Hospice Offers More Palliative Care but Costs Less Than Usual Care for Terminal Geriatric Hepatocellular Carcinoma Patients: A Nationwide Study

Abstract

Background:

Hospice care is important for patients with terminal hepatocellular carcinoma (HCC), especially in endemic areas of viral hepatitis. Differences between hospice care and usual care for geriatric HCC inpatients have not yet been explored in a nationwide survey.

Objective:

The study's purpose was to analyze differences between hospice care and usual care for geriatric HCC inpatients in a nationwide survey.

Methods:

This nationwide, population-based study used data obtained from the Taiwan National Health Insurance Database. Patients with terminal HCC who were ≥65 years old and received their end-of-life care in the hospital between January 2001 and December 2004 were recruited. The comparison group was selected by propensity score matching from patients receiving usual care in acute wards.

Results:

We enrolled 729 terminal HCC patients receiving inpatient hospice care and 729 matched controls selected from 2482 HCC patients receiving usual care. Hospice care patients were treated mainly by family medicine doctors (36%) and oncologists (26%), while usual care patients were treated mainly by gastroenterologists (60.2%). The natural opium alkaloids were used more in the hospice care group than in the usual care group (72.7% versus 25.5%, P<0.001), whereas the length of stay (8±7.7 days versus 14.1±14.3 days, P<0.001), aggressive procedures (all P<0.005), and medical expenses (all P<0.001) were significantly less in the hospice care group.

Conclusion:

HCC patients in hospice wards received more narcotic palliative care, underwent fewer aggressive procedures, and incurred lower costs than those in acute wards. Hospice care should be promoted as a viable option for terminally ill, elderly HCC patients.

Introduction

Hepatocellular carcinoma (HCC) is a leading cause of cancer death worldwide.¹ In Taiwan, an area endemic for hepatitis B virus, HCC ranks number two among all solid tumors.² Most patients are in advanced stages with a median survival of 18.2 months.³ Patients with terminal HCC frequently suffer from abdominal discomfort, cachexia, anorexia/vomiting, esophageal variceal bleeding, and a variety of decompensated conditions such as ascites, peripheral edema, and hepatic encephalopathy.⁴ Symptomatic treatment is important to help relieve discomfort in terminally ill HCC patients.

In 1967 Dame Cicely Saunders introduced hospice care in London.⁵ Hospice care is designed to provide supportive care to patients who are in the final phase of a terminal illness. Hospice focuses on comfort and quality of life, rather than cure. Hospice programs generally use a multidisciplinary team approach, including the services of doctors, nurses, social workers, and clergy, to offer holistic care to patients. Due to the provision of such comprehensive care, patients in hospice care have a better quality of life, compared to patients with similar conditions in usual care.⁶ In Taiwan, the hospice movement was launched in 1983, and the first hospital hospice unit was set up in 1990. There were 26 inpatient hospice units established in Taiwan by 2004.⁷

Similar to Western countries such as the United States, the United Kingdom, and Australia, hospice care in Taiwan is covered by government-run health insurance programs and is offered in many medical center hospitals or medical school affiliated hospitals.⁸ However, only a minority of terminal patients use hospice.^9,10 The rate of inpatient hospice use among cancer patients who died between 2000 and 2004 was just above 10%.¹⁰ Although the trend of hospice use increased from 2.2% in 2000 to 10.1% in 2004, the rate of Taiwanese hospice use is still far below the statistics from countries like the United States, United Kingdom, and Australia.^9–12 This underuse may be due to cultural concerns, physician preference, and individual patient misperceptions.^11,12 A previous study performed in an inpatient hospice care unit in a medical center hospital in Taiwan revealed that advanced cancer inpatients in hospice have shorter lengths of stay compared to inpatients who were treated before the introduction of hospice care.¹³ Nevertheless, there is no comprehensive, nationwide comparison between inpatient hospice care and usual care in acute wards. Therefore, we compared the characteristics of patients, medical procedures, prescriptions, and medical expenses between hospice care and usual care of elderly HCC inpatients during the period from January 2001 to December 2004.

