OC 54.3 - Validation of the IMPROVE Hospital-Acquired Venous Thrombosis Risk Assessment Model in The Medical Inpatients Thrombosis and Hemostasis Study (MITH) Population

Tuesday, July 12, 2022

11:15 AM – 11:30 AM

Location: ExCel Center, ICC Capital Suite Room 14-16

Presentation Format: In-Person

Presenting Author(s)

Andrew D. Sparks, MS

Statistician
The Larner College of Medicine at the University of Vermont
Burlington, Vermont, United States

Co-Author(s)

KW

Katherine S. Wilkinson, MS

Statistician
The Larner College of Medicine at the University of Vermont
Burlington, Vermont, United States
AR

Allen Repp

Internal Medicine Hospitalist, Professor
University of Vermont
Burlington, Vermont, United States
AL

Ang Li

Physician, Assistant Professor
Baylor College of Medicine
Houston, Texas, United States
RT

Ryan Thomas

Resident Physician
University of Vermont
Burlington, Vermont, United States
NR

Nicholas S. Roetker

Epidemiologist
Hennepin Healthcare Research Institute
Minneapolis, Minnesota, United States
Neil A. Zakai, MD, MSc

Professor
Hematology/Oncology Division, Department of Medicine, The Larner College of Medicine at the University of Vermont
Burlington, Vermont, United States

Background: Governmental and professional societies recommend assessing risk of hospital-associated (HA) venous thrombosis (HA-VTE) in medical patients, defined as a VTE occurring during hospitalization or up to 3 months after discharge. The IMPROVE risk assessment model (RAM) predicts HA-VTE risk, and we sought to validate this RAM in an independent population.

Aims: To assess the performance of the IMPROVE RAM & IMPROVE score in an independent, external population of hospitalized medical patients.

Methods: The IMPROVE RAM was calculated for all medical admissions between 2010-2019 at the University of Vermont Medical Center (UVM) as part of the MITH study. Cox regression was used to estimate the hazard ratios (HR) for HA-VTE risk factors and to examine model performance. This research is Institutional Review Board approved and funded by the National Institutes of Health, USA.

Results: Over 23,911 admissions there were 314 HA-VTE events: 94 (30%) in-hospital and 220 (70%) after discharge. Table 1 compares the results from the original IMPROVE-VTE development cohort to the present UVM validation cohort, including the number of admissions and HA-VTE events, as well as the prevalence of IMPROVE risk factors and their associations with HA-VTE. Multivariable HR estimates were in general lower in the UVM population compared with the original IMPROVE study; one exception was active cancer, which showed a similar association in both studies. Assessing RAM performance, the AUC of the IMPROVE model in the UVM cohort was 0.64. HA-VTE risks were overestimated at lower scores and underestimated at higher scores in the UVM cohort relative to the original IMPROVE cohort (Table 2).

Conclusion(s): The IMPROVE score, while associated with HA-VTE risk, demonstrated reduced accuracy in the UVM validation cohort when compared to the IMPROVE development cohort. HA-VTE RAMs with better external validity are needed that potentially assess in-hospital and post-discharge events as separate outcomes.