Dive Brief:
- A digital and privacy rights nonprofit is suing the CMS in an attempt to get more information on the agency’s pilot program that uses artificial intelligence to manage prior authorization requests in Medicare.
- The Electronic Frontier Foundation argues the CMS has not responded to a Freedom of Information Act request it submitted relating to the technology companies participating in the pilot and any evaluations on accuracy, bias or hallucinations in their technology.
- The EFF wants a district court in California to order the CMS to disclose the records, arguing the pilot has already delayed care for seniors and increased administrative strain on providers since it launched in January. “The public has a right to know more about the algorithms driving decisions around their healthcare,” Tori Noble, staff attorney at the EFF, said in a statement.
Dive Insight:
The CMS’ Innovation Center announced the pilot program, called the Wasteful and Inappropriate Service Reduction, or WISeR, Model, last year.
The experiment runs through 2031. Under the model, doctors in six states will have to receive AI-backed prior authorization before offering some care in traditional Medicare, as part of an attempt to cut down on “low-value” services that CMS says have limited clinical benefit.
However, prior authorization is uncommon in fee-for-service Medicare, and providers have long argued the practice delays medically necessary care. Clinicians and Democrat lawmakers have already raised concerns about the pilot, saying it could reduce healthcare access and add more administrative work for providers.
Using AI for prior authorization has also come under fire. Some insurers have faced lawsuits for using AI to allegedly improperly deny claims, including in Medicare Advantage plans, where prior authorization is more common.
The lawsuit from EFF, filed Wednesday in district court in California, said CMS’ model is already causing harm. In Texas, only 62% of requests under the pilot were initially approved, rising to 84% once a human reviewed them, the nonprofit said, citing a Washington Post article published this month.
In comparison, 92% of prior authorization requests were fully or partially approved nationwide in MA.
Meanwhile, the public has limited information on how the AI algorithms used in the pilot operate, EFF said. For example, little information is available on the data used to train the AI, according to the suit.
“Outdated, incomplete, or biased training data can result in disproportionate denials of care to underserved populations and protected classes,” the EFF wrote. “And it is unclear whether the WISeR Model has any safeguards against systemic flaws such as algorithmic bias, privacy violations, and wrongful denials of care.”
In an attempt to find more information on the pilot, the nonprofit filed a FOIA request in January, which allows individuals to ask for federal and state agencies for access to government records.
The EFF requested agreements with software vendors participating in the model, documents describing how the companies would be paid, data sharing agreements between the CMS and vendors, and information about how their technology is tested, among other information.
Though the CMS acknowledged it had received the EFF’s FOIA request in February, it hasn’t released any records in response, according to the complaint. That violates FOIA, which typically requires agencies to respond to requests within 20 business days, the group argued.
The EFF wants the court to order the CMS to disclose the records, as well as waive fees associated with the FOIA request and award the nonprofit costs related to the lawsuit.
The CMS did not respond to a request for comment by press time.
