Figure. Proportion of hospice enrollees who transferred from home to another location before death by intensity of care received.
Barclay JS, Kuchibhatla M, Tulsky JA, Johnson KS. Association of Hospice Patients' Income and Care Level With Place of Death. JAMA Intern Med. 2013;173(6):450–456. doi:10.1001/jamainternmed.2013.2773
Author Affiliations: Division of General Medicine, Geriatrics, and Palliative Care, Department of Medicine, University of Virginia, Charlottesville (Dr Barclay); and Departments of Medicine (Drs Barclay, Tulsky, and Johnson) and Biostatistics and Bioinformatics (Dr Kuchibhatla), Division of Geriatrics, Center for the Study of Aging and Human Development (Drs Kuchibhatla and Tulsky), and Center for Palliative Care (Drs Barclay, Tulsky, and Johnson), Duke University, and Geriatrics Research Education and Clinical Center (Dr Johnson) and Center for Health Services Research in Primary Care (Dr Tulsky), Veterans Affairs Medical Center, Durham, North Carolina.
Importance Terminally ill patients with lower incomes are less likely to die at home, even with hospice care.
Objectives To examine the relationship between income and transfer from home before death and the interaction between income and level of hospice care as a predictor of transfer from home in patients admitted to routine home hospice care.
Design We matched zip codes to US census tracts to generate median annual household incomes and divided the measure into $10 000 increments (≤$20 000 to >$50 000). We abstracted data from the central administrative and clinical database of a hospice care provider. We analyzed the relationship between income and transfer from home before death using logistic regression adjusted for demographics, diagnosis, region, and length of stay. Level of hospice care was examined as any continuous care vs none. Unlike routine care, which includes periodic visits by hospice, continuous care is a higher level of care used for short periods of crisis to keep a patient at home and includes hospice services in the home at least 8 hours in a 24-hour period.
Setting A for-profit hospice provider, VITAS Healthcare, operating 26 programs in 8 states.
Participants Hospice patients admitted to routine care in a private residence from January 1, 1999, through December 31, 2003.
Main Outcome Measure Transfer from hospice care in a private residence to hospice care in a site outside the home before death.
Results Of the 61 063 enrollees admitted to routine care in a private residence, 13 804 (22.61%) transferred from home to another location (ie, inpatient hospice unit or nursing home) with hospice care before death. Patients who transferred had a lower mean median household income ($42 585 vs $46 777; P < .001) and were less likely to have received any continuous care (49.38% vs 30.61%; P < .001). The median number of days of continuous care was 4. For patients who did not receive continuous care, the odds of transfer from home before death increased with decreasing median annual household incomes (odds ratio range, 1.26-1.76). For patients who received continuous care, income was not a predictor of transfer from home.
Conclusions and Relevance Patients with limited resources may be less likely to die at home, especially if they are not able to access needed support beyond what is available with routine hospice care.
Most Americans report wanting to die at home.1,2 However, despite these preferences and better outcomes for care at home compared with other settings, in 2007 only 30% of decedents younger than 65 years and 24% of decedents 65 years or older died at home.3- 6 Even when patients want to die at home, lack of caregiver support,7 lack of health care provider knowledge of preferences,1 and poor symptom control8 may result in transfer to subacute or acute care settings before death.9,10
Some patients face additional challenges in dying at home. For example, compared with wealthier patients, those with lower incomes are less likely to die at home7,8 owing to poorer access to health care, less knowledge of resources, less communication with providers about care preferences, lack of resources to assist with caregiving, and greater symptom burden at the end of life.3,11,12 In addition, those with lower incomes are less likely to enroll in hospice, which facilitates dying at home.13,14 In 2003, approximately 50% of hospice enrollees died at home compared with 25% in the general population.15 By providing an interdisciplinary team of health care professionals for symptom management, personal care, psychosocial and emotional support, and medications and equipment related to the terminal illness, hospice may help decrease some barriers to dying at home for those with limited resources.16,17
Of importance for indigent patients, the standard hospice benefit is defined for most patients by Medicare or Medicaid, and most hospices provide unreimbursed care for those without coverage. Private insurance plans generally provide similar benefits.18,19 Hospice staff are available 24 hours a day8 and, when needed, can provide continuous care in the home to treat symptoms not easily managed with routine hospice care. Routine hospice care consists primarily of periodic home visits by staff, whereas continuous care is a short-term intense period of care that includes the presence of hospice staff providing care for a minimum of 8 hours in a 24-hour period, with at least half the care provided by a nurse. Continuous care helps patients stay in their homes by providing the care they might otherwise seek in acute care settings.
