Thursday, February 3, 2011

Getting the Incentives Straight

Good policy is about getting the incentives straight. My post yesterday on the recent paper in JAMA looking at differences between For Profit and Non Profit hospice providers noted that adding quality of care to discussions of cost was needed to make the best policy.

I think the overall goal of health care is to extend people's lives and/or to improve the quality of their life (QOL). Since everyone dies, there are eventually diminishing returns on the productivity (here defined as extends life and/or improves QOL) on what we spend to forestall death and improve QOL. It is hard to determine when this takes place, and harder still to decide what to do when this point is reached. But, it is inevitable. Hospice is one option when patients begin to move into this murky territory when treatments may no longer be 'worth it' as defined by them and their doctor(s). However, by improving QOL, hospice is most certainly still productive care as I have defined the term.

QOL is hard to measure on a prospective basis, meaning as you move through time. Using retrospective quality assessment, especially family members saying hospice helped my loved one and me through a hard time, hospice passes the market test as around half of all Medicare beneficiaries who die do so while using hospice. We know that. However, we need better measurement of quality of hospice care if we are going to get the incentives right in Medicare hospice policy. How quality changes and improves by length of hospice use is very important if we are going to determine the optimal period of hospice use from a QOL perspective.

The worries about increasing length of hospice in the Medicare program are focused on the long periods of use, which is understandable because hospice has long held out hope (and some evidence) of an intervention that helps patients and reduces Medicare expenditures. I think that often groups such as MEDPAC and Congress focus so much on long hospice stays because it helps drive a narrative that one way to reduce costs is just to stop people from abusing the system. I am against abuse of the system, but we won't 'waste fraud and abuse' (WFA) our way out fo the overall cost problem our country faces, and too much focus here just lets us avoid harder issues.

Focus on WFA and long stays in hospice generally is inevitable, and correct, but we need to keep in mind that there is more than one problem (long stays) to be considered in trying to get the incentives correct in hospice. Short stays in hospice may also represent a problem.

A paper I did with colleagues found that hospice reduces Medicare expenditures by around $2,200 in the last year of their life; this means that after choosing hospice, Medicare paid about $2,200 less for their care until death than they would have had they not chosen hospice. Maximum cost savings to Medicare occurred with a length of hospice use of around:
  • 60-100 days for Cancer
  • 50-110 days for diseases other than Cancer
And cost savings at the maximum level were around $7,000 for Cancer, about double that for non-Cancer. Keep in mind the definition of saving here is what Medicare paid for care, not the profit margin of a hospice, which is a related, but not the same issue. And this paper, like most focused on cost didn't measure quality, which is a limitation and what we really need to know (cost and quality together).

The median length of hospice use in the study in JAMA this week was 20 days for those treated in a FP hospice and 16 days for those treated in a NP. The years and data sources are not the same as those that produced the cost savings data noted above, but to illustrate the incentives lets stick those numbers (they have changed some, but not that much).

If the median length of use (half the patients use less than amount, half more) is 20 days, then doubling the median length of use still doesn't produce a length of use that has achieved maximum savings for the Medicare program. For around 75% of the patients using hospice, we see that the profit motive incentive, and the incentive of the Medicare program from an overall cost perspective is the same: we need to find a way to increase length of hospice use. What we wish we knew was how quality of life/care improved for patients when moving from a length of hospice use from the median (20 days) toward the maximum cost savings for Medicare (2-3 months). The shape of this curve is needed to know how hard we should work to try and incentivize stays on the short end of the spectrum to be longer. If maximum quality of life improvement is obtained at 20 days, it might not be that important. However, if QOL and patient benefit is not achieved until 60 day length of use, then we should work very hard to increase shorter periods of hospice use.

We shouldn't focus policy only on the very long periods of hospice use, we must make sure that whatever we do there doesn't shorten the length of use for more common periods of usage. If the median fell from 20 days to 15 days, you would be reducing the cost savings to Medicare and probably harming patient QOL. Note that the publicly available data on Medicare hospice margin (profitability) shows that periods of hospice use less than 10 days or so are money losers for hospice providers, so the profit incentive at the lowest end of hospice use (shortest 30% of stays) is also aligned with both quality improvement and the finding that hospice reduces Medicare expenditures, with the biggest daily effects being closest to death. Thus, everyone has an incentive to transform 5 day lengths of hospice use into at least 15-20 day periods of use.

Both the long periods of hospice stay and the short ones may represent policy problems. We need to make sure that Medicare hospice policy pays attention to both ends of the distribution, and doesn't only seek to shorten very long periods of hospice use.

Updated: just added a bit in the last two paragraphs to further clarify aligning of incentives to extend length of short periods of hospice use.

9 comments:

  1. Don
    Does your work in evaluating hospice savings take into account the potential denominator of eligible hospice patients?

    By that I mean, costs savings (put aside QOL for this question) are relative to churning in an acute stay hospital, and SNFs?

    Could a hospice "enlightened" community, that was once use to dying at home with suboptimal services (read: low cost) suddenly see the light, and ramp up use.

