Abstract:Objective To construct a predicting model for acute venous thromboembolism(VTE) in patients with sepsis. Methods Clinical data of patients with sepsis in the Medical Information Mart for Intensive Care Ⅲ(MIMIC Ⅲ) database from 2010 to 2012 were retrospectively analyzed. The patients were randomly divided into two groups based on whether they had VTE during hospitalization. A total of 26 variables were initially screened, including demographic characteristics, comorbidities, laboratory results, and interventions. LASSO regression and Logistic regression were utilized for selecting predictors and constructing prediction models(nomogram). Discrimination and calibration of nomogram was evaluated by Harrell’s concordance index(C-index) and receiver operating characteristic(ROC) curve. The data from 297 cases of sepsis patients treated in the intensive care unit(ICU) of Tongji Hospital Affiliated to Tongji University from January 2020 to December 2022 were collected for external validation of the model. Results Multivariate Logistic regression analysis showed that D-dimer(OR=9.183, 95%CI: 8.124-10.251, P=0.005), mean arterial pressure(MAP)(OR=0.096, 95%CI: 0.042-0.217, P<0.001), blood lactate(OR=5.077, 95%CI: 3194-8.071, P<0.001), congestive heart failure(OR=2.452,95%CI: 1.566-3.839, P<0.001) and mechanical ventilation(OR=2.061, 95%CI: 1.319-3.221, P=0.001) were independent predictors of VTE in sepsis patients. The variables D-dimer, MAP, congestive heart failure, blood lactate, and mechanical ventilation were integrated to construct a predictive model. Internal validation showed a C-index of 0.922(95%CI: 0.891-0.953, P<0.001), with an area under the ROC cure(AUC) of 0.92. External validation revealed a C-index of 0.832(95%CI: 0.744-0.920, P<0001), with an AUC of 0.76. Conclusion The model established in this study, which is composed of D-dimer, MAP, congestive heart failure, blood lactate and mechanical ventilation, can effectively predict the early risk of acute VTE in patients with sepsis.
