[Construction and analysis of early warning and prediction model for risk factors of sepsis-associated encephalopathy]

Abstract

Objective: To investigate the epidemiological characteristics of sepsis-associated encephalopathy (SAE) in patients with sepsis, analyze its risk factors and build a prediction model, which provides evidence for early clinical identification of SAE patients and improvement of clinical outcomes.

Methods: A retrospective observational study was conducted. Sepsis patients admitted to the critical care medical center of the First Affiliated Hospital of Xinjiang Medical University from February 2022 to February 2023 were enrolled. According to whether SAE occurred, the patients were divided into sepsis group and SAE group. The 24 patients without sepsis in the same period were used as controls (non-sepsis group). Demographic data, relevant scores and laboratory test indicators at admission to intensive care unit (ICU), and prognostic indicators were collected. Univariate and multivariate Logistic regression analysis was used to analyze the risk factors for sepsis and SAE. Receiver operator characteristic curve (ROC curve) was drawn. The predictive value of each risk factor for sepsis and SAE.

Results: A total of 130 patients with sepsis were included, of which 52 had SAE, and the incidence of SAE was 40.00%. There were significant differences in the length of ICU stay and total length of stay among all groups, while there were no significant differences in hospitalization cost and mechanical ventilation time. Multivariate Logistic regression analysis showed that pulmonary infection [odds ratio (OR) = 46.817, 95% confidence interval (95%CI) was 5.624-389.757, P = 0.000], acute physiology and chronic health evaluation II (APACHE II: OR = 1.184, 95%CI was 1.032-1.358, P = 0.016), sequential organ failure assessment (SOFA: OR = 9.717, 95%CI was 2.618-36.068, P = 0.001), Charson comorbidity index (CCI: OR = 4.836, 95%CI was 1.860-12.577, P = 0.001), hemoglobin (Hb: OR = 0.893, 95%CI was 0.826-0.966, P = 0.005), glutamyltranspeptidase (OR = 1.026, 95%CI was 1.008-1.045, P = 0.006) were independent risk factors for sepsis in ICU patients. Pulmonary infection (OR = 28.795, 95%CI was 3.296-251.553, P = 0.002), APACHE II score (OR = 1.273, 95%CI was 1.104-1.467, P = 0.001), SOFA score (OR = 8.670, 95%CI was 2.330-32.261, P = 0.001), CCI (OR = 5.141, 95%CI was 1.961-13.475, P = 0.001), Hb (OR = 0.922, 95%CI was 0.857-0.993, P = 0.031), glutamyltranspeptidase (OR = 1.020, 95%CI was 1.002-1.038, P = 0.030) were independent risk factors for SAE in sepsis patients. ROC curve analysis showed that the area under the curve (AUC) of pulmonary infection, APACHE II score, SOFA score, CCI, Hb, and glutamyltranspeptidase for predicting sepsis were 0.792, 0.728, 0.987, 0.933, 0.720, and 0.699, respectively; the AUC of the combined prediction of the above 6 variables for sepsis was 1.000, with a sensitivity of 100% and a specificity of 100%. The AUC predicted by pulmonary infection, APACHE II score, SOFA score, CCI, and Hb for SAE were 0.776, 0.810, 0.907, 0.917, and 0.758, respectively; the AUC of the combined prediction of the above 5 variables for SAE was 0.975, with a sensitivity of 97.3% and a specificity of 93.1%.

Conclusions: Sepsis is more severe when accompanied by encephalopathy. Pulmonary infection, Hb, APACHE II score, SOFA score and CCI were independent risk factors of SAE. The combination of the above five indicators has good predictive value for early screening and prevention of the disease.