1. Patel and colleagues conducted a secondary economic analysis of a trial that assigned high-risk cancer patients to critical illness consultations (SIC).
2. Machine learning algorithms effectively identified patients at high risk of mortality. SIC for these patients was associated with reduced healthcare costs.
Level of evidence: 2 (good)
Research Summary: Patients with advanced cancer often receive treatments that do not meet their preferences at the end of life. However, serious illness discussions (SICs) are rarely initiated, even though they can improve quality of life and mood and reduce healthcare utilization. In a randomized controlled trial (RCT), a machine learning (ML) algorithm identified cancer patients with a 180-day mortality risk of 10% or higher. These patients were then randomly assigned to a SIC intervention group or a standard care control group. Patel and colleagues conducted a secondary analysis of the study using total and average daily healthcare expenditures in the last 6 months of life as primary outcomes. Secondary outcomes included average expenditures in the last 3 months and last month of life. They analyzed 1,187 patients and found that average daily healthcare expenditures in the last 6, 3, and last month of life were lower in the intervention group than in the control group. Savings were achieved for systemic therapy and outpatient care. The study demonstrated that an ML algorithm could predict patient mortality and prompt physicians to initiate SIC, leading to significant cost savings and improved goal-aligned care.
Click here to read the NEJM AI study
Related reading: The Economics of Artificial Intelligence in Healthcare: Diagnosis and Treatment
detail [randomized controlled trial]: 26,525 adult cancer patients were used to train an ML algorithm to predict 180-day mortality using demographic variables, Elixhauser comorbidities, clinical tests and selected electrocardiogram data. The algorithm was then adjusted for risk ≥10% to identify patients eligible for the RCT. The intervention included sending SIC reminders to clinicians before seeing the patient, and clinicians' SIC rates were compared with their colleagues. Patel and colleagues' secondary analysis included 1,187 patients from the RCT, and per-patient expenditure information was collected from the hospital accounting system. Primary outcomes were mean total medical expenditure and medical expenditure per day during the last 6 months of life. Secondary outcomes were mean per-day expenditure during the last 3 and 1 months of life. All outcomes were stratified by visit type (e.g., acute care, outpatient). The intervention group saved $75.33 (95% CI, -$136.42 to -$14.23) in mean daily expenditures during the last 6 months of life compared with the control group. Stratification by visit type showed significant savings for systemic and outpatient care. Total cost savings during the last 6 months were $13,747 (95% CI, -$24,897 to -$2,598). In addition, the intervention group experienced reduced expenditures for both secondary outcomes ($431.80 vs. $473.20 in the last 3 months and $814.46 vs. $947.18 in the last month). Although the authors found significant savings, their results were limited by the sample from a single academic health system and by substantial baseline differences between participants in the two groups.
Image: PD
©2024 2 Minute Medicine, Inc. All rights reserved. No work may be reproduced without the express written consent of 2 Minute Medicine, Inc. Please contact us here for licensing inquiries. Articles should not be construed as medical advice, and are not intended as such by the authors or 2 Minute Medicine, Inc.