Presentation Details
| How to Define Low Risk? Kenneth C.Cummings1, Maureen Keshock1, Preethi Patel1, Mar-Qilia Kelly1, Stewart Richardson2, Scott Greenwald2, Josh Gray2, Daniel Sessler3, Nassib Chamoun2. 1Cleveland Clinic, Cleveland, OH, USA.2Health Data Analytics Institute, Dedham, MA, USA.3UTHealth Houston, Houston, TX, USA |
Abstract
BACKGROUND: Cleveland Clinic partnered with Health Data Analytics Institute to implement a risk stratification platform using validated risk models. These models use administrative data, patient demographics, and the proposed procedure to predict risks for multiple outcomes. These predictions then direct patients to several perioperative pathways. A key question is how to define risk strata. Because unanticipated admission and mortality are associated with many complications, we chose a composite of risks of these outcomes. Initial empirical thresholds were chosen based on capacity and projected volumes.
METHODS: Using the platform described above, patients are now stratified into 4 risk groups using a composite of risk predictions for mortality and readmission. This allows tailored intensity of preoperative optimization: low-risk patients may be seen virtually (or not at all) while high-risk patients have a much more intensive optimization and potentially postoperative hospitalist care (Figure 1). To more objectively define risk groups, we used prior-year data for one of the outcomes of interest (90-day readmission) as a function of the cut point definitions for the high-, intermediate-, and low-risk groups. The goal was to identify a point at which further raising the risk threshold for the low-risk group would appreciably increase unanticipated readmissions beyond the empirically chosen 35th percentile.
RESULTS: Retrospective data from 178,833 procedures in 2024 were included. Doubling the “low-risk” threshold from the 35th to the 70th percentile does not appreciably increase readmission rates (Figure 2), supporting a broader definition of patients who may be seen virtually or even bypass the preoperative process.
CONCLUSIONS: Appropriately focusing resources on truly high-risk patients has the greatest opportunity to improve outcomes while improving financial stewardship. As the low-risk threshold is raised, we will monitor outcomes of interest to ensure there are no untoward changes. A similar but much larger analysis looking at risk thresholds using Medicare data is underway and will be presented at the Summit.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
METHODS: Using the platform described above, patients are now stratified into 4 risk groups using a composite of risk predictions for mortality and readmission. This allows tailored intensity of preoperative optimization: low-risk patients may be seen virtually (or not at all) while high-risk patients have a much more intensive optimization and potentially postoperative hospitalist care (Figure 1). To more objectively define risk groups, we used prior-year data for one of the outcomes of interest (90-day readmission) as a function of the cut point definitions for the high-, intermediate-, and low-risk groups. The goal was to identify a point at which further raising the risk threshold for the low-risk group would appreciably increase unanticipated readmissions beyond the empirically chosen 35th percentile.
RESULTS: Retrospective data from 178,833 procedures in 2024 were included. Doubling the “low-risk” threshold from the 35th to the 70th percentile does not appreciably increase readmission rates (Figure 2), supporting a broader definition of patients who may be seen virtually or even bypass the preoperative process.
CONCLUSIONS: Appropriately focusing resources on truly high-risk patients has the greatest opportunity to improve outcomes while improving financial stewardship. As the low-risk threshold is raised, we will monitor outcomes of interest to ensure there are no untoward changes. A similar but much larger analysis looking at risk thresholds using Medicare data is underway and will be presented at the Summit.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
Content Locked. Log into a registered attendee account to access this presentation.