- Email: [email protected]
Clinical and Socio-Demographic Predictors of Home Hospice Patients Dying at Home: A Retrospective Analysis of Hospice Care Association's Database in Singapore
Accepted Manuscript Clinical and socio-demographic predictors of home hospice patients dying at home: A retrospective analysis of Hospice Care Association’s database in Singapore Yee Song Lee, MBBS, MPH, Ramaswamy Akhileswaran, MD, FAMS, MBA, Marcus Ong Eng Hock, MBBS, FRCS, MPH, Win Wah, MBBS, MPH, EPH, David Hui, MD, MSc, Sheryl Hui-Xian Ng, BSc (Hons), Gerald Koh, MBBS, MMed, FCFP, FAMS, MGer, PhD PII:
S0885-3924(17)30056-8
DOI:
10.1016/j.jpainsymman.2017.01.008
Reference:
JPS 9380
To appear in:
Journal of Pain and Symptom Management
Received Date: 25 August 2016 Revised Date:
9 January 2017
Accepted Date: 10 January 2017
Please cite this article as: Lee YS, Akhileswaran R, Ong Eng Hock M, Wah W, Hui D, Ng SH-X, Koh G, Clinical and socio-demographic predictors of home hospice patients dying at home: A retrospective analysis of Hospice Care Association’s database in Singapore, Journal of Pain and Symptom Management (2017), doi: 10.1016/j.jpainsymman.2017.01.008. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Original Article
16-00533R1
Clinical and socio-demographic predictors of home hospice patients dying at home: A retrospective analysis of Hospice Care Association’s database in Singapore
RI PT
Yee Song Lee, MBBS, MPH, Akhileswaran Ramaswamy, MD, FAMS, MBA, Marcus Ong Eng Hock, MBBS, FRCS, MPH, Win Wah, MBBS, MPH, EPH, David Hui, MD, MSc, Sheryl Hui-
SC
Xian Ng, BSc (Hons), and Gerald Koh, MBBS, MMed, FCFP, FAMS, MGer, PhD
National University Health System (NUHS) (Y.S.L.); Khoo Teck Puat Hospital (R.A.);
M AN U
Department of Emergency Medicine (M.O.E.H.), Singapore General Hospital; and Saw Swee Hock School of Public Health (W.W., S.H.-X.N., G.K.), National University of Singapore, Singapore, Singapore; and University of Texas M.D. Anderson Cancer Center (D.H.), Houston,
TE D
Texas, USA
Address correspondence to:
Yee Song Lee, MBBS, MPH
EP
45 Mandalay Road #07-04 Mandale Heights
AC C
Singapore 308225
E-mail: [email protected]
1
ACCEPTED MANUSCRIPT
ABSTRACT Context: Hospice care can be delivered in different settings but many patients choose to receive
RI PT
it at home because of familiar surroundings. Despite their preferences, not every home hospice patient manages to die at home.
Objective: To examine the independent factors associated with home hospice patient dying at
SC
home.
Method: Retrospective analysis of Hospice Care Association (HCA)’s database. HCA is the
M AN U
largest home hospice provider in Singapore. The study included all patients who were admitted into home hospice service from 1 Jan 2004 to 31 Dec 2013. Cox Proportional Hazards modelling with time as constant was used to study the relationship between independent variables and home death.
TE D
Result: 19,721 patients were included in the study. Females (Adjusted Risk Ratio, ARR 1.09, 95% CI 1.04 – 1.15), older patients (ARR 1.01, 95% CI 1.00 - 1.01), shorter duration of home hospice stay (ARR 0.88, 95% CI 0.82 – 0.94), fewer episodes of hospitalisation (ARR 0.81,
EP
95% CI 0.75 – 0.86), living with caregivers (ARR 1.54, 95% CI 1.05 – 2.26), doctor (ARR 1.05, 95% CI 1.01 – 1.08) and nurse (ARR 1.06, 95% CI 1.04 – 1.08) visits were positive
AC C
predictors of dying-at-home. Diagnosis of cancer (ARR 0.93, 95% CI 0.86 – 1.00) was a negative predictor of dying-at-home. Conclusion: Female, older age, living with a caregiver, non-cancer diagnosis, more doctor and nurse visits, shorter duration of home hospice stays and fewer episodes of acute hospitalisations are predictive of dying-at-home for home hospice patients.
2
ACCEPTED MANUSCRIPT
Keywords Palliative care; hospice care; home care; home hospice; neoplasm; cancer; place of death
RI PT
Running head: Predictors of Home Death for Home Hospice
AC C
EP
TE D
M AN U
SC
Accepted for publication: January 10, 2017
3
ACCEPTED MANUSCRIPT
Introduction Palliative care has increased in prominence in recent decades and has now become an integral
RI PT
part of many health service systems around the world. The increase in demand for palliative care is significant, especially in countries with profound demographic shift such as an ageing
population. In addition, the global change in disease patterns from communicable to noncommunicable diseases has contributed to the rise in demand for palliative care (1). The
SC
evidence on the benefits of palliative care is strong: several studies have shown that palliative
M AN U
care improves patient symptoms, quality of life, patient and family satisfaction and possibly the survival of patients (2, 3). Patients with a life-limiting illness can access palliative care for symptom relief even if they are still receiving curative treatment. However, for patients who have a life expectancy of 1 year or less and curative treatment is no longer indicated, palliative care is the most appropriate treatment to keep them comfortable and die in their preferred place
TE D
of death. This type of palliative care is known as hospice care and it can be delivered in different settings such as an in-patient hospice, a day hospice and a patients’ home. Patients’ preferred place of death is normally one of the major factors that decide the place of hospice care. In
EP
reality, despite the wishes of dying at home, a large proportion of home hospice patients still die
AC C
in hospitals and in certain cases, utilise a hospital bed inappropriately (4). A systematic review by Gomes et al on studies from United States, United Kingdom, Canada and Australia that included more than 1 million participants found that the determinants associated with dying at home were low functional status, living with relatives, having extended family support, preference to die at home and receiving home care (5). The result of this systematic review had illustrated the importance of patient living status, caregiver status and home hospice care. 4
ACCEPTED MANUSCRIPT
The provision of palliative care is intricately linked to the sociocultural environment of a society. In Singapore, a multi-ethnic Asian country with diverse cultural background, the need for palliative care research in order to provide culturally-appropriate palliative services could not
RI PT
be greater. There were two studies published in Singapore that examined the determinants of place of death for the general population and cancer patients. According to Beng et al, a
retrospective database analysis of Singapore Registry of Birth and Death, female, Malay ethnic
SC
group and higher age group were more likely to die at home (6). The study had also shown that elderly who died of stroke were more likely to die at home compared to heart disease, cancer
M AN U
and respiratory disease. Another retrospective database analysis study by Hong et al had also yielded similar results (7). This study analysed the Singapore Cancer Registry data and showed that older and female patients were more likely to die at home.
Among demographic factors, the findings in Singapore were consistent with other countries.
TE D
Age and gender were major determinants with female and older patients more likely to die at home based on studies conducted in Taiwan, England and United States (8-10). Another determinant of place of death is socio-economic status and a country's healthcare financing
EP
model. Some studies have shown that high economic status was associated with higher
AC C
propensity of dying at home (10, 11) while others had shown the contrary (8, 9). A study has also shown that geographical location influenced the place of death of home hospice patients, especially in larger countries with an urban-suburban divide (11). Other studies have shown that patients who had shorter length of stay in home hospice service were more likely to die at home (10). Both Singapore studies were retrospective database analysis of national registries. However, these studies did not examine the significance of socio-economic status, length of stay in home 5
ACCEPTED MANUSCRIPT
hospice service, caregiver status, living status, home medical support and distance between patient’s home and hospital. Hence, it is the intention of this research to address this knowledge gap and provide new evidence for policy making by studying the independent factors associated
RI PT
with home hospice patient dying at home in Singapore, a multi-ethnic Asian country.
