At last week’s National Hospice and Palliative Care Organization leadership meeting, data analysts commissioned by the Hospice Action Network offered a second opinion on a CMS-commissioned report released recently by Abt Associates. More of a critique than a second opinion, the team from Avalere drew different conclusions than Abt had published and then went so far as to question Abt’s having broken accepted analysis protocol by offering opinions on its own conclusions. No reaction from an Abt representative was available by press time.
Avalere Vice President Eric Hammelman described his team’s analysis before a large NHPCO audience, explaining that the team had come to its alternate view by starting with sounder assumptions than Abt had used. CMS, he said, was looking for ideas on how reform the hospice payment system. Avalere believes Abt’s perspective, based on its set of assumptions, may not lead the government in the best direction.
For example, Abt reported that $750 million (5%) of last year’s $15 billion hospice cost was paid for services provided outside of hospice benefit days. However, Hammelman countered, Abt used a one-day overlap assumption. When a hospice admission occurs on the same day as hospital discharge, a common, even beneficial, practice, Abt assumed duplicate payments were being made to hospital and hospice for the same service day.
“But the hospital gets the same DRG payment regardless of the patient’s length of stay,” Hammelman reminded the audience. “They are not specifically paid a separate amount for the day of discharge. The issue is about how to properly define ‘boundary days’ in such a way that the data meaningfully informs CMS decision makers.” He further insisted that discharge dates can be, and often are, entered erroneously, as often happens when data entry personnel record a weekend discharge when they return to work on Monday morning.
“We determined it was more accurate to use a 3-day overlap to identify duplicate payments,” he explained. “By doing that, we found 20% fewer incidents of duplicate payments.”
Other common occurrences that can skew data happen when a patient moves in the other direction, from hospice to hospital. Live discharges to hospital happen 40% of the time, Hammelman said. “If the hospice sees a patient on Mondays and Thursdays, and the patient goes back to the hospital on Friday, they may learn of the transfer and record the discharge from hospice on Monday, at the time of their next scheduled visit.
“Or, if the patient has a one- or two-day hospitalization and is back home in time for the next visit, it can and does happen that the hospice billing staff does not learn of the hospitalization until they file the claim.” He added that one advantage of hospital and ACO partnerships with post-acute providers is that they will tell you when your patient shows up.
Concern over hospital vs. hospice competition
Among other notable findings, two stood out. Abt reported that 38% of emergency department visits are related to falls, under a category CMS calls “injuries and poisonings,” and that an unusually high percentage of physician encounters take place in the hospital.
Hammelman noted that CMS wants to know why so many patients are falling and causing hospital costs within a hospice episode. “Can’t hospice be doing something more to keep terminal patients safer? Plus, they want to know why so many routine physician examinations happen in the hospital instead of in the home or in a skilled nursing facility, leading to excessive non-hospice expenses.”
Avalere analysts looked deeply into these two situations and saw problems but not the hospice billing errors Abt identified. “Most ED visits are coded as ‘unrelated to the patient’s terminal diagnosis,’” Hammelman observed. Why? Because over 90% of the time, the ED coder selects a diagnosis that will make sure the hospital gets paid. “Hospitals are very good at getting paid,” he opined. “20% of emergency department visits are coded for ‘ill-defined conditions.’”
Regarding the second problem, physicians making home or SNF hospice visits frequently code that they saw the patient in the hospital. “I don’t have any idea why they do this,” he admitted, “but it happens quite frequently.”
These two coding practices should be of concern to hospice providers, he warned, because CMS is going to question hospices about both. If CMS makes a payment rule modification that reduces hospice payments in these cases, hospices will be improperly harmed. If it stops paying the hospital when coding is suspicious, hospitals might very well begin to demand that hospices cover costs that CMS begins to deny.
After reading Abt’s report, CMS is concerned about overlapping payments that amounted to $50 million (.3%) in 2013. Avalere discovered that 6.4% of hospice-to-hospital discharges are two days or less. In 21.8% of cases hospital stays are between two and fourteen days.
“Forty percent of hospice patients experience hospital stays,” Hammelman explained, “but often the hospice discharge does not happen until two days after the actual transfer. Conversely, at the point of hospice admission, hospitals, SNFs, and HHAs might record the hospital discharge later by a few days than the actual transfer into hospice, each for different reasons.”
If Abt had used a three-day overlap view, as Avalere did, instead of a one-day overlap, they would have come to a completely different conclusion about double billing, Hammelman insisted.
According to Avalere, the data also disproves both of two opposing assumptions. One theory is that families are more prone to call 911 because they are afraid and unfamiliar with hospice, especially during the first seven days, when the lion’s share of ED visits happen. Others believe that skilled nursing facilities are trigger happy with 911 over cost and payment reasons.
“It turns out,” Hammelman said, “the rate of hospice patients visiting hospital emergency departments is exactly the same whether the patient is at home or in a SNF or in an in-patient hospice.”
Where hospice providers should be focusing efforts, he reasoned, is on those first seven days after admission. “There should be something hospices could be doing to onboard patients that will discourage their families or their institutions from calling 911 so often during those early days.”
Abt’s surprise opinion
Hammelman made it quite clear that it is unusual for a CMS contractor to make an editorial comment with its data analysis. Abt issued a critical opinion in its report regarding a perceived 14% decrease in hospice services during the last two days of life, implying that hospices are either doing an inadequate job of guessing when death will occur or simply cutting costs at the patient’s expense.
Hammelman pointed out the fact that Abt does not count physician, clergy, social worker, or volunteer visits, though they do count therapy visits. “What is a therapist going to do for a person on the last day of life?” he asked, letting his frustration show for the first time. “When Abt says no hospice services are provided, they ignore that hospices gradually – and appropriately – shift from medical to bereavement services at the end, as comfort grows more important than medical care. Hospices provide plenty of visits in the final week of life but Abt does not score them. We brought the measurement of so-called ‘abandoned patients’ down to 5% from 14% by including physicians and clergy.”
What Abt did score on the hospice side of the ledger was quite a lot of spending that occurs just before the hospice start date. By using a three-day overlap to measure boundary days instead of Abt’s one-day overlap, Avalere calculated a 15% difference in estimated non-hospice spending in the days just before hospice admission for patients who died in 2012 while on hospice.
Lastly, the Avalere cautions
Hospice owners need to be hyper-aware of certain aspects of this Medicare Technical Report data, whether reported by Abt or re-interpreted by Avalere.
As a summary piece of advice, Hammelman related the story of a physician who once yelled at him during a survey phone call, insisting, “You analysts can’t understand what I do until you sit in my office for a day.”
“I couldn’t disagree with him,” Hammelman said. “And CMS needs the same kind of help. They need you to make sure they know what goes on in a dying patient’s home.”