Palliative Medicine 2006; 20: 81
/86

Shelter-based palliative care for the homeless terminally ill

Tiina Podymow Medical Director, Inner City Health Project, University of Ottawa, Department of Medicine,
Ottawa Hospital, Ottawa, Ontario, Jeffrey Turnbull Medical Director, Inner City Health Project, Chairman of
Medicine, Ottawa Hospital, University of Ottawa and Doug Coyle Clinical Epidemiology Program of the Ottawa
Health Research Institute and the Department of Medicine, University of Ottawa, Ottawa, Ontario

Background: The homeless have high rates of mortality, but live in environments not
conducive to terminal care. Traditional palliative care hospitals may be reluctant to accept
such patients, due to behavior or lifestyle concerns. The Ottawa Inner City Health Project
(OICHP) is a pilot study to improve health care delivery to homeless adults. This is a
retrospective analysis of a cohort of terminally ill homeless individuals and the effectiveness
of shelter-based palliative care. As proof of principle, a cost comparison was performed.
Methods: 28 consecutive homeless terminally ill patients were admitted and died at a
shelter-based palliative care hospice. Demographics, diagnoses at admission and course
were recorded. Burden of illness was assessed by medical and psychiatric diagnoses,
addictions, Karnofsky scale and symptom management. An expert panel was convened to
identify alternate care locations. Using standard costing scales, direct versus alternate care
costs were compared. Results: 28 patients had a mean age 49 years; average length of
stay 120 days. Diagnoses: liver disease 43%, HIV/AIDS 25%, malignancy 25% and other
8%. Addiction to drugs or alcohol and mental illness in 82% of patients. Karnofsky
performance score mean 409/16.8. Pain management with continuous opiates in 71%. The
majority reunited with family. Compared to alternate care locations, the hospice projected
$1.39 million savings for the patients described. Conclusion: The homeless terminally ill
have a heavy burden of disease including physical illness, psychiatric conditions and
addictions. Shelter-based palliative care can provide effective end-of-life care to terminally
ill homeless individuals at potentially substantial cost savings. Palliative Medicine 2006;
20: 81
/86

Key words: cost; harm reduction; homeless; hospice; Karnofsky; palliative care; shelter

Introduction found dead in public places, others are found in

Homeless persons have high rates of early mortality from
AIDS, cancer and hepatic disease.1
3 Mean age at death
is reported to be 34
/47 years, with age-adjusted death
rates 2.3
/4 times higher than the general population in
North American and European cities.1,2,4,5 Despite
diagnoses at death with diseases amenable to palliative
treatment, the homeless often live in environments not
conducive to terminal care. The majority of homeless
people have co-existing diagnoses of substance abuse or
mental illness6
9 and behavior or lifestyle concerns may
preclude treatment of these patients in a hospice envir-
onment. Even with terminal illness, some remain in
shelters due to its familiar environment, or to continue
drug, tobacco or alcohol use. With inappropriate care
and housing, a large number of homeless people are
Address for correspondence: Dr. Jeffrey Turnbull, The Ottawa
Hospital, 501 Smyth Road, Room LM12, Ottawa Ontario,
Canada K1H 8L6. Tel: 613-737-8755. Fax: 613-737-8851.
E-mail: turnbull@ottawahospital.on.ca
Dr. Tiina Podymow, E-mail: tpodymow@uottawa.ca
residential dwellings, at home, arrive dead to the emer-
gency department, or die shortly afterward3,5 Others
require hospitalization at end-of-life,3,10,11 and the
frequency of in-hospital deaths is reported to range
from 34 to 59% in homeless cohorts.2,3,5 Hospitalization
costs associated with homelessness due to increased
length of stay are known to be high9 though the problem
remains as to where to treat these individuals in the
context of ongoing substance abuse. Despite the growing
numbers of homeless, those terminally ill are not
represented in any clinical literature.
The Ottawa Inner City Health Project (OICHP) was a
pilot project designed to provide comprehensive health
care services to homeless adults. This project utilizes a
harm reduction policy to reduce the adverse conse-
quences from substance use without requiring abstinence.
We hypothesized that effective shelter-based palliative
care could be provided to terminally ill homeless
individuals at substantial cost savings. This is a retro-
spective description and analysis of 28 homeless people
who died having received palliative care at a shelter-based

