Figure 1. Flowchart of Health and Retirement Study decedents. *Inpatient hospice, assisted-living facility, rest home, retirement home, or senior care home. SNF indicates skilled nursing facility. Reported percentages incorporate survey weights; percentages may not sum to 100.0%.
Figure 2. Adjusted prevalence of skilled nursing facility (SNF) admission in the last 6 months of life by age group. Prevalence of SNF admission in the last 6 months of life was calculated with adjustment for groups of age at death and year of death. Reported values incorporate survey weights to account for the complex survey design.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Aragon K, Covinsky K, Miao Y, Boscardin WJ, Flint L, Smith AK. Use of the Medicare Posthospitalization Skilled Nursing Benefit in the Last 6 Months of Life. Arch Intern Med. 2012;172(20):1573–1579. doi:10.1001/archinternmed.2012.4451
Author Affiliations: Divisions of Palliative Care (Dr Aragon) and Geriatrics (Drs Covinsky, Boscardin, Flint, and Smith), Department of Medicine, and Division of Biostatistics, Department of Epidemiology and Biostatistics (Dr Boscardin), University of California, and San Francisco Veterans Affairs Medical Center (Drs Covinsky, Boscardin, Flint, and Smith and Ms Miao), San Francisco.
Background In the last 6 months of life, many older adults will experience a hospitalization, followed by a transfer to a skilled nursing facility (SNF) for additional care. We sought to examine patterns of Medicare posthospitalization SNF use in the last 6 months of life.
Methods We used data from the Health and Retirement Study, a longitudinal survey of older adults, linked to Medicare claims (January 1994 through December 2007). We determined the number of individuals 65 years or older at death who had used the SNF benefit in the last 6 months of life. We report demographic, social, and clinical correlates of SNF use. We examined the relationship between place of death and hospice use for those residing in nursing homes and the community before the last 6 months of life.
Results The mean age at death among 5163 individuals was 82.8 years; 54.5% of the cohort were female, and 23.2% had resided in a nursing home. In total, 30.5% had used the SNF benefit in the last 6 months of life, and 9.2% had died while enrolled in the SNF benefit. The use of the SNF benefit was greater among patients who were 85 years or older, had at least a high school education, did not have cancer, resided in a nursing home, used home health services, and were expected to die soon (P < .01 for all). Of community dwellers who had used the SNF benefit, 42.5% died in a nursing home, 10.7% died at home, 38.8% died in the hospital, and 8.0% died elsewhere. In contrast, of community dwellers who did not use the SNF benefit, 5.3% died in a nursing home, 40.6% died at home, 44.3% died in the hospital, and 9.8% died elsewhere.
Conclusions Almost one-third of older adults receive care in a SNF in the last 6 months of life under the Medicare posthospitalization benefit, and 1 in 11 elders will die while enrolled in the SNF benefit. Palliative care services should be incorporated into SNF-level care.
The period leading up to death is a vulnerable time for patients and for families, with many symptoms that are difficult to witness and manage. This often leads to hospitalization and, for some, discharge to a skilled nursing facility (SNF).
Skilled nursing facility use has exploded in recent decades. As hospital policy shifted to favor early discharge from the hospital in the 1980s and 1990s, a large proportion of nursing homes became certified to provide posthospitalization skilled nursing care as a bridge between the hospital and the home.1,2 Medicare beneficiaries are entitled to up to 100 days of SNF care if they have been hospitalized for 3 or more days and have a skilled need at discharge, such as rehabilitation, intravenous medications, and wound care. Medicare covers 100% of the cost for the first 20 days; after 20 days, if approved, the patient has a copayment of $144.50 a day.3 Most enroll in SNF care for rehabilitation or life-prolonging care, but experience suggests that some dying patients are discharged to a SNF for end-of-life care.
Hospice services are often the only formal end-of-life care available in nursing homes.4 If a patient is in a nursing home under the Medicare SNF benefit, Medicare regulations prohibit concurrent enrollment in the hospice benefit for the same diagnosis.5 Rarely, a patient may receive SNF care while receiving hospice services if the SNF care is for a diagnosis unrelated to the hospice diagnosis.6 In addition, switching a patient from Medicare coverage under the SNF benefit to the hospice benefit has financial implications for the patient and for the nursing home. Medicare reimburses nursing homes at a higher rate for skilled services. Unlike the SNF benefit, the hospice benefit does not pay for room and board. Patients who transition to the hospice benefit must pay for room and board out of pocket or by enrollment in Medicaid, for which many patients do not qualify.
Research has focused on the end-of-life experience in hospitals, but little is known about SNF use near the end of life. If such use is common near the end of life, there may be clinical implications for patients in a SNF; these may include requiring greater engagement in advance care planning and in goals of care and setting up reasonable expectations of what SNF care may be able to provide a person near the end of life. This may mean that SNF care needs to be reevaluated to incorporate a greater focus on end-of-life care.
