The place and cause of death of Japanese peritoneal dialysis patients

Background Although 70% of Japanese individuals wish to die at home, the in-hospital death rate in Japan is 71%. The in-hospital death rate is more than 80% in dialysis patients, most of whom are hemodialysis patients. Few reports exist on the relationship between place and cause of death in peritoneal dialysis (PD) patients. Methods We conducted a retrospective study of PD patients who died between January 2008 and June 2022 at three facilities. Results Ninety-four patients died, 71% in hospital, 25% at home, and 4% in elderly care facilities. The in-hospital death rates were 67%, 74%, and 71% ( P = 0.90) in the three facilities. Of the 67 in-hospital deaths, infection was the most common cause of death (30%), followed by sudden cardiac death (SCD) (15%), senility/cachexia (15%), heart failure (12%), and malignancy (9%). In the out-of-hospital deaths, there were no cases of infection, and SCD accounted for the majority (59%), followed by senility/cachexia (11%), heart failure (7%), stroke (4%), and malignancy (4%). Limiting to SCD, patients who died outside the hospital were significantly younger and had a higher prevalence of ischemic heart disease than those who died in hospital. Conclusions The in-hospital death rate for PD patients was lower than that previously reported for dialysis patients and similar to that of the Japanese population. Based on the place and cause of death in PD patients, to further increase the death rate at home, which many PD patients would have preferred, it may be necessary to prevent infectious diseases, strengthen interventions for patients at high risk of SCD, and shift to home care for patients with senil-ity/cachexia and malignancies.


Background
Until 1953, the death rate at home in Japan was over 80%; however, it had decreased to 12.2% by 2005.Although the death rate at home has gradually increased since then, in 2021, the death rate at home was only 17.2%, while the in-hospital death rate was 71% [1].This high rate of inhospital deaths and low rate of deaths at home is evident in comparison with Europe, the USA, and South Korea [2].By contrast, approximately 70% of Japanese individuals wish to die at home [3]; therefore, a large gap exists between reality and individual hopes.This gap should be corrected, and the Japanese government is promoting physician-led home visits and end-of-life care at home to control medical costs, which are rising year by year, and to prevent hospitals from exceeding their capacity limits [4].
Large-scale data on hospital and home death rates among dialysis patients in Japan are scarce.A single institution reported a high rate of in-hospital deaths (81.7%) among 180 dialysis patients [5].Since 97% of dialysis patients in Japan are hemodialysis (HD) patients [6], these data predominantly relate to HD patients.HD patients are highly dependent on the hospital because they visit it three times a week and have a low threshold for hospitalization, which naturally leads to a high inhospital death rate.By contrast, peritoneal dialysis (PD) patients are treated at home and visit the hospital only once or twice a month.To the best of our knowledge, no reports exist on the place of death specific to PD patients anywhere in the world.Hence, we conducted a retrospective study to determine the place and cause of death of patients who died during PD management in Japan.

Study design and subjects
Three emergency hospitals in Ehime Prefecture managing at least ten PD patients participated in a multicenter, retrospective, observational study.Patients who died between January 2008 and June 2022 while on PD alone or a combination of PD and HD were included.Patients who were hospitalized because of complications and changed from PD to HD during hospitalization but could not be discharged and died were also included.

Clinical parameters and definitions
All clinical measurements were obtained from medical records.Information such as place and cause of death, age at death, sex, the primary disease causing dialysis induction, comorbidities, duration of PD, and presence of concomitant HD therapy were also obtained from medical records.In addition, for patients with sudden cardiac deaths, the last measurements of plasma B-type natriuretic peptide and left ventricular ejection fraction (LVEF) before death were also collected.The place of death was classified into four categories: emergency hospitals, hospitals for rehabilitation and long-term care, elderly care facilities, and home.Out-of-hospital deaths were those in elderly care facilities and at home.Out-of-hospital cause of death was determined by the physician in charge of each hospital on the basis of the autopsy findings and the circumstances at the last hospital visit.In this study, acute myocardial infarction and fatal arrhythmia were classified as sudden cardiac deaths.Deaths due to hypotension or malnutrition were classified as senility or cachexia.

Outcomes
The primary outcome was place of death, and the secondary outcome was cause of death.We compared cause of death and patient characteristics by place of death.For sudden cardiac deaths, we also compared patient characteristics between the in-hospital and out-of-hospital death groups.