# 2006 Edward Arnold (Publishers) Ltd 10.1191/0269216306pm1103oa

82 T Podymow et al.

hospice, and a cost analysis was performed as proof of admission, pain management, medication compliance,
principle. self-care assistance, religious counsel, family contacts and
location of patient death was obtained, retrospectively
after the patient’s death using a form completed by the
Methods registered nurse that cared for the patient, and the
patient’s hospice chart and electronic medical record
were reviewed. The outcomes of pain and symptom
The Hospice is a 15-bed shelter-based palliative care pilot
management were recorded as documented in the chart.
program created in June 2001 as part of the OICHP to
A Karnofsky performance score13 was determined at
house and treat terminally ill homeless individuals.
admission, to classify functional impairment. Descriptive
Referral of patients was made through the hospital,
statistics using the mean and standard deviation were
community physicians and shelter staff. Criteria for
used for continuous outcomes.
admission included chronic homelessness, having no
As proof of principle, a cost estimation was performed
natural caregivers, lacking financial resources to provide
by comparing actual care received to alternate probable
for care, and diagnosis with a life-threatening illness.
care. To determine alternate likely care settings, a five-
Patients were housed at the shelter in a separate area
member panel consisting of a hospital-based social
designated for the Hospice program, and were provided
worker, a registered nurse, a palliative care physician
with shelter beds and meals. The program employed a
and two internists were convened. The panel was non-
registered nurse full-time seven days per week, and a
blinded; the patients were known to some or all of the
registered practical nurse to work in the evening and
members. For each patient, the total number of palliative
overnight shift five days per week. The program also
days from admission to death was calculated. With
employed a client care worker 24 hours per day, seven
attention to social conditions, drug and alcohol addic-
days per week, whose intended role was as a substitute
tions, medical and psychiatric diagnoses, comorbidities,
family member to supervise the patients and to help with
course of illness, character and behavior during the
activities of daily living, application for social benefits,
course of palliation, a consensus was reached regarding
and to attend medical appointments. One of two project
the probable alternate location for care for each patient.
physicians supervised medical care for the patients
Alternate locations included: general shelter, ambulance-
24 hours per day, seven days per week and patients
delivered emergency visits, acute care hospital, palliative
were seen weekly in-shelter on rounds. Intercurrent
care hospital, supportive housing with or without visit
illnesses were treated by same. Palliative care and pain
nursing support, and domiciliary hostel (supervised
and symptom management consults were completed
housing). The panel also considered the likelihood that
on-site by a specialist physician collaborating with the
the patient may have died on the street or shelter with no
project. Hospital care appointments were coordinated
pain and symptom management. Actual program costs
with the hospice to allow continuity of care, and patients
were direct per diem costs of the OICHP Hospice, which
were transported and accompanied to appointments.
included housing and food, regular nursing, client care
Physical and occupational therapy, chiropody, and nutri-
worker salary, medications, transportation, physician
tional support were accommodated. Psychiatric illnesses
care, extra shift nursing, oxygen therapy, emergency
were monitored and treated by project physicians and a
department, hospital and ambulance costs. Alternate
psychiatric nurse practitioner. Harm reduction was
location costs were calculated based on standard per
achieved by providing clean needles, safe syringe dis-
diem unit health care costs obtained from the Hospital
posal, a smoking area outside the shelter, and dispensing
Management Information System.
standard 14 g drinks of alcohol on-demand to alcohol-
addicted patients. Record-keeping was performed using a
secure, on-line medical record system developed by the
OICHP.12 Results
Twenty-eight consecutive homeless patients were
admitted to the Hospice, received terminal care and Twenty-eight consecutive patients were cared for and
died between July 2001 and August 2003, and all were died in the Hospice program for 3
/523 days, mean 1209/
included for retrospective analysis with approval from the 140 days. The hospice functioned at full capacity; the
Ottawa Hospital research ethics board. At the program 28 patients described were only those who died, e.g., 28/
entry a consent and confidentiality statement was read to 15 /43 admissions in the period described.
each subject, and written consent to chart access was The characteristics of the patients are listed in Table 1.
obtained. Demographic data, medical and psychiatric The mean age at death was 49.1 years, with 25 males, two
diagnoses along with drug and alcohol use were extracted females and one transgender patient. The majority was
from the physician and nurse’s patient intake history and Caucasian with one Aboriginal and two African-Amer-
medical transfer notes. The main symptoms during ican patients. Housing prior to hospice admission was a
 Shelter-based palliative care for the homeless terminally ill 83