SC
Methodology
We retrospectively analysed a database of patients who were admitted to HCA Hospice Care
M AN U
(HCA) from 1st January 2004 to 31st December 2013. HCA Hospice Care is a registered nonprofit organization in Singapore which provides home hospice care and operates a day hospice centre. HCA provides home hospice service to any patient who has a life limiting illness with a life expectancy of less than 12 months regardless of age, religion, ethnicity, nationality and
TE D
financial status. HCA is also the largest local home hospice provider in Singapore; providing 60% of all home hospice service in Singapore with services spanning across the island. Independent variables were chosen based on literature review, clinical reasoning and the
EP
information available in the database. Patient demographic profile, clinical diagnosis, living
AC C
status, caregiver relationship, length of stay in home hospice service, hospital utilisation, mean testing, hospice staff visits and place of death were selected as independent variables. Place of death was selected as the dependent variable and was categorised as home versus non-home. For patients with repeated admissions, the episodes of hospital admissions were summed up together to produce a total number of hospitalisation. Patient living status was categorised as living alone, living with a caregiver or living in an institution. HCA looks after a small number of patients who lived in institutions such as nursing homes that did not provide hospice care. This group of
6
ACCEPTED MANUSCRIPT
patients received the standard end of life care service from HCA as those who lived at home. Those who died at the original institutions were categorised as home deaths.
RI PT
Means-testing which is routinely performed in HCA was used as a proxy for economic status and it was categorised as 0%, >0% to 50% and >50%. 0% indicates no subsidy (highest socioeconomic status) and >50% indicate the highest subsidy (lowest socio-economic status).
Caregiver relationship was categorised as spouse, non-spouse relative, unrelated and none. The
SC
relationship between place of death and distance from patient home to primary referral hospital
M AN U
was also studied. The distance was generated through a software system known as Network Analyst Extension in ArcMap 10.1 (ESRI, USA) by calculating the travel distance along the roadways from the postal codes of patients’ home to the primary referral hospital. Detailed recording of symptom control in the database was only initiated in recent years so this variable was not used in the analysis. Cause of death was classified according to the International
TE D
Classification of Disease, version 10 (ICD-10) and it was dichotomised into non-cancer and cancer groups. Data on length of stay in hospice services was found to be skewed with a wide range from 1 to 3978 and hence, was log-transformed in the analysis. The research protocol was
EP
approved by the Institutional Review Board (IRB) of National University of Singapore.
AC C
Patient characteristics were described in frequencies for categorical variables, while continuous variables were reported in mean and standard deviation. Bivariate and multivariate analyses were conducted using Cox Proportional Hazards modelling with time as a constant to study the relationship between the independent variables and place of death (12). Risk ratios were reported using robust standard errors. Presence of collinearity between independent variables in the final model was tested by examining variance inflation factors (13). The conventional
7
ACCEPTED MANUSCRIPT
P<0.05 was used as the cut-off for statistical significance. All statistical analyses were
Results
RI PT
performed using SPSS version 22 and STATA 13 (14).
There were 19,721 unique patients who were registered with the HCA during the 10 year period
SC
died either at home or in hospitals or in-patient hospices.
As shown in Table 1, there were 10,858 (55.0%) patients died at home, 8,335 (42.3%) patients
M AN U
died in hospitals, 397 (2.0%) patients died in in-patient hospices and 131 (0.7%) died in nursing home. The age of patients ranged from 12 to 105 years with mean of 71.0 years. The gender of patients was quite evenly distributed with the number of female patients at 9,629 (48.8%). Home hospice patients had an average of 0.85 hospital admissions and the mean distance between
TE D
hospital and patients’ home was 10.9 kilometres. Patients registered with HCA receive home visits from doctors and nurses but not necessarily from medical social workers unless there were unresolved social issues. Nurses formed the highest number of hospice staff visits with mean of
EP
3.22 visits per patient whereas medical social worker formed the least number of visits with mean of 0.08 visits per patient. Majority of the patients in this study had cancer (95.6%) and
AC C
lived at home with their caregivers (97.2%). Similarly, most of the patients were related to their caregivers, spouses (65.8%) and non-spouse relatives (33.3%), while only a small proportion of patients were looked after by unrelated caregiver (0.7%) or did not have any caregiver (0.1%). In term of socio-economic status, 28.5% received >50% subsidies, 26.0% received >0 to 50% subsidies and 45.5% received no subsidy.
8
ACCEPTED MANUSCRIPT
Table 2 showed the unadjusted and adjusted risk ratio of the independent variables. Under the unadjusted model, females have a higher likelihood of home death than males (Crude Risk Ratio, CRR=1.12, 95%CI: 1.09–1.15). Higher age group (CRR=1.01, 95%CI: 1.01–1.01),
RI PT
shorter duration of hospice service (CRR=0.94, 95%CI: 0.92–0.96) and fewer episodes of
hospitalisation (CRR=0.83, 95%CI: 0.81–0.85) were associated with higher chance of dying at home. Patients who received more doctor (CRR=1.16, 95%CI: 1.15–1.17) and nurse visits
SC
(CRR=1.08, 95%CI: 1.08–1.09) were also more likely to die at home. However, more frequent medical social worker visits were associated with lower chance of dying at home (CRR=0.95,
M AN U
95%CI: 0.92–0.99). The bivariate analysis also demonstrated that patients who lived further away from hospital, cared for by spouses, living with caregivers and had non-cancer diagnosis were more likely to die at home. In term of socio-economic status, patients from the middle means-testing group (>0% to 50%) were more likely to die at home compared to patients from
TE D
highest (0%) and lowest means-testing group (>50%).
Under the adjusted model, the independent determinants of death at home for demographic characteristics were female gender (Adjusted Risk Ratio, ARR=1.09, 95%CI: 1.04–1.15) and
EP
older age (ARR=1.01, 95%CI: 1.00-1.01). From a health services perspective, length of stay in
AC C
home hospice service (ARR=0.88, 95%CI: 0.82–0.94), home visit from healthcare personnel such as doctor (AOR=1.05, 95%CI: 1.01–1.08) and nurse (AOR=1.06, 95%CI: 1.04–1.08) were independent predictors of dying at home. Having cancer (ARR=0.93, 95%CI: 0.86–1.00) decreased the likelihood of dying at home whereas living at home with caregivers (ARR=1.54, 95%CI: 1.05–2.26) increased the likelihood of dying at home. Examination of VIF revealed no significant collinearity between the independent variables.
9
ACCEPTED MANUSCRIPT
Table 1. Profile of study population (N=19,721). N (%)
Place of death At home Hospital In-patient hospice Nursing home
10,858 (55.0) 8,335 (42.3) 397 (2.0) 131 (0.7)
RI PT
Variables
Gender Male Female
SC
10,090 (51.2) 9,629 (48.8)
71.00 (13.34)
38 (1-3978, 14-100)
Log-transformed length of stay in service, days (Mean, SD)
1.57 (0.63)
Number of hospitalisation episodes(Mean, SD)
0.85 (1.55)
Number of doctor visits (Mean, SD)
1.12 (1.08)
Number of nurse visits (Mean, SD)
3.22 (2.37)
Number of medical social worker visits (Mean, SD)
0.08 (0.45)
Distance from hospital, kilometre (Mean, SD)
10.9 (5.71)
Diagnosis Non-cancer Cancer
650 (4.4) 14,194 (95.6)
AC C
EP
Length of stay in home hospice service in days Median (minimum-maximum, 25th – 75th interquartile range )
TE D
M AN U
Age in years (Mean, SD)
Caregiver status Spouse Non-spouse relative Not related None
12,412 (65.8) 6,279 (33.3) 135 (0.7) 26 (0.1)
Living arrangement Alone
87 (1.7) 10
ACCEPTED MANUSCRIPT
5,034 (97.2) 58 (1.1)
Mean-tested subsidy level 0% >0% to 50% >50%
7,967 (45.5) 4,549 (26.0) 4,987 (28.5)
AC C
EP
TE D
M AN U
SC
RI PT
At home with caregivers Institution
11
ACCEPTED MANUSCRIPT
RI PT
Table 2. Unadjusted and adjusted risk ratio relating the independent variables to place of death for patients registered under Hospice Care Association (HCA) from 1st January 2004 to 31st December 2013.