Because little is known about SNF use at the end of life, we used data from the Health and Retirement Study (HRS), a nationally representative longitudinal survey of older adults, linked to Medicare claims data to examine the use of the Medicare posthospitalization SNF benefit at the end of life. The objective of this study was to evaluate these data to look at the prevalence of use of the Medicare SNF benefit in the last 6 months of life, as well as the predictors of use and the care following use.
The HRS is a survey conducted by the Institute for Social Research at the University of Michigan, Ann Arbor, and is sponsored by the National Institute on Aging. The goal is to study changes in health and wealth as people age. Since 1992, it has collected information on a representative sample of Americans 50 years or older; additional participants are added every 6 years. Participants are interviewed every 2 years. After a participant dies, his or her next of kin provides information about the end-of-life period in an exit interview. Data are collected about financial status, physical health and functioning, and place of death. More information about HRS sampling, data collection procedures, and measures can be found on the HRS website (http://hrsonline.isr.umich.edu/).
We looked at 6721 HRS respondents 65 years or older who had died between January 1994 and December 2007, and who had Medicare data for the last 12 months of life. We excluded 1149 participants who had been enrolled in Medicare managed care because claims data are incomplete for these individuals. For an additional 409 individuals, Medicare data were incomplete. The study sample consisted of 5163 HRS decedents enrolled in fee-for-service Medicare. The institutional review board at the University of California, San Francisco, approved the study.
Using Medicare claims data, we determined the proportion of HRS decedents who had used the SNF benefit in the last 6 months of life. Admitting diagnosis to the SNF was determined using diagnosis related group codes. We used Medicare claims to examine the use of the hospice benefit while enrolled in the SNF benefit, the duration of use of the SNF benefit, the time between hospital discharge and SNF benefit use, and the proportion of decedents who had died during the use of the SNF benefit.
We then looked for factors that might be associated with SNF benefit use at the end of life based on prior research in end-of-life care and clinical experience.7-10 Demographic factors included age, sex, and race/ethnicity. Social factors included marital status, education level, household net worth, geographic region at death, and any activities of daily living (ADL) dependence before the last 6 months of life (ADLs include eating, toileting, dressing, bathing, transferring, and walking across a room), as well as nursing home residence before the last 6 months of life and, if community dwelling, the use of home health services before the last 6 months of life. For clinical factors, we examined 7 health conditions (cancer, hypertension, diabetes mellitus, lung disease, heart disease, stroke, and cognitive impairment) and whether the patient was expected to die soon as reported by the next of kin of the decedents during the exit interview.
From 5163 HRS decedents, we determined the percentage who had used the SNF benefit in the last 6 months of life. We report age-adjusted prevalence of use of SNF benefit over time for those residing in a nursing home and for those living in the community before SNF enrollment.
Participants had to be residing in the community at the time of enrollment in the HRS, which may have led to lower SNF use in the first several years of the study. Therefore, we report the mean use of SNF from 1994 to 1996, 1997 to 1999, 2000 to 2002, 2003 to 2005, and 2006 to 2007. We stratified the use over time between those living in a nursing home and those living in the community before SNF use. We further divided the total number of SNF benefit users into 3 age groups (65-74, 75-84, and ≥85 years) to look for differences in the use by age over time (an interaction effect).
To assess the association between demographic, social, and clinical factors and SNF benefit use at the end of life, we used a log-binomial regression analysis and adjusted for age, sex, race/ethnicity, marital status, educational level, household net worth, geographic region at death, history of ADL dependence, and the presence or absence of chronic conditions to estimate relative risk. To look at clinical care differences between those who used the SNF benefit in the last 6 months of life and those who did not, we examined the place of death and hospice use at the time of death. All reported analyses were weighted for the differential probability of selection and account for the complex design of the HRS. Reported percentages may not sum to 100.0% because of the use of survey weights. Statistical analyses were performed using commercially available software (STATA 11; StataCorp LP; and SAS 9.2; SAS Institute, Inc).
The final study sample consisted of 5163 HRS decedents enrolled in fee-for-service Medicare during the last 12 months of life. The mean age at death was 82.8 years, while 41.3% were 85 years or older at the time of death. Among the cohort, 45.5% were male, 86.6% were of non-Hispanic white race/ethnicity, 41.7% were married or partnered, 45.8% had less than a high school education, and 23.2% resided in a nursing home before the last 6 months of life. Characteristics of decedents who did and did not use the SNF benefit are summarized in Table 1.
