Primary source: Cohere Health via PR Newswire, November 16, 2026, 09:00 ET

Clinicians in Cohere's Survey Report Prior-Auth Abandonment Fell From 55% to 28%.

On November 16, 2026, Cohere Health released its 2026 National Provider Survey on Prior Authorization. Among clinicians in the survey, reports that patients abandoned recommended treatment because of authorization delays fell from 55% to 28%. Ninety-six percent of clinicians and 94% of office administrators said they continue to trust AI in prior authorization. If you run an auth desk, that is a vendor telling you the delay problem eased, in a survey the vendor paid for. The sample is 200. Cohere sells prior-authorization AI to health plans.

55% to 28%

Clinicians reporting that patients abandoned recommended treatment because of authorization delays, year over year. Wakefield Research fielded the survey. This is a share of clinicians, not a count of patients.

What the survey actually measured

ItemWhat November 16 published
Who askedCohere Health, which the release calls a clinical intelligence and operations platform for health plans. Second annual survey. Wakefield Research fielded it.
Who answered200 clinicians and office administrators across U.S. hospitals and medical systems. The white paper page, opened the same day and dated November 13, 2026, splits that as 100 clinicians and 100 office administrators.
AbandonmentReports of patients abandoning recommended treatment because of authorization delays fell by roughly half, from 55% to 28% among clinicians. The release also says there were fewer reports of avoidable ER visits or hospital care tied to delays, and it does not give that second number.
Trust96% of clinicians and 94% of office administrators continue to trust AI in prior authorization. The release says that trust is becoming more conditional, and that providers want transparency, consistency, and human oversight.
What it does not sayA denial rate, a turnaround time, a plan's name, or a result from Cohere's own platform. The 55% and 28% are shares of clinicians reporting an outcome.

“Fewer patients are abandoning care or ending up in the emergency room because of prior authorization delays, and that's meaningful progress,” said Dr. Brian Covino, chief medical officer. “But providers are still advocating strongly for patients who face barriers to care. They want AI that eases manual work and speeds care while clinical judgment and human oversight stay in the driver's seat. We share that view completely: AI should never be used to deny care, only to help streamline administrative work so providers can focus on the decisions that matter most for their patients.”

The release is blunt about the work that remains. Providers are still navigating manual, opaque processes. Unclear documentation requirements and inconsistent plan rules produce avoidable denials when the office does not know what is needed. Cohere's own sentence: “More automation alone will neither ease that friction nor earn providers' trust if the process remains unchanged.”

A patient survey four days earlier asked a different question

On November 12, Penguin Ai released a consumer survey it commissioned and fielded in November 2026 among 500 U.S. adults. Every respondent needed at least one prior authorization for themselves or a family member in the past two years. Of the 273 who reported a delay, denial, or medication restriction, 62% said they have given up on care because fighting the insurer was not worth it. Across all 500, Penguin says 48% say the same.

Those are not the same statistic. Cohere asked clinicians and administrators at hospitals and medical systems whether patients abandoned recommended treatment. Penguin asked adults who had been through prior auth whether they themselves gave up. One share can fall while the other stays high. Both surveys were paid for by companies that sell prior-authorization tools. Penguin's about says payers, providers, and risk-bearing organizations use its platform for prior authorization, HCC coding, claims adjudication, payment integrity, and appeals.

What this survey is not

Cohere's about says the platform produces “up to 9x ROI and 94% provider satisfaction.” That sentence is company boilerplate in the release. It is not a finding of this survey, and the release does not show the math. A drop in the share of clinicians who report abandonment is not a drop in authorization volume, and it is not a measured reduction in denials.

RCAI analysis: fewer walk-aways is not fewer auths

This is RevCycleAI's read.

An auth team lives in volume, turnaround, and the appeal. Cohere's clinicians are reporting that fewer of their patients are walking away, and that they still trust AI if a person stays accountable. That can be true at the same time the queue gets longer. Menlo's healthcare report, out the same morning and covered here, says 91% of payers expect more prior authorization requests over three years, and that prior auth is their top AI priority at 82%. If plans automate the review and providers automate the request, the work can get faster and more numerous together. This survey does not measure volume.

Two other prior-auth records are easy to mash into this one. ONC and CMS opened a voluntary electronic prior authorization test on November 12. Passing it is not compliance and not a CMS determination. The JAMA research letter on WISeR found targeted traditional-Medicare services fell in six states over six months, and the letter cannot say the forgone care was waste. Cohere is reporting what clinicians say about patients abandoning treatment. WISeR measured services delivered. Different programs, different questions.

The RCAI signal

In a 200-person survey Cohere commissioned, clinicians reporting that patients abandoned treatment over auth delays fell from 55% to 28%, and 96% of clinicians still say they trust AI in prior auth. That is not a patient census, and it is not a result from Cohere's platform.

RCAI take

Use the 55-to-28 as what clinicians in this sample say they are seeing. Then look at your own auth queue and ask whether patients are still dropping off the schedule. The open item is the “fewer” ER visits and hospital stays: the release asserts the direction and does not print the number, and the white paper landing page does not print it either. Penguin's 62% is a different respondent answering a different question, four days earlier, for a different vendor.

Sources: PR Newswire, “National Survey Shows Progress on Prior Authorization Delays as Providers Push for More Transparent AI,” Cohere Health, November 16, 2026, 09:00 ET, Boston (page opened). Cohere Health, “Better AI, not less AI: 2026 provider survey,” white paper page dated November 13, 2026 (page opened; the page confirms the second annual survey, Wakefield Research, and the 100 / 100 split, and it does not restate 55%, 28%, 96%, or 94%). PR Newswire, Penguin Ai consumer survey, November 12, 2026, 08:11 ET (page opened). Dr. Brian Covino's quotation is verbatim from the Cohere release. The 9x ROI and 94% provider satisfaction lines are the Cohere About, not survey findings.