Objective  To develop and validate a nomogram for predicting the risk of weaning failure in elderly patients with severe pneumonia undergoing mechanical ventilation. Methods  A retrospective analysis was conducted on the clinical data of 330 elderly patients with severe pneumonia undergoing mechanical ventilation who were hospitalized in our hospital from July 2021 to July 2023. According to their weaning outcomes, they were divided into a successful group (n=213 ) and a failure group (n=117). Univariate analysis and multivariate non-conditional logistic regression analysis were used to explore the factors influencing the weaning failure of mechanical ventilation in elderly patients with severe pneumonia. Results  Univariate analysis showed that there were significant differences in age, smoking status, chronic obstructive pulmonary disease, ventilation time, albumin, D-dimer, and oxygenation index levels between the two groups (all P<0.05). Multivariate logistic regression analysis revealed that age ≥65 years, smoking, presence of chronic obstructive pulmonary disease, ventilation time ≥7 days, D-dimer ≥2 000 μg/L, and reduced oxygenation index were risk factors for weaning failure in the elderly patients with severe pneumonia. The nomogram model constructed based on these factors had an area under ROC curve of 0.970 (95%CI 0.952 - 0.989), and the calibration curve demonstrated good agreement between predicted and observed values. Conclusions  Age, smoking status, chronic obstructive pulmonary disease, ventilation time, D-dimer, and oxygenation index are influencing factors for weaning failure in elderly patients with severe pneumonia receiving mechanical ventilation. The nomogram model constructed based on these factors exhibits good discrimination and accuracy.

Citation： HU Miaomiao, LIU Li, ZHANG Shuqin. Establishment and validation of a risk prediction model for weaning failure in elderly patients with severe pneumonia undergoing mechanical ventilation. Chinese Journal of Respiratory and Critical Care Medicine, 2024, 23(5): 321-326. doi: 10.7507/1671-6205.202310039 Copy