Construction of a Risk Prediction Model for Feeding Intolerance in Mechanically Ventilated Children in the Pediatric Intensive Care Unit

doi:10.12104/j.issn.1674-4748.2025.10.007

Abstract：Objective: To analyze the independent risk factors for feeding intolerance (FI) in mechanically ventilated children in the pediatric intensive care unit (PICU), and to construct a risk prediction model. Methods: A total of 204 mechanically ventilated children in the PICU of a tertiary children's hospital in Dongguan from January 2020 to September 2023 were selected as the study subjects, and were divided into the tolerance group (60 cases) and the intolerance group (144 cases) according to whether FI occurred. The general information, mechanical ventilation-related indicators, and occurrence of FI were investigated and analyzed. Univariate analysis was used to identify the influencing factors of FI in PICU mechanically ventilated children. Logistic regression analysis was used to construct a risk prediction model for FI. A nomogram was drawn using the rms package in R software. The goodness of fit of the model was evaluated by the Hosmer-Lemeshow test. The predictive accuracy of the model was assessed by the area under the receiver operating characteristic (ROC) curve. The Bootstrap method was used to calibrate the prediction model. Results: The incidence of FI was 70.59%. The final predictive factors included serum potassium concentration, positive end-expiratory pressure (PEEP), C-reactive protein (CRP), fasting blood glucose, serum albumin concentration, and duration of mechanical ventilation. The risk prediction model formula is: P = 1 / [1 + exp(4.404 - 2.747 × serum potassium concentration + 1.475 × PEEP + 0.484 × fasting blood glucose - 0.199 × serum albumin concentration + 0.020 × CRP + 0.015 × duration of mechanical ventilation)]. The Hosmer-Lemeshow test showed that the model had good fitting (χ² = 6.079, P = 0.638). The calibration curve showed good consistency between the standard curve and the calibrated curve. The area under the ROC curve was 0.897, with a sensitivity of 89.6% and a specificity of 78.3%. Conclusion: The incidence of FI in mechanically ventilated children in the PICU is relatively high. Serum potassium concentration, PEEP, CRP, fasting blood glucose, serum albumin concentration, and duration of mechanical ventilation are all influencing factors for FI in PICU mechanically ventilated children. The risk prediction model established based on these risk factors has high predictive accuracy and good clinical applicability.