Padua Prediction Score was Inappropriate for Venous Thromboembolism Risk Assessment in Acute Medical Inpatients: A Cohort Study

Abstract Background Padua Prediction Score (PPS) still lacks of effectively external validation in Chinese population. To assess the accuracy of PPS to predict venous thromboembolism (VTE) risk in acute medical inpatients, and to compare the predictive value with Caprini risk assessment model (RAM). Methods This consecutive cohort study included acute internal medicine inpatients in Beijing Institute of Respiratory Medicine, Beijing, China from January to August 2019. The PPS and Caprini RAM were assessed for all the patients. The occurrence rate of VTE predicted by PPS and Caprini RAM was compared using Chi-squared test. Discrimination was evaluated by the area under curve using the Delong tests. Reclassification parameters were also used to compare the PPS's clinical utility with that of the Caprini RAM. Results: A total of 1647 inpatients of internal medicine were retrospectively registered. The overall incidence of VTE during hospitalization was 13.2%. The incidence of VTE in low-risk patients was 6.5% assessed by PPS, which was significantly higher than that by Caprini RAM (2.6%, p < 0.001). The area under the receiver operating characteristic curve (ROC) for PPS and Caprini RAM was 0.747 (95%CI, 0.725–0.768) and 0.773 (95%CI, 0.752–0.793), respectively (p = 0.0453). Based on the best cut-off, the specificity of the PPS appeared higher than that of the Caprini RAM (82.45 vs 60.07, p < 0.001), but its sensitivity was lower (59.45 vs 83.87, p < 0.001). Compared with Caprini RAM, the net reclassification index was estimated at 0.094 (p = 0.016) and integrated discrimination improvement was 0.037 (p = 0.098) by PPS. The ROC curve of the Caprini RAM and PPS in the subgroup of patients without pharmacological prophylaxis during their hospitalization was 0.776 of the Caprini RAM, which was significantly higher than that of the PPS (p = 0.005). Conclusions PPS seems inappropriately to predict VTE risk in acute internal medicine inpatients. It has limitations not only in the performance of identifying “VTE high-risk” patients, but also in the ability of discrimination. An accurate, widely applicable, validated RAM for prediction of the VTE risk needs to be further constructed in Chinese hospitalized medical patients.