PB1484 - The use of artificial intelligence for preventing venous thromboembolism in hospitalized patients: A national qualitative study

Tuesday, June 27, 2023

18:30 – 19:30 ET

Room: Exhibition

Presenting Author(s)

SZ

Sabrina Zerbey

Research Assistant
Beth Israel Deaconess Medical Center
Boston, Massachusetts, United States

Co-Author(s)

Barbara Lam, MD

Fellow
Beth Israel Deaconess Medical Center
Boston, Massachusetts, United States
LD

Laura Dodge

Assistant Professor
Beth Israel Deaconess Medical Center
Boston, Massachusetts, United States
SD

Siddhant Datta

Hospitalist
Beth Israel Deaconess Medical Center
Boston, Massachusetts, United States
Rachel Rosovsky, MD, MPH

Hematologist, Associate Professor of Medicine
Massachusetts General Hospital and Harvard Medical School
Newton, Massachusetts, United States
WR

William Robertson

Program Director
National Blood Clot Alliance
Philadelphia, Pennsylvania, United States
LL

Leslie Lake

President
National Blood Clot Alliance
Philadelphia, Pennsylvania, United States
PE

Pavania Elavalakanar

Research Coordinator
Division of Hematology and Hematologic Malignancies, Beth Israel Deaconess Medical Center, Boston, MA
Boston, Massachusetts, United States
Jeffrey I. Zwicker, MD

Chief of the Hematology Service
Department of Medicine, Hematology Service, Memorial Sloan Kettering Cancer Center, New York City, NY
New York, New York, United States
MS

Mara Schonberg

Associate Professor of Medicine
Beth Israel Deaconess Medical Center, Harvard Medical School
Boston, Massachusetts, United States
Rushad Patell, MD (he/him/his)

Assistant Professor of Medicine
Beth Israel Deaconess Medical Center, Boston, MA, USA
Boston, Massachusetts, United States

Background: Venous thromboembolism (VTE) in hospitalized patients is a major public health challenge. Guidelines recommend using risk stratification tools to personalize who receives VTE prophylaxis. Artificial intelligence (AI) is a promising tool for helping make these complex decisions, but clinician attitudes towards AI for VTE prevention are unknown.

Aims: Conduct focus groups of clinicians to understand 1) current practices around VTE prevention in adult hospitalized patients and 2) attitudes towards using AI for VTE prevention.

Methods: Hospital-based physicians, physician trainees, pharmacists, and nurses across the country were invited via email to participate in 60-minute virtual focus groups, each with up to 10 participants. Basic demographic information was collected and all recordings were transcribed. Two investigators reviewed the transcripts to derive a codebook, then iteratively coded all interviews and discussed discrepancies until consensus was achieved. Participants were recruited until data saturation was achieved

Results: Of the 33 participants, the majority were physicians and under 40 years old (Table 1). Clinicians described prescribing VTE prophylaxis “by default” and only using risk assessment models (RAMs) for challenging scenarios. They reported that gaps in prophylaxis tend to happen during transitions in care and identified bleeding risk as particularly difficult to assess. Clinicians reported that the electronic health record (EHR), as well as hospital policy and patient preference, influence their choice of prophylaxis.

Clinicians believed that AI can help with VTE prevention but also voiced concerns around accuracy, transparency, and liability. They reported that more research is needed, and how a hospital deploys an AI system may influence clinician acceptance. Several clinicians described how AI should supplement rather than replace clinical decision-making (Table 2)

Conclusion(s): Clinician respondents describe using VTE prophylaxis in adult hospitalized patients with minimal premeditation. They perceived AI as a promising tool albeit with multiple barriers to real-life use.