Journal Logo

Research Article: Observational Study

The survival time of end-of-life home care patients in Fukui prefecture, Japan

A retrospective observational study

Kosaka, Makotoa; Miyatake, Hirotomo MDa,∗; Kotera, Yasuhiro PhDb; Masunaga, Hidehisa MSca; Arita, Satoshi MSca; Tsunetoshi, Chie BAa; Nishikawa, Yoshitaka MD, PhDc; Ozaki, Akihiko MD, PhDa,d; Beniya, Hiroyuki MDa

Editor(s): Bush., Eric

Author Information
doi: 10.1097/MD.0000000000027225


1 Introduction

Patient-centered care has become central in clinical medicine to improve patients’ quality of care throughout their treatment course.[1] Accordingly, shared decision-making (SDM), a communication process by which patients and clinicians work together to make optimal health care decisions, has become of key significance.[2] In this context, specifically in patients with life-limiting illnesses, such as advanced cancer and dementia, the concept of advance care planning (ACP)—a process that enables patients and their caregivers to devise a crucial care plan for the time the patient has left—is essential.[3] Some reports indicate that ACP did not improve quality of life (QOL), while others reported positive effects on QOL.[4,5] These findings suggest the importance of individualizing ACP depending upon the unique medical history and needs of each patient, rather than conducting it in one uniform manner. Additionally, the values and preferences of patients and their family, including community and cultural contexts, should be also considered.

ACP is particularly crucial in end-of-life care in home care, because this unique care has been used among chronically ill patients and those with severe conditions, including those in end-of-life periods. However, it is not always easy to facilitate ACP in home care setting. For example, in home care, where family members make a predominant part of care relating to patients, perspectives of family members highly influence overall directions of patients’ care, which would potentially cause conflicts with healthcare professionals and patients. Further, some families do not wish to tell patients about their true conditions throughout end-of-life care, and patients may suffer from the fact that they do not know their conditions, and the patients themselves may suffer because they do not know their condition. Also, the presence of patients’ families during home visits can make it difficult for patients to express their true feelings, such as wish to stay home. Further, there are limited frequencies of interactions between healthcare professionals and patients and their family members in home care setting, which would hamper an initiation and development of ACP in these relevant stakeholders. Given these limitations, in facilitating ACP discussion in home care setting, accurate understanding of patients’ survival would be critical because it would facilitate the healthcare professionals to schedule a timing to proceed with ACP discussion. Although a previous review article on this theme indicated that actual prognosis of the patient does not matter because ACP should be facilitated in advance,[6] an accurate prediction of the remaining time of a patient with life-limiting illness might have become a key component in more properly individualizing ACP. However, little is known about survival outcome of home care patients.

Japan is one of the few countries with a super-aged population. As of 2017, the rate of the population aged 65 years and older (i.e., aging rate) was 27.7%, whereas the aging rate globally was only 8.3% in 2015.[7] Further, older adults requiring chronic care has been rising, as has the number of people who prefer to pass away at home.[8] This is probably because hospitalization is generally burdensome for elderly patients.[6]

However, many older people in Japan still pass away while hospitalized against their will, indicating a need to support end-of-life care at home. Only 12.7% of Japanese patients died at home in 2015, although over 60% wanted to receive terminal care at home.[9] In Japanese home care, doctors visit patients at their homes or at facilities where patients reside, and provide regular care. This helps manage the increasing number of chronic diseases in the Japanese population, enabling older adults with terminal prognoses to choose where to pass their remaining days. This treatment structure was developed to reduce medical expenses and fulfill patients’ wishes to spend the end of their lives at home. Moreover, from an economic standpoint, home care is superior to hospital care and conserves hospital resources.[10] Despite the importance of prognoses in home care and the increasing need for end-of-life care among older adults, research exploring these matters is scarce.

Accordingly, the present study aimed to identify factors that could better predict the survival period among the patients receiving home care in Japan and its relationship with their disease prognoses. We believe that this study would provide fundamental information to help improve a process of ACP among patients receiving home care.

