Risk prediction model for ventilator-associated pneumonia in ICU patients:a systematic review

YU Shenyan; MAO Xiaorong; LI Tao; PANG Shuwen; ZENG Xia

doi:10.12102/j.issn.2095-8668.2025.10.004

Abstract：Objective:To systematically evaluate the risk prediction model of VAP in patients with mechanical ventilation in the ICU and to provide a reference for the clinical selection of relevant risk assessment tools.Methods:Relevant literatures were retrieved from CNKI,WanFang Database,VIP,CBM,PubMed,Web of Science,the Cochrane Library,and EMbase.The search time limit was from the establishment of the database to April 1,2024.Literature screening and data extraction were independently conducted by two researchers.The risk of bias assessment tools related to the prediction model were used to evaluate the risk of bias and applicability of the included literature.Results:A total of 11 articles were included,involving 15 VAP prediction models for ICU patients with mechanical ventilation.The total sample size ranged from 192 to 10 431 cases,and the candidate predictors ranged from 10 to 42.The AUC of the 15 models ranged from 0.722 to 0.966,and the AUC of all of them was>0.7,showing good predictive performance.The evaluation results of the risk of bias showed that the model as a whole had a relatively high risk of bias.Conclusion:There are still many deficiencies in the predictive risk model for VAP in ICU patients with mechanical ventilation.In the future,more large-sample and multi-center prospective studies should be developed,and predictive models should be established through machine learning algorithms.Moreover,after its establishment,the validation research of the model should be carried out to further optimize and promote the high-quality model.