Would you refer this patient to hospice? An evaluation of tools for determining life expectancy in end-stage dementia

Research output: Contribution to journalArticlepeer-review


There will be approximately 5.7 million people with Alzheimers dementia in the United States by 2020, growing to 13.2 million by 2050. Because hospice enrollment under Medicare for patients with end-stage dementia hinges on the ability to prognosticate, it is important that clinicians have accurate, validated prognostic tools. Unfortunately, the available published tools are valid for specific populations, and therefore have limited generalizability. Application of published prognostic tools to our nursing home patient produced probabilities of surviving less than 6 months ranging from 0% to 80%. (Table 3). This wide variability demonstrates the need for better prognostic tools to aid clinicians caring for the almost 50% of institutionalized patients with dementia. When evaluated as a group, these tools demonstrate a correlation between life expectancy of 6 months or less and poor functional status, poor oral intake or nutritional status, and a bed-bound lifestyle with ADL dependence. The ideal tool should incorporate these features. While the Medicare hospice admission guidelines for patients with dementia currently incorporates some of these features, it is a poor predictor of prognosis due to dependence on the ordinal FAST staging, which is itself a poor prognostic tool. Furthermore, hospice dementia guidelines are better at predicting who will not die in 6 months than who will. Data predicting life expectancy after an acute illness are useful, but may be difficult to use if no such illness occurs. Life tables reinforce the relationship between age, poor functional status, disease comorbidities, and limited life span. However, they may be too broad for prognosticating life span less than 6 months. Two tools currently incorporate the negative prognostic indicators for inclusion in the ideal instrument: poor functional status, poor oral intake, and a bed-bound status with ADL dependence. The first is the MRI, validated for patients newly admitted to longterm care. The MRI should be studied in dementia patients who have resided in long term care for months or years, since hospice care in the nursing home is shown to reduced symptom burden. The second tool, the PPS, could be useful for community dwelling or institutionalized dementia patients. It is easy to complete and may have greater sensitivity than the MRI. Neither tool requires physician participation. Both tools would benefit from greater measurement precision in the domains of poor functional status and poor oral intake, since patients with advanced dementia often live for years bed-bound with complete ADL dependence, including inability to self-feed. This case demonstrates a common challenge for clinicians: the need to use available tools for determining prognosis when their patient differs from the population in which the tool was developed. While we did not review every prognostic measure in the literature we chose those with application to patients with end-stage dementia, or those in long-term care. Considered together (Table 3), these may provide a prognostic prospective that is useful for clinicians. In summary, applying available prognostic tools to a nursing home resident with end-stage dementia demonstrates a wide range of prognoses, demonstrating the limitations of current methods for determining prognosis in this patient population. No tool has been developed for use specifically with long-term nursing home residents, of whom almost half have dementia. With the numbers of patients with Alzheimers disease expected to increase dramatically, the importance of studies on prognosis in end-stage dementia cannot be overvalued.

Original languageEnglish (US)
Pages (from-to)1410-1419
Number of pages10
JournalJournal of palliative medicine
Issue number6
StatePublished - Dec 1 2007

ASJC Scopus subject areas

  • Nursing(all)
  • Anesthesiology and Pain Medicine

Fingerprint Dive into the research topics of 'Would you refer this patient to hospice? An evaluation of tools for determining life expectancy in end-stage dementia'. Together they form a unique fingerprint.

Cite this