Place of Death and Its Determinants for Patients with Cancer in Singapore: An Analysis of Data from the Singapore Cancer Registry,2000–2009

Introduction

Place of death is a much-studied topic in recent years. This is because knowing where people die and what determines the place of death have implications for health care planners and service providers, as well as for patients and their families. Place of death, especially death at home, has been used as an indicator of whether a patient had received appropriate care at the end of life. Death in a hospice has also been used as a proxy for indication of access to and utilization of palliative care services. ¹ For patients with terminal cancer, dying at home is the expressed choice of many, and is viewed as being able to achieve “good death.” ² For patients who need more complex care near the end of life, who could not die at home for various reasons, dying in in-patient hospices would be an acceptable alternative.

In population-level studies on place of death among patients with cancer, between 12.8% and 45.4% of deaths had occurred at home, and between 3.8% and 20.0% had been in hospices.^1,3–12 As for Asian populations, studies in Taiwan^14,14 reported deaths at home to be between 32.4% and 60.0%, and in Singapore, Beng and colleagues reported that among patients aged≥65 years who had died of cancer, 35.8% of the deaths had been at home, and 12.4% in a “hospice and licensed sick receiving home.” ¹⁵

As for determinants of place of death, these could be grouped into individual factors (demographic and personal variables), factors related to illness (type of cancer, length of disease, etc.), and environmental factors (type of health care services available, social support, etc.), according to Gomes and Higginson's comprehensive review. ¹² Among the individual factors, age was a major determinant. Some studies had reported that older patients were more likely to die at home,^1,6–8 whereas others had reported the reverse.^9,13 Similarly, for gender, some studies had reported that females were more likely to die at home,^4,13 whereas other had reported the opposite. ¹ As for ethnicity, there were not many reports other than on decedents of “white” or “black” descent in the United States, where whites were more likely to die at home, and blacks more likely to die in a hospital.^4,11 In Singapore, determinants of death at home for patients with cancer were older age (≥75 years compared with 65–74 years), being female, and belonging to the Malay ethnic group. ¹⁵

Singapore is an island nation with a total population of 5.08 million, of which 3.77 million are “residents” (Singpore citizens and permanent residents). Its residents comprise an ethnic mix of 74.2% Chinese, 13.4% Malay, and 9.2% Indian. ¹⁶ Culturally speaking, as a nation that has become “developed” rapidly in the past few decades, Singapore has an interesting combination of Western culture being embraced by a large section of the younger generation, while the older population still retains much of the traditional Asian social norms and values. Cancer is the top cause of death in Singapore. In the period 2003 to 2007 inclusively, deaths due to cancer comprised 27.1% of deaths from all causes, and the mean annual number of cancer deaths was 4,432.0. ¹⁷

Besides the study by Beng and colleagues, not much is known, at a national level, about where patients with cancer had died, and what factors determined location. Information on place of death will contribute toward closing the knowledge gap for Singaporean policy makers regarding the planning of health care services for patients with terminal cancer, to enable them to be cared for and die in the community.

Studies on place of death using death certificate or population-based data have been conducted in Europe,^1,9,18–20 the United States,^4,21 Australia, ⁵ and Asia (Taiwan).^13,14 In Singapore, the Singapore Cancer Registry (SCR) is a population-based registry that began its operations in 1968. It holds the records of all Singapore residents (citizens and permanent residents) diagnosed with cancer in the country. Sources of data include pathology records, hospital records, notification from the medical professions, and death records. Cancer registration is comprehensive because all cases diagnosed by histology and all cases with diagnoses of cancer in the hospital records and death records are included.

This population-level study therefore aimed to examine the places of death among patients with cancer in Singapore, and the relationship between demographic characteristics and disease-related factors with death at home and death in in-patient hospices.

Methods

This was a cross-sectional analysis of all patients with cancer in the SCR who had died from year 2000 to 2009 inclusively. The diagnosis of cancer (and therefore the inclusion in the registry) could occur before year 2000. Patients with invasive cancer were included, whereas patients with two or more concurrent primary cancers were excluded (there were 77 such cases). For patients with second or third primary sites, only the first site was listed for the purposes of this analysis.

Cancer cases identified from the sources mentioned above were checked against registered cases in a computerized system using national identifiers that are unique to each Singapore resident. For new cases that were identified in this manner, trained registry coordinators subsequently completed the registration process by viewing patient medical records to extract additional information required by the registry, and verifying information that had been obtained through the various data sources.