Many studies have evaluated factors associated with the location of death20- 23 and with the use of health care services and the place of death among patients who leave hospice care.24- 26 However, the association of income and/or the intensity of care provided by hospice with transfer from home to another location before death among those continuing to receive hospice care remains largely unexplored. The purpose of this study was to examine the association between income and transfer from home to another location and how this association differs according to the intensity of care provided by hospice (any continuous care vs none) in a large cohort of patients who continued to receive hospice care until death. Understanding the association of income and the intensity of hospice care with transfer from home to another location may provide information about the type of services, beyond those currently available as part of the hospice program, that patients with lower incomes may need to die in the location of their preference.
Data were obtained from VITAS Healthcare, a for-profit hospice provider. During the study period, VITAS operated 26 hospice programs in 8 states (Florida, Illinois, Ohio, California, Texas, Wisconsin, New Jersey, and Pennsylvania). After approval from the Duke University Health System institutional review board and VITAS Healthcare, data were abstracted from the hospice provider's central administrative and clinical database.
We included hospice enrollees who were admitted to routine home hospice care, who lived in a private residence at the time of admission, and who died during the interim from January 1, 1999, through December 31, 2003. We excluded 1 patient missing location of death, 220 lacking median household income (owing to missing zip codes), 337 without caregiver information, and 2 without admission diagnoses. The final sample included 61 063 hospice enrollees.
The primary outcome was transfer from hospice care in a private residence to hospice care in a site outside the home before death. We defined a transfer as any enrollee with a location of death other than a private residence, including an assisted living facility, nursing home, hospital, or inpatient hospice setting. We dichotomized the primary outcome variable as no transfer (remained home until death) vs transfer (transferred to any location outside the home before death).
The predictor of interest was median annual household income. Because the database did not include the income of individual enrollees, we obtained median annual household income by matching enrollees' zip codes to US Census tract data from 2000. This method has been used in other research.27,28 We categorized median annual household income as follows: $0 to $20 000, greater than $20 000 to $30 000, greater than $30 000 to $40 000, greater than $40 000 to $50 000, and greater than $50 000. Because we focused on resource limitations, incomes above the US median household income were consolidated into a single category (>$50 000). Owing to limited subjects in zip codes with median annual incomes of $10 000 or less, we combined them to form a singe income group of $20 000 or less.
Because we wanted to determine whether the relationship between income and transfer from home to another location before death was different based on the level of hospice care provided, we created an interaction term for income × level of care. As described in the introduction, hospice provides routine care and continuous care in the home setting. We dichotomized the level-of-care variable into no continuous care (the enrollee did not receive continuous care at any time during hospice enrollment) vs any continuous care (the enrollee received continuous care at any time during hospice enrollment). The predictor for the multivariate model included an interaction term for each income category × any continuous care or each income category × no continuous care.
We selected covariates on the basis of availability in the database and the potential association with transfer from home to another location before death or receiving a higher level of care. The final model included sex, age (<65, 65 to <75, 75 to <85, or ≥85 years), race (African American, white, Hispanic, or other), marital status (married or not married), disease type (cancer or other), payment source (charity, Medicaid, Medicare, or other), enrollment in a health maintenance organization, relationship of the primary caregiver to the enrollee (spouse, child, other relative, nonrelative, or other), days in hospice care (0-7, 8-30, 31-180, or >180), and hospice program location by region (West, Midwest, Northeast, or South).
To compare patients who did not transfer from home before death with those who did, we used χ2 tests for categorical variables and Wilcoxon rank sum tests for continuous variables. We used logistic regression to evaluate predictors of transfer from home to another location, including all covariates as main effects and an interaction term between each category of income and any or no continuous care. We used the likelihood ratio test and the Wald test to assess the goodness of fit of the model. For all tests, differences were considered statistically significant at P < .05. All analyses were performed using commercially available software (SAS, version 9.1; SAS Institute, Inc).
The final sample included 61 063 patients admitted to routine care in a private residence. Of these, 13 804 (22.61%) transferred from their home before death. Among those who transferred, 7.93% went to an assisted living facility; 13.49%, to contract beds in area hospitals; 62.10%, to an inpatient hospice unit; and 16.53%, to a nursing home. The sample characteristics are in Table 1.
The median annual household income for the sample was $42 573, similar to the US median annual household income for 2000 of $42 148.29 Patients who transferred from home before death were more likely to be in the lower income categories (<$40 000) and, consistent with income category, had a lower mean median annual household income than those who did not transfer ($42 585 vs $46 777; P < .001). The absolute difference in probability of dying at home was 0.17 (P < .001) between those in the highest income group (0.83) compared with the lowest income group (0.66).
