Correlation between intestinal flora characteristics and immune function in COPD patients treated with non-invasive ventilator and its value in predicting ventilator-associated pneumonia: A prospective study.

To explore the relationship between intestinal flora characteristics and immune function in patients with chronic obstructive pulmonary disease (COPD) treated with noninvasive ventilator (NIV), and to analyze the factors affecting the occurrence of ventilator-associated pneumonia (VAP) in patients. A prospective study analyzed 280 COPD patients on noninvasive ventilation from August 2023 to August 2024, divided into VAP group and non-VAP group based on 48-hour VAP occurrence. The study explored links between gut microbiota and immune function and utilized binary logistic regression and a random forest model to predict VAP risk. A correlation was observed between intestinal flora characteristics and immune features in COPD patients, revealing complex interactions between bacterial subgroups and immune cell levels. Re-intubation, mechanical ventilation time ≥ 4 days, smoking, Escherichia coli, Enterococcus faecium, and Enterococcus faecalis were risk factors for VAP in COPD patients (OR = 2.800, 3.079, 4.665, 1.781, 1.342, and 1.600, all P < .05). In addition, Bifidobacterium, Lactobacillus, CD3+, CD4+, and CD8+ were protective factors (OR = 0.557, 0.801, 0.534, 0.349, and 0.134, all P < .05). Among these, the top 3 important factors were Lactobacillus, Enterococcus faecalis, and CD3+, (%IncMse × 10-2: 37.962%, 31.552%, 29.141%, respectively). The random forest model demonstrated significant predictive capability (P = .01, r2 =0.538); with the highest diagnostic performance under 10 factors (area under the curve = 0.908), including flora and immune characteristics. In addition, the area under the curve of the random forest model for predicting the occurrence of VAP in COPD patients was 0.857. This study preliminarily suggests that the characteristics of intestinal flora in COPD patients treated with NIV may be associated with immune function and may be involved in the occurrence of VAP. The random forest model based on intestinal flora has shown preliminary predictive value in predicting the occurrence of VAP, but more large-scale and multi-center studies are needed to further verify its clinical applicability.