The registry captures patient demographic characteristics, clinical diagnosis such as cancer histology, cancer staging, treatment information, and outcome of patients. For this study, variables could be grouped into two of the three categories as classified by Gomes and Higginson. ¹² For “individual factors” the variables studied were age at death, gender, and ethnic group. For “factors related to illness,” the variables studied were primary site of tumor, stage at diagnosis, duration between diagnosis and death, primary cause of death, and place of death. Place of death was categorized as (private) “home,” “in-patient hospice,” “hospital,” and “others,” which included community hospitals and nursing homes. The SCR did not contain information on “environmental factors” such as the availability of health care services and social support. Information on utilization of health care services such as hospice services was also not captured by the registry.

Data analysis was performed using the statistical software SPSS version 17 (SPSS Inc., Chicago, IL). χ² tests were used to analyse associations between categorical variables, while independent t tests were used for continuous variables. P values reported were two-sided assuming 5% level of significance. Multivariate logistic regressions were used to generate odds ratios (OR) and 95% confidence intervals (95% CI) for factors associated with death at home and in in-patient hospices compared with death in hospitals. The covariates adjusted for in the models were age group, gender, ethnic group, primary tumor site, stage at diagnosis, duration between diagnosis and death, and cause of death.

Results

Between 2000 and 2009, there were 52,120 deaths among patients with cancer registered with the SCR. The age at death of these patients ranged from 0 to 109 years, with the mean and median at 69 and 71 years, respectively. Six of 10 deaths (65.2%) were of individuals aged 65 years and older. Hospital was the most common place of death for about half the patients (52.9%). Three in 10 patients died at home (30.3%), and one in 10 died in an in-patient hospice (10.7%). In two of 10 patients with cancer (18.8%), “neoplasms” was not the primary cause of death. Causes of death included pneumonia, myocardial infarction, and cerebrovascular disease, among others. A descriptive profile of all deaths among patients with cancer is shown in Table 1.

The characteristics of patients with cancer by places of death are provided in Table 2. The proportion of patients who died at home increased with age. More females died at home compared with males (32.6% vs. 28.5%). More Malay patients died at home (47.3%) compared with patients of other ethnic groups (Chinese 28.7%, Indians 27.3%), whereas more Chinese patients died in in-patient hospices (11.6% compared with 3.5% for Malays and 8.0% for Indians).

Among patients who died in in-patient hospices, a higher proportion had distant metastasis (14.6%) compared with regional metastasis (11.3%) and with localized tumors (8.0%). In addition, a higher proportion of patients with cancer whose primary cause of death was “neoplasms” had died in in-patient hospices (13.0% compared with 1.1% that were due to other causes).

Discussion

This study is the first in Singapore to examine the profile of all patients with cancer who had died in the study period (2000–2009), with respect to the place of death, and the determinants of deaths at home compared with deaths in hospitals. The data from the SCR also allowed us to examine determinants of deaths in in-patient hospices, an indicator for use of hospice services.

Singapore is a small nation with a geographical area of 710.3 km². ²² There is no urban-rural differentiation, and the nearest hospital is at most half an hour away from any residential home. Hence, differences in terms of urban/rural location of death^1,5,13 and in areas with poorer access to health care facilities would not be expected.^6,20

What is present in Singapore, as in any multiethnic and multicultural society, is the ethnically different social and cultural influences on health-seeking behaviors and preferences for medical care. Our finding that, compared with the Chinese, Malays were more likely to die at home and less likely to die in in-patient hospices could possibly be due to sociocultural (including religious) differences between the Malays and other ethnic groups.

In Singapore, 99.6% of Malays are Muslims, and the tenets of Islam have strong influence on the day-to-day lives of the Malays. In a paper entitled “Palliative care for Muslim patients” published in a 2005 issue of the Journal of Supportive Oncology, the authors explained that although Islamic teachings encourage Muslims to seek treatment when they fall sick, the perception of suffering as a way of atonement for one's sins helps patients and families cope with serious and life-limiting illness. ²³ This could possibly explain why Malays were more likely to die at home.

In addition to religious influences promoting greater community cohesiveness, ²³ Malays in Singapore may also have better social and caregiver support because they have larger family units, as indicated by household sizes. According to the Singapore Census of Population 2000, Malay households have an average of 4.2 persons, compared with 3.6 for Chinese and 3.7 for Indians. ²⁴ This could contribute to more caregivers being available to look after Malay patients with cancer in the terminal phase of their illness, thereby facilitating end-of-life care at home.

