Purpose

This is a pragmatic, prospective, randomized clinical trial to test the effectiveness of large language model (LLM)-generated serious illness conversation (SIC) summaries delivered to patients' inpatient and outpatient clinicians at the time of hospital admission on the primary outcome of days in the hospital in the next 90 days.

Condition

Eligibility

Eligible Ages
Over 18 Years
Eligible Sex
All
Accepts Healthy Volunteers
No

Inclusion Criteria

  • DFCI Patients, defined as patients who have a DFCI medical record number. - Admitted to an inpatient solid tumor medical oncology service at BWH (including beds that are considered to be DFCI beds within BWH) - Age ≥ 18 years - Mortality prediction ≥ 25% - Admitted on a Sunday after 4pm through Friday at 4pm. These time restrictions are necessary for the workflow of this pragmatic pilot trial, as the intervention requires a staff member to send intervention emails during the work week but within <= approximately 18 hours of admission. Accordingly, we will not enroll patients during holiday time periods (e.g., for a Monday holiday, enrollment would start on Monday at 4pm instead of Sunday at 4pm) or on days where staff absences (e.g., illness) preclude enrollment.

Exclusion Criteria

• Patients with elective inpatient admissions (typically for planned administration of chemotherapy or other treatments)

Study Design

Phase
N/A
Study Type
Interventional
Allocation
Randomized
Intervention Model
Single Group Assignment
Primary Purpose
Supportive Care
Masking
None (Open Label)

Arm Groups

ArmDescriptionAssigned Intervention
Experimental
Arm 1: Intervention Arm
An email is sent to patients' inpatient and outpatient clinical teams containing an LLM-generated summary of prior SIC documentation, prompting clinicians to review and discuss care preferences with the patient and align care with patient goals.
  • Behavioral: LLM-generated SIC summary email to clinical teams
    An email containing a summary of prior serious illness conversations, generated by a large language model, is sent to the patient's clinical teams at admission, prompting review and discussion of care preferences.
No Intervention
Arm 2: Control Arm
Patients receive standard of care; SICs may be initiated by patients or clinicians as usual.

Recruiting Locations

Brigham and Women's Hospital
Boston, Massachusetts 02214
Contact:
Alexi A Wright, MD.MPH
617-632-2334
alexi_wright@dfci.harvard.edu

More Details

Status
Recruiting
Sponsor
Dana-Farber Cancer Institute

Study Contact

Alexi A Wright, MD, MPH
6176322334
alexi_wright@dfci.harvard.edu

Detailed Description

The study will automatically enroll up to 600 DFCI patients with solid tumor cancers admitted to BWH inpatient medical oncology services with a predicted 90-day mortality risk of ≥ 25%. Patients will be randomized 1:1 to intervention vs control arms. The intervention arm will receive LLM-generated SIC summaries sent to their clinical teams, prompting incorporation of patient preferences into care plans. The control arm will receive standard of care. Outcomes will be followed for 90 days post-admission. The study uses a waiver of informed consent and HIPAA authorization due to minimal risk and the need for real-time identification and intervention delivery.

Notice

Study information shown on this site is derived from ClinicalTrials.gov (a public registry operated by the National Institutes of Health). The listing of studies provided is not certain to be all studies for which you might be eligible. Furthermore, study eligibility requirements can be difficult to understand and may change over time, so it is wise to speak with your medical care provider and individual research study teams when making decisions related to participation.