Venous thromboembolism risk assessment in hospital-at-home setting: A prospective cohort study (the TROMBODOM study).

Abstract

Venous thromboembolism (VTE) is a significant cause of preventable mortality in hospitalized patients. While risk assessment models such as Padua and IMPROVE-VTE scores have been validated in conventional hospitalization (CH), their applicability to hospital-at-home (HaH) remains uncertain. To evaluate thrombotic risk stratification in HaH by comparing the 90-day VTE incidence and mortality between HaH and CH and developing an HaH-specific VTE predictive model. This prospective observational cohort study comprised 2062 patients (1689 HaH; 373 CH) admitted for acute medical illnesses between 2022 and 2024. Thrombotic risk was assessed using Padua and IMPROVE-VTE scores. Primary outcomes were 90-day VTE incidence and mortality. Predictors of VTE in HaH were identified using multivariable logistic regression and incorporated into a novel model (TROMBODOM). HaH patients were older (mean, 71.7 vs. 64.5 years, respectively) and more often admitted for infections (90.1% vs. 63.0%, respectively) and cardiorespiratory failure (36.6% vs. 25.7%, respectively). A larger proportion of patients in HaH were classified as high-risk based on Padua (75.1% vs. 52.0%, respectively) and IMPROVE-VTE (46.3% vs. 27.1%, respectively). Ninety-day VTE incidence was lower in HaH (1.2% vs. 3.5%, respectively; OR, 0.33; 95%CI, 0.16-0.67), whereas mortality was higher (15.8% vs. 9.9%, respectively; OR: 1.70, 95%CI: 1.18-2.44). Padua and IMPROVE-VTE scores showed modest discriminatory capacity (AUC: 0.65 and 0.75, respectively) in HaH. TROMBODOM achieved superior predictive performance (AUC: 0.86). Padua and IMPROVE-VTE overestimated the thrombotic risk of HaH. The newly developed TROMBODOM model improved VTE risk stratification in HaH and requires external validation.