2 Methods

2.1 Setting and participants

This was a retrospective observational study including patients who were examined for the first time at the Orange Home-Care Clinic between January 1, 2017 and December 31, 2018. The Orange Home-Care Clinic is located in Fukui City, Fukui Prefecture, and provides home care to about 300 patients annually, of which approximately 100 receive end-of-life care. Many patients are referred to this facility from other general hospitals or clinics.

2.2 Data collection

Patient information was extracted from electric medical records: sex, date of birth and first examination by the clinic, primary diseases, Care level, availability of caregivers, date of death (if occured before December 31 2019), and place of death

However, data regarding Care level and availability of caregivers in the medical records may not accurately reflect the situation at the start of treatment; only the most recent information was available.

The date of first visit was the date when the clinic made its first visit to the patient. The primary disease was extracted using the International Classification of Diseases-11th edition classifications. Of all 14 classifications, the diseases that coincide with their diagnoses were recorded for each patient: infectious diseases, cancer, collagen diseases, endocrine diseases, psychiatric diseases, cerebral nerve diseases, cardiovascular diseases, respiratory diseases, gastrointestinal diseases, musculoskeletal diseases, renal diseases, perinatal conditions, developmental abnormalities, and trauma.

Cerebral nerve diseases were further divided into dementia, diseases related to cerebral blood vessels, and other.

According to Japan's long-term care insurance system, certified by the local government, Care level was classified as Not certified, Support level 1-2, and Care level 1, Care level 2, Care level 3, Care level 4, and Care level 5.

As physical and cognitive functions decline, the Care level increases.[11] The availability of caregivers was classified as “No caregiver” if the person lived alone, “Informal caregiver” if the person lived with family, and “Formal caregiver” if the person lived in a care facility.

Place and date of death were categorized as either “living” or “died at facilities” on December 31, 2019. Death cases were further categorized by location: “died at residences”, “died at facilities”, and “died at hospitals”. When patients die in a hospital, hospital authorities or the patient's family report the death to the clinic. A few cases were identified from an obituary.

2.3 Ethical review

This research adhered to the Ethical Guidelines for Medical and Health Research Involving Human Subjects. The study was approved by The Ethics Committee of the Medical Governance Research Institute on June 4, 2020 (MG2018-18-20200604).

2.4 Data analysis

2.4.1 Survival probability curve

The survival rate was evaluated using the Kaplan–Meier product-limit method and was plotted on a survival curve using the date of the first visit as the starting point. In cases where the outcome (whether dead or alive) was not available, the follow-up period was terminated on the last date of treatment. To analyze survival probability according to primary diseases, we categorized diseases into cancer, dementia, neurological diseases other than dementia, and other diseases because cancer and dementia are representative of life-limiting diseases.

2.4.2 Cox proportional hazard regression model

A Cox proportional hazards univariate regression analysis was performed. In order to analyze factors affecting survival rates (i.e., dependent variable), age, sex, primary diseases, and availability of caregiver were used as factors (i.e., independent variables). All variables, except age, were recorded as categorical variables. To avoid overfitting, we used the four primary disease categories described for the survival probability curve. Likewise, Care level was categorized into two groups: not certified and certified. Multivariable regression analysis was conducted for the factors that were identified as significant in the univariate regression analysis (P < .05 as the significance level). All analyses were performed using Stata/MP 15.0 (Stata Corporation, College Station, Texas, USA) and Microsoft Excel Version 16.16.8 (Microsoft Corporation, Redmond, WA, USA).