Statistical analysis
All continuous variables and categorical data are presented as median and interquartile range or percentage, respectively.Fisher's exact test was used to compare proportions of two or more groups, the Mann-Whitney U test was used to compare continuous variables of two groups, and the Kruskal-Wallis test was used to compare continuous variables of three or more groups.All statistical analyses were performed with EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria) [7].A two-tailed P value of < 0.05 was considered significant.

Results
A total of 94 patients died during PD treatment from the three facilities.The place of death for the 94 patients was as follows: 62 (66%) at emergency hospitals, 5 (5%) at hospitals for rehabilitation and long-term care, 4 (4%) at elderly care facilities, and 23 (25%) at home.Although there was a difference of more than three times in the number of deaths of PD patients between facilities (15-56 patients), there were no significant differences in median age at death (72-76 years; P = 0.71) or in-hospital death rate (67-74%; P = 0.90).The distribution of age at death was as follows: 25 (27%) in their 70s, 24 (26%) in their 80s, 19 (20%) in their 60s, 12 (13%) in their 90s, 10 (11%) in their 50s, and 4 (4%) in their 40s.
The characteristics of the 94 patients are presented in Table 1.Overall, the median age at death was 76 years and the primary disease causing dialysis induction was nephrosclerosis (38%), diabetic kidney disease (28%), and nephritis (23%).The prevalence of ischemic heart disease was high at 39%, the median duration of PD was 31 months, and the rate of concomitant HD therapy was 24%.Comparing patient characteristics by place of death (out-of-hospital and in-hospital), median age of deaths was 67 years and 76 years (P = 0.21), prevalence of diabetes was 52% and 30% (P = 0.06), and prevalence of ischemic heart disease was 59% and 31% (P = 0.02), respectively.The causes of death are presented in Table 2. Cardiovascular disease was the most common cause at 44%, with sudden cardiac death accounting for 27% of the cases.Infectious diseases were the next most common at 21%, with pneumonia at 7% and peritonitis at 5%.Senility/cachexia was the third most common cause at 14%, and malignancy was the fourth at 7%.There were only two deaths (2%) caused by encapsulated peritoneal sclerosis (EPS).
The distribution of causes of death by place of death is shown in Fig. 1.The most common cause of in-hospital death was infection (30%), followed by sudden cardiac death (15%), senility/cachexia (15%), heart failure (12%), and malignancy (9%).Sudden cardiac death accounted for most out-of-hospital deaths (59%), followed by senility/cachexia (11%) and heart failure (7%).There were no deaths caused by infectious diseases.A comparison of patient characteristics limited to sudden cardiac death is presented in Table 3.The median age of out-of-hospital deaths was 61 years, which was significantly younger than the median 81 years of in-hospital deaths (P = 0.03).The prevalence of diabetes and ischemic heart disease in outof-hospital deaths was 69% and 81%, respectively, which tended to be higher than in-hospital deaths.The median B-type natriuretic peptide value and LVEF last measured before death were 674 pg/mL and 48%, respectively, for out-of-hospital deaths, and 613 pg/mL and 43%, respectively, for in-hospital deaths, with no significant differences observed.