Unadjusted Model Crude RR 95% CI
Gender, n (%) Male Female
5,252 (52.1) 5,605 (58.2)
4,838 (47.9) 4,024 (41.8)
1.00 1.12
Mean age in years (SD)
72.76 (13.05)
Mean of log-transformed length of stay in the service in days (SD)
1.54 (0.66)
68.63 (13.37) 1.60 (0.60)
P-value
SC
Variables
Place of Death Home Non-home
M AN U
1.09 - 1.15 <0.001
Adjusted Model Adjusted 95% CI RR
P-value
1.00 1.09
1.04 - 1.15 <0.001
1.01 - 1.01 <0.001
1.01
1.00 - 1.01 <0.001
0.94
0.92 - 0.96 <0.001
0.88
0.82 - 0.94 <0.001
0.83
0.81 - 0.85 <0.001
0.81
0.75 - 0.86 <0.001
1.16
1.15 - 1.17 <0.001
1.05
1.01 - 1.08 0.003
TE D
1.01
1.18 (1.73)
Mean number of doctor visits (SD)
1.32 (1.14)
0.89 (0.94)
Mean number of nurse visits (SD)
3.68 (2.42)
2.65 (2.18)
1.08
1.08 - 1.09 <0.001
1.06
1.04 - 1.08 <0.001
Mean number of medical social worker visits (SD)
0.07 (0.42)
0.09 (0.48)
0.95
0.92 - 0.99 0.005
0.97
0.93 - 1.01 0.092
AC C
EP
Mean number of episodes of 0.58 (1.32) hospitalisation (SD)
12
ACCEPTED MANUSCRIPT
10.80 (5.75)
1.00
1.00 - 1.00 0.009
Caregiver status, n (%) Spouse Non-spouse Not related None
7,902 (60) 3,220 (51.3) 42 (31.1) 9 (34.6)
5,320 (40) 3,059 (48.7) 93 (68.9) 17 (65.4)
1.00 0.90 0.54 0.61
<0.001 0.87 - 0.92 <0.001 0.42 - 0.70 <0.001 0.36 - 1.03 0.063
Diagnosis, n (%) Non-cancer Cancer
445 (68.5) 7,676 (54.1)
205 (31.5) 6,518 (55.9)
Living arrangement, n (%) Alone Home with caregiver Institution
25 (28.7) 3,089 (61.4) 3 (5.2)
62 (71.3) 1,945 (38.6) 55 (94.8)
Mean-tested subsidy level, n (%) 0% >0% to 50% >50%
4,078 (51.2) 2,779 (61.1) 2,736 (54.9)
3,889 (48.8) 1,770 (48.9) 2,251 (55.1)
SC
M AN U
1.00
1.00 - 1.00 0.866
1.00 0.98 0.84 1.13
0.870 0.93 - 1.04 0.548 0.48 - 1.47 0.546 0.15 - 8.66 0.907
0.86 - 1.00 0.042
1.00 0.79
0.75 - 0.83 <0.001
1.00 0.93
1.00 2.14 0.18
<0.001 1.53 - 2.98 <0.001 0.06 - 0.57 0.003
1.00 1.54 0.11
<0.001 1.05 - 2.26 0.026 0.03 - 0.45 0.002
1.00 1.19 1.07
<0.001 1.16 - 1.23 <0.001 1.04 - 1.11 <0.001
1.00 1.03 0.99
0.503 0.97 - 1.10 0.334 0.94 - 1.05 0.861
TE D
EP
AC C
RI PT
11.02 (5.67)
Mean distance from hospital, kilometre (SD)
13
ACCEPTED MANUSCRIPT
Discussion
RI PT
This is the first study in Singapore that examined the determinants of place of death among home hospice patients. We found that female gender and age were associated with home death. This is consistent with multiple studies demonstrating that male
SC
patients are often treated more aggressively at the end-of-life, and are often more likely to die in hospitals (15, 16, 17). However, other investigators did not find gender to be a
M AN U
key determinant of home death (18, 19, 20). This discrepancy may be related to differences in patient characteristics, healthcare system and/or cultural factors. Apart from that, living arrangement was found to be an important determinant of place of death. This finding is consistent with a large systematic review conducted by Gomes
TE D
et al which showed that home hospice patients who lived with their caregiver were more likely to die at home (5). In addition, a study conducted in China has concluded that living arrangement as a determinant of place of death is not just limited to living in
EP
the same household but also extends to those who live around the home. The China
AC C
study has shown that having a caregiver living nearby decreases the risk of dying in hospital by 74% among elderly home hospice patients (21). Caregiver status was shown to be a significant determinant of place of death for home hospice patients in bivariate analysis (i.e. home hospice patients who had spouse or nonspouse relative caregivers were more likely to die at home compared to non-related
14
ACCEPTED MANUSCRIPT
caregivers), but not on multivariate analysis. However, it is important to note that
RI PT
caregiver is not a prerequisite for the enrolment into HCA home hospice services. Although Singapore has seen rapid economic development and westernisation in recent decades, Asian values are still deeply entrenched in the communities. Asian values put
SC
strong emphasis on the responsibility of the children to show filial piety and care for
their elderly parents. These value and belief systems were reflected in our study where
M AN U
majority of the home hospice patients had caregivers that comprised mainly of their spouses or relatives. In addition, there is a large population of foreign workers in Singapore who work as domestic maid who sometimes also function as caregivers. The Singapore government not only provides financial subsidy for patients or family
TE D
members to hire foreign domestic workers (FDWs) but also arranges caregiving courses for FDWs to attend (22).
Principal diagnosis for admission into home hospice care service was also found to be a
EP
determinant of place of death and non-cancer patients were more likely to die at home. Nevertheless, this result is not consistent with the two previous similar studies done in
AC C
Singapore where patients with cancer were more likely to die at home (6, 7). One possible explanation could be the smaller sample size of non-cancer patients in our study (4.4%) compared to 95.6% being cancer patients. Another factor could be a selection bias by the referring doctor, i.e., only non-cancer patients who were deemed to be fit to be cared for at home were referred to HCA. Other determinants of place of
15
ACCEPTED MANUSCRIPT
death were length of stay in home hospice service and number of hospitalisations prior
RI PT
to death. It is uncertain if the majority of home hospice patients included in this study were referred late to the home hospice service, i.e., the patients were already very sick when they were referred, which could have contributed to the shorter duration of stay in
SC
the home hospice service. Some of the patients were also “terminal discharges”, i.e.,
patients discharged just before they died at home. Hence, future research should be done
M AN U
to identify the impact of ‘terminal discharges’ and also the condition of the patient at the time of referral to the service by studying the reasons for admission into the home hospice service. On the other hand, patients who had more episodes of hospitalisation were more likely to die in hospital. This could be because caregivers were unable to
TE D
cope from emotional stress or feeling of burnout. Some patients might even wish to die in hospital to avoid being a burden on their family. The number and type of healthcare personnel visits have shown to be a statistically significant determinant of place of
EP
death. Higher number of home visits by doctors and nurses were associated with higher likelihood of dying at home.
AC C
Means-testing was used as a proxy for determining the socio-economic status and it was found to be statistically significant in bivariate analysis but not in the logistic regression. The result of bivariate analysis showed that middle-income group was more likely to die at home compared to lowest and highest socio-economic status group. The reason behind this finding is probably multi-factorial. Patients from higher socio-
16
ACCEPTED MANUSCRIPT
economic status group had the most financial means to utilise the secondary healthcare
RI PT
facilities in Singapore which involved significant portion of out-of-pocket payment due to government ethos on personal responsibility and deterrence against moral hazards.
On the other hand, out-of-pocket payments might pose a significant financial barrier for
SC
patients from the middle socio-economic status group. On the contrary, financial cost might not be a significant barrier to patients from the lowest socio-economic status
M AN U
group because of the high amount of subsidies they received for their inpatient stay. Similar findings were also observed in China where those who received free medical care were 68% more likely to die in hospital (21). Variables that were found to be not statistically significant after adjustments are distance, mean testing, medical social
TE D
worker visit and caregiver relationship.
