Introduction
On December 28, 2023, Carol Clemens received an alarming notice: Only ten days after she had entered a skilled nursing facility following a life-threatening collapse, her health insurance company was refusing to pay for her continued stay.
Given that Clemens was unable to eat solid foods, speak more than a few words at a time, or walk without assistance, she immediately appealed the decision, asking the insurer to reconsider letting her finish the rehabilitation program.
Clemens’s doctor had prescribed a stay until at least January 18, 2024.
The insurer, however, was not convinced. Just one day after Clemens submitted her appeal, the company denied the request, stating that a continued stay was “not medically necessary.”
With no one to pay for her stay, Clemens returned home on January 3.
Just three days later, no longer having access to the supervision of the skilled nursing facility staff, Clemens fell again, this time resulting in a “traumatic subarachnoid hemorrhage . . . [and] severe brain bleed.”
She needed to be readmitted to the hospital, where she spent over two weeks recovering before returning to the skilled nursing facility to restart her rehabilitation, this time with a traumatic head injury.
Clemens is not alone in her experience. Clemens serves as one of multiple plaintiffs in a class action suit filed in Minnesota against insurance giant UnitedHealth Group that makes multiple claims related to Medicare Advantage plans incorrectly denying coverage to aging adults.
How did anyone seeing the facts of Clemens’s case—her inability to eat, talk, and walk—determine that continued rehabilitation was “not medically necessary?” Well, Clemens claims, no one did. As the plaintiffs allege, UnitedHealth Group is not relying on human expertise to make these determinations but is instead deploying “artificial intelligence (AI) in place of real medical professionals[,] . . . overriding . . . physicians’ determinations as to medically necessary care.”
This Note explores the current state of coverage determinations in Medicare Advantage. It views insurers’ use of AI—and the harmful effects that accompany such use—as the latest development in a Medicare Advantage program that has been plagued for decades with inaccuracy and opacity. To combat this problem, CMS should enhance information sharing about denials while also increasing meaningful access to the appeals process for individual beneficiaries. Not only will such mechanisms improve access to care, but they will also reflect an important return of power and autonomy to beneficiaries, enabling the individuals most affected by determinations to take control of their own coverage and care.
The Note proceeds in three Parts. Part I explores the concept of medical necessity in Medicare, asking what it is and who decides it. Part II outlines the problems present in the coverage determination process today, noting the high rates of inaccuracy and opacity in determinations. It also summarizes CMS’s most recent attempt to combat these problems, concluding that more is needed. Part III charts a path forward, presenting two important accountability mechanisms that CMS can enact to enhance insurer accountability.