Abstract:
Objective To establish and validate a risk prediction model for the occurrence of enteral nutrition intolerance (ENI) in adult patients with severe burns. Methods This study was a retrospective cohort study. A total of 155 adult patients with severe burns who met the inclusion criteria and hospitalized at the Affiliated Hospital of Jiangnan University between November 2020 and November 2024 were enrolled in modeling group, including 127 males and 28 females, aged 19 to 85 years. An additional 40 adult patients with severe burns who met the inclusion criteria and hospitalized at the Affiliated Hospital of Nantong University between November 2022 and November 2024 were enrolled in validation group, including 30 males and 10 females, aged 25 to 79 years. The gender, age, body mass index, number of underlying diseases, total burn area, modified early warning score, mechanical ventilation, white blood cell count, C-reactive protein level, serum albumin level, and fasting blood glucose level within 24 hours of admission, number of antibiotic types, intestinal probiotics, and sedatives and analgesics used during treatment were compared between the two groups of patients. Based on the occurrence of ENI, patients in modeling group were divided into two categories: those who developed ENI (96 cases) and those who did not (59 cases). The aforementioned data of these two categories of patients were compared to identify independent predictive factors for ENI of patients in modeling group. Accordingly, a risk prediction model for ENI of patients in modeling group was constructed, and both static and web-based dynamic nomograms were developed. The performance of the prediction model was evaluated using receiver operating characteristic (ROC) curves. The modeling group data was repeatedly sampled 1 000 times using Bootstrap method for internal validation, and external validation was conducted in validation group. Calibration curves and clinical decision curve analysis were used to analyze the calibration accuracy and clinical practicality of the prediction model, respectively. Results There were statistically significant differences between modeling group and validation group in terms of total burn area, number of antibiotic types used during treatment, mechanical ventilation and serum albumin levels within 24 h of admission of patients (with Z values of -2.35 and -2.68, χ2 values of 4.58 and 4.63, respectively, P<0.05). No statistically significant differences were observed in other variables between the two groups (P>0.05). There were statistically significant differences between patients developed ENI and those who did not in the number of underlying diseases, fasting blood glucose level within 24 h of admission, the number of antibiotic types used during treatment, white blood cell count within 24 h of admission, and the use of intestinal probiotics and that of sedative and analgesic agents (with Z values of 2.04, 4.24, and 3.36, χ2 values of 26.02, 24.13, and 4.49, respectively, P<0.05). Multivariate logistic regression analysis showed that the number of underlying diseases, total burn area, white blood cell count and fasting blood glucose level within 24 h of admission, and the use of intestinal probiotics during treatment were independent predictive factors for ENI of patients in modeling group (with odds ratios of 2.33, 1.03, 0.11, 1.22, and 0.08, 95% confidence intervals of 1.25-4.32, 1.00-1.06, 0.04-0.30, 1.04-1.42, and 0.03-0.24, respectively, P<0.05). Based on the aforementioned independent predictive factors, a risk prediction model for ENI in patient of modeling group was successfully established, and both a static and a web-based dynamic nomogram were developed. The area under the ROC curve of the prediction model was 0.90 (with 95% confidence interval of 0.84-0.95), with a sensitivity of 86.50%, a specificity of 84.70%, a maximum Youden's index of 0.712, and an optimal threshold of 61.4%. In internal and external validation, the areas under the ROC curves of the prediction model were 0.88 and 0.91, respectively (with 95% confidence intervals of 0.82-0.94 and 0.81-0.99, respectively), the calibration curves of the prediction model were all near the reference line, and the threshold probability ranges displayed by the clinical decision curves were 2%-96% and 18%-98%, respectively, with net returns>0. Conclusions The independent predictive factors for the occurrence of ENI in adult patients with severe burns include the number of underlying diseases, total burn area, white blood cell count and fasting blood glucose level within 24 h of admission, and the use of intestinal probiotics during treatment. The nomogram prediction model constructed on this basis has a good predictive efficiency for the risk of the occurrence of ENI in adult patients with severe burns.
Sun Dan,Lyu Guozhong,Cao Ling,et al.Establishment and validation of a risk prediction model for the occurrence of enteral nutrition intolerance in adult patients with severe burns[J].Chin J Burns Wounds,2025,41(12):1-10.DOI: 10.3760/cma.j.cn501225-20250820-00360.
Abstract
PDF