Methods

Data sources

The National Health Insurance (NHI) program is a compulsory universal health insurance program, which offers comprehensive medical care coverage to all Taiwanese residents. Approximately 97% of the people in Taiwan were enrolled in the NHI program at the end of 2004. The NHI dataset, constructed and managed by the National Health Research Institutes (NHRI), consists of comprehensive utilization of health care information for all NHI beneficiaries.¹⁴

In this study we obtained all inpatient claims data, corresponding medical order files, and hospital administration files for beneficiaries aged ≥65 years, from January 2001 to December 2004. The information included encrypted inpatient identification numbers; dates of admission and discharge; one major and four minor disease ICD-9-CM diagnoses; and costs of examinations, procedures, surgeries, prescribed drugs, and special medical materials used during admission. The corresponding medical order files contained details of doctors' orders, special medical materials, prescribed drug items, medical procedures, and laboratory tests. The information in hospital administration files contained the accreditation levels of each hospital, such as medical center hospitals, regional hospitals, and district hospitals. Medical center hospitals had a full complement of sophisticated medical services, in terms of diagnostic facilities and treatment capabilities, based on the standards of the Department of Health in Taiwan. A complete database of the coding numbers, which corresponded to the orders of prescribed drugs and procedures, were obtained from the NHI website.¹⁵ The study was in accordance with the Helsinki Declaration and approved by the institutional review board of the National Yang-Ming University (IRB No. 1000046).

Study population

In this study we enrolled 3236 HCC patients who died in the hospital and who were admitted from January 2001 through December 2004. HCC was defined by a compatible ICD-9-CM code (155.0) from the Registry for the Catastrophic Illness Patient Database, which is a separate subpart of the NHI Database. Those who were diagnosed with cancer could apply for a Catastrophic Illness Card in Taiwan. Cardholders were exempt from cost sharing required under the NHI program. Among them, 10 patients who were hospitalized due to injury (5) and natural disasters (5), 10 patients admitted to a respiratory care unit, and five patients admitted for tuberculosis (1) and chronic hepatitis trial projects (4) were excluded. The final sample of our study was 3211 patients. Of these patients, we analyzed 729 (23%) patients receiving care in hospice wards (hospice care group) and 2482 (77%) patients receiving usual care in acute wards (usual care group). The effects of comorbidities were estimated according to the Charlson comorbidity index.¹⁶ To minimize the potential influence of selection bias, we performed propensity score matching, using logistic regression to create a propensity score for hospice care and usual care groups.¹⁷ The covariates included patient characteristics (age and gender), Charlson comorbidity index, hospital location, and hospital accreditation level. Subsequently, a one-to-one match between these two groups was done using the nearest-neighbor matching method.

Statistical analysis

Data linkage, processing, and computation were performed using Microsoft Structured Query Language (SQL) Server 2008 (Microsoft Corp., Redmond, WA). SPSS 18.0 software (SPSS Inc., Chicago, IL) was used to analyze the data in descriptive and inferential statistics. The Mann-Whitney U test was used to analyze numerical data between groups. The generalized linear model with gamma distribution and log link function was used to analyze medical costs that are typically right-skewed.¹⁸ The Pearson's χ² test was used to compare categorical variables. Results were considered statistically significant at P<0.05.

