Brief Reports
JOURNAL OF PALLIATIVE MEDICINE Volume 10, Number 5, 2007 © Mary Ann Liebert, Inc. DOI: 10.1089/jpm.2007.0071
Differences in Hospice Care between Home and Institutional Settings DAVID G. STEVENSON, Ph.D., HAIDEN A. HUSKAMP, Ph.D., DAVID C. GRABOWSKI, Ph.D., and NANCY L. KEATING, M.D., M.P.H.
ABSTRACT Objective: To compare hospice care delivered at home with hospice care delivered in institutional settings, such as the nursing home. Data source: Secondary data from the 1994, 1996, 1998, and 2000 waves of the National Home and Hospice Care Survey, a nationally representative survey of home health and hospice care agencies in the United States and their current and discharged patients. Study design: We describe recipients, use of services, and length of enrollment for hospice received at home compared to hospice received in institutional settings. Comparisons of service use and lengths of enrollment are adjusted for age, gender, race/ethnicity, location, Charlson score, payer status, and linear time trends. Principal findings: Hospice use in the United States has grown considerably over the last decade, especially among individuals in institutional settings. Institutional hospice users were older than home hospice users; more likely to be female, unmarried, and dually eligible for Medicare and Medicaid; and more likely to have primary diagnoses other than cancer. Although institutional hospice users were more likely to have received certain types of services compared to recipients at home, they were much more likely to be enrolled for 1 week or less. Conclusions: These national data point to significant differences across hospice settings and a growing need to analyze their implications. Yet, these data also leave many questions about hospice use across settings unanswered, including whether agency costs differ in institutional compared to home settings. As policymakers seek to assess the quality and appropriateness of hospice utilization and the methods used for its payment, further empirical work is needed, including how the growing use of hospice outside the home affects options for reform.
INTRODUCTION
T
HE USE OF HOSPICE CARE in the United States has grown substantially in recent years and is currently one of the fastest growing service categories in the American health care system.1 The U.S. Centers for Medicare and Medicaid Services projects that Medicare hospice spending—which accounts for the majority of U.S. hospice spending—will increase at an average rate of 9% per year from 2004 to 2015, faster
than spending growth for hospitals, physicians, skilled nursing facilities and home health.2 As hospice use has expanded, the population of hospice users has grown more diverse, with a much larger proportion having noncancer diagnoses.2 Although the majority of hospice is still provided at home, use of hospice in other settings has grown dramatically in recent years.3,4 For example, On-Line Survey, Certification, and Reporting (OSCAR) data show that the number of current nursing home residents receiving hospice services in-
Department of Health Care Policy, Harvard Medical School, Boston, Massachusetts.
1040
1041
HOSPICE CARE ACROSS SETTINGS
creased from 13,000 in 1996 to almost 41,000 in 2004 (authors’ calculations using OSCAR data). Hospice services in the United States are provided by a variety of agency types, including freestanding agencies (61% of agencies) and those based in home health agencies (20%), hospitals (18%) and skilled nursing facilities (1%).5 Although precise data are not available, the large majority of hospice care is provided by freestanding and home health-based agencies, and these providers serve individuals across settings of care (e.g., contracting with nursing homes and hospitals). Medicare beneficiaries are eligible for the Medicare hospice benefit if their physician certifies that they have less than 6 months to live and if they agree to forgo treatment intended to cure their terminal illness. The Medicare hospice benefit defines four levels of care: routine home care (which can be received at home or in an institutional setting), 24-hour continuous home care, inpatient hospice care, and respite care. In 2002 and 2003, 93% of Medicare hospice days were paid at the routine home care rate.2 There is currently little information comparing hospice care delivered at home to hospice administered in other settings at the national level. Extant studies generally focus on nursing home hospice and have been somewhat limited in scope, for example focusing on a single hospice provider6 or parent company,7 or examining nursing home hospice use alone.8,9 These studies concur that hospice recipients in nursing homes are more likely to be older, female, unmarried, and have noncancer diagnoses than home-based hospice users. Data about service use and duration of enrollment have been less conclusive. Understanding better how hospice care differs across settings could have important implications for future decisions about hospice payment, including decisions about whether to adjust payment for settings of care or patient case mix.2 Using multiple waves of a national hospice survey administered across settings, we seek to address some of the limitations of the current literature on hospice use across settings and to highlight additional questions for study. We describe characteristics of individuals using hospice in the home versus institutional settings, and we compare use of services and duration of hospice enrollment for patients in these settings.
