Abstract
BACKGROUND: The Caprini risk assessment model (RAM) is a classical score that evaluates the risk of developing venous thromboembolism (VTE) aimed at patients with lower extremity fractures (LEF). However, the influencing factors of RAM pay more attention to clinical characteristic variables and recognized thrombophilia markers such as prothrombin, while ignoring the influence of other laboratory indicators on thrombosis. This is the first paper to screen predictive risk factors associated with VTE to modify the Caprini model and determine its diagnostic value. RESULTS AND METHODS: We included 953 LEF patients from January 2019 to December 2021 and randomized them in a 7:3 ratio into a training set (n = 667) and a validation set (n = 286). Logistic regression and LASSO regression were used to screen for independent predictors of patients in the training set. Logistic regression screened for: age (OR 1.04, 95% CI 1.02-1.06), D-dimer (OR 1.06, 95% CI 1.03-1.08), Caprini (OR 1.17, 95% CI 1.08-1.27) and fasting blood glucose (FBG) (OR 1.18, 95% CI 1.03-1.35), which also appeared in the LASSO seven-factor model. In addition to Caprini and age, FBG and D-dimer may be important predictors of deep vein thrombosis in patients with LEF. Therefore, we constructed three models. caprini + D-dimer, caprini + FBG, and caprini + D-dimer + FBG. We compared them with caprini alone. Caprini + FBG + D-dimer showed the greatest ROC (0.784, 95% CI 0.730-0.837, Delong test P = 0.002 vs. Caprini's ROC). D-dimer and FBG have a significant increase in the predictive effect for predicting DVT in terms of NRI (59.98% improvement, P < 0.001) and IDI (9.55% improvement, P < 0.001), showing the best net benefit for predicting DVT when the threshold probability was between 27 and 91%. CONCLUSIONS: Caprini RAM combined with D-dimer and FBG can improve the diagnostic efficiency of Caprini alone and provide a new method to better predict the risk of VTE in LEF patients.