Results

Baseline characteristics

Of the 3211 terminally ill elderly HCC patients, 729 (23%) were treated in hospice wards (486 male and 243 female, mean age 75 years); and 2482 (77%) were treated in acute wards (1764 males and 716 females, mean age 75 years) (see Table 1). All physicians passed standardized educational training courses to be a hospice specialist in Taiwan. The hospice care group was primarily looked after by family medicine doctors (36%) and oncologists (26%), whereas the usual care group was primarily looked after by internal medicine physicians (77%, P<0.001), who were mostly gastroenterologists (53.4%). A significantly higher percentage of hospice care patients (60%) received end-of-life care in medical center hospitals than usual care patients (45%, P<0.001). The percentage of hospice care use differed among hospitals in different regions (P<0.001). The Charlson comorbidity index, representing effects of comorbidities, was not different between the two groups, but the length of stay was significantly shorter in the hospice care group than in the usual care group (8.0±7.7 days versus 14.5±13.7 days, P<0.001). After one-to-one propensity score matching, 729 matched controls were selected, and the baseline patient characteristics in the two groups became similar (see Table 1).

Table 1.

Baseline Characteristics of Terminally Ill Elderly Patients with Hepatocellular Carcinoma Receiving Hospice Care or Prematched/Postmatched Usual Care

		Usual care
		Prematched Sample		Postmatched Sample
Characteristics	Hospice care No. (%)	No. (%)	P	No. (%)	P
Number	729 (100)	2482 (100)		729 (100)
Gender			0.021		0.539
Male	486 (67)	1764¹ (71)		497 (68)
Female	243 (33)	716 (29)		232 (32)
Physician specialty			<0.001		<0.001
Family Medicine	263 (36)	108 (4)		32 (4)
Oncology	187 (26)	180 (7)		56 (8)
Internal Medicine	143 (20)	1915 (77)		562 (77)
General Medicine	138 (18.9)	467 (18.8)		73 (10)
Gastroenterology	3 (0.4)	1326 (53.4)		439 (60.2)
Cardiology	0 (0)	19 (0.8)		6 (0.8)
Chest Medicine	1 (0.1)	66 (2.7)		28 (3.8)
Nephrology	0 (0)	14 (0.6)		7 (1)
Immunology	0 (0)	6 (0.2)		3 (0.4)
Endocrinology	0 (0)	3 (0.1)		1 (0.1)
Infection	1 (0.1)	14 (0.6)		5 (0.7)
Radiation Oncology	119 (16)	47 (2)		13 (2)
Surgery	1 (0.1)	159 (6)		61 (8)
Others	16 (2)	72 (3)		4 (0.5)
Accreditation level of hospital			<0.001		0.703
Medical center hospitals	435 (60)	1109 (45)		431 (59)
Regional hospitals	249 (34)	937 (38)		245 (34)
District hospitals	45 (6)	436 (18)		53 (7)
Region of hospital			<0.001		0.068
North	422 (58)	1508 (61)		442 (61)
Central	25 (3)	302 (12)		28 (4)
South	251 (34)	627 (25)		245 (34)
East	31 (4)	45 (2)		14 (2)
Age (year, mean±SD)	75±6	75±6	0.745	75±6	0.788
Charlson comorbidity index (mean±SD)	4.9±2.9	4.8±2.6	0.345	4.7±2.6	0.23
Length of stay (day, mean±SD)	8±7.7	14.5±13.7	<0.001	14.1±14.3	<0.001

Genders of two patients were missing.

SD, standard deviation.

Comparisons of aggressive procedures implemented

Aggressive procedures used to treat terminally ill elderly HCC patients during admission in the hospice care group and in the matched usual care group are shown in Table 2. Patients in the usual care group underwent significantly more aggressive procedures, including urinary catheterization, nasogastric tube feeding, central venous catheter insertion, intensive care unit admission, endotracheal intubation, cardiopulmonary resuscitation, abdominal drainage, hemodialysis, percutaneous transhepatic cholangiography and drainage, chest tapping or intubation, defibrillation/cardioversion, total parenteral nutrition, epinephrine injection, and esophageal balloon insertion, than those in the hospice care group (all P<0.005). Notably, there were no incidences of aggressive procedures, such as endotracheal intubation, epinephrine injection, and esophageal balloon insertion, in the hospice care group.

Table 2.