Discussion
Compared with the Japanese dialysis patients previously reported [5], the PD patients in Ehime Prefecture had a lower in-hospital death rate (71%) and a higher out-of-hospital death rate (29%).Since the government wants to increase the number of deaths at home [4] and most of the population also wants to die at home [3], the present results support that PD is a better match  than HD for kidney replacement therapy at the end of life for patients with end-stage kidney disease.Surprisingly, sudden cardiac deaths accounted for the majority of out-of-hospital deaths and no out-of-hospital deaths from infectious diseases occurred.Furthermore, the rate of out-of-hospital deaths from conditions that would ordinarily lead to end-of-life care at home was low.For example, the out-of-hospital death rate was 23% (n = 3) among 13 patients who died of senility/ cachexia and 14% (n = 1) among 7 patients who died of malignancy.
The death rate at home in Japan was more than 80% until 1953 but had decreased to 12.2% by 2005 [1].There are several possible reasons for this decline.First, the universal health insurance system began in 1961, which made it possible for people to receive inpatient care for a small cost.Second, nuclear families have become more common, which has reduced the capacities or willingness of family members to take on the responsibility of end of life care at home for family members.Since then, criticism of in-hospital deaths, in which patients die with "spaghetti syndrome" while connected to many tubes has increased, and deaths at home have begun to be re-evaluated.Although the rate of death at home has been gradually increasing since 2016, the rate of death at home was still only 17.2% in 2021, while the rate of death in hospital was 71% [1].
Large-scale data regarding in-hospital and out-ofhospital death rates among dialysis patients in Japan are limited.Furthermore, to the best of our knowledge, there are no other reports exclusively about PD patients.The only report comparing the place of death between HD and PD is limited to dialysis patients aged 80 years or older in a single center in Japan [8].Six of seven (85.7%) PD patients but only one of ten (10%) HD patients died at home, indicating a significantly higher rate of death at home among PD patients (P = 0.002).However, this report had limitations because it involved a single center and a small number of patients.Although the current study was limited to Ehime Prefecture, the study was conducted at three facilities that cover roughly 80% of PD patients in the prefecture.Two were located in a city with a population of 500,000, and the other in an aging and depopulated region.In addition, the departments of physicians managing PD treatment included both nephrology and urology, which is similar to the actual situation Fig. 1 The distribution of causes of death by the place of death.The most common cause of in-hospital death was infection at 30%, followed by sudden cardiac death at 15%, senility/cachexia at 15%, heart failure at 12%, and malignancy at 9%.Sudden cardiac death accounted for the majority of out-of-hospital deaths at 59%, followed by senility/cachexia at 11%, and heart failure at 7%.There were no deaths caused by infectious disease or EPS and intestinal disease in out-of-hospital deaths.EPS, encapsulated peritoneal sclerosis in Japan.The fact that the in-hospital death rate varied very little (67%, 71%, and 74%; P = 0.90) among the three hospitals with the different backgrounds and number of PD patients suggests that the findings are generalizable.Regarding causes of death, heart failure and infection were the leading causes of death in Japanese dialysis patients, at 22% each, followed by malignancy at 8%, cerebrovascular disease at 6%, and myocardial infarction at 4% [6].Senility/cachexia accounted for only 7%, unknown accounted for 12%, and hyperkalemia and sudden death accounted for 2%.Although sudden cardiac death was the most common cause of death in the present study, it was classified as heart failure by the Japanese Society for Dialysis Therapy Renal Data Registry [6] and likely does not reflect the actual situation.In a Japanese report [9] on HD patients, which examined causes of death in more detail, cardiovascular disease accounted for 36%, infectious diseases for 26%, and malignancy for 14%, and among cardiovascular diseases, sudden cardiac death was the most frequent (35%), while heart failure was the cause of death in only 18% of cases.Among infections, pulmonary infections accounted for about half of the cases (46%).There are only a few reports on causes of death among PD patients.In a PD registry study [10] conducted in the Tokai region of Japan from 2010 to 2012, infections other than peritonitis were the most common cause of death at 28%, followed by cardiac and aortic diseases at 19%, cerebrovascular diseases and others at 14% each, sudden death at 9%, and peritonitis at 2%.Reports differ as to whether sudden death in patients with acute myocardial infarction, death from fatal arrhythmias, or preexisting ischemic heart is classified as cardiac disease or sudden death.This fact is considered a limitation of studies on cause of death and requires standardization across registries [11].It is surprising that the rate of infectious disease deaths in the Tokai PD registry [10] was 30% when peritonitis was included, higher than that reported for dialysis patients as a whole [6] or for HD patients only [9], and very different from the 21% of infectious disease deaths in the present study.In a Canadian study [12], patients with PD had an lower risk of pneumonia and sepsis as the cause of hospitalization than HD patients (hazard ratio 0.54, 95% confidence interval 0.37-0.80;and hazard ratio 0.30, 95% confidence interval 0.18-0.51,respectively).Therefore, the risk of fatal infections such as pneumonia and sepsis is lower in PD patients.
