An article in the June 27, 2011 raised questions about inappropriate use of hospice in the Medicare program, especially by for profit providers. I am at Myrtle Beach this week and trying to stay off grid, but I have gotten several calls and emails from people asking me about this (so have failed to stay fully off). Below is a post that is reproduced from my old blog back in February, 2011.
The news story in the NY Times is factually correct, but misses the main point. What we really need to know is the relationship between use of hospice, cost and quality of care. In many ways this is true for the entire health care system. We often focus on cost, or on quality, but you need to look at cost and quality at the same time to figure out if care is ‘worth it’ or not. The post below from February 2, 2011 provides a fuller context in which to discuss hospice and the overall health care system. This follow up post from the following day (that is not reproduced below) provides even more context.
Back to the beach….
Do For Profit Hospices Milk the System? That is essentially the background question asked by a paper out today in JAMA that looks at the association between the hospice agency type (for profit [FP], not for profit [NP]) and how they differ in terms of patients cared for and for how long. A related question that is framed by the paper is whether the payment policy used by Medicare to reimburse hospice care should be changed. Before discussing the paper, let me say that the paper by Wachterman et al., like most work in this area (including work I have done showing the hospice reduces Medicare expenditures) leaves out something very important: the quality of care that is provided for the money spent. The authors readily note this as a limitation.
The paper has two basic findings. FP hospices are more likely than are NP hospices to:
- treat patients who do not have Cancer. [17.2% in FP, v. 8.4% in NP had a diagnosis of Dementia]. Patients with Cancer represented 34.2% of FP patients v. 48.4% in NP.
- provide hospice for a longer period of time prior to death. The median length of use (half the patients stayed shorter, half longer) was 20 days for patients treated in FP v. 16 in NP. The 75% percentile showed even larger differences with 1 in 4 users staying longer than 88 days for FP v. 52 days for NP. And 6.9% of patients cared for by FP hospice providers used hospice for longer than 365 days v. 2.8% who were cared for by NP providers.
The paper further shows that patients with primary diagnoses other than Cancer received less intensive services while receiving hospice than did hospice patients with Cancer. This would make the FP hospice patient mix more profitable than the NP patient mix since the payment would be the same, while the amount spent providing care lower for FP providers (this is not inevitable, but was observed).
Hospice provides interdisciplinary care for persons who are dying and have decided to forego curative treatments in order to receive hospice based palliative care, which is care that focuses on treatment of physical symptoms, psycho-social concerns, planning and family concerns. A phyisican must certify that death in likely within 6 months for the initiation of the hospice benefit in Medicare (though patients can receive services for longer).
The Medicare program began covering hospice in 1983, which helped to legitimize the concept in the U.S. health care system. Currently, nearly half of the decedents in the Medicare program (which insured around 8 in 10 persons who died last year) choose to use hospice prior to death. Hospice was initially conceived of primarily to care for persons with Cancer, but has spread to other diseases, which have a less predictable time course to death. The spread of hospice beyond Cancer provides the opportunity to provide needed care to more patients, while also making more likely longer periods of hospice use prior to death. Most hospice care is provided in patient’s homes, but can also be provided in institutional settings.
Hospice is one of the few remaining types of care that is reimbursed on a per diem basis, which means that a fixed amount of money is provided to a hospice provider for each day a patient is enrolled. The incentive with such a payment approach (for the provider) is to provide more care, since the costs of hospice are higher in the first few days (intake period) and last few days prior to death, when physical symptoms tend to be acute. So, a longer length of hospice use means providers have more days with lower costs, while receiving the same per diem payment, thus maximizing their payment. Long periods of hospice use raise concerns about inappropriate, wasteful or perhaps even fraudulent use (the current Medicare hospice benefits stipulates 6 months as the presumptive period of eligibility, though you can receive such care for longer).
It is worth noting that very short periods of hospice use raise quality concerns. In the study noted above, 28.1% of persons using a FP hospice died less than a week after initiating such care, as compared to 34.3% of those care for by a NP hospice. In fairness, that could be taken as evidence of FP hospices outperforming NPs. Very short periods of hospice use lead to worries that patients are not able to realize the full benefits of what hospice could provide in the way of symptom relief and other benefits, and may be a signal of poor quality care. Colleagues of mine who are hospice and palliative care providers tell me that 4-6 weeks of hospice is likely needed for a patient to achieve the maximum benefits.
There has been an ongoing discussion and recommendation from MEDPAC to change the Medicare hospice payment to a so-called U-shaped approach in which higher per diem rates are paid for the first several and last several days of hospice care, acknowledging that the care needs and therefore costs of patients during those days are higher. And dropping the per diem costs for the in between periods when patients are stable, and costs of care are lower.
This general change in payment policy seems reasonable to me, and could lessen the incentive to over-provide, but there is something missing from this calculus.
What about quality?
In one sense, hospice has passed the market test, and there is research that shows that patients and families are satisfied with the care they receive in hospice. However, a good deal of this work is based on retrospective reports of families who basically say that hospice was there when my family needed them, when we were in a difficult situation. But, the question remains, what is the length of hospice use that maximizes benefits? Can an empirical basis for the gut of my colleagues that 4-6 weeks of hospice use maximizes benefits be obtained? How steep is the slope of the curve? How much extra benefit would be gained from moving a person from a 5 day to a 15 day length of use prior to death? Likewise, if a payment policy reduces the extreme long outlier cases, how bad will it be in quality terms if the median drops from 20 days to 15 days?
These questions cannot be answered with any certainty given the current evidence.
The figure above shows that the Medicare margin for hospice providers (avg. margin in 2001-06, figure is an approximation of a MEDPAC figure) increases with length of use, meaning simply that it is more profitable for hospice providers the longer a patient uses in hospice. This shows hospices losing money on the lowest 3 deciles of length of hospice use (say, less than 10 day stays). Margin rises steadily beyond that, showing that longer lengths of hospice use are more profitable for hospice providers. What remains unknown is the quality curve, which is hypothetical in the graph above. Knowing the shape of the quality by length of use decile relationship is really needed in order to make Medicare hospice payment policy with full information.
I recently received a grant along with an oncologist colleague from Duke Amy Abernethy, to essentially estimate the quality by length of use relationship shown in the figure above for both hospice care but also for non-hospice palliative care in patients covered by Medicare and other insurers. This is possible because of painstaking work (not done by me) to develop a cohort of patients receiving such care in a variety of settings, with quality being prospectively measured in patients covered by different types of insurance.
Amy and I have written that end of life is the fools gold of health reform, meaning that focusing on time to death will likely disappoint as a cost savings strategy because of how difficult it can be to make an accurate prognosis. However, a focus on value of care (does this extend life? does this improve quality of life? how much does this cost?) can likely improve quality while reducing costs. And the Medicare program and end of life situations is an important place to focus such questions since diminishing returns to expenditures designed to forestall death eventually set in given that everyone eventually dies, and 8 in 10 decedents were insured by Medicare last year.
To engage this discussion in an evidence based manner means we have got to not only look at cost or quality, but consider both together. This is true for hospice policy, and for health policy generally.