In this study, over 40% of patients enrolled in the home hospice service did not have a home death. Importantly, we only examined the actual place of death instead of
EP
preferred place of death. The role of our home hospice program is to support patients’ goals of care. Although home death is often considered the gold standard, a substantial
AC C
minority may prefer to die at a hospital instead of home because of severe symptom distress or high care needs (e.g. agitated delirium) (23). Many patients are concerned about being a burden for their loved ones at home in the last days of life. Reassuringly, we found that the rate of home death has been steadily increasing over time, suggesting that our program may be helping to shift attitudes about home death (24). Further
17
ACCEPTED MANUSCRIPT
research is needed to assess the preferred place of death and how to overcome the
RI PT
barriers to a ‘good death’ in Singapore. Due to the retrospective nature of this study involving a large number of patients, we
were not able to capture some potentially important determinants of home death, such
SC
as patient/caregiver preference, prognostic awareness, level of caregiver support,
insurance coverage, and intensity of home hospice visits. Prospective studies are needed
M AN U
to examine the association between these variables and place of death. There was a significant amount of missing data in the living arrangement variable because it was only recorded in the database during the later years so results related to this variable should be interpreted with care. The distance was computed by using postcode of
TE D
patient’s home and primary referral hospital instead of hospital where patient died as this information was not captured by the database which is more likely to be closer correlated with dying-at-home.
EP
Nevertheless, the strengths of this study are its large sample size of 19,723 and the long
AC C
duration of data collection spanning a period of ten years. Moreover, HCA is the largest home hospice service provider in Singapore which serves all regions in the city-state and this makes the sample in the study nationally representative. It is also the first study in Singapore that examined the effect of caregiver status, living arrangement, mean testing and clinician visits on the place of death of home hospice patients.
18
ACCEPTED MANUSCRIPT
There are several implications can be drawn from this study. Policymakers could
RI PT
increase the manpower in home hospice sector and equip them with appropriate skills and knowledge to optimize their productivity. With higher manpower capacity, doctors and nurses could afford to visit patients more frequently which could translate into
SC
better symptom management and less hospitalisation. Besides, policymaker should
strive to help relatives of home hospice patients who want to be caregivers. Resources
M AN U
should be made available to caregivers especially to enable them to look after their love ones under the same roof as patients who lived with their caregivers had higher chance of dying at home. While helping patients to fulfil their wishes of dying at home, policy makers should also focus on empowering caregivers and looking after their wellbeing.
TE D
Caregivers will be able to perform their job better by learning the right caregiving skills and knowledge. Moreover, it is crucial to provide psychological support to caregivers as
AC C
Conclusion
EP
they are prone to suffer from emotional stress and burnout.
End of life care brings value and hope to patients who are at the last moment of their life. Being able to help these patients to achieve the highest quality of life possible and die at their preferred place should be the goal for all healthcare providers. It is the intention of this study to help achieve that by gaining insights into the different factors
19
ACCEPTED MANUSCRIPT
affecting the place of death of home hospice patients. The evidence provided by this
RI PT
study which was based in an urban, multicultural and Asia country will assist policymakers in making informed decisions and also contribute to the body of
SC
knowledge in hospice care sector.
M AN U
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
TE D
Conflict of interest
EP
The Authors declare that there is no conflict of interest.
AC C
Acknowledgements
The authors wish to thank Ms Lim Hui Ling, IT Manager, HCA Hospice Care for helping with the database analysis.
20
References
World Health Organization. The top 10 causes of death [Internet]. 2016. [cited 8
AC C
(1)
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
July 2016]. Available from: http://www.who.int/mediacentre/factsheets/fs310/en/index2.html (2)
Lorenz K. Evidence for Improving Palliative Care at the End of Life: A
Systematic Review. Annals of Internal Medicine. 2008;148(2):147.
21
ACCEPTED MANUSCRIPT
(3)
Temel J, Greer J, Muzikansky A, et al. Early Palliative Care for Patients with
RI PT
Metastatic Non–Small-Cell Lung Cancer. New England Journal of Medicine. 2010;363(8):733-742. (4)
Brogaard T, Neergaard M, Sokolowski I, Olesen F, Jensen A. Congruence
SC
between preferred and actual place of care and death among Danish cancer patients.
(5)
M AN U
Palliative Medicine. 2012;27(2):155-164.
Gomes B. Factors influencing death at home in terminally ill patients with
cancer: systematic review. BMJ. 2006;332(7540):515-521. (6)
Beng AKL, Fong CW, Shum E, et al. Where the Elderly Die: The Influence of
TE D
Socio-Demographic Factors and Cause of Death on People Dying at Home. Annals of
(7)
EP
Academy Med Singapore. 2009;38:676-83
Hong C, Chow K, Poulose J, et al. Place of Death and Its Determinants for
AC C
Patients with Cancer in Singapore: An Analysis of Data from the Singapore Cancer Registry, 2000–2009. Journal of Palliative Medicine. 2011;14(10):1128-1134. (8)
Lee L, Hu C, Loh E, Hwang S. Factors Affecting the Place of Death Among
Hospice Home Care Cancer Patients in Taiwan. American Journal of Hospice and Palliative Medicine. 2013;31(3):300-306.
22
ACCEPTED MANUSCRIPT
(9)
Gao W, Ho Y, Verne J, Glickman M, Higginson I. Changing Patterns in Place of
RI PT
Cancer Death in England: A Population-Based Study. PLoS Med. 2013;10(3):e1001410. (10)
Barclay J, Kuchibhatla M, Tulsky J, Johnson K. Association of Hospice Patients'
SC
Income and Care Level With Place of Death. JAMA Internal Medicine.
(11)
M AN U
2013;173(6):450.
Gatrell A, Harman J, Francis B, et al. Place of death: analysis of cancer deaths
in part of North West England. Journal of Public Health. 2003;25(1):53-58. (12)
Barros, Aluísio JD, and Vânia N. Hirakata. "Alternatives for logistic regression
TE D
in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio." BMC medical research methodology 3.1 (2003): 1. (13)
O’Brien RM. A caution regarding rules of thumb for variance inflation factors.
StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX:
AC C
(14)
EP
Quality and Quantity Springer. 2007;41:673–690.
StataCorp LP. (15)
Earle CC, Neville BA, Landrum MB et al. Trends in the aggressiveness of
cancer care near the end of life. Journal of Clinical Oncology. 2004 Jan 15;22(2):315321.
23
ACCEPTED MANUSCRIPT
(16)
Tang ST, Wu SC, Hung YN, et al. Determinants of aggressive end-of-life care
RI PT
for Taiwanese cancer decedents, 2001 to 2006. Journal of Clinical Oncology. 2009 Sep 20;27(27):4613-8. (17)
Hui D, Karuturi MS, Tanco KC, et al. Targeted agent use in cancer patients at
Fukui S, Fukui N, Kawagoe H. Predictors of place of death for Japanese
M AN U
(18)
SC
the end of life. Journal of Pain and Symptom Management. 2013 Jul;46(1):1-8.
patients with advanced-stage malignant disease in home care settings: a nationwide survey. Cancer. 2004 Jul 15;101(2):421-9. (19)
Sato K, Hashimoto K, Uchiumi J, et al. Factors influencing death or the
TE D
cessation of palliative care in home-based setting among patients with cancer. Available from: https://www.jstage.jst.go.jp/article/jspm/10/2/10_116/_pdf (20)
Fukui S, Kawagoe H, Masako S, et al. Determinants of the place of death among
EP
terminally ill cancer patients under home hospice care in Japan. Palliative Medicine.
AC C
2003 Jul;17(5):445-53. (21)
Gu D, Liu G, Vlosky D, Yi Z. Factors Associated With Place of Death Among
the Chinese Oldest Old. Journal of Applied Gerontology. 2007;26(1):34-57. (22)
Agency for Integrated Care. Foreign Domestic Worker Grant [Internet]. 2016
[cited 8 July 2016]. Available from: http://www.aic.sg/FDWGrant/
24
ACCEPTED MANUSCRIPT
(23)
Kristian Pollock. Is home always the best and preferred place of death? British
(24)
RI PT
Medical Journal 2015; 351:h485 Ho BJCC, Akhileswaran R, Pang GSY, Koh GCH. An 11-year study of home
AC C
EP
TE D
M AN U
SC
hospice service trends in Singapore from 2000 to 2010. Journal of Palliative Medicine.