Of 5163 HRS decedents, 30.5% had used the SNF benefit in the last 6 months of life. Most SNF stays (87.5%) represented direct transfers from the hospital to the facility, and 98.0% of SNF stays were within 30 days of the hospital discharge. In total, 58.2% were enrolled in an SNF for 1 to 20 days, and 41.8% stayed for longer than 20 days. Only 0.5% of all the decedents had used the SNF benefit and hospice benefit concurrently, and 1.5% had enrolled in hospice on the day they were discharged from the SNF benefit.
Death during the use of the SNF benefit was common. In total, 9.2% of HRS decedents died while enrolled in the SNF benefit: 3.2% died within 1 week, and 7.2% died within 1 month of admission.
We then explored the place of death and hospice use for decedents who had resided in a nursing home before the last 6 months of life. These variables were similar between nursing home residents who had used the SNF benefit and those who had not (Figure 1).
We then compared decedents who had been residing in the community before the last 6 months of life. As shown in Figure 1, death in a nursing home was more common for community dwellers who had used the SNF benefit in the last 6 months of life, occurring among 42.5% vs among 5.3% of community-dwelling non-SNF users. Of community dwellers who had used the SNF benefit and died in a nursing home, only 19.3% enrolled in hospice, whereas 44.5% of community dwellers who had not used the SNF benefit and died in a nursing home used the hospice benefit. In total, 53.5% of community dwellers who had used the SNF benefit and died in a nursing home were enrolled in the SNF benefit at the time of death.
The use of the SNF benefit was highest in those 85 years or older for each year of the study, and no interaction effect with age was found (P > .19) (Figure 2). The initial rise in rates of SNF benefit use were driven by community dwellers. From 1994 to 1996 through 1997 to 1999, the mean age-adjusted prevalence of use of the SNF benefit at the end of life among community dwellers increased from 20.3% to 30.6% (P < .002). However, after this initial rise, the use of the SNF benefit remained between 29.0% and 32.3% from 2000 to 2007. No significant change from 1994 to 2007 was observed among nursing home residents who had used the SNF benefit. Table 2 gives the top 10 Medicare Provider Analysis Review File diagnosis related group admission diagnoses to a SNF. After adjustment for demographic factors, social factors, chronic conditions, and history of ADL dependence, we found greater use of the SNF benefit among older adults (≥85 years), those with at least a high school education, those who did not have cancer, and those who resided in a nursing home (Table 3). Greater use of the SNF benefit was observed among those expected to die and among individuals residing in the community, those who had used home health services before the last 6 months of life.
In this nationally representative study of older individuals, almost one-third of individuals 65 years or older had received SNF-level care in the last 6 months of life under the Medicare benefit. Strikingly, 1 in 11 elders died while enrolled in the SNF benefit. Individuals who had resided in a nursing home before the last 6 months of life died in similar locations and had similar rates of hospice use whether they used the SNF benefit or not. However, those who lived in the community and then went on to use the SNF benefit differed from those who did not use the SNF benefit. Almost half of community dwellers who had used the SNF benefit subsequently died in a nursing home, and more than half were enrolled in the SNF benefit at the time of death; less than one-quarter were enrolled in hospice at the time of death. By comparison, less than 10% of community dwellers who did not use the SNF benefit died in a nursing home, but twice as many (41.7%) were enrolled in hospice. Rates of SNF benefit use increased among community-dwelling individuals until 1999 and then leveled out, likely due to minimal growth in the number of SNF beds since that time.11
Our finding that Medicare decedents commonly used SNF care at the end of life suggests a need to better understand who is using the SNF benefit and whether they are receiving care that matches their goals. After hospitalization, elderly patients often experience a functional decline in 1 or more ADLs.12,13 When planning for discharge, the primary focus for these individuals is often to provide skilled nursing or rehabilitative services that will allow a patient to return to the prior level of health and functioning, an important aspect of caring for older adults. Individuals who receive SNF care are likely seen as having more potential for recovery. What may not be recognized is that the decline during a hospitalization may reflect that the patient's overall health condition will continue to deteriorate, despite adequate rehabilitation. In fact, the needs that necessitate SNF use are the same indicators of an end-of-life trajectory seen in frail elders.14 Honest and frank discussions about goals of care not only in the hospital but once they are admitted to a SNF may allow an earlier introduction to palliative care.
The SNF benefit users had greater disability and required more home health and nursing home care. Studies14,15 have shown a noticeable, although variable, decline in function leading up to death. Patients transferring to SNF-level care often have high care needs and are medically complex.11 At discharge, patients' goals may be inconsistent with hospice care, or health care providers may not recognize that patients are on an end-of-life path. Incorporating a palliative care focus into SNF-level care may allow earlier recognition of when hospice referral is appropriate. While rehabilitation is an important aspect of elder care, other issues near the end of life (eg, symptom management or discussions about goals of care) may not be addressed when a patient is enrolled in the SNF benefit.