In addition to Malay ethnic group, our study found that increasing age was also an independent predictor of death at home and in an in-patient hospice, when compared with death in a hospital. This was similar to the findings among residents in New York, but contrasted with those among residents in London. ¹⁰ The authors of that study postulated that the divergent relationship between the association of age and death at home in the two cities could be due to differences in “environmental factors” such as health care financing system and the availability of health and social services. In Singapore, a possible explanation could be the preference for place of death among the elderly. Beng and colleagues reported that 83% of patients with cancer aged 65 years and above preferred to die at home, compared with 50% of those younger than 65 years of age (relative risk 1.67, CI 1.052.64). The reasons cited were better emotional and physical support or “the belief that elderly patients would generally like to be at home when they die.” ¹⁵

In this study, we found that females were more likely to die at home. This was similar to reports from the United States (Connecticut) ⁴ and Taiwan ¹³ but different from five of the six European countries in Cohen's study (except for Italy). ¹ Even though Singapore is “Westernized,” the traditional Asian social mores are still very much espoused by the older Singaporeans of all ethnic groups. Older Singaporean women tended to be housewives, who stay at home to take care of their husbands and children. They are therefore more likely to choose to remain at home in their final days surrounded by their family members.

Further, about one in five (18.8%) patients with cancer had died of noncancer causes, and these patients were more likely to die in the hospital. Patients who died of noncancer causes were more likely to be older, long-term survivors of cancer, ²⁵ although some may have died of complications of cancer treatment or of comorbidities arising from the cancer itself, which may explain why the odds of dying in hospital were higher compared with dying at home or in in-patient hospices. Further study on this group of patients would enable us to know if the cause of death was associated with the underlying malignant condition, whether hospitalization could have been avoided, and what supportive care would be needed to enable these patients to die at home.

The only other study in Singapore that examined the place of death at the national level was the study by Beng and colleagues, ¹⁵ on all Singaporean residents aged 65 and above who had died in the year 2006, 26% of whom had died of cancer. Besides the difference in the time periods, the study population was also different from our study, as our study was on Singapore residents of all ages who were diagnosed with cancer, and who might have died of noncancer causes. In Beng et al.'s study, 35.8% of patients who had died of cancer as the primary cause of death had died at home, compared with our study of 30.3% home deaths. This was not surprising on two counts. First, Beng and colleagues' study population was older, and increasing age was a determinant of home death, found in both our studies. Second, we found in our study that cancer (“neoplasms”) as a cause of death was also a determinant of home death. Hence as Beng et al.'s study reported on patients who had died of cancer, it was not surprising that more of these pateints had died at home. Nevertheless, although different databases were used in these two studies, the similarities in the findings corroborate the validity of the study results.

There are strengths and limitations in this study, which analysed data from the SCR. Death registrations are used worldwide to study cancer trends and are considered reasonably reliable, ²⁶ although biases may exist due to miscoding and misclassification of causes of deaths. The SCR has been in existence since 1968. Data collected by the registry are robust, as the collection and storage are subjected to stringent quality assurance checks.

However, as with any registry, the information collected is limited in areas such as clinical aspects of care, and the variables in our analysis could account for only about 8% to 10% of all possible factors. We were also not able to study the impact of hospice care on place of death, as the information on place of care was not available from the SCR. Although it could be assumed that all patients who had died in in-patient hospices had received hospice care, we did not know what proportion of patients who had died at home had received palliative or hospice care.

Finally, due to intrinsic differences that exist in the population demographics as well as the medical, social, and cultural systems, the findings of this study cannot be generalized or compared with studies conducted in other countries.

The study of place of death as an indicator of whether a patient has received appropriate care at the end of life is complex, as there are many aspects that need to be examined, such as patient and caregiver preferences, consistency between place of end-of-life care and eventually where patients die, availability of health care infractructure and support for care outside the hospital and in the community and homes, among others. Many studies need to be conducted to provide the information needed to address these aspects, and the findings from our study provide but one piece of information in the large tapestry of knowledge. This study provided information on some of the two groups of factors that determine place of death (individual factors and factors related to illness). Further studies could be conducted, linking health care utilization data (including use of palliative and hospice services) to the SCR, thereby enabling the examination of the environmental factors that determine the place of death, for the planning of health care services appropriate for the care of patients with terminal cancer.