25
S0885-3924(17)30056-8
DOI:
10.1016/j.jpainsymman.2017.01.008
Reference:
JPS 9380
To appear in:
Journal of Pain and Symptom Management
Received Date: 25 August 2016 Revised Date:
9 January 2017
Accepted Date: 10 January 2017
Please cite this article as: Lee YS, Akhileswaran R, Ong Eng Hock M, Wah W, Hui D, Ng SH-X, Koh G, Clinical and socio-demographic predictors of home hospice patients dying at home: A retrospective analysis of Hospice Care Association’s database in Singapore, Journal of Pain and Symptom Management (2017), doi: 10.1016/j.jpainsymman.2017.01.008. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Original Article
16-00533R1
Clinical and socio-demographic predictors of home hospice patients dying at home: A retrospective analysis of Hospice Care Association’s database in Singapore
RI PT
Yee Song Lee, MBBS, MPH, Akhileswaran Ramaswamy, MD, FAMS, MBA, Marcus Ong Eng Hock, MBBS, FRCS, MPH, Win Wah, MBBS, MPH, EPH, David Hui, MD, MSc, Sheryl Hui-
SC
Xian Ng, BSc (Hons), and Gerald Koh, MBBS, MMed, FCFP, FAMS, MGer, PhD
National University Health System (NUHS) (Y.S.L.); Khoo Teck Puat Hospital (R.A.);
M AN U
Department of Emergency Medicine (M.O.E.H.), Singapore General Hospital; and Saw Swee Hock School of Public Health (W.W., S.H.-X.N., G.K.), National University of Singapore, Singapore, Singapore; and University of Texas M.D. Anderson Cancer Center (D.H.), Houston,
TE D
Texas, USA
Address correspondence to:
Yee Song Lee, MBBS, MPH
EP
45 Mandalay Road #07-04 Mandale Heights
AC C
Singapore 308225
E-mail: [email protected]
1
ACCEPTED MANUSCRIPT
ABSTRACT Context: Hospice care can be delivered in different settings but many patients choose to receive
RI PT
it at home because of familiar surroundings. Despite their preferences, not every home hospice patient manages to die at home.
Objective: To examine the independent factors associated with home hospice patient dying at
SC
home.
Method: Retrospective analysis of Hospice Care Association (HCA)’s database. HCA is the
M AN U
largest home hospice provider in Singapore. The study included all patients who were admitted into home hospice service from 1 Jan 2004 to 31 Dec 2013. Cox Proportional Hazards modelling with time as constant was used to study the relationship between independent variables and home death.
TE D
Result: 19,721 patients were included in the study. Females (Adjusted Risk Ratio, ARR 1.09, 95% CI 1.04 – 1.15), older patients (ARR 1.01, 95% CI 1.00 - 1.01), shorter duration of home hospice stay (ARR 0.88, 95% CI 0.82 – 0.94), fewer episodes of hospitalisation (ARR 0.81,
EP
95% CI 0.75 – 0.86), living with caregivers (ARR 1.54, 95% CI 1.05 – 2.26), doctor (ARR 1.05, 95% CI 1.01 – 1.08) and nurse (ARR 1.06, 95% CI 1.04 – 1.08) visits were positive
AC C
predictors of dying-at-home. Diagnosis of cancer (ARR 0.93, 95% CI 0.86 – 1.00) was a negative predictor of dying-at-home. Conclusion: Female, older age, living with a caregiver, non-cancer diagnosis, more doctor and nurse visits, shorter duration of home hospice stays and fewer episodes of acute hospitalisations are predictive of dying-at-home for home hospice patients.
2
ACCEPTED MANUSCRIPT
Keywords Palliative care; hospice care; home care; home hospice; neoplasm; cancer; place of death
RI PT
Running head: Predictors of Home Death for Home Hospice
AC C
EP
TE D
M AN U
SC
Accepted for publication: January 10, 2017
3
ACCEPTED MANUSCRIPT
Introduction Palliative care has increased in prominence in recent decades and has now become an integral
RI PT
part of many health service systems around the world. The increase in demand for palliative care is significant, especially in countries with profound demographic shift such as an ageing
population. In addition, the global change in disease patterns from communicable to noncommunicable diseases has contributed to the rise in demand for palliative care (1). The
SC
evidence on the benefits of palliative care is strong: several studies have shown that palliative
M AN U
care improves patient symptoms, quality of life, patient and family satisfaction and possibly the survival of patients (2, 3). Patients with a life-limiting illness can access palliative care for symptom relief even if they are still receiving curative treatment. However, for patients who have a life expectancy of 1 year or less and curative treatment is no longer indicated, palliative care is the most appropriate treatment to keep them comfortable and die in their preferred place
TE D
of death. This type of palliative care is known as hospice care and it can be delivered in different settings such as an in-patient hospice, a day hospice and a patients’ home. Patients’ preferred place of death is normally one of the major factors that decide the place of hospice care. In
EP
reality, despite the wishes of dying at home, a large proportion of home hospice patients still die
AC C
in hospitals and in certain cases, utilise a hospital bed inappropriately (4). A systematic review by Gomes et al on studies from United States, United Kingdom, Canada and Australia that included more than 1 million participants found that the determinants associated with dying at home were low functional status, living with relatives, having extended family support, preference to die at home and receiving home care (5). The result of this systematic review had illustrated the importance of patient living status, caregiver status and home hospice care. 4
ACCEPTED MANUSCRIPT
The provision of palliative care is intricately linked to the sociocultural environment of a society. In Singapore, a multi-ethnic Asian country with diverse cultural background, the need for palliative care research in order to provide culturally-appropriate palliative services could not
RI PT
be greater. There were two studies published in Singapore that examined the determinants of place of death for the general population and cancer patients. According to Beng et al, a
retrospective database analysis of Singapore Registry of Birth and Death, female, Malay ethnic
SC
group and higher age group were more likely to die at home (6). The study had also shown that elderly who died of stroke were more likely to die at home compared to heart disease, cancer
M AN U
and respiratory disease. Another retrospective database analysis study by Hong et al had also yielded similar results (7). This study analysed the Singapore Cancer Registry data and showed that older and female patients were more likely to die at home.
Among demographic factors, the findings in Singapore were consistent with other countries.
TE D
Age and gender were major determinants with female and older patients more likely to die at home based on studies conducted in Taiwan, England and United States (8-10). Another determinant of place of death is socio-economic status and a country's healthcare financing
EP
model. Some studies have shown that high economic status was associated with higher
AC C
propensity of dying at home (10, 11) while others had shown the contrary (8, 9). A study has also shown that geographical location influenced the place of death of home hospice patients, especially in larger countries with an urban-suburban divide (11). Other studies have shown that patients who had shorter length of stay in home hospice service were more likely to die at home (10). Both Singapore studies were retrospective database analysis of national registries. However, these studies did not examine the significance of socio-economic status, length of stay in home 5
ACCEPTED MANUSCRIPT
hospice service, caregiver status, living status, home medical support and distance between patient’s home and hospital. Hence, it is the intention of this research to address this knowledge gap and provide new evidence for policy making by studying the independent factors associated
RI PT
with home hospice patient dying at home in Singapore, a multi-ethnic Asian country.
