Palliative Care for End-Stage Dementia: The Interface Between Practice, Policy, and Profits
In medicine, we tend to think about end-of-life issues as ethical or legal concerns. What would the patient want? What would be in his or her best interest? If he or she cannot decide, who is empowered to make these decisions? We believe that if such issues are addressed and the appropriate legal papers are signed, then the clinical care will correspond to these decisions. But in actual practice this is usually not the case. If there is an available intervention, it will be applied.
Perhaps the best evidence for this comes from a randomized controlled study of palliative care in 100 patients with advanced dementia admitted to a tertiary care academic center.1 Patients with families who were agreeable to the intervention of a palliative care team were randomized to this intervention or control. Over the ensuing three years, despite the families’ willingness and the efforts of a palliative nurse specialist and a geriatrician, there were almost no differences in the care that was actually provided: no difference in the number of rehospitalizations; length of hospitalization; or the proportion who died in hospital. Perhaps more distressing is that there was no difference in the proportion who received systemic antibiotics (75% overall), intravenous (IV) therapy (74% overall), long-term feeding tubes (69% overall), or new feeding tubes (44% overall). One of the few differences was that those in the intervention group were more likely to have palliative care plans (11 vs 2), but in nine of those 11 cases, this was not adopted until discharge, and thus never acted on.
What makes this finding even more perplexing is that on quiet reflection, most physicians who care for elderly persons with dementia and most of the families of such people want the very opposite. This is evidenced by a national survey of over 800 physician members and 1000 nonphysician members of the Gerontological Society of America, as well as 500 families of persons with dementia recruited through the Alzheimer’s Association.2 Persons in end-stage dementia were described in the following terms:
• Need complete assistance with eating and toileting
• Can no longer recognize themselves or their loved ones
• Cannot do any of the things that used to make them happy
• Cannot talk anymore
• Have medical complications of dementia, such as falls, urinary incontinence, or pneumonia
Participants were then asked which of the following five levels of treatment they thought would be most appropriate:
1. Do everything possible to keep the person alive. Treat every illness such as hypertension or diabetes. Use feeding tube, cardiac resuscitations, and respirators
2. Same as level 1, but no cardiac resuscitation
3. Same as level 1, but no cardiac resuscitation or respirator
4. Same as level 3, but also stop treating acute conditions except pain
5. Same as level 4, continue treatment of chronic conditions and pain, but no tube feeding
Over 60% of physicians and 70% of family members chose level 5. So, why is the care that is actually provided so different from what the great majority of clinicians and family members believe is appropriate? Probably because clinicians do what they are paid to do. Our system pays for interventions and does not pay for withholding them. Almost everyone in the system benefits from this arrangement, except possibly the patient, the family, and the taxpayer. Case in point: a patient with end-stage dementia spikes a fever. If he or she is at home, imagine the family trying to arrange an emergency home visit from a physician! They have no choice but to call an ambulance. If this occurs at a nursing home, it is easiest for the staff to arrange a transfer to the Emergency Room. There is no financial disincentive. If the nursing home holds the bed, Medicaid will pay for it even with the patient gone. In the meantime, the patient can be admitted, worked up, and placed on IV antibiotics, thus generating Medicare billings for the hospital and its staff. Such an approach is not only driven by payments but also by a regulatory and legal environment that assumes that the highest level of care is the best care. Nursing homes are hesitant to take patients with advanced dementia without feeding tubes because the facility gets penalized if a patient loses significant weight prior to death. This is a quality-of-care measure.
If you have a system that pays for interventions and not for withholding them, you need an alternative system that pays even when you withhold them. The Medicare Hospice Program is just such a system; it pays a flat amount per day for providing palliative care but does not pay for acute interventions. In 1986, Volicer,3 a geriatrician at a Boston VA facility, described an inpatient hospice program for end-stage dementia. There clearly was a need for this approach. In the survey of gerontologists and family members of patients with dementia mentioned previously, authors also assessed attitudes toward hospice care. The following description was provided to participants:
Hospice care has had as its main purpose to assist people who are terminally ill in the easing of their death and not the prolonging of their lives. This is done with limited medical procedure and a staff of professionals such as doctors, nurses, social workers, pastors, etc who provide support for the dying person and his/her relatives. Hospice care can be provided in special areas of hospitals, nursing homes and the home of the patient.2
After receiving this description, 91% of professional and 90% of family members viewed hospice care as appropriate for end-stage dementia. Yet, although 87% of the sample were aware of existing hospice programs for cancer, only 13% were aware of hospice programs for dementia. This low rate of awareness suggested the unavailability of hospice services for patients with dementia. This was confirmed by a national survey of hospices that was subsequently carried out.4 It found that less than 1% of patients in hospice nationwide had a primary diagnosis of dementia, and only 21% of hospices treated any patients with dementia. Regulations require that a physician attest that a patient has 6 months or less to live before being eligible for the Medicare Hospice Benefit. Eighty percent of hospices listed this as the major obstacle to enrolling persons with dementia.
