Analysis of risk factors and construction of prediction model for lower extremity deep vein thrombosis after liver transplantation.

Abstract

Background: Lower extremity deep venous thrombosis (LEDVT) is a serious and potentially fatal complication with a high incidence following liver transplantation, significantly affecting patient prognosis. This study aimed to investigate the characteristics and risk factors associated with LEDVT post-transplantation and to develop an effective clinical prediction model.

Methods: A retrospective analysis was conducted on 298 liver transplant recipients at Hebei Medical University Third Hospital between January 2021 and April 2024. The cohort was randomly divided into a training set and a validation set in a 7:3 ratio. Baseline variables, including demographics, smoking history, comorbidities, surgical data, and biochemical indicators at admission, were collected. The training set data were used to construct the predictive model. Relevant predictors were identified using non-parametric rank-sum tests, chi-square tests, and least absolute shrinkage and selection operator (LASSO) regression. A multivariate logistic regression model was then developed to optimize these predictors and generate a nomogram. Model performance was assessed through receiver operating characteristic (ROC) analysis, calibration curve analysis, and decision curve analysis (DCA). Validation of the model was conducted using the independent validation set.

Results: LEDVT occurred in 28 (13.5%) of the 208 patients in the training cohort. LASSO regression identified smoking history, hyperlipidemia, intraoperative blood loss, and elevated D-dimer levels as independent predictors of LEDVT after liver transplantation. A nomogram was constructed based on these predictors, with risk scores assigned to each variable (as depicted in Fig. 3). Higher total scores were associated with an increased likelihood of LEDVT. The predictive model demonstrated satisfactory discrimination and calibration, with an area under the ROC curve (AUC) indicating good predictive accuracy. The calibration plot, Hosmer-Lemeshow test, and DCA further confirmed the model's clinical utility.

Conclusions: The developed prediction model exhibited excellent performance in identifying patients at high risk for LEDVT following liver transplantation. Early identification of at-risk individuals allows for timely intervention, potentially reducing LEDVT incidence and improving patient outcomes. Furthermore, this model has significant implications for reducing healthcare costs and optimizing resource allocation.