On October 17, the Senate Committee on Homeland Security and Governmental Affairs, Permanent Subcommittee on Investigations (PSI) chaired by Senator Richard Blumenthal (D-CT) released  “Refusal of Recovery: How Medicare Advantage Insurers Have Denied Patients Access to Post-Acute Care,” a report examining the  prior authorization and denials data from three of the largest Medicare Advantage (MA) organizations: United Healthcare, Humana, and CVS/Aetna. Combined, they cover 60 percent of MA plan enrollees.

The report’s findings underscore the on-the-ground experiences that LeadingAge skilled nursing and home health members share with us repeatedly: frequent denials of necessary care; huge increases in prior authorization requests overall; short durations of care and services, when they are approved; and the inability to speak to a human involved in the review of clinical documentation.

Methodology

The PSI sought data on prior authorization requests and denials from the three largest MA organizations between 2019 and 2022. This time frame was chosen because it aligned with increases in concerns from patients and providers, including LeadingAge members, on the rising prevalence of prior authorization and its threat to resident and patient wellbeing.

PSI sought, and received, year-by-year breakdowns of prior authorization requests and denials by setting by year to understand how utilization of prior authorization varied by post-acute care (PAC) setting. The report looked at long term acute care hospitals (LTCHS), inpatient rehabilitation facilities (IRFs), skilled nursing facilities (SNFs), and home health agencies (HHAs), though the bulk of the reporting and data collected focused on LTCHs, IRFs, and SNFs.

PSI also obtained documentation including internal plan meeting notes and presentations around the use of algorithms, artificial intelligence (AI), and other predictive technologies including how the companies were using these algorithms and technologies in utilization management practices. In addition, PSI reviewed plan training materials for prior authorization reviewers and the metrics by which their performance was measured. They were also briefed by the companies and spoke with stakeholders, including LeadingAge.

Key Findings

From the more than 280,000 pages reviewed thus far, the PSI report arrived at some dramatic findings that mirror the experiences of LeadingAge members and the older adults. These include

PAC denials increased dramatically across the Top 3 MA organizations between 2019 and 2022 and are proportionally higher than their overall service denial rates. 

In 2022, United Healthcare (UHC) and CVS denied post-acute care (PAC) prior authorization requests at rates 3x higher than their overall rates of denial. Humana denied post-acute care service requests at a rate 16x higher than its overall rate of denial, denying 24.6% of PAC requests. CVS (Aetna) denial rates remained stable but were already significantly higher than UHC and Humana at 25.9% across all PAC services. Simultaneously, the number of prior authorization requests for CVS enrollees increased 57.5% making their PAC denials higher than all other services.

In addition, UHC’s denial rate for skilled nursing facility care increased by a factor of nine, from an initial denial rate of 1.4% (3,016 requests) in 2019 to an initial denial rate of 12.6% (34,359 requests) in 2022.

From 2019 to 2022, UHC shifted care approvals away from SNF and other PAC providers to HHAs to generate savings.

While home health service denials were down during this period, the number of prior authorization requests for home health rose from 19,283 in 2019 to 356,606 requests in 2022–an 18-fold increase. This resulted in a significant increase in the administrative burden placed on these providers. While initial service approvals were approved for home health, UHC generated savings by simultaneously launching a “home health management solution” that reduced the number of home health visits and episodes.

Artificial intelligence and predictive technologies were deployed primarily for the purposes of reducing plan costs even when plans knew the algorithms made errors, and increased denial rates. 

During this time, the report documents these plans testing and deploying a series of artificial intelligence, automation and/or predictive technology tools. The result was most of these tools generated more denials. For example, SNF denials increased once NaviHealth, a post acute care, AI-driven, utilization management product purchased by Optum (a division of UHC) in 2020 and used by UHC and Humana, was deployed. The NaviHealth “nhPredict” algorithm, which is used to determine denials, estimates the number of days of SNF care a beneficiary is expected to need based on an AI evaluation of similar beneficiaries, rather than relying on the actual clinical assessment of the patients by the care provider.

NaviHealth employees were specifically instructed “Do NOT guide providers or give providers answers to the questions” used to collect information that guided prior authorization decisions–a finding that helps to explain the frustration we’ve heard from members who’ve told us that plan representatives that offer no help in getting services approved. Those representatives were instructed not to help.

In another case, UHC realized one tool had faster review times for requests and an “increase in adverse determination rate,” which they felt was “validated based on findings contraindicated evidence missed in the original review.” This would suggest these new tools will require providers to be hypervigilant in documenting information submitted to the plans to ensure the information cannot be misconstrued by a machine as contradictory, or risk more requests being denied.

UHC wasn’t alone. “CVS developed a data-driven strategy of focusing on prior authorization requests with ‘a significant probability to be denied.’ ”  These companies all engaged in practices around technology-automated denials with the goals of reducing the time spent in prior authorization request evaluations and the money spent on human reviewers.

This is in contrast to the burden being put on LeadingAge provider members to deliver more clinical documentation that is seemingly ignored or not appropriately considered, which reinforces members’ experience that it was increasingly difficult to get a human on the phone to discuss a request.

The report findings demonstrate that plans’ prior authorization practices are not designed to improve beneficiary experience. Rather, they are designed to deny and deter beneficiaries from accessing medically necessary services that are covered under Medicare. For providers, the administrative burden resulting from these prior authorization requests is exponential as the number of requests increases, and as predictive technology make careful documentation submissions paramount.

The report concludes that the “data obtained so far is troubling regardless of whether the decision reflected in the data was the result of predictive technology or human discretion. 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.”

While the report takes insurers to task, all of its recommendations are squarely focused on the Center for Medicare and Medicaid Services (CMS). In essence, the report finds that there is evidence suggesting wrongdoing by the plans and that although CMS has the authority to remedy this situation, the agency has yet to act.

Recommendations that CMS begin collecting MA prior authorization data broken down by service category as a matter of transparency as well as to improve oversight and the ability to identify trends. With that data in hand, CMS should then identify insurers for targeted audits based upon their prior authorization practices.

However, the report further states that the authors of the report are not confident that current CMS oversight activities, as outlined in a 2024 memo, will be sufficient. “The coverage criteria for many stays in post-acute care facilities are flexible enough to enable denials whose wrongfulness would be difficult to detect without close inspection.”

Lastly, the report suggests that regulatory actions to date don’t go far enough to limit inappropriate use of predictive technologies to deny care. Therefore, it recommends CMS expand the responsibilities of MA plan utilization management committees to prevent predictive technologies from unduly influencing human reviewers.

The PSI has no authority to pursue its own legislation but does have the ability to refer matters to the appropriate Senate committees that do have jurisdiction over the matters raised by the report.

How much influence this report has will likely be impacted by the outcome of the election. Nonetheless, it will be difficult for the MA plans to suggest that the findings aren’t true when it is based upon information that they provided.

LeadingAge will use the findings of this report to support its advocacy in pursuit of further reforms to the MA program.