SC
Methodology
We retrospectively analysed a database of patients who were admitted to HCA Hospice Care
M AN U
(HCA) from 1st January 2004 to 31st December 2013. HCA Hospice Care is a registered nonprofit organization in Singapore which provides home hospice care and operates a day hospice centre. HCA provides home hospice service to any patient who has a life limiting illness with a life expectancy of less than 12 months regardless of age, religion, ethnicity, nationality and
TE D
financial status. HCA is also the largest local home hospice provider in Singapore; providing 60% of all home hospice service in Singapore with services spanning across the island. Independent variables were chosen based on literature review, clinical reasoning and the
EP
information available in the database. Patient demographic profile, clinical diagnosis, living
AC C
status, caregiver relationship, length of stay in home hospice service, hospital utilisation, mean testing, hospice staff visits and place of death were selected as independent variables. Place of death was selected as the dependent variable and was categorised as home versus non-home. For patients with repeated admissions, the episodes of hospital admissions were summed up together to produce a total number of hospitalisation. Patient living status was categorised as living alone, living with a caregiver or living in an institution. HCA looks after a small number of patients who lived in institutions such as nursing homes that did not provide hospice care. This group of
6
ACCEPTED MANUSCRIPT
patients received the standard end of life care service from HCA as those who lived at home. Those who died at the original institutions were categorised as home deaths.
RI PT
Means-testing which is routinely performed in HCA was used as a proxy for economic status and it was categorised as 0%, >0% to 50% and >50%. 0% indicates no subsidy (highest socioeconomic status) and >50% indicate the highest subsidy (lowest socio-economic status).
Caregiver relationship was categorised as spouse, non-spouse relative, unrelated and none. The
SC
relationship between place of death and distance from patient home to primary referral hospital
M AN U
was also studied. The distance was generated through a software system known as Network Analyst Extension in ArcMap 10.1 (ESRI, USA) by calculating the travel distance along the roadways from the postal codes of patients’ home to the primary referral hospital. Detailed recording of symptom control in the database was only initiated in recent years so this variable was not used in the analysis. Cause of death was classified according to the International
TE D
Classification of Disease, version 10 (ICD-10) and it was dichotomised into non-cancer and cancer groups. Data on length of stay in hospice services was found to be skewed with a wide range from 1 to 3978 and hence, was log-transformed in the analysis. The research protocol was
EP
approved by the Institutional Review Board (IRB) of National University of Singapore.
AC C
Patient characteristics were described in frequencies for categorical variables, while continuous variables were reported in mean and standard deviation. Bivariate and multivariate analyses were conducted using Cox Proportional Hazards modelling with time as a constant to study the relationship between the independent variables and place of death (12). Risk ratios were reported using robust standard errors. Presence of collinearity between independent variables in the final model was tested by examining variance inflation factors (13). The conventional
7
ACCEPTED MANUSCRIPT
P<0.05 was used as the cut-off for statistical significance. All statistical analyses were
Results
RI PT
performed using SPSS version 22 and STATA 13 (14).
There were 19,721 unique patients who were registered with the HCA during the 10 year period
SC
died either at home or in hospitals or in-patient hospices.
As shown in Table 1, there were 10,858 (55.0%) patients died at home, 8,335 (42.3%) patients
M AN U
died in hospitals, 397 (2.0%) patients died in in-patient hospices and 131 (0.7%) died in nursing home. The age of patients ranged from 12 to 105 years with mean of 71.0 years. The gender of patients was quite evenly distributed with the number of female patients at 9,629 (48.8%). Home hospice patients had an average of 0.85 hospital admissions and the mean distance between
TE D
hospital and patients’ home was 10.9 kilometres. Patients registered with HCA receive home visits from doctors and nurses but not necessarily from medical social workers unless there were unresolved social issues. Nurses formed the highest number of hospice staff visits with mean of
EP
3.22 visits per patient whereas medical social worker formed the least number of visits with mean of 0.08 visits per patient. Majority of the patients in this study had cancer (95.6%) and
AC C
lived at home with their caregivers (97.2%). Similarly, most of the patients were related to their caregivers, spouses (65.8%) and non-spouse relatives (33.3%), while only a small proportion of patients were looked after by unrelated caregiver (0.7%) or did not have any caregiver (0.1%). In term of socio-economic status, 28.5% received >50% subsidies, 26.0% received >0 to 50% subsidies and 45.5% received no subsidy.
8
ACCEPTED MANUSCRIPT
Table 2 showed the unadjusted and adjusted risk ratio of the independent variables. Under the unadjusted model, females have a higher likelihood of home death than males (Crude Risk Ratio, CRR=1.12, 95%CI: 1.09–1.15). Higher age group (CRR=1.01, 95%CI: 1.01–1.01),
RI PT
shorter duration of hospice service (CRR=0.94, 95%CI: 0.92–0.96) and fewer episodes of
hospitalisation (CRR=0.83, 95%CI: 0.81–0.85) were associated with higher chance of dying at home. Patients who received more doctor (CRR=1.16, 95%CI: 1.15–1.17) and nurse visits
SC
(CRR=1.08, 95%CI: 1.08–1.09) were also more likely to die at home. However, more frequent medical social worker visits were associated with lower chance of dying at home (CRR=0.95,
M AN U
95%CI: 0.92–0.99). The bivariate analysis also demonstrated that patients who lived further away from hospital, cared for by spouses, living with caregivers and had non-cancer diagnosis were more likely to die at home. In term of socio-economic status, patients from the middle means-testing group (>0% to 50%) were more likely to die at home compared to patients from
TE D
highest (0%) and lowest means-testing group (>50%).
Under the adjusted model, the independent determinants of death at home for demographic characteristics were female gender (Adjusted Risk Ratio, ARR=1.09, 95%CI: 1.04–1.15) and
EP
older age (ARR=1.01, 95%CI: 1.00-1.01). From a health services perspective, length of stay in
AC C
home hospice service (ARR=0.88, 95%CI: 0.82–0.94), home visit from healthcare personnel such as doctor (AOR=1.05, 95%CI: 1.01–1.08) and nurse (AOR=1.06, 95%CI: 1.04–1.08) were independent predictors of dying at home. Having cancer (ARR=0.93, 95%CI: 0.86–1.00) decreased the likelihood of dying at home whereas living at home with caregivers (ARR=1.54, 95%CI: 1.05–2.26) increased the likelihood of dying at home. Examination of VIF revealed no significant collinearity between the independent variables.
9
ACCEPTED MANUSCRIPT
Table 1. Profile of study population (N=19,721). N (%)
Place of death At home Hospital In-patient hospice Nursing home
10,858 (55.0) 8,335 (42.3) 397 (2.0) 131 (0.7)
RI PT
Variables
Gender Male Female
SC
10,090 (51.2) 9,629 (48.8)
71.00 (13.34)
38 (1-3978, 14-100)
Log-transformed length of stay in service, days (Mean, SD)
1.57 (0.63)
Number of hospitalisation episodes(Mean, SD)
0.85 (1.55)
Number of doctor visits (Mean, SD)
1.12 (1.08)
Number of nurse visits (Mean, SD)
3.22 (2.37)
Number of medical social worker visits (Mean, SD)
0.08 (0.45)
Distance from hospital, kilometre (Mean, SD)
10.9 (5.71)
Diagnosis Non-cancer Cancer
650 (4.4) 14,194 (95.6)
AC C
EP
Length of stay in home hospice service in days Median (minimum-maximum, 25th – 75th interquartile range )
TE D
M AN U
Age in years (Mean, SD)
Caregiver status Spouse Non-spouse relative Not related None
12,412 (65.8) 6,279 (33.3) 135 (0.7) 26 (0.1)
Living arrangement Alone
87 (1.7) 10
ACCEPTED MANUSCRIPT
5,034 (97.2) 58 (1.1)
Mean-tested subsidy level 0% >0% to 50% >50%
7,967 (45.5) 4,549 (26.0) 4,987 (28.5)
AC C
EP
TE D
M AN U
SC
RI PT
At home with caregivers Institution
11
ACCEPTED MANUSCRIPT
RI PT
Table 2. Unadjusted and adjusted risk ratio relating the independent variables to place of death for patients registered under Hospice Care Association (HCA) from 1st January 2004 to 31st December 2013.