METHODS Data and patients We use the 1994, 1996, 1998, and 2000 waves of the National Home and Hospice Care Survey
(NHHCS), a nationally representative survey of home and hospice care agencies and their current and discharged patients conducted by the National Center for Health Statistics. The data were collected using a two-stage sampling process. In the first stage, agencies were randomly selected (by size) from 24 strata according to agency type (home health, hospice, or mixed), region (Northeast, Midwest, West or South), and location in a metropolitan statistical area. An interviewer contacted the designee for each sampled agency and collected general agency information. In the second stage, up to six current and six discharged patients were randomly chosen from each selected agency. Our analyses focus on the discharge cohort only. Patient-level data were obtained through interviews with the agency staff member most familiar with the patient’s care, along with a review of the patient records, if necessary. Hospice patients were not interviewed directly. We excluded all home health patients and focused on individuals who received hospice services either in an institutional setting or at home. The final sample included 7288 discharged patients.
Hospice setting For each patient, agency staff recorded patient residence as (1) private residence (home or apartment), (2) rented room/boarding house, (3) retirement home or apartment, (4) board and care/assisted living/residential care facility, (5) nursing home/hospital/other inpatient health facility, or (6) other. We categorized residents as living at “home” if their residence was listed as (1), and we categorized residents as being “institutionalized” if their residence was listed as (5). Importantly, the NHHCS does not distinguish across institutional settings. However, one could infer through analyses of other datasets that the vast majority of individuals receiving hospice in institutional settings are in nursing homes as opposed to hospitals or other inpatient settings. For example, 2000 OSCAR data show that almost 18,000 current nursing home residents were receiving hospice services, a number quite similar to the estimate of 18,547 institutional users from the 2000 NHHCS current patient file. Other researchers have reached similar conclusions using National Nursing Home Survey data,3 and precise data on the number of hospice recipients in freestanding and hospital-based hospice units are unavailable. As a result, our discussion of the results primarily centers on nursing home hospice use in comparison to home hospice use.
1042
Hospice services, assistance with activities of daily living, and length of enrollment For each patient, agency staff recorded services received from the hospice agency during the last 30 days or since admission (if enrolled less than 30 days). Services included physician and skilled nursing services; medications and durable medical equipment; personal care and homemaker services; various therapies; counseling and other social services; volunteer services; and respite care. Individuals might receive additional services unrelated to the terminal condition from institutions in which they reside that are not captured in the survey. Agency staff also reported the activities of daily living (ADLs) for which patients received some help from the agency. We summed the number of ADLs (bathing, dressing, eating, transferring, walking, and toileting). Length of enrollment was defined as the number of days from admission to death or discharge.
Patient characteristics Agency staff reported demographic characteristics, including gender, age, race, Hispanic ethnicity, region of residence, marital status, and insurance status (Medicare, Medicaid, both Medicare and Medicaid, private, other). They also reported a primary admission diagnosis and up to five additional diagnoses. We present information on primary diagnoses across five categories, including cancer, heart disease, neurologic/psychological conditions (predominantly Alzheimer’s and dementia), lung disease (predominantly COPD), and “other.” To characterize comorbid illness, we used this list of diagnoses to calculate a Charlson comorbidity score based on the presence of one or more of 18 chronic medical conditions.10
Analyses We first described hospice use overall and by setting during 1994 through 2000. Next, we used 2 and t tests to compare characteristics of institutionalized and home hospice patients on individual and agency characteristics. We then described use of specific hospice services by setting. We used logistic regressions to calculate unadjusted and adjusted odds ratios for each type of service used in the institutional versus home setting. We used a linear regression to compare (the natural log of) overall length of enrollment for discharged patients and logistic regressions to compare various durations of hospice enrollment (specifically enrollments of 3 days, 7 days, and 90 days). We also performed a robustness check for length of enrollment using a Cox regression frame-
STEVENSON ET AL.