Physicians show a markedly conservative bias when pressed to predict death. For example, in patients with severe cardiac illness, death most often comes within 1 day of their having been estimated by their physicians as having a 50% chance of living 6 months.5 Possibly because of the reluctance to make this determination, the average cancer patient lives less than 3 weeks (not months) after being enrolled in hospice. Since this bias affects end-stage cancer, imagine how it must influence physicians’ willingness to make such an attestation in a more chronic and insidious condition such as Alzheimer’s disease. Until 1995, it was not even listed as a cause of death by the National Center for Health Statistics. We now know it as the seventh most frequent cause of death, ahead of influenza and pneumonia.6
Since predicting death in 6 months seemed to be such an obstacle to provision of hospice care, a series of small prospective studies was undertaken to determine prediction criteria.7-9 It was shown that when a set of criteria developed by the National Hospice and Palliative Care Organization in 1996 could be fully met, the mean survival time was 3.2 months. The criteria were the following:
• Functional Assessment Staging Scale (FAST) Stage 7c10
• Presence of a medical complication such as aspiration pneumonia or difficulty swallowing
These criteria were subsequently adopted by Medicare regulators and produced both anticipated and unanticipated consequences. First (as might have been predicted), there have been questions about whether these are the right criteria. For example, in a study that examined longevity in 165 patients in hospice, it was shown that although 77% of those who met these criteria died within 6 months, 80% who had been enrolled in hospice without meeting these guidelines also died within 6 months.11 These authors believed that factors related to alimentation might be most predictive, with anorexia and dry mouth being associated with death within 6 months. Other research groups have turned to data from the Minimum Data Set (the 12-item Cognitive Performance Scale) and have been able to show a much higher positive predictive value than the FAST.12 Despite continued research to define appropriate criteria, patients with dementia are currently still more likely than others to live past 6 months, leading researchers in this area to agree that it would make sense for Medicare to adopt more flexible guidelines than for clinicians to apply ever more rigid criteria.
Second (as anticipated), with wider appreciation of the need for hospice care for dementia and the adoption of Medicare guidelines for this group, the proportion of patients with dementia has expanded dramatically. According to a 1995 survey, less than 1% of patients in hospice had a primary diagnosis of dementia,4 while 10 years later in 2005, another survey showed this to be the primary diagnosis in over 11% of patients in hospice.13 This dramatic increase in numbers, and therefore costs, led to a push-back by Medicare with increased scrutiny of hospices that enroll large numbers of patients with dementia, denial of claims, and allegations of fraud.
What was not anticipated is that the same fiscal forces that previously impeded the adoption of hospice care for dementia are currently distorting its expansion. One factor pointing toward this outcome was detected in the 1995 survey when it was noted that 42% of for-profit but only 22% of not-for-profit and 15% of publicly-run hospices were providing care to persons with dementia. Caring for patients with dementia, because of their longer stays, can be very profitable. This has been one factor fueling a dramatic expansion of the for-profit hospices industry.14 Initially, one of the factors that made hospice attractive to Medicare regulators was that it was perceived as providing cost savings, or at least being cost-neutral. Early studies supported that view. More recent studies continue to demonstrate that hospice is cost-neutral to Medicare for cancer, but it is 20-42% more expensive for a patient with dementia to receive hospice care than the alternative.15
Another distortion of hospice care for dementia due to our payment system is its extensive use in patients dying in nursing homes.16 From 1992 to 2000, the percentage of patients in hospice residing in nursing homes has increased from 11% to 36%.15 The Medicare Hospice Benefit pays the same per diem whether a patient is at home or in a nursing home. Since Medicaid does not reduce its nursing home per diem when a patient is enrolled in hospice, there are incentives to do just that. We therefore have the somewhat bizarre situation where nursing homes—the facilities in which most persons with dementia will actually be living at time of death—are not expected to be able to provide appropriate end-of-life care and need to bring in hospice to make this possible. A 1997 survey of nursing homes showed that the only hospice-like bereavement service that they routinely provided was sending a condolence card to the family of the deceased.17 It seemed obvious at that time that nursing homes should be expected to do a better job at such care. But this has not occurred. Nursing homes see themselves as “skilled nursing facilities,” not places where people die. Furthermore, now that both hospice and nursing homes are frequently paid to treat the same dying patient, changing nursing home regulations will probably involve fighting both the nursing home and the hospice lobby.
So there you have it: practice, policy, profits, and palliative care in end-stage dementia.
The author reports no relevant financial relationships.
Dr. Luchins is Chief Mental Health Research, Jesse Brown VAMC, Chicago, IL, and Visiting Associate Professor of Clinical Psychiatry, University of Illinois at Chicago.