Chinese Hepatolgy ›› 2025, Vol. 30 ›› Issue (8): 1133-1137.

• Other Liver Diseases • Previous Articles     Next Articles

Establishment and validation of a column chart model for individualized prediction of parenteral nutrition associated liver disease in elderly patients after gastrointestinal surgery

ZHANG Zao-chun1,LEI Ping2, SUN Zu-jian3   

  1. 1. Department of Pharmacy,Chongzhou People′s Hospital,Chongzhou 611230,China;
    2. Department of Oncology, Chongzhou People′s Hospital,Chongzhou 611230,China;
    3. Department of Nutrition, Chongzhou People′s Hospital,Chongzhou 611230,China
  • Received:2025-02-20 Published:2025-09-19

PDF (PC)

8

Abstract: Objective To construct and validate a column chart prediction model for postoperative parenteral nutrition (PN) associated liver disease(PNALD) in elderly gastrointestinal surgery patients. Methods The complete data were collected from 368 elderly postoperative PN patients in our hospital from March 2018 to March 2022. Patients were divided into two groups based on whether PN related liver injury occurred: the injury group and the control group. A column chart model was established based on data of 368 patients. Data from 158 patients from March 2022 to March 2024 were collected for external validation of the established model. The multi factor logistic regression model was used to analyze the influencing factors. The R3.6.3 software and RMS package were used to construct a column chart prediction model for PN related liver injury in elderly patients after gastrointestinal surgery. The discrimination and consistency of the column chart model ere evaluated using receiver operating characteristic (ROC) curves and calibration curves. Results The albumin (ALB)[(31.75±7.92)g/L] and prealbumin (PAB)[(128.73±32.56)mg/L] in the injury group were lower than those in the control group[(37.82±8.71)g/L、(175.96±38.71)mg/L], while urea nitrogen (BUN), serum creatinine (Scr), C-reactive protein (CRP), fasting time[(15.68±4.28)d], and PN time[(12.85±2.57)d] were higher than those in the control group[(11.04±2.63)d、(9.11±1.84)d] (P<0.05). Multivariate logistic regression showed that ALB(OR=0.910,95CI%=0.842~0.983), PAB(OR=0.951,95CI%=0.931~0.972), PN time(OR=2.225,95CI%=1.535~3.226), and fasting time(OR=1.725,95CI%=1.379~2.157) were all influencing factors of PN related liver injury (P<0.05). The column chart model showed that for every 1g/L decrease in ALB, the weight increased by 3.42 points; for every 10 mg/L decrease in PAB, the weight increased by 7.14 points; the weight increased by 7.81 points for every additional day of fasting time; the weight increased by 5.73 points for every 1-day increase in PN time. The H-L fit test showed a χ2 of 8.129 and a P-value of 0.421, indicating good consistency. The area under the ROC curve was 0.881 (95% CI: 0.832~0.929), indicating good discrimination of the model. The external validation results showed good model consistency (χ2=5.586, P=0.694) and discriminability, with an AUC of 0.931 (95% CI: 0.893~0.969). Conclusion Alb, PAB, fasting time, and PN time are all influencing factors of PN-associated liver disease in elderly PN patients undergoing gastrointestinal surgery. The column chart risk prediction model established in this study has high practicality.

Key words: Gastrointestinal surgery, Parenteral nutrition, Parenteral nutrition-associated liver disease, Column chart