Unadjusted Model Crude RR 95% CI
Gender, n (%) Male Female
5,252 (52.1) 5,605 (58.2)
4,838 (47.9) 4,024 (41.8)
1.00 1.12
Mean age in years (SD)
72.76 (13.05)
Mean of log-transformed length of stay in the service in days (SD)
1.54 (0.66)
68.63 (13.37) 1.60 (0.60)
P-value
SC
Variables
Place of Death Home Non-home
M AN U
1.09 - 1.15 <0.001
Adjusted Model Adjusted 95% CI RR
P-value
1.00 1.09
1.04 - 1.15 <0.001
1.01 - 1.01 <0.001
1.01
1.00 - 1.01 <0.001
0.94
0.92 - 0.96 <0.001
0.88
0.82 - 0.94 <0.001
0.83
0.81 - 0.85 <0.001
0.81
0.75 - 0.86 <0.001
1.16
1.15 - 1.17 <0.001
1.05
1.01 - 1.08 0.003
TE D
1.01
1.18 (1.73)
Mean number of doctor visits (SD)
1.32 (1.14)
0.89 (0.94)
Mean number of nurse visits (SD)
3.68 (2.42)
2.65 (2.18)
1.08
1.08 - 1.09 <0.001
1.06
1.04 - 1.08 <0.001
Mean number of medical social worker visits (SD)
0.07 (0.42)
0.09 (0.48)
0.95
0.92 - 0.99 0.005
0.97
0.93 - 1.01 0.092
AC C
EP
Mean number of episodes of 0.58 (1.32) hospitalisation (SD)
12
ACCEPTED MANUSCRIPT
10.80 (5.75)
1.00
1.00 - 1.00 0.009
Caregiver status, n (%) Spouse Non-spouse Not related None
7,902 (60) 3,220 (51.3) 42 (31.1) 9 (34.6)
5,320 (40) 3,059 (48.7) 93 (68.9) 17 (65.4)
1.00 0.90 0.54 0.61
<0.001 0.87 - 0.92 <0.001 0.42 - 0.70 <0.001 0.36 - 1.03 0.063
Diagnosis, n (%) Non-cancer Cancer
445 (68.5) 7,676 (54.1)
205 (31.5) 6,518 (55.9)
Living arrangement, n (%) Alone Home with caregiver Institution
25 (28.7) 3,089 (61.4) 3 (5.2)
62 (71.3) 1,945 (38.6) 55 (94.8)
Mean-tested subsidy level, n (%) 0% >0% to 50% >50%
4,078 (51.2) 2,779 (61.1) 2,736 (54.9)
3,889 (48.8) 1,770 (48.9) 2,251 (55.1)
SC
M AN U
1.00
1.00 - 1.00 0.866
1.00 0.98 0.84 1.13
0.870 0.93 - 1.04 0.548 0.48 - 1.47 0.546 0.15 - 8.66 0.907
0.86 - 1.00 0.042
1.00 0.79
0.75 - 0.83 <0.001
1.00 0.93
1.00 2.14 0.18
<0.001 1.53 - 2.98 <0.001 0.06 - 0.57 0.003
1.00 1.54 0.11
<0.001 1.05 - 2.26 0.026 0.03 - 0.45 0.002
1.00 1.19 1.07
<0.001 1.16 - 1.23 <0.001 1.04 - 1.11 <0.001
1.00 1.03 0.99
0.503 0.97 - 1.10 0.334 0.94 - 1.05 0.861
TE D
EP
AC C
RI PT
11.02 (5.67)
Mean distance from hospital, kilometre (SD)
13
ACCEPTED MANUSCRIPT
Discussion
RI PT
This is the first study in Singapore that examined the determinants of place of death among home hospice patients. We found that female gender and age were associated with home death. This is consistent with multiple studies demonstrating that male
SC
patients are often treated more aggressively at the end-of-life, and are often more likely to die in hospitals (15, 16, 17). However, other investigators did not find gender to be a
M AN U
key determinant of home death (18, 19, 20). This discrepancy may be related to differences in patient characteristics, healthcare system and/or cultural factors. Apart from that, living arrangement was found to be an important determinant of place of death. This finding is consistent with a large systematic review conducted by Gomes
TE D
et al which showed that home hospice patients who lived with their caregiver were more likely to die at home (5). In addition, a study conducted in China has concluded that living arrangement as a determinant of place of death is not just limited to living in
EP
the same household but also extends to those who live around the home. The China
AC C
study has shown that having a caregiver living nearby decreases the risk of dying in hospital by 74% among elderly home hospice patients (21). Caregiver status was shown to be a significant determinant of place of death for home hospice patients in bivariate analysis (i.e. home hospice patients who had spouse or nonspouse relative caregivers were more likely to die at home compared to non-related
14
ACCEPTED MANUSCRIPT
caregivers), but not on multivariate analysis. However, it is important to note that
RI PT
caregiver is not a prerequisite for the enrolment into HCA home hospice services. Although Singapore has seen rapid economic development and westernisation in recent decades, Asian values are still deeply entrenched in the communities. Asian values put
SC
strong emphasis on the responsibility of the children to show filial piety and care for
their elderly parents. These value and belief systems were reflected in our study where
M AN U
majority of the home hospice patients had caregivers that comprised mainly of their spouses or relatives. In addition, there is a large population of foreign workers in Singapore who work as domestic maid who sometimes also function as caregivers. The Singapore government not only provides financial subsidy for patients or family
TE D
members to hire foreign domestic workers (FDWs) but also arranges caregiving courses for FDWs to attend (22).
Principal diagnosis for admission into home hospice care service was also found to be a
EP
determinant of place of death and non-cancer patients were more likely to die at home. Nevertheless, this result is not consistent with the two previous similar studies done in
AC C
Singapore where patients with cancer were more likely to die at home (6, 7). One possible explanation could be the smaller sample size of non-cancer patients in our study (4.4%) compared to 95.6% being cancer patients. Another factor could be a selection bias by the referring doctor, i.e., only non-cancer patients who were deemed to be fit to be cared for at home were referred to HCA. Other determinants of place of
15
ACCEPTED MANUSCRIPT
death were length of stay in home hospice service and number of hospitalisations prior
RI PT
to death. It is uncertain if the majority of home hospice patients included in this study were referred late to the home hospice service, i.e., the patients were already very sick when they were referred, which could have contributed to the shorter duration of stay in
SC
the home hospice service. Some of the patients were also “terminal discharges”, i.e.,
patients discharged just before they died at home. Hence, future research should be done
M AN U
to identify the impact of ‘terminal discharges’ and also the condition of the patient at the time of referral to the service by studying the reasons for admission into the home hospice service. On the other hand, patients who had more episodes of hospitalisation were more likely to die in hospital. This could be because caregivers were unable to
TE D
cope from emotional stress or feeling of burnout. Some patients might even wish to die in hospital to avoid being a burden on their family. The number and type of healthcare personnel visits have shown to be a statistically significant determinant of place of
EP
death. Higher number of home visits by doctors and nurses were associated with higher likelihood of dying at home.
AC C
Means-testing was used as a proxy for determining the socio-economic status and it was found to be statistically significant in bivariate analysis but not in the logistic regression. The result of bivariate analysis showed that middle-income group was more likely to die at home compared to lowest and highest socio-economic status group. The reason behind this finding is probably multi-factorial. Patients from higher socio-
16
ACCEPTED MANUSCRIPT
economic status group had the most financial means to utilise the secondary healthcare
RI PT
facilities in Singapore which involved significant portion of out-of-pocket payment due to government ethos on personal responsibility and deterrence against moral hazards.
On the other hand, out-of-pocket payments might pose a significant financial barrier for
SC
patients from the middle socio-economic status group. On the contrary, financial cost might not be a significant barrier to patients from the lowest socio-economic status
M AN U
group because of the high amount of subsidies they received for their inpatient stay. Similar findings were also observed in China where those who received free medical care were 68% more likely to die in hospital (21). Variables that were found to be not statistically significant after adjustments are distance, mean testing, medical social
TE D
worker visit and caregiver relationship.