work utilized by other analyses of hospice data.11,12 Multivariate models adjusted for demographic traits (age, gender, and race/ethnicity), geography (region and MSA), payer status, health status (Charlson score, difficulty seeing, and difficulty hearing), and length of enrollment. Given the complex survey design, the “svy” commands in STATA software (version 9.2, Stata Corp, College Station, TX) were used to incorporate the NHHCS weights to account for the unequal probability selection of patients and also correct the standard errors for clustering within agencies. Because strata were not included on the NHHCS public use file, we constructed this measure using the agency type, region and metropolitan statistical area variables.13
RESULTS The number of individuals receiving hospice care in the United States has grown substantially in recent years, from 328,026 discharged hospice users in 1994 to 621,133 in 2000 (Fig. 1). Most individuals received hospice services at home, but the number in institutional settings grew at a relatively rapid rate. Between 1994 and 2000, the number of institutionalized hospice users grew from 36,943 to 216,963, and the proportion of discharged hospice users who were in institutional settings grew from 11% to 35%. Individuals who received hospice care in institutional settings differed from home-based hospice recipients along demographic, health status, and provider dimensions (Table 1). Institutional users were significantly more likely to be older (76.6 versus 70.3 years, p 0.001), female (55% versus 47%, p 0.01), and unmarried (69% versus 45%, p 0.001). Charlson scores based on primary diagnoses were significantly lower for institutional users compared to individuals at home (3.98 vs. 5.27, p0.001), a difference stemming largely from the lower prevalence of cancer in the institutionalized population (49% versus 72% for home users, p 0 .001). Institutional users were more likely to have a neurologic condition (12% versus 4%, p 0.001) and lung disease (9% versus 5%, p 0.02). Although the proportion of individuals who relied solely on Medicare, Medicaid, or private funds for their hospice care did not differ across settings, the proportion of individuals who relied on both Medicare and Medicaid (i.e., dual eligibles) was substantially higher for institutional hospice users compared to home hospice users. Institutional users were also significantly more likely to receive hospice ser-
1043
HOSPICE CARE ACROSS SETTINGS
Discharged Hospice Recipients
700,000 600,000 500,000 400,000 300,000 200,000 100,000 0 1994
1996
1998
2000
Year All
Home
Institution
FIG. 1. Change in hospice use over time, by setting. Based on National Home and Hospice Care Survey Data from 1994, 1996, 1998, 2000.
vices from agencies that were for-profit (23% versus 15%, p 0.02). Service use varied across hospice settings for some service categories even after adjustment for patient
TABLE 1.
CHARACTERISTICS
OF
characteristics and length of enrollment (Table 2). A caveat to interpreting these variables is that service use measures whether the hospice agency reported delivering any category-specific services during the last 30
HOSPICE USERS
Individual Age (years) Female Married White Black Hispanic Primary diagnosis Cancer Heart disease Neuro/psych conditions Lung disease Other Charlson score (mean) Payer Medicare only Medicaid only Medicare and Medicaid Private only Other Agency traits For-profit Certified as a home health provider
AND
AGENCIES,
Institution
Home
p value
76.6 55.3% 31.5% 82.6% 7.4% 2.3%
70.3 47.4% 55.4% 80.3% 8.8% 4.2%
0.00 0.01 0.00 0.43 0.61 0.05
48.9% 14.8% 12.4% 8.7% 15.3% 3.98
71.8% 9.9% 3.9% 4.5% 9.8% 5.27
0.00 0.03 0.00 0.02 0.02 0.00
48.9% 7.0% 13.9% 7.5% 22.7%
49.8% 5.7% 4.2% 13.% 27.0%
0.82 0.42 0.00 0.02 0.18
23.0% 44.8%
15.2% 52.8%
0.02 0.14
BY
SETTING
Based on National Home and Hospice Care Survey data from 1994, 1996, 1998, and 2000. Based on total sample of 7288 (1665 in institution; 5622 at home).
1044
STEVENSON ET AL. TABLE 2.