In our study, 1 in 11 elders died while enrolled in the SNF benefit, most during the first 30 days of admission. The expectation of patients' imminent death by their next of kin was predictive of SNF benefit use, and for community dwellers, more than half who died in a nursing home had been enrolled in the SNF benefit at the time of death. This suggests that patients are being discharged from hospitals to nursing homes under the Medicare SNF benefit for end-of-life care. While it is possible to be enrolled in the hospice and SNF benefits concurrently if the services are billed for separate conditions (eg, SNF for rehabilitation following a fall and hospice for cancer), only 0.5% of decedents were enrolled in both programs simultaneously in our study. Besides functional decline, the end of life is associated with many other issues, such as pain, shortness of breath, spiritual distress, caregiver burnout, and grief.16 Although our study provides no evidence about the quality of care received, the literature suggests that nursing home residents have little access to palliative care outside of hospice services.17-19 Hospice care in the nursing home can be variable, although patients and families report better symptom control and satisfaction with care when nursing home residents receive hospice care.20,21 Hospice services provide symptom management, bereavement assistance, social work, and dedicated chaplaincy.17 The SNF benefit does not provide the depth of end-of-life services that hospice offers.
Unfortunately, financial issues may contribute to why patients near the end of life are using the SNF benefit and not the hospice benefit. Elderly patients living in the community who are becoming more symptomatic or are functionally declining may be admitted to the hospital because families cannot manage them at home. Our study suggests that older, more clinically complex patients are using the SNF benefit near the end of life. They may require care that cannot be provided in the home even with hospice services. Unless an individual already has Medicaid coverage, hospice may not be an option because the cost of room and board in a nursing home is too burdensome. Families often face an uncomfortable choice: either they pay for room and board out of pocket to have access to hospice services, or they continue under the Medicare SNF benefit, relying on nursing home services for palliative and end-of-life services.
Likewise, nursing homes receive higher reimbursement for patients enrolled in the SNF benefit compared with long-term care payments for patients via Medicaid. Given the Medicaid bed-holding policies for nursing home residents, these institutions have a financial incentive to hospitalize their patients.22,23 After hospitalization, nursing home residents may return under the Medicare SNF benefit, which will reimburse the SNF more than if they came back for custodial care paid for by Medicaid.24 The SNF care may be appropriate for some of these patients, but for others who cannot receive SNF and hospice care concurrently for the same condition, this practice may lead to care that is not consistent with a patient's goals or to poorer-quality end-of-life care.
Limitations of our study are noted. Although we report the admission diagnosis, we were unable to determine the true circumstances that led each individual to use the SNF benefit after hospitalization. For example, a patient with heart failure who is near the end of life could have been transferred to a nursing home under the Medicare SNF benefit for the management of intractable shortness of breath that cannot be managed at home. Hospice may have been considered. However, we found that rehabilitation was a top diagnosis related group for admission to the SNF, suggesting that some patients are not transferred to a SNF for end-of-life care. Because we studied decedents, we were unable to make comparisons with participants who use the SNF benefit and go on to live longer lives. However, the incorporation of a palliative focus alongside rehabilitation would likely be a benefit for patients whether or not they were clearly on an end-of-life trajectory. Finally, we were unable to describe SNF use for the 17.1% of patients with Medicare managed care who were excluded from the study.
In conclusion, Medicare SNF benefit use is high at the end of life, highlighting the need to incorporate quality palliative care services in nursing homes. The hospice benefit is the primary way in which palliative care services are provided in nursing homes. A growing focus is on the development of palliative care in nursing homes alongside the current goals of functional improvement. Perhaps having Medicare pay concurrently for postacute SNF care and hospice services for the same condition could allow earlier incorporation of palliative care for these medically complex patients.
Correspondence: Alexander K. Smith, MD, MS, MPH, Division of Geriatrics, Department of Medicine, University of California, San Francisco, 4150 Clement St, 181G, San Francisco, CA 94121 (firstname.lastname@example.org).
Accepted for Publication: June 29, 2012.
Published Online: October 1, 2012. doi:10.1001/archinternmed.2012.4451
Author Contributions:Study concept and design: Aragon, Covinsky, Flint, and Smith. Acquisition of data: Aragon and Smith. Analysis and interpretation of data: Aragon, Covinsky, Miao, Boscardin, and Smith. Drafting of the manuscript: Aragon and Smith. Critical revision of the manuscript for important intellectual content: Aragon, Covinsky, Miao, Boscardin, Flint, and Smith. Statistical analysis: Aragon, Covinsky, Miao, Boscardin, and Smith. Obtained funding: Smith. Study supervision: Boscardin and Smith.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported in part by grants K24AG029812 from the National Institute on Aging (Dr Covinsky) and UL1 RR024131 from the National Center for Research Resources University of California, San Francisco–Clinical and Translational Science Institute (Dr Smith).