Abstract:

Objective To discuss the effect factors of enteral nutrition feeding intolerance （FI） of the patients with severe acute pancreatitis （SAP）， and to construct and verify the risk prediction model. Methods The data of 246 SAP patients were collected and the patients were divided into tolerance group （n=143） and non-tolerance group （n=103） according to whether the enteral nutrition FI occurred.Age， intra-abdominal pressure， hypertriglyceridemia， hypoproteinemia， and other related data of the patients in two groups were compared， the predictive factors of FI were screened out by Logistic regression analysis，and the prediction model was constructed.The area under receiver operating characteristic （ROC） curve （AUC） and Hosmer-Lemeshow test were used to verify the prediction effect of the model. Ninety-two SAP patients were selected to verify the predictive effect of the model. Results A total of 6 predictors，such as hypertriglyceridemia （OR=1.962， 95%CI：1.088－3.538，P=0.025），hypoproteinemia（OR=1.920， 95%CI：1.049－3.516，P=0.034），acute physiological and chronic health status Ⅱ（APACHEⅡ）≥score 20 points （OR=3.118， 95%CI：1.720－5.653，P<0.01），intra-abdominal pressure ≥12 mmHg （OR=2.826， 95%CI：1.534－5.205，P=0.001），time to start enteral nutrition ≥72 h（OR=2.350，95%CI：1.179－4.683，P=0.015）， and addition of microecological agent （OR=0.379， 95%CI：0.209－0.685，P=0.001） were included. The formula of the model was Z=－1.206+0.674×hypertriglyceridemia+0.652×number of patients with hypoproteinemia+1.137×number of patients with APACHEⅡ score ≥20 points +1.039×number of patients with intra-abdominal pressure ≥12 mmHg+0.855×number of patients with time to start enteral nutrition ≥72 h－0.971×number of patients with　addition of microecological agent. The Hosmer-Lemeshow test results showed that χ²=5.985 and P=0.901， and the AUC of the prediction model was 0.793 （95%CI： 0.735－0.851， P<0.01），the Yoden index was 0.498，the optimal critical value was 0.499， the sensitivity was 0.818， and the specificity was 0.680. The agent verification results of the model showed that the AUC was 0.880 （95%CI： 0.811－0.948，P<0.01），the sensitivity was 0.774，the specificity was 0.867， and the accuracy rate was 81.52%. Conclusion The constructed risk prediction model of SAP enteral nutrition FI has high sensitivity and specificity， and good prediction efficiency.

Key words: Severe acute pancreatitis, Enteral nutrition, Feeding intolerance, Risk factor, Prediction model