Table 1 Patient characteristics shelter in the majority, with none receiving palliative
services elsewhere. Aside from those admitted to the
Patient characteristics N /28 %
Hospice from the hospital, none of the patients were
Age receiving palliative services elsewhere. The diagnoses at
Range 32
/74
Mean9/standard deviation 499/10.5 admission and most responsible for death were alcoholic
Sex cirrhosis, malignancy and HIV. Other active problems
Male 25 89 contributing to morbidity but not directly responsible for
Female 2 7
Transgender 1 4 death included ascites, chronic obstructive lung disease,
Race skin ulcers, hepatitis B and hepatitis C. The mean
Caucasian 25 89 Karnofsky performance score at admission was 409/17,
African American 2 7
Aboriginal 1 4 which indicates the inability to care for self, with
Prior housing disability requiring special care and assistance equivalent
Shelter 14 50 to that provided in institutional or hospital care.13 Most
Supportive housing/rooming house 9 32
Street 2 7 patients (86%) suffered symptoms consistent with a
Hospital 2 7 diagnosable mental illness, including depression, schizo-
Jail 1 4
Number of days admission to death
phrenia, and anxiety. Patients were addicted to alcohol in
Range 3
/523 50%, tobacco in 96%, intravenous drugs in 21% and any
Mean (days) 120 (9/140) street drug in 46%, most commonly marijuana.
0
/7 4
8
/30 5 The clinical course and care of the patients is presented
31
/90 8 in Table 2. The majority of patients initially experienced
91
/180 4
/180 7 pain, and in most cases this required ongoing adminis-
Karnofsky performance scale score at admission tration of long-acting or intravenous opiates with pain
80
/100 0 0
60
/79 4 14
ultimately controlled in all but one patient. Other
30
/59 14 50 common symptoms treated during the course of pallia-
0
/29 10 36
Mean9/standard deviation 409/16.8
tion included nausea, confusion, stupor or coma and
Most responsible diagnosis at death
shortness of breath. A palliative care consult, very useful
Alcoholic cirrhosis 12 43 in successfully optimizing pain and symptom manage-
Cancer 7 25
Lung 3
ment, was obtained for nearly half of the patients.
Prostate 1 Patients were prescribed 2
/37 medications, mean 159/9,
Anal 1
Skin 1 and compliance defined as taking medications as pre-
Pancreas 1 scribed /80% of the time was reported in most.
HIV 7 25
Chronic obstructive lung disease 1 4 The patients required considerable assistance for self-
Peripheral vascular disease 1 4 care during the course of palliation, requiring extensive
Secondary diagnoses assistance or being totally dependent for eating, toileting,
Hepatitis C 18 64.3
Skin ulcers 12 42.9 transfers and ambulation.
Ascites 12 42.9 A harm reduction approach was taken for actively
Hepatitis B 11 39.3
Pneumonia 8 28.6 addicted patients, and hospice patients were exempt from
Encephalopathy 6 21.4 the usual shelter policy of barring those using drugs or
UTI 4 14.3
Deep vein thrombosis 3 10.7 alcohol. Alcohol addictions were successfully managed
Diabetes 2 7.1 by medically prescribed alcohol in 32%, which allowed
HIV 2 7.1
Malignancy 2 7.1 consumption of modest amounts of alcohol provided in-
Stroke 1 3.6 shelter by hospice staff, to prevent seeking, binging and
Other 3 10.7
behavior harmful to self. A further 32% continued to use
Psychiatric disease
Depression 12 43 intravenous drugs with clean needle exchange. Once pain
Schizophrenia 11 39 and symptoms were controlled, illicit intravenous and
Anxiety 5 18
None 4 14 street drug use was observed to significantly diminish.
Addiction Religious counsel was received by most during their
Tobacco 27 96 stay. The hospice received permission from the patients to
Alcohol 14 50
IV Drug 6 21 contact family in many; ultimately most received visits
Any street drugs 13 46 from family members, many of whom had infrequent or
More than one street drug 6 21
Benzodiazepines 2 7 no contact for many years.
Cocaine 5 18 End-of-life issues were discussed with each patient and
Demerol 1 4
Marijuana 7 25 most requested or agreed to ‘do-not-resuscitate’ orders,
Morphine 4 14 whereby emergency ambulance and hospital transfer
No addiction 5 18
were specifically declined by the patient in advance, and
84 T Podymow et al.
end-of-life scenarios were discussed with an emphasis
on the provision of comfort measures. Some patients
requested transfer to hospital for potentially reversible
conditions, and ultimately 82% died at the hospice, and
18% at the hospital.
Reasons that may have precluded transfer of the
patient to a regular hospital or palliative care hospice
included patient preference, active drug or alcohol use
precluding institutional care, and behavioral reasons
(e.g., poor impulse control, behavior surrounding drug
or alcohol use). The panel considered that without
palliative in-shelter hospice care, 68% of the subjects
discussed might not have sought care nor received
