Livanta, the CMS contractor that reviews short hospital stays under Medicare’s two-midnight rule, has opened the door to greater use of the case-by-case exception and seems to have a more generous view of inpatient admissions than in its audits—including for appendectomies and gallbladder removals—according to its July publication.[1]
That’s the take of physician advisors who reviewed it. “They are much more liberal in allowing inpatient admissions on one-day stay patients,” said Ronald Hirsch, M.D., vice president of R1. Livanta’s medical directors clarified they’re talking about “emergent or urgent operations, not any admission from the emergency room for non-urgent conditions,” according to an email exchange with Hirsch. But he said people generally don’t show up at the emergency department (ED) for non-emergency appendectomies and cholecystectomies. Livanta also green-lit Part A payment for a one-day stay for a patient with melena although it denied claims along those lines last year. “I was shocked because it seems so different than what we have been told in audits,” Hirsch said.
A CMS spokesperson told RMC it reviewed Livanta’s document before publication. The spokesperson noted that “Livanta’s reviews must comply" with CMS’s two-midnight guidelines and regulatory requirements.
What to do with the Livanta development is an open question. At a minimum, hospitals should work with physicians to improve documentation of high-risk, one-day stays and push back on denials if they conflict with Livanta’s publication, Hirsch said. “They wouldn’t publish it if they didn’t expect people to use the examples of what they could do.”
But Stephanie Van Zandt, M.D., medical director of physician advisor services at a large health system in Florida, said it will continue to be “super conservative” with Medicare admissions under the two-midnight rule.
“Livanta is stepping out on a limb,” Van Zandt said. “When you read the Livanta cases, you go, what? I would never agree to that.” They don’t seem consistent with Medicare’s Program Integrity Manual and the risk of audit from any payer is too high and not worth the hassle.
In fact, Van Zandt has approved only one use of the case-by-case exception to the two-midnight rule—which allows Part A payment even when the physician only expects a one-night stay—in three or four years.
In its publication, Livanta—a Beneficiary and Family Centered Care-Quality Improvement Organization (QIO)—explained it relies on CMS’s two-midnight guidelines “to identify cases where resource utilization best justifies inpatient payment” and makes decisions based on the documentation available when the inpatient order was written. Under step 4 of the guidelines, Livanta assesses whether it was reasonable for the admitting physician to expect the patient to require medically necessary hospital services for two midnights or more, including all outpatient/observation and inpatient time. Under step 6, the QIO evaluates whether claims for patients who stayed fewer than two midnights (i.e., one night in the hospital) support the physician’s determination that inpatient care was necessary based on complex medical factors (e.g., risk of an adverse event, severity of signs and symptoms).
Livanta cited these examples of adverse events:
-
metabolic abnormalities (e.g., diabetic ketoacidosis, symptomatic hyperkalemia or hypercalcemia);
-
acute medical conditions (e.g., crescendo angina or life-threatening arrythmia requiring urgent intervention or high-risk medication);
-
pulmonary embolism with right ventricular strain;
-
acute surgical conditions (e.g., cholecystitis or appendicitis where early intervention may be associated with next-day discharge); and
-
“other use of high-risk medication that can only be given on an inpatient basis.”
Some of those examples drove physician advisors to distraction. “For them to come out and say all appendectomies and cholecystectomies can be inpatient was shocking,” Hirsch said.
For example, a 65-year-old Medicare beneficiary in good health except for hypertension who has an appendectomy after presenting to the ED with acute appendicitis “doesn’t fit the case-by-case exception as we previously interpreted it,” Hirsch said.
In their email response to Hirsch, Livanta’s medical directors noted that “if the procedure is not emergent then this would not apply. If we encounter evidence of a pattern of abuse of the emergency room during our reviews, we are required to report this to CMS for further investigation.”
It’s one thing for Livanta to say this, it’s another for hospitals to apply it. Van Zandt noted that her health system may have more than 100 Medicare patients a year who require emergency appendectomies and stay one night. “How am I going to justify that’s inpatient?” The same difficulty confronts hospitals with gallbladder removals. That’s an outpatient procedure unless something unexpected happens in the operating room or they spike a fever afterward, “but more than 99% of patients recover as expected,” she said.
And how could diabetic ketoacidosis be considered an appropriate use of the case-by-case exception? Patients with this condition are all over the place. Some are at risk of imminent death and some easily controlled with an adjustment in their insulin regimen and sent home, Van Zandt explained. Conservative health systems don’t use the case-by-case exception “because it’s too vague and ambiguous and the risk of short-stay audit is too high.”
