Insurers’ AI Denials of Postacute Care Face Senate Scrutiny

A new report from the Senate Permanent Subcommittee on Investigations levies sharp criticism against UnitedHealthcare, Humana, and CVS for placing profits ahead of patient care, through their use of artificial intelligence (AI) automation and predictive technologies to deny prior authorization requests for Medicare Advantage plan holders requiring postacute care.¹

“For beneficiaries of Medicare Advantage...it has become not just a bureaucratic maze, but a potential threat to their health," according to the report.

Between 2019 and 2022 alone, denial rates for postacure care were far higher than the denial rates for other types of care, the report states.² Further, for 2022 alone, UnitedHealthcare and CVS were called out for having postacute care denials that were 3 times higher than their overall rate of prior authorization denials and Humana’s postacute care denial rate was16 times that of its overall prior authorization denial rate. UnitedHealthcare, CVS, and Humana are the 3 largest Medicare Advantage insurers—covering close to 60% of all Medicare Advantage enrollees—according to a report on the committee’s findings.¹

This is not the first time such actions have come under intense scrutiny, with previous Senate hearings decrying the use of algorithms for coverage determinations³ and recent research demonstrating disparities in prior authorization decisions.^4,5

Highlights From the Report

Each insurer was called out in the report, culled from the more than 280,000 pages of documents the 3 insurers provided, for their abysmal approval rates of prior authorization requests in favor of saving millions down the line and for using AI to do so.²

For UnitedHealthcare, the following were determined:

• From 2020 to 2022, the denial rate for postacute care more than doubled: from 10.9% to 22.7%; this occurred while Humana was working to automate its prior authorization process
• A model, “Machine Assisted Prior Authorization,” approved in April 2021 was cited for reducing the average time to review an approval request from 10 minutes to 6 minutes while also requiring a nurse or doctor to review the evidence behind a decision
• Another model tested in 2021, “Healthcare Economics Auto Authorization Model,” was cited for its faster review times and for producing more adverse determinations (or denials)
• nH Predict, a third model, was called out for its past connection to care denials6 and for care coordinators using the model to make postacure care decisions for hospitalized patients, who were then instructed to not inform care providers how and why decisions were made over the phone by naviHealth, a care management company that provides postacute care management services

The Senate Permanent Subcommittee on Investigations is rebuking the top 3 Medicare Advantage insurers following findings on prior authorizations that show high denial rates from the use of automation and predictive technologies. Image Credit: © Dmitry - stock.adobe.com

For CVS, the following were determined:

• Prior authorization review requests for postacute care requests alone leaped 57.5% between 2019 and 2022 compared with 40% overall
• One of the care giant’s models, a program it developed, said case reviews of requests likely to be denied should be prioritized
• Data from 2018, as seen in documents from May 2019, demonstrate that CVS saved upwards of $660 million by denying prior authorizations requests for its Medicare Advantage beneficiaries
• “Mistake” approvals for postacute care admission requests were reversed at a rate that was 10 times higher compared with acute hospital admissions, following the use of a predictive model that handed down too many approvals CVS thought should have been denied
• Data following the use of artificial intelligence to reduce skilled nursing facility spend showed potential savings of $77 million vs the $4 million CVS had predicted

For Humana, the following were determined:

• Denials after a prior authorization request was submitted jumped 54% in 2 years (2020-2022)
• Training on how to “properly” review prior authorization requests for long-term acute care were implemented after data from during the COVID-19 public health emergency period showed lower rates of coverage denials compared with before the pandemic, with specific language cited for how to explain denials
• Changes made to templates meant to direct prior authorization decisions were shown to be for denial purposes and to be able to reinforce denial decisions should they come up for appeal
• Policies may give contractors (eg, naviHealth) more say in the use of predictive technologies, with evidence unclear on Humana’s use of artificial intelligence to make prior authorization decisions

“The data obtained so far is troubling regardless of whether the decisions reflected in the data were the result of predictive technology or human discretion,” according to the report. “It suggests Medicare Advantage insurers are intentionally targeting a costly but critical area of medicine—substituting judgment about medical necessity with a calculation about financial gain.”

Its primary recommendation for turning things around centers on greater oversight from CMS that focuses on more detailed analysis of prior authorization by service category, targeted audits following sharp increases in denial rates, and for regulations that serve to restrict the influence of predictive technologies on human decision-making.

References

1. Vogel S. Senate report slams Medicare Advantage insurers for using predictive technology to deny claims. Healthcare Dive. October 21, 2024. Accessed October 28, 2024. https://www.healthcaredive.com/news/medicare-advantage-AI-denials-cvs-humana-unitedhealthcare-senate-report/730383/

2. Refusal of recovery: how Medicare Advantage Insurers have denied patients access to post-acute care. US Senate Permanent Subcommittee on Investigations. October 17, 2024. Accessed October 28, 2024. https://www.hsgac.senate.gov/wp-content/uploads/2024.10.17-PSI-Majority-Staff-Report-on-Medicare-Advantage.pdf

3. Pifer R. Senate scrutinizes MA payment denials, including use of algorithms. Healthcare Dive. May 18, 2023. Accessed October 28, 2024. https://www.healthcaredive.com/news/medicare-advantage-denials-prior-authorization-ai-senate/650578/

4. McCormick B. Exploring racial, ethnic disparities in cancer care prior authorization decisions. AJMC^®. October 24, 2024. Accessed October 28, 2024. https://www.ajmc.com/view/exploring-racial-ethnic-disparities-in-cancer-care-prior-authorization-decisions

5. Ukert B, Schauder S, Cullen D, Debono D, Eleff M, Fisch MJ. Racial and ethnic disparities in prior authorizations for patients with cancer. Am J Manag Care. 2024;30(10):494-499.

6. Ross C, Herman B. Denied by AI: How Medicare Advantage plans use algorithms to cut off care for seniors in need. STAT. March 13, 2023. Accessed October 28, 2024. https://www.statnews.com/2023/03/13/medicare-advantage-plans-denialartificial-intelligence/