Venous thromboembolism (VTE) is a major contributor to hospital mortality and disability-adjusted life-year (DALY) loss. Multiple guidelines recommend using the Padua or IMPROVE scores to stratify VTE risk in hospitalized medical patients. However, the IMPROVE score is not recommended in Chinese guidelines, and there is very little evaluation of its clinical application and effectiveness in the Chinese population. The objective of this study is to compare the efficacy of the Padua and IMPROVE scoring models for assessing VTE risk in Chinese medical inpatients. We conducted a retrospective analysis of the clinical characteristics and thrombotic risk of 42,257 medical inpatients at a tertiary hospital in Guangdong, China, between 2021 and 2022. Logistic regression was used to assess thrombotic risk factors. The Receiver Operating Characteristic (ROC) curves, Area Under the Curve (AUC), sensitivity, and specificity were employed to evaluate the performance of the two models. Of the 42,257 patients included, 948 (2.24%) experienced VTE during hospitalization. According to the Padua score, 3,7513 (88.78%) of patients were considered low risk, while 4,744 (18.22%) were classified as high risk. The IMPROVE score identified 20,744 (49.09%) of patients as low risk, 20799(49.22%) as intermediate risk, and 714(1.69%) as high risk. The AUC for the Padua score was 0.735 (95% CI: 0.717–0.753), with a sensitivity of 49.4% and specificity of 89.6%. For the IMPROVE score, the AUC was 0.711 (95% CI: 0.693–0.729), with a sensitivity of 32.5% and specificity of 99.0%. The DeLong test, used to compare the AUCs, yielded a z-value of 1.886 with a P-value of 0.059, indicating no statistical difference. When assessing VTE risk in patients with stroke, cancer, nephrotic syndrome, and critical illness (ICU/CCU stay), both scoring models showed comparable predictive performance with AUCs ranging between 0.7 and 0.8. Both the Padua score and IMPROVE score have good predictive ability for VTE events during hospitalization in medical patients. Among them, the IMPROVE score has objective assessment items, simpler operation, and more detailed risk stratification, which is beneficial for clinicians to take physical and pharmacological preventive measures at different levels.

ChiCTR2200056903, February 22, retrospectively registered.