Those who stayed at home were more likely to have received continuous care (49.38% vs 30.61%; P < .001). Among those who received any continuous care, 83.82% were receiving it at the time of death; the mean duration of continuous care was 6 (median, 4) days. Those who transferred from home had a longer mean length of stay (70.3 vs 48.4 days; P < .001), were more likely to use Medicaid (7.66% vs 4.85%; P < .001) or to be African American (18.05% vs 12.54%; P < .001), and were less likely to be cared for by their spouse (41.55% vs 46.18%; P < .001).
Results of the multivariate model are in Table 2. The income × continuous care interaction was significant (P < .001). Among those who did not receive continuous care, the odds of transferring from home to another location before death increased as median annual household income decreased. Those with a median annual household income of $20 000 or less compared with more than $50 000 had almost twice the odds of transferring from home before death (odds ratio [OR], 1.76 [95% CI, 1.48-2.09]). The Figure shows how income and level of care was related to transfer from home. Among those receiving continuous care, no significant difference was found in rates of transfer from home across income levels; for all income levels, a smaller proportion of those receiving any continuous care (vs no continuous care) transferred from home before death (Table 2).
Other variables associated with a higher odds of transferring from home included a payment source of Medicaid vs Medicare (OR, 1.25 [95% CI, 1.12-1.39]), a primary caregiver who was a relative other than a child (1.19 [1.09-1.31]) or a nonrelative (1.64 [1.42-1.90]) vs a spouse, and longer hospice enrollment, with those enrolled more than 180 days (vs ≤7 days) having almost 4 times greater odds of transfer (3.92 [3.60-4.28]). In contrast, Hispanic vs white patients (OR, 0.83 [95% CI, 0.77-0.88]), those not enrolled in an HMO (0.78 [0.75-0.82]), women (0.90 [0.86-0.94]), and those living in the West, Midwest, or New England vs the South (OR range, 0.26-0.66) were less likely to transfer from home.
In this analysis of hospice enrollees admitted to routine care in a private residence, more than one-fifth did not die at home. Among those who did not receive continuous care, those with lower median annual household incomes were more likely to transfer from home to another location before death. However, among those who received any continuous care, rates of transfer from home were similar across income levels. These findings suggest that enrollees with limited socioeconomic resources may be less likely to die at home even with the support of routine home hospice care. Even short periods of more intense support, such as that provided by a higher level of care (continuous care), may help overcome socioeconomic resource disparities and allow patients to die at home when consistent with their preferences.
These findings are congruent with other research demonstrating that patients in lower-income groups, including those enrolled in hospice, are more likely to die in institutional settings.4,30 Although hospice provides substantial resources, most direct caregiving is provided by family and friends, with hospice support provided intermittently and for short periods. During the last year of life, more than two-thirds of patients require informal caregiving assistance.31 Many caregivers pay out-of-pocket for caregiving help, and many report needing help but being unable to afford it. In 1 estimate, care in the last 6 months of life totaled more than $14 000, with almost 20% of caregivers purchasing home health assistance on their own.32 Additional costs outside those required to manage the terminal illness may add to out-of-pocket expenses. Costs increase over time, and our study indicates that longer lengths of stay were associated with greater odds of transfer. Because Medicaid covers the cost of room and board in a nursing home, indigent patients who qualify for Medicaid may be more likely to seek care in a nursing facility as the burdens of care in the home setting increase. However, in this study, when controlling for Medicaid, lower income in the absence of continuous care remained an independent predictor of transfer from home, suggesting that resource limitations may play a significant role in patients' ability to die at home.
In addition to physical and monetary costs of caregiving, the emotional toll is also great. In general, patients of lower socioeconomic status have a lower quality of life,33 and their caregivers have worse health and are more likely to have depression.34- 37 The emotional difficulties become greater when symptoms are difficult for informal caregivers to manage,38 and patients with lower incomes are more likely to have uncontrolled symptoms at the end of life.39 Less availability of home care and fewer supportive services for caregivers are associated with a decreased probability of dying at home.9
More than 90% of hospice is delivered at the routine level of care, which includes medications and equipment and intermittent visits from nurses, home health aides, chaplains, and social workers. These services may not reduce the burden on informal caregivers enough to allow them to continue caring for their loved ones at home, especially those who lack financial resources to pay for additional care to supplement that provided by hospice.40,41 In contrast to routine hospice care, continuous care constitutes a short period of intense care in a patient's home to manage acute symptoms. In this study, continuous care was associated with an increased likelihood of dying at home regardless of income. These short periods of intense care may be most useful when death is near because 83.82% of those who received any continuous care were receiving it at the time of death; the median number of days of continuous care was 4. Families of patients dying at home report significant issues with symptom management and care burden; changes in hospice care to provide formal caregiving or improved symptom management through increased access to continuous care could ameliorate those issues.42 The additional costs would have to be offset by decreased use of acute care, such as emergency department visits.31
In addition to the costs and demands of caregiving, some groups may not desire in-home death or hospice care because of cultural beliefs and values. Prior research has demonstrated that low-income and African American patients are less likely to desire in-home death and to have favorable attitudes toward hospice care.43,44 Therefore, although dying at home is preferred by many patients, addressing spiritual, emotional, and physical symptoms and providing support at the location desired is of primary importance.
