Health Letter, January 2024
By Robert Steinbrook, M.D.
Director, Public Citizen's Health Research Group
If you’re not outraged,
you’re not paying attention!
Read what Public Citizen has to say about the biggest blunders and outrageous offenses in the world of public health, published monthly in Health Letter.
UnitedHealth Group has about 90,000 employed or affiliated doctors, which account for about 10% of all physicians in the United States. UnitedHealthcare, the health benefits business of UnitedHealth Group, is the nation’s largest health insurance company.
A recent investigation by Stat news showed how a UnitedHealthcare subsidiary called NaviHealth pressured its employees to follow the calculations of a computer algorithm in determining when to cut off payments to seriously ill Medicare patients receiving rehabilitation care. The goal was to keep rehabilitation stays of patients in Medicare Advantage plans within 1% of the days projected by the algorithm, regardless of whether additional days of rehabilitation were justified under Medicare coverage rules, thereby cutting costs, and maximizing profits. A subsequent class action lawsuit alleged that the companies knew that the algorithm had a high error rate, citing the frequency of payment denials “reversed through internal appeals processes or administrative law judge rulings.” Although UnitedHealthcare and NaviHealth have disputed the allegations, the Centers for Medicare & Medicaid Services (CMS) is looking into the matter to see if enforcement or compliance actions are needed.
There are numerous examples of Medicare Advantage insurers’ use of artificial intelligence and related computer systems to avoid covering needed care. As 30.8 million eligible Medicare beneficiaries, about half of all Medicare beneficiaries, are enrolled in Medicare Advantage plans, this deeply troubling problem requires urgent attention.
Although recent CMS regulations seek to prevent Medicare Advantage insurers from denying care inappropriately, the effectiveness of these regulations remains to be seen. A final Medicare rule, issued in 2023 and applicable to coverage beginning on January 1, 2024, affirms that Medicare Advantage plans cannot deny coverage of medical products or services that would otherwise be covered if a beneficiary were enrolled in the traditional Medicare program. The rule also states that “[Medicare Advantage] organizations must ensure that they are making medical necessity determinations based on the circumstances of the specific individual…as opposed to using an algorithm or software that doesn’t account for an individual’s circumstances. Moreover, plans “must make the evidence that supports the internal criteria used by (or used in developing) these tools publicly available, along with the internal coverage policies themselves.”
CMS has not prohibited the use of artificial intelligence and algorithmic tools by Medicare Advantage plans. Measures beyond the new CMS regulations may be needed, as well as proactive monitoring of the effects of the new regulations based on how frequently Medicare Advantage insurers deny needed care. A November 2023 letter to CMS from 32 House Democrats provides constructive suggestions about how to ensure proper oversight. The suggested measures include: (1) Reporting data about the denial of services; (2) Comparing “guidance” from algorithmic tools with coverage decisions; (3) Assessing how frequently initial determinations of coverage by artificial intelligence are adjusted for unanticipated changes in a patient’s condition; and (4) Determining whether algorithms are “self-correcting.” This means that when a premature termination of services is reversed on appeal, the reversal is fed back to the software so that it learns what care should be covered and then factors what it has learned into subsequent decisions.