Related Experiment Video
Updated: May 6, 2026
Standardized Colon Ascendens Stent Peritonitis in Rats - a Simple, Feasible Animal Model to Induce Septic Acute Kidney Injury
Published on: February 15, 2022
Construction and Validation of a Risk Prediction Model for Acute Gastrointestinal Injury in Non-ICU Elderly
1General Medical Department, The Third Hospital of Shanxi Medical University (Shanxi Bethune Hospital), Taiyuan City, China. xujiajia392@163.com.
Background:
Acute gastrointestinal injury (AGI) has a relatively high prevalence among elderly critically ill patients in non-intensive care units (non-ICUs), and significantly influences their clinical outcomes. Therefore, it is important to identify people at risk for AGI and take preventive measures as early as possible.
Objective:
We aimed to construct and validate a risk prediction model for AGI in non-ICU elderly critically ill patients.
Design:
Case-control study.
Participants:
In total, 538 elderly critically ill patients admitted to the general medical department of a tertiary hospital in Shanxi from April 2021 to May 2024.
Main Measures:
Influential factors for AGI were determined using univariate and multifactorial logistic regression analyses. We constructed a risk prediction model and created a nomogram. The bootstrap resampling method was utilized for internal validation. A total of 151 patients from different time periods were selected for the external validation.
Key Results:
The multifactorial logistic regression analysis revealed that the independent predictors for AGI were the duration of antibiotic use, number of vasoactive drugs, delayed enteral nutrition, age-corrected Charlson comorbidity index, and white blood cell count, all of which were included in the model and created a nomogram. The Omnibus test showed that the overall efficacy of the model was good (P < 0.001). The area under the receiver operating characteristic curve (AUC) was 0.807, the corrected AUC was 0.806, and the AUC was 0.796 for external validation, indicating good model discrimination. The calibration curves and Hosmer-Lemeshow tests revealed that the model was well calibrated (P = 0.627, Brier = 0.172 in internal validation; and P = 0.366, Brier = 0.182 in external validation). The clinical decision curves showed that the model had good clinical utility.
Conclusions:
AGI is common in non-ICU elderly critically ill patients. This AGI risk prediction model can be used as a screening tool to identify high-risk patients for AGI and assist clinical decision making.