3 Results

We included 277 patients who were receiving end-of-life care during the study period. Their characteristics are presented in Table 1. Patients’ median age was 82 years (min <1 year, max 101 years) at initial examination and 123 (44.4%) were men. The most common primary disease was cancer (n = 92, 33.2%), followed by dementia (n = 76, 27.4%). Regarding Care level, 81 patients (29.2%) were not certified and 83 patients were at Care level 4 or 5 (30.0%). As for the caregivers, 179 patients (64.6%) were living with family members (i.e., an informal, untrained caregiver available), and 69 (30%) were living in an institution (a formal, trained caregiver available). Regarding outcomes on December 31, 2019, there were 99 patients (35.7%) who continued to receive home care provided by our clinic. A further 96 (34.7%) had died at home and 19 (6.9%) died at institutions while receiving home care. There were 27 (9.7%) patients who had died in hospital after our home care had been suspended because the patient concerned had opted for emergency transfer. Of the remaining 36 patients with unknown outcomes (13.0%), at least 17 were hospitalized or began attending another hospital.

Table 1 - Sociodemographic and clinical characteristics of patients who received home care at the clinic for the first time in 2017 and 2018.
Number (%)(N = 277)
 Male 123 (44.4)
 Female 154 (55.6)
 100- 2 (0.7)
 90–99 47 (17.0)
 80–89 106 (38.3)
 70–79 48 (17.3)
 60–69 35 (12.6)
 50–59 13 (4.7)
 40–49 3 (1.1)
 30–39 6 (2.2)
 20–29 4 (1.4)
 10–19 1 (0.4)
 0–9 12 (4.3)
Level of the need for care or support
 Not certified 81 (29.2)
 Support level 1/2 12 (4.3)
 Care level 1 29 (10.5)
 Care level 2 34 (12.3)
 Care level 3 38 (13.7)
 Care level 4 49 (17.7)
 Care level 5 34 (12.3)
Availability of caregivers
 Informal caregiver available 179 (64.6)
 Formal caregiver available 69 (24.9)
 No caregiver available 28 (10.1)
 Uncertain 1 (0.4)
Primary disease
 Cancer 92 (33.2)
 Dementia 76 (27.4)
 Other nervous diseases 19 (6.9)
 Cerebral vascular diseases 18 (6.5)
 Developmental anomalies 10 (3.6)
 Trauma 8 (2.9)
 Circulation 7 (2.5)
 Respiratory 7 (2.5)
 Kidney 7 (2.5)
 Others 34 (12.3)
Outcomes on December 31, 2019
 Died at their own residences 96 (34.7)
 Died at facilities 19 (6.9)
 Died at hospitals 27 (9.7)
 Survived 99 (35.7)
 Uncertain 36 (13.0)

Figure 1 shows the Kaplan–Meier curve for the overall survival rate of patients. The survival probability was 82.5% at 30 days, 67.8% at 90 days, 52.7% at 1 year, and 39.1% at 3 years (see Table S1 Supplemental Digital Content, which shows the survival probability according to primary diseases).

Figure 1:
Kaplan–Meier survival estimate of the patients.

Figure 2 shows the Kaplan–Meier curve of the survival rates according to primary diseases, namely: cancer patients, dementia patients, patients with other nervous and cerebral vascular diseases, and other patients. As shown in Table S1 Supplemental Digital Content, among 76 patients with dementia, the survival rates were 91.9% at 30 days, 86.4% at 90 days, 78.1% at 1 year, and 47.0% at 3 years. Among the 92 cancer patients, the survival rates were 64.6% at 30 days, 33.4% at 90 days, 9.5% at 1 year, and 4.1% at 3 years. For patients with other nervous and cerebral vascular diseases, the survival rates were 91.9% at 30 days, 83.6% at 90 days, 72.5% at 1 year, and 60.8% at 3 years. The log-rank test showed that the difference between disease groups was significant (P < .05).

Figure 2:
Kaplan–Meier survival estimate of the patients according to main diseases.

