Performance of Western HA-VTE Risk Models in East Asian Populations: A Prospective Multicenter Study

Area of Science:

• Clinical Medicine
• Epidemiology
• Health Services Research

Background:

• Hospital-associated venous thromboembolism (HA-VTE) is a significant cause of patient morbidity.
• Existing risk prediction models for HA-VTE lack consistent validation, particularly in non-Western populations.

Purpose of the Study:

• To prospectively validate five established HA-VTE risk scores in a large cohort of Chinese inpatients.
• To assess the performance and applicability of Western-derived HA-VTE risk models in an East Asian population.

Main Methods:

• Prospective, multicenter cohort study involving 2,948 adult inpatients across 17 Chinese hospitals.
• Validation of five HA-VTE risk scores: Caprini, Padua, IMPROVE, Kucher, and Intermountain.
• Symptomatic VTE events were tracked for 3 months post-discharge, with 90-day incidence calculated.

Main Results:

• The 90-day HA-VTE incidence was 2.54%. All five models demonstrated moderate discrimination (AUC 0.726-0.759).
• Caprini score showed the highest sensitivity (0.83) but over-identified high-risk patients (42.6%). Padua and Intermountain offered better specificity (0.75-0.81) but lower sensitivity (0.64-0.67).
• Performance varied across subgroups, with attenuated results in patients receiving prophylaxis and limitations in practical use for complex models like Caprini in surgical settings.

Conclusions:

• Western-derived HA-VTE risk models possess moderate predictive utility in Chinese inpatients but require local recalibration.
• Model performance differs across patient subgroups, necessitating careful consideration for clinical application.
• Development of simpler, East Asian population-specific HA-VTE risk scores is recommended to improve clinical utility and accuracy.