Patients in Case Studies ‘Were Not in Distress’
Livanta’s case summaries of inpatient admissions were surprising because they sound more like observation cases with the potential for admission, Hirsch said. “These were patients who received treatment in the ED and improved dramatically. They were not in distress,” he said. That runs counter to CMS’s “drumbeat” that hospitals should routinely start with observation services if they’re not sure how long the patient requires medically necessary hospital care and upgrade to inpatient admission as appropriate.
Here’s the first example in the Livanta publication: “This 86-year-old male with multiple prior episodes of angioedema without clear causation presented with onset of voice change and dysphagia. He self-administered steroids and an EpiPen at home prior to the ED visit. He still had symptoms in the ED, but his vital signs and oxygenation were normal and there was no stridor or trouble with secretions. Imaging confirmed the diagnosis of pharyngeal and neck edema. Documentation in the record shows he was admitted to inpatient status for close monitoring. He was never in distress clinically and improved symptomatically in the ED prior to inpatient admission. He was admitted, treated with steroids and Pepcid, continued to improve overnight, and was discharged after a one-midnight stay.”
Livanta explained that the medical records supported the two-midnight expectation based on complex medical factors. “There is a real concern for a serious adverse event. This prompted his change in status to inpatient admission despite there not being a reasonable expectation of a two-midnight stay,” the QIO said. Therefore, Part A payment was appropriate.
Hirsch emailed Livanta to question its rationale. “The guy was nearly asymptomatic,” he wrote. In response, the medical directors responded: “The record in this case clearly documented the physician’s concern that there might be a deterioration in the patient’s condition, and this was the reason for the admission. The patient also received close monitoring. As you know, the QIO does not second guess clinical decisions made in the field if there are data to support them documented in the record. In this case the data were clearly explicated and there was no need to impute them. Had the record been unclear or suggested that there was minimal danger of recurrence, and the patient was admitted for routine observation, the decision might have been different.”
‘It’s a Conundrum Wrapped in an Enigma’
In the second example, Livanta also found inpatient admission under the two-midnight rule was appropriate even though the hospital didn’t meet the two-midnight benchmark. “This 79-year-old female has a history of gastrointestinal bleed, the most recent of which was six weeks prior to this presentation. At that time, a colonoscopy showed inflammation and ulceration in the sigmoid and rectum as well as likely diverticular bleed. She presented to the ED with several episodes of melena within the last 12 hours. Her hemoglobin (Hgb) was 10.8 and her vital signs were stable. The plan was to admit, monitor her hemoglobin, and consider endoscopy depending on the clinical course. She was stable throughout and was discharged after a one-midnight stay.”
Again, Hirsch emailed Livanta, expressing surprise because the patient wasn’t actively bleeding and had stable vital signs. In response, the medical directors explained: “This is a patient with a suspected diverticular bleed and several episodes of melena. As you are aware, these types of bleeds can be episodic and massive. While the patient was not bleeding at the moment, there was no way to be sure that there would not be another significant bleed prior to the endoscopy. We agree that this COULD have been handled as observation, but there was an Inpatient order that can be reasonably supported by the data in the record. Since we do not question inpatient orders that can be supported by the record, we would approve this case.”
What’s confusing is that the QIO and other auditors do, in fact, question inpatient orders, Hirsch said. “It’s a conundrum wrapped in an enigma.” When hospitals are faced with defending their one-day stays and other admissions, he suggests showing auditors these case summaries. A diagnosis of melena really throws him off because while it can be serious, Livanta denied 11 one-day stays for melena, according to its first year of short-stay reviews when they resumed after CMS paused them for almost two years.[2] They also denied 10 cases of ventricular tachycardia and 10 patients with complete heart block.
Another statement in the Livanta publication—that some drugs “can only be given on an inpatient basis”—makes no sense, Hirsch said. He told Livanta as much, saying in the email that “anything can be inpatient or outpatient,” and asked for an example. In response, Livanta’s medical directors said they disagreed with his comments. “There are a number of high-risk medications that might be associated with potentially physiologically unstable conditions and require close monitoring when given as a continuous infusion. One example would be continuous infusion of an intraarterial thrombolytic agent.”
Wary of Inpatient Orders Close to Discharge
It still doesn’t ring true for Hirsch. Patients with a pulmonary embolism arrive at the ED, are rushed to the cath lab and administered the medication that Livanta is referring to, Hirsch said. “No one stops and says, ‘I better put my orders in the electronic medical records to make sure it’s inpatient,’” he said. The surgeon writes the admission order while the patient is recovering. “The doctor knows they will admit the patient as inpatient, but the actual order that formally admits the patient doesn’t get written until after,” he noted. “It’s allowed.”