The impact of each sociodemographic variable on survival was determined using the Cox proportional hazard regression model (Table 2). Cancer patients had significantly higher mortality rates than dementia patients (hazard ratio [HR], 6.53; 95% confidence interval [CI], 4.16–10.28)). In addition, when age was analyzed as a continuous variable, the mortality rate was significantly higher as age increased (HR 1.01 [95% CI 1.00–1.02]). Other factors such as sex, Care level, and caregiver availability were not significant. Multivariable analysis was performed for age and primary disease, which were significant factors in the univariate analysis. We found that the HR for cancer and age was 7.34 (95% CI 4.63–11.62) and 1.01 (95% CI 1.00–1.02), respectively; also, both were factors that independently affected survival probability.

Table 2 - Univariate and multivariate cox proportional hazard models for survival.
Univariate hazard ratio(95% confidence interval) Multivariable hazard ratio(95% confidence interval)
Main diseases
 Dementia 1.00 1.00
 Cancer 6.53 (4.16–10.28) 7.34 (4.63–11.62)
 Other nervous and cerebral vascular diseases 0.88 (0.45–1.73) 1.12 (0.56–2.25)
 Other diseases 0.88 (0.51–1.73) 1.11 (0.63–1.95)
 Male 1.00
 Female 0.76 (0.55–1.05)
Age 1.01 (1.00–1.02) 1.01 (1.00–1.02)
Care level
 Not certified 1.00
 Certified 0.78 (0.55–1.11)
Availability of caregiver
 Not available 1.00
 Informal caregiver 1.74 (0.96–3.16)
 Formal caregiver 0.54 (0.26–1.11)
P < .05.
P < .01.
P < .001.

4 Discussion

In this study, the estimated survival rate was 52.7% at one year after the initial examination at the clinic. The patient's primary disease had a significant impact on prognosis, particularly in patients with cancer who had a significantly worse outcome than those with other diseases. One possible reason is that in Japan, home care is mainly provided as palliative care for terminally ill patients who have no possibility of recovery.[12] Further, palliative care had been traditionally covered by health insurance only for cancer patients, although palliative care for terminal stage heart failure was covered from 2018.

The survival rates for patients with dementia and other nervous and cerebral vascular diseases were not significantly different from those with other diseases in terms of long-term survival. One of the barriers to providing palliative care to patients with dementia has been shown to be a lack of prior ACP,[13] despite an increasing recognition that early ACP is especially important in cases of cognitive impairment. Therefore, a well-informed and accurate prognosis can be helpful in decision-making for end-of-life care. In fact, their treatment and management may also vary depending on the predicted prognosis. According to a randomized controlled trial of patients with dementia aged >65 years and scoring 7 on the Global Deterioration Scale who resided in a nursing home for longer than 90 days, the findings revealed that interventional treatment was significantly reduced when the proxy, who was the residents’ formally or informally designated medical decision maker, believed that the patient had a life expectancy of less than 6 months.[14]

When considered from a prognostic standpoint, the majority of home care patients follow two patterns: patients who wish to receive end-of-life care at home as an alternative to hospital-based palliative care, and patients who wish to be treated in order to manage chronic diseases at home. For the former group, ACP and other issues should preferably be discussed among all concerned at the outset. However, it may be difficult for physicians to respond directly to patients at home in the event of emergencies. Therefore, it is important to create an environment for SDM with formal and informal caregivers to enable them to respond to emergency situations appropriately. In contrast, for the latter group, ACP does not have to be performed hastily, but this concept underlines a process to reach death during patients’ end-of life period more than their ancestry concepts, such as advance directives. Following this principle, ACP should be preceded step by step for these patients as well.

Of the patients whose medical care ended by December 31, 2019, 64.6% (115/178) died at home or in a resident facility: among cancer and non-cancer patients, 77.0% (67/87) and 52.7% (48/91) died, respectively. This is relatively high considering a study by Ko et al.[15] in Taiwan; the percentage of patients in home palliative care who died at home was 43.6% (43.7% were cancer patients and 42.9% non-cancer patients). However, considering that some patients/families would prefer to be transported to the hospital in case of an emergency, future research needs to identify the preferences of the patients and their families to further refine interpretation of these findings and how to determine the most appropriate ACP.