Servicesb Physician services Skilled nursing Medications Diet/nutrition Durable medical equipment Personal care Homemaker Intravenous therapy Physical therapy Enterostomal therapy Respiratory therapy Counseling Psychological services Social services Respite care Volunteer services Spiritual care
Assistance received ADL assistancec Length of enrollment Enrollment (days)d
SERVICE USE
IN
HOSPICE,
BY
SETTING OR (adjusted)a
95% CI (adjusted)a
Institution
Home
OR
95% CI
0.40 0.92 0.62 0.21
0.22 0.91 0.51 0.10
2.42 1.10 1.58 2.37
1.56–3.77 0.65–1.84 1.15–2.17 1.35–4.19
2.55 0.91 1.60 2.37
1.68–3.87 0.54–1.54 1.16–2.20 1.44–3.90
0.45 0.61 0.11 0.09 0.03 0.02 0.13 0.33 0.12 0.74 0.03 0.32 0.49
0.45 0.61 0.16 0.05 0.04 0.03 0.13 0.36 0.10 0.69 0.06 0.30 0.45
0.99 0.97 0.66 2.01 0.91 0.56 1.02 0.88 1.19 1.30 0.46 1.13 1.19
0.74–1.33 0.70–1.36 0.42–1.04 1.10–3.64 0.39–2.11 0.22–1.43 0.58–1.79 0.60–1.28 0.76–1.87 0.89–1.90 0.19–1.11 0.79–1.61 0.85–1.68
1.02 0.90 0.71 2.33 0.75 0.55 1.15 0.90 1.21 1.47 0.50 1.12 1.23
0.76–1.36 0.65–1.25 0.43–1.18 1.21–4.46 0.29–1.94 0.24–1.28 0.66–2.00 0.61–1.33 0.73–1.99 1.01–2.15 0.19–1.30 0.78–1.62 0.88–1.72
Institution
Home
Coefficient
p value
Coefficient (adjusted)a
p value (adjusted)a
4.29
4.31
0.01
0.89
0.01
0.70
56.01
64.46
0.42
0.00
0.50
0.00
aAdjusted for age, gender, race/ethnicity, marital status, location, Charlson, length of enrollment, payer status, and year. Length of enrollment regression also adjusted for discharge status. bLogistic regression of whether any services were received in last 30 days or since admission (reported by hospice agency). cPoisson regression on the number of ADL categories for which assistance was received (i.e., not ADL limitations). dLinear regression of ln (length of enrollment). Cox regression produced similar results. Based on National Home and Hospice Care Survey data from 1994, 1996, 1998, and 2000. Based on total sample of 7,288 (1665 in institution; 5622 at home). CI, confidence interval; OR, odds ratio; ADL, activities of daily living.
days and does not measure the intensity of usage or account for services delivered by other providers. Institutional users were significantly more likely than home hospice users to have received several types of services from the hospice agency, including physician services (adjusted odds ratio [ORadj] 2.55; 95% confidence interval [CI]: 1.68–3.87), prescription medications (ORadj 1.60; 95% CI: 1.16–2.20), diet and nutritional services (ORadj 2.37; 95% CI: 1.44–3.90), intravenous therapy (ORadj 2.33; 95% CI: 1.21–4.46); and social services (ORadj 1.47; 95% CI: 1.01–2.15). Home hospice users were not significantly more likely to have received any categories of services compared to institutional hospice users. Among discharged patients, average length of enrollment differed by hospice setting (56 days for institutional users versus 64 days for home users). After adjusting for demographic traits, geography, discharge status, and health status, length of enrollment for institutional hospice users was significantly shorter than
for home hospice users (p 0.001), a difference robust to use of a Cox regression framework. In large part, this difference reflects the significantly higher proportion of very short periods of enrollment among institutionalized users (Fig. 2). Twenty-two percent of institutional hospice users had enrollments of 0–3 days (compared to 12% at home) and another 20% had enrollments of 4–7 days (compared to 10% at home). Viewed another way, hospice recipients in institutions were much more likely to have lengths of enrollment of 3 days or less (ORadj 1.70; 95% CI: 1.12–2.57) and also much more likely to have enrollments of 1 week or less compared to home hospice users (ORadj 2.26; 95% CI: 1.63–3.12). At the other end of the spectrum, there was no significant difference across sites in the proportion of enrollments longer than 90 days (OR 0.82; 95% CI: 0.55–1.21). Given the NHHCS’ lack of precision in characterizing institutional residence, we also conducted sensitivity analyses limiting the cohort to the 411 institu-
1045
HOSPICE CARE ACROSS SETTINGS 0.25
Proportion of Sample
0.20
0.15
0.10
0.05
0.00 0-3 days
4-7 days
8-14 days Home
15-30 days
31-60 days
61-90 days
91 days
Institutional
FIG. 2. Hospice length of enrollment, by setting. Based on National Home and Hospice Care Survey Data from 1994, 1996, 1998, 2000.
tional users referred to hospice by nursing homes (a subset of institutional hospice users). Results comparing individual and agency characteristics and length of enrollment across settings were quite similar. Service use differences were similar but no longer significant, likely resulting from limited power due to the small sample size.
