Utility of a Caprini-combined prediction model in patients with gynecological venous thromboembolism in China.

Objective: Explored the risk factors for venous thromboembolism in gynecological inpatients in western China, and developed an improved model to predict the VTE of this patient population. Methods: The records of 6897 patients hospitalized in the Gynecology Department of Xinjiang Autonomous Region People's Hospital were retrospectively reviewed, during January 1, 2021 to July 31, 2022 and meet the inclusion criteria are selected. The efficacy of the Caprini-combined prediction model was evaluated, and the Caprini-combined prediction model and independent risk factor-combined prediction model for predicting VTE were assessed using receiver operating characteristic (ROC) curve analysis. Results: The study cohort was divided into two groups: a VTE group (n = 229) and a non-VTE group (n = 6,668). Univariate analysis was performed on all patients, followed by collinearity diagnostics for variables that showed statistically significant differences. Variables with a variance inflation factor (VIF) <2 were included in the multivariate logistic regression analysis. The results identified the following independent risk factors for VTE: length of hospital stay, age ≥70 years, Caprini score ≥3, respiratory or heart failure, body mass index (BMI) ≥30 kg/m2, high platelet count, low serum albumin level, and elevated D-dimer level. Among these, the Caprini score demonstrated the strongest association with VTE (OR = 9.939). The areas under the ROC curve for the Caprini scale, the combined Caprini prediction model (incorporating Caprini score, serum albumin, and D-dimer levels), and the combined model of all independent risk factors were 0.671, 0.973, and 0.979, respectively. We had developed a simple nomogram to predict VTE. Conclusion: Our study constructed a nomogram for predicting the risk of VTE based on age, BMI, Caprini scale score, respiratory/heart failure, length of hospital stay, PLT, serum albumin, and D-dimer levels. This model has good discrimination and prediction effects, and has certain clinical value.