Aggressive Procedures Implemented in Terminally Ill Elderly Patients with Hepatocellular Carcinoma Receiving Hospice Care or Usual Care

	Hospice care (N=729)		Usual care (N=729)
Procedures	%	No.	%	No.	P
Urinary catheterization	30.9	225	53.5	390	<0.001
Nasogastric tube feeding	16.5	120	38.3	279	<0.001
Central venous catheter insertion	3.0	22	31.1	227	<0.001
Intensive care unit admission	0.1	1	21.8	159	<0.001
Endotracheal intubation	0	0	24.4	178	<0.001
Cardiopulmonary resuscitation	0.1	1	16.6	121	<0.001
Abdominal drainage	6.3	46	10.8	79	0.002
Hemodialysis	0.5	4	2.9	21	0.001
Percutaneous transhepatic cholangiography and drainage	1.0	7	5.8	42	<0.001
Chest tapping or intubation	1.1	8	4.3	31	<0.001
Defibrillation / cardioversion	0.1	1	4.4	32	<0.001
Total parenteral nutrition	0.5	4	2.9	21	0.001
Epinephrine injection	0	0	1.4	10	0.002
Esophageal balloon insertion	0	0	1.5	11	0.001

Drug prescribing patterns in hospice care and usual care groups

The prescription patterns in both care groups are presented in Table 3. The medications were categorized by the Anatomical Therapeutic Chemical Classification. In the hospice care group, the most commonly prescribed medication was natural opium alkaloids (72.7%), followed by solutions affecting the electrolyte balance (72.6%), plain sulfonamide diuretics (53.8%), propulsives (50.9%), and osmotically acting laxatives (50.3%). In the matched usual care group, the medications prescribed in order of decreasing frequency were as follows: solutions affecting the electrolyte balance (94.0%), solutions for parenteral nutrition (79.3%), plain sulfonamide diuretics (74.6%), adrenergic and dopaminergic agents (62.7%), and cephalosporins (61.2%). Patients in the hospice care group used more natural opium alkaloids and benzodiazepine derivatives, but less adrenergic and dopaminergic agents and cephalosporins than those in the usual care group (all P<0.001).

Table 3.

Prescribed Medicine in Terminally Ill Elderly Patients with Hepatocellular Carcinoma Receiving Hospice Care or Usual Care

	Hospice care (N=729)			Usual care (N=729)
Ranking	Name of the drug class (ATC code)	%	No.	Name of the drug class (ATC code)	%	No.
1	Natural opium alkaloids (N02AA)	72.7	530	Solutions affecting the electrolyte balance (B05BB)	94.0	685
2	Solutions affecting the electrolyte balance (B05BB)	72.6	529	Solutions for parenteral nutrition (B05BA)	79.3	578
3	Plain sulfonamide diuretics (C03CA)	53.8	392	Plain sulfonamide diuretics (C03CA)	74.6	544
4	Propulsives (A03FA)	50.9	371	Adrenergic and dopaminergic agents (C01CA)	62.7	457
5	Osmotically acting laxatives (A06AD)	50.3	367	Cephalosporins (J01DA)	61.2	446
6	Benzodiazepine derivatives (N05CD)	45.0	328	Propulsives (A03FA)	59.1	431
7	Contact laxatives (A06AB)	41.3	301	Osmotically acting laxatives (A06AD)	58.6	427
8	Solutions for parenteral nutrition (B05BA)	38.8	283	Electrolyte solutions (B05XA)	56.8	414
9	Glucocorticoids (H02AB)	38.0	277	Contact laxatives (A06AB)	43.8	319
10	Magnesium compounds (A02AA)	32.9	240	Aldosterone antagonists (C03DA)	42.4	309

ATC, Anatomical Therapeutic Chemical.

Comparisons of medical expenses between hospice care and usual care groups

Comparisons of medical expenses between the hospice and matched usual care group are shown in Table 4. The total medical expenses were significantly higher in the usual care group (adjusted mean=US$326/day, 95% confidence interval=290–366 in the usual care group versus adjusted mean=US$114/day, 95% confidence interval=102–128 in the hospice care group, P<0.001). Individual daily medical expenses, including diagnoses, laboratory examinations, radiologic examinations, therapies, medications, and hemodialyses were all significantly higher in the usual care group than in the hospice care group (all P<0.001).

