Background: Risk adjustment indices are frequently used for hospital billing and reimbursements in the United States. Severity of illness (SOI) and risk of mortality (ROM) are two risk adjustment indices used in determining the APR-DRG, a classification system commonly used by insurers to determine hospital reimbursements. Recently, many health systems implemented AI-powered clinical documentation assistants to enhance clinical workflows. The impact of these tools in measurements for billing has not been well studied.
Purpose: To evaluate the impact of an AI clinical documentation tool on risk of mortality (ROM) and severity of illness (SOI) indices.
Description: We analyzed 7,659 inpatient encounters at a single, urban 595-bed medical center over a 6-month period between January 1 and June 30, 2024. 41 hospitalists were trained on the use of a commercially-available AI clinical documentation tool and allowed to use the tool in their regular clinical practice. Of the total number of patient encounters, 43.1% had documentation assisted by the AI tool. Average ROM score for encounters utilizing the tool was 2.35, compared to 2.29 without the tool (p = 0.009). Average SOI was 2.59 with the tool versus 2.49 without the tool (p < 0.001).
Conclusions: Use of AI clinical documentation tools impact risk adjustment indices such as ROM and SOI. The increasing adoption of AI-powered clinical documentation tools may have implications for U.S. hospital billing and for the cost of care nationally.