Artificial intelligence for better goals of care documentation

Area of Science:

• Healthcare disparities
• End-of-life care research
• Medical informatics

Background:

• Racial disparities exist in goals of care (GOC) conversations for hospitalized patients.
• These disparities may contribute to inequities in end-of-life care.
• Targeted initiatives are needed to address documentation gaps.

Purpose of the Study:

• To evaluate the association between a targeted initiative and GOC documentation rates across different racial groups.
• To assess if the initiative reduced racial disparities in GOC documentation.

Main Methods:

• Retrospective analysis of GOC documentation during a targeted initiative.
• Inclusion of adult patients with AI-predicted elevated mortality risk.
• Data collected from an urban academic medical center between July 2021 and December 2022.

Main Results:

• Overall GOC documentation rate was 28% (1019/3643 patients).
• Documentation rates were similar across racial groups: 30% Black, 28% White, 24% Other (p=0.3933).
• No statistical difference in GOC documentation rates among races over time (p=0.5142).

Conclusions:

• The targeted initiative for GOC documentation in high-risk patients was associated with similar rates across racial groups.
• Further research is required to confirm if this initiative promotes racial equity in GOC documentation in diverse settings.