services and died homeless with no pain and symptom
management.
Average costs per patient stay were calculated for the
actual cost of the Hospice program and these were
compared to costs of care at alternate probable locations
(Table 3). Unit health care costs and inpatient per diems
as obtained from the Hospital Management Information
System are reported in Canadian dollars. The average
health care cost per admission to the Hospice program
was $15 0009/17 600. This was substantially less than the
estimated costs of traditional care for the same patient
sample based on the judgment of the expert panel,
$64 6009/76 800, P B/0.001. The major difference in costs
was due to the considerable difference in the bed-day
costs associated with traditional palliative care and
tertiary care hospitals ($684 and $633, respectively)
compared to the Hospice program ($125). Hospice
program costs included housing, food, nursing, client
care worker, medications and medical supplies, and
physician costs. Compared with probable alternate
care locations, the Hospice program saved an estimated
$1.39 million for the patients described.
Discussion
We report on 28 patients who received palliative care
and died in the Hospice program. Patients were homeless
with an extremely high burden of illness complicated
by co-morbidities, mental illness and addictions. Most
were admitted from the general shelter. The mean
Karnofsky score alone at admission indicated a level of
physical disability requiring institutional or hospital
care.13 The cohort described was typical of the homeless
population as it is known in the literature, and diagnoses,
addictions and age at death were consistent with those
previously described.1
3,5,7,9 Palliative care provided at
the hospice was consistent with that of the World Health
Organization definition14 and patients received pain and
symptom management, support and spiritual care, with
many dying in the hospice surrounded by family and
friends.
Adherence to treatment in this population is often
difficult, resulting in symptom intensification and the
need for crisis services. It is likely that medication and
pain control, supervised and administered by the hospice
staff, prevented the need for emergency care. Many of the
patients would have been excluded from traditional
palliative care because of behavior or lifestyle, however,
the harm reduction approach did not appear to increase
substance use.
The balance between treating pain and not abetting
addiction required vigilance in this program. As much
as possible, co-analgesics were administered e.g., non-
steroidal anti-inflammatory drugs or tricyclic antidepres-
sants in conjunction with opioids in order to minimize
opioid use. In cases where opioid dose escalation was
required to adequately treat pain, rotation to methadone
was often made. It was noted that some patients with
co-existing addiction and anxiety sought illicit drugs to
placate their anxiety, and patient histories were taken
carefully in this regard. Addressing anxiety with benzo-
diazepine or antidepressant treatment for example, was
useful in eliminating illicit drug use in many cases. Non
medicinal treatment was an adjunct in most; a hot tub
was installed in the shelter, chiropractic and physical
therapy were used, as were traditional healing practices
e.g., ‘smudging’ for Aboriginal patients.
The program had strict rules and policies forbidding
injection in the shelter, consistent with the shelter’s own
rules forbidding this practice. All medications were kept
under lock and key, and administered solely by program
staff, with directly observed therapy for oral opioid
medications. Drug delivery was by patch or by pump in
those requiring high opioid doses. There were attempts
by some patients to illegally sell or use the narcotics
provided by the program, for example attempts to use the
fentanyl patches for injection or to tamper with the
narcotic pump. The most severe consequence for these
behaviors was to be barred from the hospice and its
services for 24 hours, with the subsequent signing of a
contract; this practice was effective in eliminating illegal
behavior. Interestingly, the patients themselves as well as
other shelter clients contributed to the integrity of the
program by reporting suspected illicit behavior directly to
program staff.
The average time from admission to death was four
months, which is long but perhaps not surprising
considering these people were most often accepted from
shelters for palliation, with no alternate caregivers or
home, and fairly high burden of illness. There was
consideration to discharge long-term patients from
the hospice, but care needs were too heavy to allow
admission from the hospice to long-term convalescence,
and life expectancy was too long to permit admission to a
palliative care hospital.
 Shelter-based palliative care for the homeless terminally ill 85

