To verify the efficacy of Caprini and Rogers risk assessment models of screening for venous thromboembolism (VTE) in patients undergoing thoracic surgery.

A single center retrospective study was performed in the department of thoracic surgery in Beijing Chao-Yang Hospital affiliated to Capital Medical University from July to December in 2016. The Caprini and Rogers risk score were recorded retrospectively for each patient. Meanwhile, by using of logistic bivariate regression analysis, we have obtained the predicted probability of Caprini and Rogers in combination diagnosis. We have drawn receiver operating characteristic curves (ROC) respectively, and calculated the area under the curve (AUC), then we have done pairwise comparison severally. The Youden index was taken as the cutoff point.

The total incidence of VTE after thoracic surgery was 13.4%. The AUC for the the Caprini model, the Rogers model and the predictive probability model were 0.713±0.043 (P<0.001), 0.577±0.062 (P=0.207) and 0.730±0.041 (P<0.001) respectively. There was no significant difference in AUC between Caprini model and the other two models (P>0.05); while the AUC of the predictive probability model was significantly higher than that of the Rogers model (P=0.015). Caprini model had a sensitivity of 0.923 and a specificity of 0.47 when the Youden index was 0.393. The Rogers model had a sensitivity of 0.385 and a specificity of 0.75 when the Youden index was 0.135. The predictive probability model had a sensitivity of 0.962 and a specificity of 0.48 when the Youden index was 0.444.

The combined use of Caprini and Rogers risk assessment model can improve the accuracy of screening for venous thromboembolism in patients undergoing thoracic surgery.