Reducing readmissions in the safety net through AI and automation

Area of Science:

• Health Informatics
• Population Health Management
• Health Equity

Background:

• Hospital readmissions pose a significant challenge, particularly in safety-net health systems.
• Reducing readmissions is crucial for improving patient outcomes and financial sustainability.
• Existing readmission reduction strategies often require enhancement with technology for greater impact.

Purpose of the Study:

• To implement and evaluate a technology-driven initiative aimed at reducing hospital readmissions.
• To assess the impact of the initiative on clinical outcomes, care equity, and financial performance.
• To determine the feasibility of this model for other resource-limited healthcare settings.

Main Methods:

• A retrospective interrupted time series analysis was conducted from October 2015 to January 2023.
• An electronic health record-integrated, automated decision-support tool standardized inpatient care.
• A predictive artificial intelligence algorithm identified high-risk patients for proactive outpatient management.

Main Results:

• Readmission rates decreased from 27.9% to 23.9% (P < .004).
• Disparities in readmission rates for Black/African American patients were eliminated.
• All-cause mortality reduced (HR, 0.82; P = .037), and the system retained $7.2 million in funding.

Conclusions:

• The technology-based initiative effectively reduced readmissions, improved equity, and enhanced survival in a safety-net system.
• This model demonstrates the potential of technology-driven, value-based care for resource-limited settings.
• The approach successfully met pay-for-performance metrics while improving patient care and financial stability.