Abstract: Objective To systematically evaluate the prediction models of enteral feeding intolerance（EFI） in critically ill patients to provide references for subsequent model development. Methods Relative articles on the prediction models of EFI in critically ill patients were retrieved from CNKI,Wanfang,VIP,CBM,PubMed,Embase,Web of Science,Cochrane Library and CINAHL from the establishment of the databases to November 19,2023. Language limited to Chinese and English. Two researchers independently screened,extracted data,and assessed bias risk and applicability by PROBAST. Results 19 studies were included,involving 24 models with an area under the subject working characteristic curve of 0.700 ~ 0.940. The independent predictors repeatedly reported were APACHE Ⅱ score,intra-abdominal pressure and serum albumin levels. 19 studies had good applicability but high risks of bias due to existing data sources,outcome definitions incorporating the predictors,small sample size,inappropriate handling of continuous variables and missing data,incomplete univariate analysis for predictors selection,and incomplete model performance assessment. Conclusion Existing predictive models for EFI in critically ill patients are still in the initial stage,as they have methodological defects. Future research should incorporate more sensitive and reproducible indicators for evaluating gastrointestinal dysfunction and follow the TRIPOD guidelines for reporting the results.

Key words: Enteral Nutrition, Feeding Intolerance, Prediction Model, Systematic Review, Critical Care Nursing