DISCUSSION Hospice use has grown considerably in the United States over the last decade, especially among individuals in institutional settings. As noted above, our assessment and that of others3 is that the growth in institutional hospice has been driven primarily by the expansion of hospice use in the nursing home. Reflecting the composition of the nursing home population, institutional hospice users from hospices across the U.S. were older than home hospice users; more likely to be female, unmarried, and dually eligible for Medicare and Medicaid; and more likely to have primary diagnoses other than cancer. Although institutional hospice users were more likely to have received several types of services compared to hospice recipients at home, they were much more likely to be enrolled for one week or less. Our findings of considerable case-mix differences across settings are consistent with previous studies of
institutional hospice use.6–8,14 For example, the prevalence of cancer was much greater in the home setting, and the prevalence of heart disease, lung disease, and neurological conditions such as Alzheimer’s disease was higher among institutional users. In addition, institutional users have fewer comorbid illnesses. Our findings on length of hospice enrollment and service use across settings add to an existing literature that is somewhat conflicting. Previous analyses of prevalence data from the mid-1990s by the U.S. federal government raised concerns about excessively long enrollment periods among nursing home residents.15 More recent analyses, however, found similar8 or shorter7 enrollment distributions among hospice enrollees across settings. Using nationally representative data, we found that institutional hospice users are more likely to be enrolled for a shorter duration of time, with 42% receiving services for a week or less (versus 22% of home users). On service use across settings, some studies report that nursing home residents received fewer services and cost less than home-based users,16 and others report no significant difference or mixed results across settings (i.e., nursing home residents receive more of some services and less of others).7 Other recent data from a single hospice chain suggest that accounting for institutionalized status of hospice enrollees added little explanatory power to Medicare’s current per
1046
diem payment system for the number of hospice visits or costs of labor provided during those visits.14 Yet, another study using detailed cost data for one hospice (including data on nonlabor costs such as medications, palliative chemotherapy and radiation, and travel costs) found that total hospice costs per day were lower for nursing home residents compared to other hospice users.17 Our analyses showed that institutional hospice users were more likely to receive some types of services compared to enrollees at home. Importantly, though, we were unable to assess service intensity, which would be necessary to make assertions about relative resource use or expenditures. Important context for understanding service use by nursing home hospice recipients is the contracting approach for hospice delivery in this setting. Under typical contracts, a hospice agency is responsible for providing all palliative care related to an individual’s terminal illness. The nursing home continues to provide room and board and compensatory long-term care services, some of which may overlap with services provided by the hospice.16 Because the vast majority of nursing home hospice recipients enroll in hospice after their admission to the nursing home,9 these individuals would be receiving supportive care services and durable medical equipment from the nursing home prior to their hospice enrollment. This could result in greater use of some services but less of others, and it might also impact resource use at the beginning of a hospice stay, which has been found to be relatively resource intensive. Our findings of differences across settings underscore a growing need to analyze their implications for clinical care and policy. For example, the short enrollment periods we observed could impede the ability to deliver the full benefits of hospice care. The Medicare Payment Advisory Commission, which advises the U.S. Congress on issues affecting the Medicare program is also interested in whether differences across settings should influence payment for hospice.2,18 Currently, Medicare, the predominant payer of hospice services in the United States, does not adjust perdiem hospice payments across settings of care, nor does it adjust payments according to a patient’s casemix or the services or supplies that are actually provided by the hospice agency.2 Instead, the Medicare program pays hospice agencies per-diem rates across four categories of care: routine home care, continuous home care, general inpatient care, and inpatient respite care. The short enrollments we observed imply lower costs per-enrollment, but they could also imply higher
STEVENSON ET AL.