    I am envisioning a Mcaid/CHIP scenario to boost child enrollment. I know not apples to apples, but you get the picture. Suddenly, the comparison group is not acute care vs hospice, but die at home (low cost "bad death") vs hospice?

    thanks
    Brad

    ReplyDelete
  2. Brad

    In one sense, the reason hospice can 'save money' is because the 'control group' is so expensive. In our paper, if you look at last 365 days of life, costs are about same for hospice v. those not dying in hospice. We used propensity score matching because is a choice and not random. Then we took hospice entrance and compared from that day to death for the hospice decedent and two controls who were similar but didn't use hospice. The savings are from the point of hospice entry. The biggest savings are in the last 7-10 days of life because hospice folks tend to avoid the last hospitalization. Our study sample was all deaths, so someone doing 'self hospice' meaning die at home with little care could have been in the control group if similar in terms of diagnosis and other observables. So, a low use community with no hospice could spend more with access to hospice. And hospice users using such care longer than ~125 days or so cost more than non hospice from point of hospice election to death. So, long periods of hospice use to cost more for Medicare.

    Real question would be if you went to concurrent care (start hospice without unelecting curative)...there are two studies that still show savings but many (MEDPAC, CMS) don't believe it I don't think.

    ReplyDelete
  3. Don - I couldn't agree more. The problem policy makers seem to be focused on is the long "stay" outliers, when the short stay outliers are at least as big a problem, if not more. The Wachterman, Marcantonio, et al paper in JAMA give us reason to be suspicious of some long stay cases, especially because there are more of these long-stay cases in for-profit than not-for-profit hospices. For example for LOS >=365 days, for profits had 6.9% of cases and not-for-profits had 2.8% of cases. The suspicion might be that the difference represents the inappropriate bias of for-profits to take advantage of the per diem -- in 4.1% of cases.

    I think the bigger news is that in the <7 day LOS category not-for-profits have 34.3 % of their cases and for-profits have 28.1% of cases. From my sense of the clinical QOL benefits of the hospice model, I don't think anyone can have much benefit in that length of time.

    In fact, while a hospice may not profit from a really short case (where they can't make up the up-front costs), I am suspicious that hospitals may be "profiting" by using hospice as a way to discharge a patient, reducing hospital costs, but in ways that leave no real potential clinical benefit to the patient.

    Anyway, looking at these data, I think the most urgent policy matter is to get the median LOS up. I think that the most simple minded intervention should be tried. Instead of requiring an MD to certify that someone will die within 6 months, lets try 9 or 10 months and see what happens. I'm not sure it will reduce the number of short stays, but I also don't see any reason to think that it will extend the number of very long stays.

    ReplyDelete
  4. @Christopher Langston. I hadn't thought of the idea that short stay hospice could be viewed as hospitals 'dumping' patients to hospice...under prospective payment hospitals get fixed amount based on diagnosis, so sending someone to hospice halfway through avg. hosp LOS could maximize profit for a hospital...it could also be what is best for the patient. Can't really sort that out with these types of data.

    I agree that more effort needed increase shorter/median LOS should be a priority. Finally, the hospice benefit generally needs to be modernized; the language of curative doesn't make so much sense (we would cure it if we could). Hard to let the best policy be front and center on EOL in the current political climate though. Concurrent care is the next step...

    ReplyDelete
  5. Just to chime in again, hospices offer different levels of service (they are not all the same--some will administer certain drugs, others wont, etc), nurses invariably screen at bedside and want documentation certifying eligibility from the referring doc, and it is possible the FP and NP centers are simply attracting different types of patients, albeit ones whose characteristics we are not adjusting for in the propensity scores.

    Thanks again,
    brad

    ReplyDelete
  6. @Brad F:
    you are correct that we didn't match or account for different practices across hospices. From standpoint of determining cost to Medicare, the different practices across hospices don't matter, at least directly, because Medicare pays per diem amount....then hospices have lots of discretion as you note. The definition of cost to Medicare is what Medicare paid. The different practices of hospice could have big effect on quality, which could affect lifespan which could indirectly effect cost. Different practices could also effect profitability.

    Also, one argument for large hospices is that they have more scope to do certain types of care that may help a patient but which is very expensive relative to the per diem.

    I will send you some of the protocols when we get closer for how we plan to measure quality, as I would be interested in getting your perspective and insight. thanks

    ReplyDelete
  7. Don- Adding quality of care to discussions of hospice payment reform is necessary, although it appears at least some significant hospice payment reform will move forward before additional quality data is available. It also appears that MedPAC recommendations target the entire industry (i.e., they do not target hospices based on incorporation status, particular states, etc.). So using currently available data, what hospice payment criteria do you believe would incentivize increasing hospice median LOS (addressing problems with short LOS), while decreasing WFA (addressing problems with long LOS)? Some thoughts include: strengthening the hospice aggregate cap definition, calculation, and repayment process; considering the percentage of hospice patients discharged alive; defining and surveying volunteer and bereavement requirements more carefully; more closely examining the location where hospice is provided (home, nursing facility, inpatient unit); and strengthening fraud and abuse measures. Your thoughts? Thanks-
    Cordt

    ReplyDelete
  8. Cordt:
    MEDPAC certainly wants a U shaped payment approach (higher per diem in first few and last few days). This make sense to me (the shape of the payment), though I am unsure of what the payment levels should be. I assume the rates will be based on cost finding studies. I would do that and up the scrutiny of long stay patients I guess. However, I am not sure any change in hospice payment is inevitable in the 112th Congress... For me, a move to concurrent care is a large priority, because thy lingo of stopping curative and starting hospice doesn't make much sense (if you could cure it you would).

    ReplyDelete
  9. I support the U shaped payment structure that MEDPAC is supporting for hospice reimbursement.
    - Zack with www.hospicebasics.com - All About Hospice Care

    ReplyDelete