Table 2 Patient care have been preferable to compare two such groups in
parallel as a randomized controlled trial, logistical,
Patient care N /28 %
population and financial constraints made such metho-
Main symptoms dology unfeasible. Similarly, information on health care
Pain 25 89.2 costs of homeless patients who died with similar illnesses
Nausea 19 67.9
Confusion/stupor/coma 17 60.7 and addiction profiles was categorically not available for
Shortness of breath 14 50.0 direct comparison as matched controls. The cost estimate
Headache 5 17.9 was performed with the patients acting as their own
Oxygen required controls, as judged by an expert panel. This allowed an
Yes 17 60.7 accurate cost estimate of the hospice program, however,
Continuous 5 17.9 projections regarding the alternate site of care by a
Pain management non-blinded panel may have over or under estimated the
Continuous 20 71.4 amount or level of institutional care. In those actually
Per required need (PRN) 8 28.6
discharged from hospital to the hospice, there was
Type of pain management no alternate site for their care, and time until death
PRN morphine/codeine 16 57.1 would have been frequently expected to be in-hospital.
Fentanyl patch 13 46.4
Regular morphine 7 25.0
For those referred from a shelter, rooming house,
Morphine infusion 4 14.3 living outdoors or from jail to the hospice, a best estimate
Methadone 1 3.6 was made retrospectively, gauged by the patient’s medical
Palliative care consult 12 57.1
and psychiatric disease burden and subsequent actual
level of required care. It is possible, if not probable that
Number of medications some may have died in public places or in-shelter, and
Range 2
/37
Mean9/standard deviation 159/9 though unacceptable as a standard of care, this would
have cost less.
Type of self care assistance Previous comparisons of medical expenditures in the
Incontinent bladder/diapers 27 96.4
Washing assist 27 96.4 last year of life observe a 13
/20% decrease in a hospice
Transfer assist 26 92.9 setting compared to hospital, and hospital services
Diabetic/low salt diet/nutritional 23 82.1 accounted for more than 50% of expenditures.15 Cost
supplements
Incontinent bowel 22 78.6 analyses of death from individual diseases are uncom-
Feeding assist 21 75.0 mon; in one, the cost estimate of non-homeless indivi-
Dressing 17 60.7 duals dying of untreated liver disease in 1997 United
Specialist/hospital appointments 12 42.9
Foley catheter 6 21.4 States dollars (USD) for each of 7 patients was $110 576
for a mean length 20.19/14.3 days.16 Those non-homeless
Religious counsel 17 60.7 dying of non-small cell lung cancer surviving less than
Family visits one year spent 30.5 days in hospital and had an average
Daily 8 28.6 cost of USD $47 280 in 1998.17 Perhaps most relevant,
Weekly 8 25.0 in a study of hospitalization costs associated with
Monthly 4 14.3
Never 8 25.0 homelessness, the homeless, though not terminally ill in
this study, were found to stay 36% longer than other
Location of death patients, with a cost for an additional 4.1 days of USD
Mission hospice 23 82.0
Hospital 5 18.0 $2400
/4000 in 1997.9
There were elements immeasurable concerning the
Reasons precluding transfer to palliative care hospital
Patient preference 23 82.1
patients’ trust, sense of security, gratitude and well
Drug/alcohol addictions 16 57.1 being during their hospice stay. The patients in this study
Behavioral 13 46.4 and the involved families expressed gratitude for their
treatment as well as appreciation of the hospice staff.
Many confirmed that they would not have wanted
This pilot program was intended as an experimental tertiary or palliative hospitalization, preferring to die in
program for terminally ill homeless people with no the shelter system to which they had become familiar.
natural caregivers as a descriptive study of the homeless Continuity of care among the homeless has previously
at end-of-life and a feasibility study for shelter-based been found to be extremely difficult. Shelter operators,
terminal care. Cost estimates were performed as proof of already capable in caring for the homeless were integrated
principle, to compliment existing cost studies in the into a shelter-based medical model of care to address
homeless literature, which describe greatly increased previously unmet needs, permitting palliative care deliv-
health care costs for this population.9 Although it would ery which was needed and potentially cost saving. Those
86 T Podymow et al.