costs per-day. In particular, the cost and intensity of hospice care is generally greater on the first and last days of service use.2 Because of the higher portion of short enrollments, these high-cost days—which are not treated differently by Medicare payment policy— could have a disproportionate impact on overall hospice costs in the nursing home setting. As noted above, NHHCS data report only whether the hospice provided a particular type of service. If institutional hospice users receive a wider array of services at a lower level of intensity (e.g., because of their proximity to one another or because they receive other complementary services from the institution), NHHCS data would not allow one to capture these details. In addition, NHHCS data do not include complete data on costs of care or, for Medicare beneficiaries, on the payment level of hospice care received. Previous analyses have shown that routine care accounts for the vast majority of hospice care across settings.2,9 Ultimately, more research on hospice use across settings is needed. Our study provides a longitudinal look at this issue on a national scale; however, our analyses are limited in the extent to which they can describe service use, and they offer no insights on the relative cost of care across settings. Moreover, the last year of available NHHCS data is 2000. One impediment to further work in this area is the inability to obtain service use and cost data from most data sources on hospice care, including Medicare administrative data. Several questions important to future hospice policy thus remain unanswered, including whether agency costs are generally lower in nursing homes compared to home settings and whether different diagnoses affect the timing of hospice use and service costs among nursing home residents. If spending for hospice care continues to rise at its current pace, there will likely be increased pressure on U.S. policy-makers to assess the quality and appropriateness of hospice utilization and the methods used for its payment. This cannot be accomplished without further empirical insights, including how the growing use of hospice outside the home affects options for reform.
REFERENCES 1. Smith C, Cowan C, Heffler S, Catlin A: National health spending in 2004: Recent slowdown led by prescription drug spending. Health Aff (Millwood). 2006;25:186–196. 2. MedPAC: Hospice in Medicare: Recent Trends and a Review of the Issues. Report to Congress: Increasing the Value of Medicare. Washington, D.C.: MedPAC; 2006: 59–78.
HOSPICE CARE ACROSS SETTINGS 3. Han B, Remsburg RE, McAuley WJ, Keay TJ, Travis SS: National trends in adult hospice use: 1991–1992 to 1999–2000. Health Aff (Millwood) 2006;25:792–799. 4. Miller SC, Mor VN. The role of hospice care in the nursing home setting. J Palliat Med 2002;5:271–277. 5. Based on the 96% of hospice providers that are Medicarecertified. Data available at (Last accessed March 26, 2007). 6. Casarett DJ, Hirschman KB, Henry MR: Does hospice have a role in nursing home care at the end of life? J Am Geriatr Soc 2001;49:1493–1498. 7. Miller SC: Hospice care in nursing homes: Is site of care associated with visit volume? J Am Geriatr Soc. Aug 2004;52:1331–1336. 8. Miller SC, Mor V: The emergence of Medicare hospice care in US nursing homes. Palliat Med 2001;15:471–480. 9. Miller SC, Gozalo P, Mor V: Use of Medicare’s Hospice Benefit by Nursing Facility Residents. Washington, D.C.: U.S. DHHS Assistant Secretary for Planning and Evaluation; 2000. 10. Charlson ME, Pompei P, Ales KL, MacKenzie CR: A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 1987;40:373–383. 11. Christakis NA, Escarce JJ: Survival of Medicare patients after enrollment in hospice programs. N Engl J Med 1996;335:172–178. 12. Christakis NA, Iwashyna TJ: Impact of individual and market factors on the timing of initiation of hospice terminal care. Med Care 2000;38:528–541. 13. Carlson MD, Gallo WT, Bradley EH: Ownership status and patterns of care in hospice: Results from the National Home and Hospice Care Survey. Med Care 2004;42: 432–438.
1047 14. Nicosia N, Reardon E, Lorenz K, Lynn J, Buntin MB: The Medicare Hospice Payment System: A Preliminary Consideration of Potential Refinements. Washington, DC: RAND (under contract for MedPAC); 2006. 15. U.S. Office of the Inspector General. Hospice and Nursing Home Contractual Relationships. Washington, D.C.: U.S. Department of Health and Human Services; 1997. OEI-0595-00251. 16. U.S. Office of the Inspector General. Hospice Patients in Nursing Homes. Washington, D.C.: U.S. Department of Health and Human Services; 1997. OEI-05-95-00250. 17. Huskamp H, Newhouse J, Cararella J, Keating N. Variation in patient-level costs for one hospice: Lessons for the Medicare per-diem payment system? (unpublished manuscript) 2007. 18. MedPAC. Hospice in Medicare: Recent Trends and a Review of the Issues. Report to Congress: New Approaches in Medicare. Washington, D.C.: MedPAC; 2004, pp. 139–154.
Address reprint requests to: David G. Stevenson, Ph.D. Department of Health Care Policy Harvard Medical School 180 Longwood Avenue Boston, MA 02115 E-mail:
[email protected]