Table 4.

Medical Expenses of Terminally Ill Elderly Patients with Hepatocellular Carcinoma Receiving Hospice Care or Usual Care

	Hospice care (N=729)		Usual care (N=729)
Items	Adjusted mean	95% CI	Adjusted mean	95% CI	P¹
Diagnosis expenses²	12.4	10.7–14.5	18.5	15.9–21.5	<0.001
Laboratory examination expenses	7.9	6.9–9.0	40.5	35.3–46.4	<0.001
Radiologic examination expenses	2.6	1.7–3.8	19.6	13.0–29.5	<0.001
Therapeutic expenses	18.1	15.7–20.8	39.1	34.0–44.9	<0.001
Medication expenses	24.8	21.4–28.8	81.3	70.1–94.3	<0.001
Hemodialysis expenses	2.7	2.0–3.7	34.9	25.8–47.3	<0.001
Total medical expenses	114	102.0–128.0	326.0	290.0–366.0	<0.001

Values in tabulation are presented in US$/day.

Analyzed by generalized linear model with gamma distribution and log link function.

CI, confidence interval.

Discussion

In this study we collected comprehensive data for terminally ill elderly HCC patients who received either inpatient hospice care or usual care in the acute wards. A total of 3211 elderly HCC patients who died during hospital admission between January 2001 and December 2004 were analyzed from the nationwide NHI Database. The results showed that the majority of elderly HCC patients were male and hospitalized in medical center hospitals. Furthermore, most patients were treated in the usual acute wards. In the hospice care group, approximately one-third were treated by family medicine physicians, with shorter lengths of stay in hospital, fewer aggressive procedures, and lower medical expenses. Hospice patients were prescribed more natural opium alkaloids and benzodiazepine derivatives than the usual care inpatients.

The types of hospice care currently available in Taiwan include inpatient hospice care and hospice home care (home visits by the nurses and other interdisciplinary staff on a regular basis). In January 2013 there are 43 hospital-based inpatient hospice care and 70 hospice home care teams.¹⁹ Overall, an estimated 13,000 cancer patients receive these services annually.²⁰ Inpatient hospice care is available and fully accessible to all beneficiaries in Taiwan. Hospice care is offered in all medical center hospitals, in selected regional hospitals, and rarely in district hospitals. HCC patients in Taiwan are frequently referred to medical center hospitals, where a full complement of diagnostic facilities and treatment capabilities is more available.²¹ Patients are then referred to the hospice in the medical center hospital when they reach a terminal stage. It is more convenient to access sophisticated medical services for a hospice inpatient in a medical center hospital, although such services are rarely used. These constitute a higher percentage of elderly HCC patients hospitalized in medical center hospitals and a higher percentage of hospice care in medical center hospitals than in regional or district hospitals.

In Taiwan, hospice care is not a widely used service. In the present survey, the majority of elderly terminal HCC patients received end-of-life care in acute wards rather than in hospice wards. Several factors might influence the choice of using hospice care, including physicians' preferences and referral practices, cultural concerns, individual patient choices and circumstances, and public and professional awareness of the benefit.^11,12,22,23 In Taiwanese traditions, inpatient hospice units are associated with the negative image of ‘death ward,’ and the strong sense of filial obligation leads people to care for their parents at all times. Additionally, caregivers and family members generally prefer life-sustaining treatments for terminally ill patients, and some physicians prefer not to discuss end-of-life issues.²⁴ These reasons might encourage Taiwanese patients and family members to choose end-of-life care in the acute wards rather than in the hospice wards.

