, a comparison of accuracy with common weaning predictors was conducted; development group vs. RSBI (AUC = 0.688 vs. 0.601, P = .004), development group vs. MIP (AUC = 0.688 vs. 0.563, P < .001), showing statistically significant improvements over RSBI and MIP.

Conclusions:

The nomogram is more accurate than RSBI and MIP, providing the medical team with a tool to evaluate and intervene timely with appropriate treatment, thereby increasing the rate of weaning from ventilators and enhancing the quality of clinical care.