5 Implications and future perspectives

As clinical implications of this study, our findings regarding the characteristics of patients would be used in home care clinics to promote discussion of ACP when introducing home care to patients and their families based on their individual prognoses. In concrete, ACP of cancer patients receiving home care should be facilitated at a relatively early timing. Indeed, physicians (in any environment) could make use of our findings to explain home care patient characteristics to patients who are considering home care for the remaining periods of their terminal illnesses. Further, it has been shown that those who accurately understood their prognosis were more likely to choose to receive preferred end-of-life care.[16] As such, the information of survival probability would be useful both for physicians and patients.

With regards to an evidence body about ACP in home care setting, it has been revealed that ACP conversations facilitated by health workers could reduce frequent hospitalization in the home care setting.[6] Also, another research showed that patient requests for information regarding prognosis could be cues to facilitate ACP.[6] These studies suggest that ACP should be initiated regardless of the actual prognosis, but we hypothesized that we should individualize ACP based on patients’ characteristics. This study additionally added to an evidence body the finding that the prognosis could be an important element in individualizing ACP for the first time. Future research can benefit from including other types of data, such as the patient's (and their caregivers’) preferences for place of death and the conditions and environment of the home care location. Prospective observational studies evaluating the outcomes associated with their expectations are necessary. Some patients have a strong preference for end-of-life care at home, while others prefer hospital care for various reasons, including high accessibility to urgent care.[17] In addition, when deciding to introduce home terminal care, broad knowledge about the potential problems and prognosis in patient management is essential for healthcare providers.

Further, when considering patients’ end-of-life care preferences, the unique characteristics of Japanese clinical practice should be considered, as they are vastly different from other countries. Japan has historically been a collectivist society, with individuals placing the group's opinion above their own as opposed to people in individualistic cultures such as the United States of America.[18] In this regard, social norms and morals have traditionally dictated that the wishes of family members and the patient are equally important when deciding on end-of-life care. Indeed, a retrospective study conducted in primary care in Japan reported that family preference was a stronger determinant than the patient's individual preference for the choice of place to die.[19] While most, if not all, of these traditional beliefs are now in a state of flux given that Japan is experiencing social upheaval and change, it is meaningful to consider differences of cultural and other contexts in future studies.

6 Limitations

As this was a retrospective observational study, using only medical records from a single home care clinic, there were a large number of unknown factors relating to subsequent outcomes after hospitalization or suspension of home care. In particular, while a discussion of the patients’ prognosis and needs up until the point of their death would require a time when the diagnosis of the primary diseases was made, we were not able to extract this information. Furthermore, because the information in medical records is updated chronologically, it does not accurately represent the patient's situation at the time of the start of treatment. It was often difficult to extract older entries of variables considered in this study because of the nature of medical records used in the clinic.

Additionally, since this study was only conducted using the data from one clinic, the findings from this small unique sample may not be representative of clinics elsewhere. However, it can be said that the participants are representative of patients in Japan in that there are many older adults suffering from cancer or dementia. Lastly, other confounding variables, including financial matters and interfamily clashes, may have been present, but not evaluated in the analysis.

7 Conclusion

In this study, we found that the patient's primary disease had a significant impact on prognosis. This is the first study to investigate prognosis in the home. The results will help to individualize ACP discussion at home. It is important to investigate further in the future.


The authors express sincere gratitude to all clinic staff members, patients, and caregivers involved in this study.

Author contributions

Conceptualization: Makoto Kosaka, Hirotomo Miyatake, Yasuhiro Kotera, Hidehisa Masunaga, Satoshi Arita, Chie Tsunetoshi, Yoshitaka Nishikawa, Akihiko Ozaki.

Data curation: Makoto Kosaka.

Project administration: Hidehisa Masunaga, Satoshi Arita, Akihiko Ozaki, Hiroyuki Beniya.

Visualization: Makoto Kosaka.