The tenets of end-of-life care have been associated with the philosophically rooted recognition of individual dignity. Helping patients die with dignity, alleviate suffering and pain, and control symptoms, along with the use of less aggressive therapies, are the most common treatments of patients in the hospice care unit.^7,12,25 Our results show that 72.7% of the hospice care inpatients were prescribed natural opium alkaloids, which is similar to the 76.4% reported in another hospital-based study.⁴ Only 25.5% (data not shown) of patients in the usual care group were prescribed natural opium alkaloids. Furthermore, patients receiving care in the hospice wards had shorter lengths of stay. These results were similar to another hospital-based study,¹³ while another retrospective study from the United States reported that hospice is associated with a longer survival for certain terminally ill patients.⁹ Our results might be attributed to rapidly growing HCC after stopping anticancer treatments and to latent poor liver function. Moreover, treatment of symptoms resulting from HCC and liver failure is difficult and is mainly empirical in the hospice wards.⁴ The hospice care inpatients also had lower medical expenses per person or per inpatient day than those in the usual care group. This corresponds with previous studies showing that hospice care costs less than usual care.^25–27

The results of this research should be viewed in light of its limitations. We could not obtain the socioeconomic and educational status of these patients, the attitudes of physicians toward hospice care, the preferences of patients and their caregivers, and life expectancies of the patients. A cross-sectional study design was not supportive of understanding the relationships among these factors. Although the covariates included in the logistic regression were very limited in number, the propensity-score matching at the patient level provided comparable characteristics between the two groups at baseline, which potentially avoided some selection biases.

Although lower cost burden and fewer aggressive procedures were desirable as seen in the hospice group, there was no data on symptom burden/pain scores between the two groups and a quality of life outcome measure was not available in the administrative data. Further research with quality of life outcome measure could help advance our knowledge in this respect.

The major strength of this study was its nationwide population-based setting that included a relatively large number of elderly HCC cases and that allowed a comprehensive observation of medical behaviors and expenses between hospice care and usual care. The findings provided epidemiological evidence that hospice care not only provided more palliative treatment for terminally ill elderly HCC patients, but also cost less than usual care. The findings also provided evidence to support changing traditional perceptions about hospice care and promoting end-of-life hospice care for elderly HCC patients.

Footnotes

Acknowledgments

This study was based in part on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health, and managed by the National Health Research Institutes in Taiwan. The interpretation and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health, or National Health Research Institutes. This work was supported by grants from Taipei Veterans General Hospital (V101C-178), National Science Council (NSC 98-2314-B-075-029), the National Research Program for Biopharmaceuticals of Taiwan (100CT202), and Aim for the Top University Plan of National Yang-Ming University and the Ministry of Education, Taiwan.

Author Disclosure Statement

No competing financial interests exist.

References

Venook

, Papandreou

, Furuse

, de Guevara

. The incidence and epidemiology of hepatocellular carcinoma: A global and regional perspective. Oncologist, 2010; 15,S4:5–13.

Department of Health, Government of Taiwan: 2011 Statistics of Causes of Death. www.doh.gov.tw/CHT2006/DM/DM2_2.aspx?now_fod_list_no=12743&class_no=440&level_no=4(Last2013 January 13.)

Chen

, Hu

, Huang

et al. Applicability of staging systems for patients with hepatocellular carcinoma is dependent on treatment method: Analysis of 2010 Taiwanese patients. Eur J Cancer, 2009; 45,9:1630–1639.

Lin

, Wu

, Tsai

, Lin

, Chen

, Hwang

. Hospice palliative care for patients with hepatocellular carcinoma in Taiwan. Palliat Med, 2004; 18,2:93–99.

Clark

. Cicely Saunders, Founder of the Hospice Movement: Selected Letters 1959–1999. London: Oxford University Press, 2005.

Kane

, Wales

, Bernstein

, Leibowitz

, Kaplan

. A randomised controlled trial of hospice care. Lancet, 1984; 1,8382:890–894.

Loke

, Rau

. Differences between inpatient hospice care and in-hospital nonhospice care for cancer patients. Cancer Nurs, 2011; 34,3:E21–E26.