Livanta’s publication also doesn’t seem to square with the Medicare Program Integrity Manual’s section on medical reviews, Van Zandt said.[3] It states, among other things, “Per the 2-midnight benchmark, Medicare contractors shall assess short stay (i.e., less than 2 midnights after formal inpatient admission) hospital claims for their appropriateness for Part A payment. Generally, hospital claims are payable under Part A if the contractor identifies information in the medical record supporting a reasonable expectation on the part of the admitting practitioner at the time of admission that the beneficiary would require a hospital stay that crossed at least two midnights.”
The problem is that patients in the case studies would typically be treated with observation services because the physicians don’t anticipate admitting them as inpatients, she said. If they’re ultimately admitted, it would be after an ED visit and observation, with hours spent there included in the countdown toward two midnights, as the manual explains. But the physician wouldn’t have authored an inpatient admission order yet. “It’s a catch-22,” Van Zandt explained, “because when the physician places the inpatient order, the physician is attesting that the patient is requiring two midnights of medical necessary care.” Upgrading a Medicare patient from observation to inpatient and then discharging them later that day is “counterintuitive” to the two-midnight rule, she said.
‘Why Am I Still in Observation?’
Hospitals are obliged to review short stays, and cases with inpatient orders placed the day before or same day of discharge must be evaluated for possible denial of Medicare Part A payment and rebilled for inpatient Part B payment only, Van Zandt said. “Physicians who write an inpatient admission order to upgrade their observation patient after a two-midnight stay in observation and then medically clear the patient for discharge later the same day don’t show an expectation for a two-midnight stay,” she explained. At least one of the two nights in the hospital should be inpatient to qualify as an inpatient stay.
It drives her crazy. “You can’t upgrade when the patient isn’t going to be there for continued medically necessary care,” Van Zandt said. “That’s not compliant.” And if she were a Medicare patient, she’d be mad. “I stayed two midnights. Why am I still in observation?”
On the flip side of the coin are physicians reluctant to write admission orders after the patients have already been in the hospital one night. They hold off upgrading the patient to inpatient until, for example, they see the next round of test results in the morning. “I remind physicians they don’t have a crystal ball,” Van Zandt said. “What if the blood test is worse or the patient requires a rapid response? It has to be real-time decision making.” Also, when utilization review (UR) nurses are unsure whether the patient is going home in the next 24 hours, the question to ask the attending physicians is if the patient requires a second midnight of hospital care for treatment, procedures or consultations—not whether they met InterQual criteria, she explained. “There was a lot of knowledge lost during the pandemic with UR nurses and bedside physicians in terms of statusing hospital patients.”
‘It Took Five Emails to Convince’ the UR Team
The approval of admissions under the case-by-case exception is vanishingly rare at her health system and that won’t change. Van Zandt recently OK’d a one-day admission but “it took five emails to convince the utilization team in charge of Medicare short-stay reviews that I really meant for the one-midnight stay to be considered appropriate for inpatient.” The patient had a significant non-ST elevation myocardial infarction (STEMI), a type of heart attack. Four hours after the patient came to the ED, he had a stent inserted, was admitted as an inpatient and discharged the next day.
Utilization review wanted to downgrade to outpatient via condition code 44 and called Van Zandt because she’s the senior physician advisor, but “I decided to keep the stay inpatient despite a probable one-midnight stay.” Between the risk factors and severity of illness, she didn’t want to downgrade the case. After a short stay review, the case was deemed appropriate to bill under the case-by-case exception. (This was not coded as an inpatient-only procedure (IPO) per her coding department. Acute myocardial infarction requiring left heart catheterization (LCH) and percutaneous coronary intervention (PCI) has been on the IPO list since 2018, but CPT 92941 has very specific criteria and is reserved for truly emergent cases—primarily STEMI but occasionally non-STEMI if the patient is still symptomatic, which requires emergent/urgent LHC with PCI, Van Zandt said.) Her health system rebills less than 5% of fee-for-service Medicare cases as inpatient Part B after short-stay reviews, Van Zandt said. “It’s all about communications and education between UR nurses, physician advisors and attending physicians and performing case reviews in real-time,” she said. “We’re trying to empower doctors to do the right thing—status patients correctly and document it.”
Contact Hirsch at rhirsch@r1rcm.com and Van Zandt at stephanie.vanzandt@baycare.org.