Leveraging Artificial Intelligence to Improve Clinical Appropriateness of Inpatient Designation in a Utilization Management Setting.

Abstract

Background: Inappropriate use of observation services for acute hospitalizations can lead to decreased reimbursement for care. Traditional evidence-based criteria are restrictive and do not consistently consider patients' preexisting conditions at the time of hospital arrival, highlighting the need for better utilization management (UM) and decision-making in assigning observation service versus inpatient admission. Objective: Guided by Neuman's Systems Model, the intervention aims to evaluate whether the implementation of an artificial intelligence (AI) tool in a UM registered nurse (RN) department can reduce health system observation service rates by enhancing the identification of patients' comorbidities, as well as improving the assessment of medical necessity and severity of illness for determining inpatient appropriateness in a large academic health system. Methods: Pre- and postimplementation observation versus inpatient volumes at discharge and observation-to-inpatient conversions were compared. Results: Postimplementation observation service discharge rates (12.75% monthly average) were lower compared with preimplementation observation service discharges (16.69% monthly average). UM RNs played a central role in the intervention, using the AI-generated Care Level Score to guide conversations with providers and advocate for appropriate patient placement. Conclusion: The implementations for nursing of an AI tool in the UM review process effectively reduced observation service discharge rates by improving the identification of comorbidities and enhancing the assessment of medical necessity. This approach demonstrated potential for better decision-making in recommending inpatient appropriateness and reducing observation service volume.