Artificial intelligence (AI) is everywhere, even when you don’t ask for it. When you do a search with Google, it now provides not only webpages but also an AI summary of what it thinks I am trying to figure out. And I will admit that I am still very wary of AI, but possibly because I have been avoiding learning to use it optimally.
Then, last week I got a question and decided to test the Google AI called Gemini. A case manager wrote to me asking for help. She states that the skilled nursing facilities (SNFs) in her area are refusing to accept patients under Part A from inpatient rehabilitation facilities. She stated that the facilities have been pushing back and asking if the patient has had a three-midnight inpatient stay in a “regular” hospital prior to being in acute inpatient rehab and refusing patients without that stay. According to them, they have been denied payment because of the three-midnight rule for access to Part A coverage of a SNF stay and that an inpatient rehabilitation facility (IRF) stay did not count.
Now I know that is wrong.
The Medicare Benefit Policy Manual, Chapter 8, states “In order to qualify for post-hospital extended care services, the individual must have been an inpatient of a hospital for a medically necessary stay of at least three consecutive calendar days.” And a patient admitted to an inpatient rehab facility is considered an inpatient of a hospital so the stay does qualify them for Part A SNF care.
So, I thought this was the perfect test for AI. So, I asked Google’s Gemini.
First, I asked, “Does an IRF stay qualify for Part A SNF benefit?” and Google answered “No, an Inpatient Rehabilitation Facility (IRF) stay does not qualify for or count toward a Part A Skilled Nursing Facility (SNF) benefit because they are two separate, distinct Medicare Part A coverage categories with their own qualification rules. However, a qualifying inpatient hospital stay before an IRF can satisfy the three-day hospital stay requirement needed if you later transition to an SNF.”
Then I asked, “Does a three-day IRF stay qualify for Part A SNF benefit?” adding only that the stay was three days. And Google answered, “Yes, a 3-day stay in an Inpatient Rehabilitation Facility (IRF) counts as a qualifying inpatient stay for Medicare Part A Skilled Nursing Facility (SNF) benefits, provided the IRF stay met formal inpatient status criteria spanning at least three consecutive midnights.”
And the most fascinating part of those results is that both of them had as their reference document the exact same webpage from a rehabilitation facility that explains Medicare coverage for rehabilitation. Same reference, two totally different conclusions.
I know I am not the first to warn that these public AI systems cannot be 100 percent trusted. It’s easy, it saves time and effort, it can compile and interpret reams of information into a usable format in seconds, but as is said, be wary when someone offers you something for free. We want to trust it, but we must proceed with caution. Ensure that the input is correct and verify the output. Don’t take the human element out of the loop.
Now why did the nursing home get denied? If I had to speculate, I would guess that the SNF did not prepare the claim properly to indicate the dates of the prior inpatient admission at the IRF with occurrence span code 70. And without that, the claims processing system would rightly reject the claim. But that’s just a guess.
And one last unrelated item of note. I reported a few weeks ago that the Centers for Medicare & Medicaid Services (CMS) is not happy with the American Medical Association’s (AMA’s) monopoly on the use of CPT® codes.
Well, last week, a patient’s rights organization sued the AMA to make the CPT codes freely available. So now we wait to see not only what CMS will do but also what the courts will decide.
Perhaps the nightmare scenario of ICD-10-PCS for every service, inpatient or outpatient, will not be in our future after all.
EDITOR’S NOTE:
The opinions expressed in this article are solely those of the author and do not necessarily represent the views or opinions of MedLearn Media. We provide a platform for diverse perspectives, but the content and opinions expressed herein are the author’s own. MedLearn Media does not endorse or guarantee the accuracy of the information presented. Readers are encouraged to critically evaluate the content and conduct their own research. Any actions taken based on this article are at the reader’s own discretion.


















