Development of a multidimensional predictive model for antimicrobial resistance in severe pneumonia in mechanically ventilated children based on metagenomics and end-tidal carbon dioxide monitoring
Objective To evaluate the performance of metagenomic sequencing(mNGS)combined with end-tidal carbon dioxide(ETCO2)monitoring in predicting antibiotic resistance in mechanically ventilated children with severe pneumonia.Methods A total of 112 children with severe pneumonia who were enrolled from July 2022 to April 2024 at Guiyang Children's Hospital.Clinical data and bronchoalveolar lavage fluid(BALF)samples were collected.Pathogens and resistance genes were detected using mNGS.A decision tree algorithm,combined with ETCO2 data,was used to construct a resistance prediction model.Results The main pathogens were Streptococcus pneumoniae,Haemophilus inflluenzae,Staphylococcus aureus,and respiratory syncytial virus.Key resistance genes included bla_TEM,mecA,ermB,and vanA.Initial and highest ETCO2 levels were significantly correlated with several resistance genes(P＜0.05).The model accuracy was 0.859,with an AUC of 0.896.Conclusion mNGS combined with ETCO2 monitoring significantly improves the accuracy of resistance prediction,providing a novel approach for individualized treatment of severe pneumonia.
NETL
NSTL
万方数据