Writing – original draft: Makoto Kosaka, Hirotomo Miyatake, Akihiko Ozaki.

Writing – review & editing: Makoto Kosaka, Hirotomo Miyatake, Yasuhiro Kotera.


[1]. Mead N, Bower P. Patient-centredness: a conceptual framework and review of the empirical literature. Soc Sci Med 2000;51:1087–110.
[2]. Charles C, Gafni A, Whelan T. Shared decision-making in the medical encounter: what does it mean? (or it takes at least two to tango). Soc Sci Med 1997;44:681–92.
[3]. Tulsky JA. Beyond advance directives: importance of communication skills at the end of life. JAMA 2005;294:359–65.
[4]. Tang ST, Chen JS, Wen FH, et al. Advance care planning improves psychological symptoms but not quality of life and preferred end-of-life care of patients with cancer. J Natl Compr Canc Netw 2019;17:311–20.
[5]. Skorstengaard MH, Brogaard T, Jensen AB, et al. Advance care planning for patients and their relatives. Int J Palliat Nurs 2019;25:112–27.
[6]. Bigger S, Haddad L. Advance care planning in home health: a review of the literature. J Hosp Palliat Nurs 2019;21:518–23.
[7]. The Ministry of Health, Labour and Welfare, Japan.: White paper on aging society. Available at: Accessed June 28, 2020.
[8]. The Ministry of Health, Labour and Welfare, Japan.: Recent developments in “The guidelines for the medical decision making process in the final stage of life”. Available at: Accessed July 1, 2020.
[9]. Tsuchida T, Onishi H, Ono Y, Machino A, Inoue F, Kamegai M. Population-based survey regarding factors contributing to expectation for death at home. Asia Pac Fam Med 2018;17:07.
[10]. Kinjo K, Sairenji T, Koga H, et al. Cost of physician-led home visit care (Zaitaku care) compared with hospital care at the end of life in Japan. BMC Health Serv Res 2017;17:40.
[11]. The Ministry of Health, Labour and Welfare, Japan.: Health and Welfare Services for the Elderly. Available at: Accessed April 10, 2020.
[12]. Ozaki A, Tsubokura M, Claire L, et al. The importance of family caregiving to achieving palliative care at home: a case report of end-of-life breast cancer in an area struck by the 2011 Fukushima nuclear crisis. Medicine 96. e8721. 10.1097/MD.0000000000008721.
[13]. Mataqi M, Aslanpour Z. Factors influencing palliative care in advanced dementia: a systematic review. BMJ Support Palliat Care 2020;10:145–56.
[14]. Loizeau AJ, Shaffer ML, Habtemariam DA, et al. Association of prognostic estimates with burdensome interventions in nursing home residents with advanced dementia. JAMA Intern Med 2018;178:922–9.
[15]. Ko MC, Huang SJ, Chen CC, et al. Factors predicting a home death among home palliative care recipients. Medicine (Baltimore) 2017;96:e8210.
[16]. Wen FH, Chen JS, Chang WC, et al. Accurate prognostic awareness and preference states influence the concordance between terminally ill cancer patients’ states of preferred and received life-sustaining treatments in the last 6 months of life. Palliat Med 2019;33:1069–79.
[17]. Choi J, Miyashita M, Hirai K, et al. Preference of place for end-of-life cancer care and death among bereaved Japanese families who experienced home hospice care and death of a loved one. Support Care Cancer 2010;18:1445–53.
[18]. Ruhnke GW, Wilson SR, Akamatsu T, et al. Ethical decision making and patient autonomy: a comparison of physicians and patients in Japan and the United States. Chest 2000;118:1172–82.
[19]. Masumoto S, Sato M, Ichinohe Y, Maeno T. Factors facilitating home death in non-cancer older patients receiving home medical care. Geriatr Gerontol Int 2019;19:1231–5.

home care; prognosis; shared decision making

Supplemental Digital Content

Copyright © 2021 the Author(s). Published by Wolters Kluwer Health, Inc.