In this study, over 40% of patients enrolled in the home hospice service did not have a home death. Importantly, we only examined the actual place of death instead of
EP
preferred place of death. The role of our home hospice program is to support patients’ goals of care. Although home death is often considered the gold standard, a substantial
AC C
minority may prefer to die at a hospital instead of home because of severe symptom distress or high care needs (e.g. agitated delirium) (23). Many patients are concerned about being a burden for their loved ones at home in the last days of life. Reassuringly, we found that the rate of home death has been steadily increasing over time, suggesting that our program may be helping to shift attitudes about home death (24). Further
17
ACCEPTED MANUSCRIPT
research is needed to assess the preferred place of death and how to overcome the
RI PT
barriers to a ‘good death’ in Singapore. Due to the retrospective nature of this study involving a large number of patients, we
were not able to capture some potentially important determinants of home death, such
SC
as patient/caregiver preference, prognostic awareness, level of caregiver support,
insurance coverage, and intensity of home hospice visits. Prospective studies are needed
M AN U
to examine the association between these variables and place of death. There was a significant amount of missing data in the living arrangement variable because it was only recorded in the database during the later years so results related to this variable should be interpreted with care. The distance was computed by using postcode of
TE D
patient’s home and primary referral hospital instead of hospital where patient died as this information was not captured by the database which is more likely to be closer correlated with dying-at-home.
EP
Nevertheless, the strengths of this study are its large sample size of 19,723 and the long
AC C
duration of data collection spanning a period of ten years. Moreover, HCA is the largest home hospice service provider in Singapore which serves all regions in the city-state and this makes the sample in the study nationally representative. It is also the first study in Singapore that examined the effect of caregiver status, living arrangement, mean testing and clinician visits on the place of death of home hospice patients.
18
ACCEPTED MANUSCRIPT
There are several implications can be drawn from this study. Policymakers could
RI PT
increase the manpower in home hospice sector and equip them with appropriate skills and knowledge to optimize their productivity. With higher manpower capacity, doctors and nurses could afford to visit patients more frequently which could translate into
SC
better symptom management and less hospitalisation. Besides, policymaker should
strive to help relatives of home hospice patients who want to be caregivers. Resources
M AN U
should be made available to caregivers especially to enable them to look after their love ones under the same roof as patients who lived with their caregivers had higher chance of dying at home. While helping patients to fulfil their wishes of dying at home, policy makers should also focus on empowering caregivers and looking after their wellbeing.
TE D
Caregivers will be able to perform their job better by learning the right caregiving skills and knowledge. Moreover, it is crucial to provide psychological support to caregivers as
AC C
Conclusion
EP
they are prone to suffer from emotional stress and burnout.
End of life care brings value and hope to patients who are at the last moment of their life. Being able to help these patients to achieve the highest quality of life possible and die at their preferred place should be the goal for all healthcare providers. It is the intention of this study to help achieve that by gaining insights into the different factors
19
ACCEPTED MANUSCRIPT
affecting the place of death of home hospice patients. The evidence provided by this
RI PT
study which was based in an urban, multicultural and Asia country will assist policymakers in making informed decisions and also contribute to the body of
SC
knowledge in hospice care sector.
M AN U
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
TE D
Conflict of interest
EP
The Authors declare that there is no conflict of interest.
AC C
Acknowledgements
The authors wish to thank Ms Lim Hui Ling, IT Manager, HCA Hospice Care for helping with the database analysis.
20
References
World Health Organization. The top 10 causes of death [Internet]. 2016. [cited 8
AC C
(1)
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
July 2016]. Available from: http://www.who.int/mediacentre/factsheets/fs310/en/index2.html (2)
Lorenz K. Evidence for Improving Palliative Care at the End of Life: A
Systematic Review. Annals of Internal Medicine. 2008;148(2):147.
21
ACCEPTED MANUSCRIPT
(3)
Temel J, Greer J, Muzikansky A, et al. Early Palliative Care for Patients with
RI PT
Metastatic Non–Small-Cell Lung Cancer. New England Journal of Medicine. 2010;363(8):733-742. (4)
Brogaard T, Neergaard M, Sokolowski I, Olesen F, Jensen A. Congruence
SC
between preferred and actual place of care and death among Danish cancer patients.
(5)
M AN U
Palliative Medicine. 2012;27(2):155-164.
Gomes B. Factors influencing death at home in terminally ill patients with
cancer: systematic review. BMJ. 2006;332(7540):515-521. (6)
Beng AKL, Fong CW, Shum E, et al. Where the Elderly Die: The Influence of
TE D
Socio-Demographic Factors and Cause of Death on People Dying at Home. Annals of
(7)
EP
Academy Med Singapore. 2009;38:676-83
Hong C, Chow K, Poulose J, et al. Place of Death and Its Determinants for
AC C
Patients with Cancer in Singapore: An Analysis of Data from the Singapore Cancer Registry, 2000–2009. Journal of Palliative Medicine. 2011;14(10):1128-1134. (8)
Lee L, Hu C, Loh E, Hwang S. Factors Affecting the Place of Death Among
Hospice Home Care Cancer Patients in Taiwan. American Journal of Hospice and Palliative Medicine. 2013;31(3):300-306.
22
ACCEPTED MANUSCRIPT
(9)
Gao W, Ho Y, Verne J, Glickman M, Higginson I. Changing Patterns in Place of
RI PT
Cancer Death in England: A Population-Based Study. PLoS Med. 2013;10(3):e1001410. (10)
Barclay J, Kuchibhatla M, Tulsky J, Johnson K. Association of Hospice Patients'
SC
Income and Care Level With Place of Death. JAMA Internal Medicine.
(11)
M AN U
2013;173(6):450.
Gatrell A, Harman J, Francis B, et al. Place of death: analysis of cancer deaths
in part of North West England. Journal of Public Health. 2003;25(1):53-58. (12)
Barros, Aluísio JD, and Vânia N. Hirakata. "Alternatives for logistic regression
TE D
in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio." BMC medical research methodology 3.1 (2003): 1. (13)
O’Brien RM. A caution regarding rules of thumb for variance inflation factors.
StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX:
AC C
(14)
EP
Quality and Quantity Springer. 2007;41:673–690.
StataCorp LP. (15)
Earle CC, Neville BA, Landrum MB et al. Trends in the aggressiveness of
cancer care near the end of life. Journal of Clinical Oncology. 2004 Jan 15;22(2):315321.
23
ACCEPTED MANUSCRIPT
(16)
Tang ST, Wu SC, Hung YN, et al. Determinants of aggressive end-of-life care
RI PT
for Taiwanese cancer decedents, 2001 to 2006. Journal of Clinical Oncology. 2009 Sep 20;27(27):4613-8. (17)
Hui D, Karuturi MS, Tanco KC, et al. Targeted agent use in cancer patients at
Fukui S, Fukui N, Kawagoe H. Predictors of place of death for Japanese
M AN U
(18)
SC
the end of life. Journal of Pain and Symptom Management. 2013 Jul;46(1):1-8.
patients with advanced-stage malignant disease in home care settings: a nationwide survey. Cancer. 2004 Jul 15;101(2):421-9. (19)
Sato K, Hashimoto K, Uchiumi J, et al. Factors influencing death or the
TE D
cessation of palliative care in home-based setting among patients with cancer. Available from: https://www.jstage.jst.go.jp/article/jspm/10/2/10_116/_pdf (20)
Fukui S, Kawagoe H, Masako S, et al. Determinants of the place of death among
EP
terminally ill cancer patients under home hospice care in Japan. Palliative Medicine.
AC C
2003 Jul;17(5):445-53. (21)
Gu D, Liu G, Vlosky D, Yi Z. Factors Associated With Place of Death Among
the Chinese Oldest Old. Journal of Applied Gerontology. 2007;26(1):34-57. (22)
Agency for Integrated Care. Foreign Domestic Worker Grant [Internet]. 2016
[cited 8 July 2016]. Available from: http://www.aic.sg/FDWGrant/
24
ACCEPTED MANUSCRIPT
(23)
Kristian Pollock. Is home always the best and preferred place of death? British
(24)
RI PT
Medical Journal 2015; 351:h485 Ho BJCC, Akhileswaran R, Pang GSY, Koh GCH. An 11-year study of home
AC C
EP
TE D
M AN U
SC
hospice service trends in Singapore from 2000 to 2010. Journal of Palliative Medicine.
25