This study encompasses several limitations. First, data were supplied by a single for-profit hospice provider servicing 8 states; our findings may not be typical of other providers. For example, VITAS Healthcare hospice had a lower rate of death in the home setting (34% vs 40%), a shorter mean length of stay (41 vs 69 days), and a higher percentage of continuous care days (5.9% vs 1.2%) than the national average.18 Consistent with national trends, Medicare paid for most of the care, and rates of charity care were similar to or greater than the national average (2.5% vs 2.2%).19 Rates of in-home deaths may vary by hospice provider, and, as in this study, prior research has shown that in-home deaths vary by state and region.45 In addition, although guidelines for the use of continuous care are included in the Medicare hospice benefit, some variation likely exists across hospices in when and how this level of care is provided based on available resources and additional guidelines defined by individual providers.
Although data included patients admitted from 1999 to 2003, the structure of the Medicare hospice benefit, per diem payment structure, and criteria for eligibility for the different levels of care remain largely unchanged.46 Patients' share of costs remains minimal, and enrollment in hospice has increased substantially.47 Although the current population of hospice patients includes a greater percentage of noncancer diagnoses than in our study, this sample included a significant proportion of noncancer diagnoses, and the logistic regression showed a minimal effect of this variable. Almost half of all hospice patients currently receive care in a private residence, the defining population for this analysis.18 More than half of all hospices are for-profit organizations, and analyses of proprietary data from a national hospice formed part of the basis for proposed changes to the current hospice payment structure.47 Large hospices, whose increased resources are often required to provide continuous care, have become more prominent, making this sample representative of the current expansion of hospice services.
Although many patients admitted to hospice in a private residence likely wanted to die at home, we do not have information about the preferences of individual hospice enrollees or reasons for transfer from home. Large prospective studies are needed to capture the many factors that determine place of death. Last, individual incomes were not available in this database; therefore, we matched patient zip codes to census tract information to generate median annual household incomes. This method has been used in other research, but its application is limited, particularly in areas where large individual variability in income may exist.28,48- 50
Indigent patients have decreased access to health care, worse outcomes, and overall worse health-related quality of life.33 In this study, we found an association between low income and a decreased likelihood of remaining at home until death among hospice patients. Our findings suggest that hospices may need to provide additional resources to help indigent patients die at home beyond those currently available via routine hospice care. Such resources likely would include increased access to short periods of more intense care during times of crisis to manage uncontrolled physical and emotional symptoms and other sources of caregiver burden, such as those currently available via continuous care. Future research is needed to develop models of care that ensure access to high-quality end-of-life care for all patients regardless of income.13
Correspondence: Joshua S. Barclay, MD, Division of General Internal Medicine, Geriatrics, and Palliative Care, Department of Medicine, University of Virginia, PO Box 800909, Charlottesville, VA 22903.
Accepted for Publication: October 26, 2012.
Published Online: February 18, 2013. doi:10.1001/jamainternmed.2013.2773
Author Contributions: Drs Barclay and Johnson had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Barclay, Kuchibhatla, Tulsky, and Johnson. Acquisition of data: Johnson. Analysis and interpretation of data: Barclay, Kuchibhatla, Tulsky, and Johnson. Drafting of the manuscript: Barclay and Johnson. Critical revision of the manuscript for important intellectual content: Barclay, Kuchibhatla, Tulsky, and Johnson. Statistical analysis: Barclay and Kuchibhatla. Administrative, technical, and material support: Tulsky and Johnson. Study supervision: Tulsky and Johnson.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by Beeson Career Development Award K08AG028975 in Aging Research and Geriatric Research Education and Clinical Center (Veterans Affairs Medical Center, Durham).
Disclaimer: This study uses data from the VITAS Healthcare Corporation. It was not funded by VITAS and does not reflect the views of the VITAS Healthcare Corporation. VITAS had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Previous Presentations: This study was presented as a poster at the American Academy of Hospice and Palliative Medicine Annual Assembly 2009; March 25-28, 2009; Austin, Texas. An abstract of this study has been published in the Journal of Pain and Symptom Management (2009;37:535).