Table 3 Costs

Location Standard cost Hospice program
/ average Alternate probable care
/ P-value
per diem* ($) total cost per patient average total cost
per patient

Ambulatory nursing care 37 8099/2420 1129/445

Hospice 125 12 6009/17 100 0
Tertiary care hospital 633 8009/1800 32 6009/44 300
Palliative care hospital 684 1719/900 29 9009/61 800
Long-term care facility 63 0 9009/3900
Hostel 83 5009/1100 8009/2100
Emergency care/ambulance service 93/532 1009/200 4009/800
Total 15 0009/17 600 64 6009/16 800 B/0.001

*Costs are in Canadian dollars based on Unit health care costs obtained from the Hospital Management Information
System. Plus-minus values are means9/SD. Twenty-eight patients were cared for in the Hospice program for 3
/523 days,
mean 1209/140 days.

responsible for the homeless within large urban areas
should consider integration of comprehensive health
services in a supportive shelter-based setting using a
harm reduction strategy.
Acknowledgements
We are indebted to the following for their assistance,
Wendy Muckle RN BSc.N MHA Ottawa Inner City
Health Project (OIHCP) Director, Dr. Pippa Hall MD,
CCFP, MEd, FCFP Palliative Care SCO Health Service
and Karen Nelson MSW Chief of Social Work, The
Ottawa Hospital for expertise on the panel, Vela Tadic
MSW for database creation OICHP, Pat Thornton RN,
and Pat Klus RN of the OICHP for data collection and
Diane Morrison Executive Director of The Union
Mission. This work was supported by a grant from the
Human Resources Development Corporation, Govern-
ment of Canada for the Inner City Health Project.
References
1 Hibbs JR, Benner L, Klugman L, et al . Mortality in a
cohort of homeless adults in Philadelphia. N Engl J Med
1994; 331: 304
/9.
2 Hwang SW. Mortality among men using homeless
shelters in Toronto, Ontario. JAMA 2000; 283: 2152
/7.
3 Hwang SW, Orav EJ, O’Connell JJ, et al . Causes of death
in homeless adults in Boston. Ann Intern Med 1997; 126:
625
/8.
4 Barrow SM, Herman DB, Cordova P, et al . Mortality
among homeless shelter residents in New York City. Am
J Public Health 1999; 89: 529
/34.
5 Nordentoft M, Wandall-Holm N. 10 year follow up
study of mortality among users of hostels for homeless
people in Copenhagen. BMJ 2003; 327: 81.
6 Gelberg L, Linn LS. Demographic differences in health
status of homeless adults. J Gen Intern Med 1992; 7:
601
/8.
7 Susan Farrell TA, Fran Klodawsky, Donna Pettey.
Describing the Homeless Population of Ottawa-Carle-
ton. Ottawa: University of Ottawa; 2000 February.
8 D’Amore J, Hung O, Chiang W, et al . The epidemiology
of the homeless population and its impact on an urban
emergency department. Acad Emerg Med 2001; 8: 1051
/
5.
9 Salit SA, Kuhn EM, Hartz AJ, et al . Hospitalization
costs associated with homelessness in New York City. N
Engl J Med 1998; 338: 1734
/40.
10 Rousseau P. The homeless terminally ill and hospice
& palliative care. Am J Hosp Palliat Care 1998; 15: 196
/
7.
11 Smith G. Palliative care in Toronto for people with
AIDS: the impact of class on poor PWAs. J Palliat Care
1994; 10: 46
/50.
12 Tadic V, Muckle W, Turnbull J. Internet-based medical
record keeping of the Ottawa homeless. Journal of Urban
Health 2003; 80: ii-79.
13 Schag CC, Heinrich RL, Ganz PA. Karnofsky perfor-
mance status revisited: reliability, validity, and guidelines.
J Clin Oncol 1984; 2: 187
/93.
14 WHO Definition of Palliative Care. In World Health
Organization; 2003.
15 Emanuel EJ, Ash A, Yu W, et al . Managed care, hospice
use, site of death, and medical expenditures in the last
year of life. Arch Intern Med 2002; 162: 1722
/8.
16 Wong LL, McFall P, Wong LM. The cost of dying of
end-stage liver disease. Arch Intern Med 1997; 157:
1429
/32.
17 Hillner BE, McDonald MK, Desch CE, et al . Costs of
care associated with non-small-cell lung cancer in a
commercially insured cohort. J Clin Oncol 1998; 16:
1420
/4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.