Goldsmith

, Dietrich

, Du

, Morrison

. Variability in access to hospital palliative care in the United States. J Palliat Med, 2008; 11,8:1094–1102.

Connor

, Pyenson

, Fitch

, Spence

, Iwasaki

. Comparing hospice and nonhospice patient survival among patients who die within a three-year window. J Pain Symptom Manage, 2007; 33,3:238–246.

10.

Tang

, Chen

, Huang

, Koong

, Lin

, Hsiao

. Hospice utilization in Taiwan by cancer patients who died between 2000 and 2004. J Pain Symptom Manage, 2007; 33,4:446–453.

11.

Tang

, Huang

, Liu

, Wang

, Chen

. A population-based study on the determinants of hospice utilization in the last year of life for Taiwanese cancer decedents, 2001–2006. Psychooncology, 2010; 19,11:1213–1220.

12.

Tang

, Liu

, Lai

, McCorkle

. Discrepancy in the preferences of place of death between terminally ill cancer patients and their primary family caregivers in Taiwan. Soc Sci Med, 2005; 61,7:1560–1566.

13.

Chiou

, Chen

, Lin

et al. A five-year analysis of the inpatient service of the hospice and palliative care unit in VGH-Taipei. Taiwan J Fam Med, 2005; 15:123–133.

14.

National Health Research Institutes: National Health Insurance Research Database. w3.nhri.org.tw/nhird//en/Data_Subsets.html(Last2013 January 13.)

15.

Bureau of National Health Insurance: Medical Material Items of National Health Insurance. www.nhi.gov.tw/webdata/webdata.aspx?menu=18&menu_id=683&WD_ID=757&webdata_id=1166(Last2013 January 13.)

16.

Charlson

, Pompei

, Ales

, MacKenzie

. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis, 1987; 40,5:373–383.

17.

D'Agostino

Jr.

Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med, 1998; 17,19:2265–2281.

18.

Nelder

, Wedderburn

. Generalized linear models. J R Stat Soc A Stat, 1972; 135:370–384.

19.

Bureau of National Health Insurance: Hospice Care. www.nhi.gov.tw/webdata/webdata.aspx?menu=20&menu_id=712&webdata_id=3650&WD_ID=902(Last2013 January 13.)

20.

Ministry of Foreign Affairs: Improving Taiwan's End-of-Life Care. taiwantoday.tw/ct.asp?xItem=173667&ctNode=1767(Last2013 January 13.)

21.

Huo

, Lin

, Hsia

et al. The model for end-stage liver disease based cancer staging systems are better prognostic models for hepatocellular carcinoma: A prospective sequential survey. Am J Gastroenterol, 2007; 102,9:1920–1930.

22.

Beccaro

, Costantini

, Giorgi Rossi

, Miccinesi

, Grimaldi

, Bruzzi

. Actual and preferred place of death of cancer patients. Results from the Italian survey of the dying of cancer (ISDOC) J Epidemiol Community Health, 2006; 60,5:412–416.

23.

Higginson

, Sen-Gupta

. Place of care in advanced cancer: A qualitative systematic literature review of patient preferences. J Palliat Med, 2000; 3,3:287–300.

24.

Keating

, Landrum

, Rogers

Jr.

et al. Physician factors associated with discussions about end-of-life care. Cancer, 2010; 116,4:998–1006.

25.

Lin

, Chiu

, Hsu

et al. Medical expenditure and family satisfaction between hospice and general care in terminal cancer patients in Taiwan. J Formos Med Assoc, 2009; 108,10:794–802.

26.

. The impact of hospices on health care expenditures: The case of Taiwan. Soc Sci Med, 2002; 54,6:981–991.

27.

Pyenson

, Connor

, Fitch

, Kinzbrunner

. Medicare cost in matched hospice and non-hospice cohorts. J Pain Symptom Manage, 2004; 28,3:200–210.