In contrast to in-hospital deaths, in which infectious diseases accounted for the largest proportion of deaths (30%), none of the out-of-hospital deaths was caused by infectious diseases.Infectious diseases, by their nature, rarely cause sudden death, and the existence of effective treatments such as antimicrobial agents often leads to hospitalization.Dialysis patients are immunocompromised, and they may die even with antimicrobial therapy, which is probably the reason why they are the most common cause of hospital deaths.However, sudden cardiac death can occur at any time, whether in the hospital or at home.PD patients are at an increased risk of cardiovascular disease, spend significantly more time at home than HD patients, and receive less intensive medical interventions for cardiovascular disease than HD patients.Given these facts, it is unsurprising that sudden cardiac death is the leading cause of out-of-hospital death in PD patients.To reduce sudden cardiac death, more aggressive screening tests and prevention and treatment interventions are needed for PD patients with diabetes and ischemic heart disease.
The death at home that most of the people hope for [3] is not a sudden death from a healthy state but a death at home following a trajectory such as senility/cachexia, in which the patient slowly becomes frail, or a trajectory such as malignancy, in which the patient maintains relatively good physical function throughout the course of the disease but rapidly declines during the last few weeks or days of life [13].However, the present results show that the out-of-hospital death rate was 23% (n = 3) among 13 PD patients who died of senility/cachexia, and 14% (n = 1) among 7 PD patients who died of malignancy, both of which are low rates.This may result from a lack of progress in cooperation with home medical care with regard to PD.The hurdles to medical coordination in PD are thought to be the lack of recognition of PD and the inadequacy of the Japanese medical insurance system itself.
PD patients account for less than 3% of dialysis patients [6], and PD treatment may still be perceived as a special treatment given to a selected few.In particular, physicians responsible for home medical care are hesitant to provide even general internal medical care to PD patients without specialized knowledge of kidney disease and dialysis.The physician-led home visit in Japan is a unique system of regular physician visits, which includes palliative care and is conducted in collaboration with multiple professionals [4].Therefore, it is essential to educate home physicians about PD to overcome these hurdles.In September 2020, the Japanese Society of Peritoneal Dialysis launched a collaborative certification system and has started to train PD providers in regional medical cooperation.Another hurdle is that PD is a treatment that can be performed by trained patients themselves or their family members; however, nonmedical caregivers are not allowed to perform the bag exchange.Requiring visiting nurses to change the bags each time would be much too expensive; therefore, an urgent revision of the system is highly desirable.Furthermore, in Japan, dialysis patients do not have to pay for their own medical expenses even if they are hospitalized, which is a main reason why inhospital deaths are common among all dialysis patients.Many PD patients are inevitably hospitalized when their medical needs increase at the end of life, leading to inhospital deaths.
Unfortunately, death at home is not widespread in Japan among the general public because advance care planning (ACP) is not common [14].A 2018 Ministry of Health, Labour, and Welfare survey found that, while 66% of the Japanese general public aged 20 years or older agreed that advance directives should be prepared, only 8.1% actually prepared a written document [15].The presence or absence of ACP can also influence the place of death for patients who are introduced to home care.As an example, in a prospective study of patients aged 65 years or older who began receiving home care, only 36% died at home [4].In this study, in contrast to the high rate of deaths at home among patients who were introduced to home care with the expectation of end-oflife care at home, patients who were introduced to home care while their activities of daily living were well maintained often did not have ACPs drawn up, resulting in lower rates of end-of-life care at home.
Another major problem noted is that the optimal timing for performing an ACP is unknown [14].The introduction of dialysis is a major turning point for a patient with chronic kidney disease and is, therefore, an optimal time to develop an initial ACP in conjunction with shared decision-making regarding the choice of kidney replacement therapy.We believe that it would be practical to update ACPs when activities of daily living or socioeconomic factors changes during PD or HD treatment.
Finally, perspectives for reducing in-hospital deaths and increasing out-of-hospital deaths among PD patients, especially end-of-life care at home, are summarized in Fig. 2. For infectious diseases, which were the most common cause of in-hospital deaths, preventive efforts should be made to reduce the incidence of pneumonia and peritonitis.For sudden cardiac death, which was the most common cause of out-of-hospital death, many patients were relatively young and had diabetes or ischemic heart disease, so aggressive screening for myocardial ischemia and therapeutic and preventive interventions are needed.Meanwhile, for senility/cachexia and malignancy, it is necessary to promote ACP and strengthen cooperation with home physicians so that PD patients can spend their end of life out of hospital.
The present study had several limitations.First, it was a small study, with only 94 patients in three facilities.Nonetheless, there were no significant differences in inhospital death rates among the three facilities, which suggests that the findings are generalizable.The second issue is the classification of cause of death.As described above, definitions of cardiovascular disease and sudden death differ from report to report, and standardization is desirable [11].Especially in the case of out-of-hospital deaths, the cause of death is often based on the subjective opinion of the attending physician.However, the dilemma is that, if subjectivity is eliminated, the cause of death is often unknown.Third, it is unclear whether ACP was performed in the PD patients in the study.Whether the majority of PD patients in Japan hope to die at home (similar to the general population) is not known and is an issue that should be investigated.Despite these limitations, this report is the first to show an association 2 Perspectives on place and cause of death in peritoneal dialysis patients.To reduce infectious diseases, which were the most common cause of in-hospital deaths, preventive efforts should be made.To reduce sudden cardiac death, which was the most common cause of out-of-hospital deaths, aggressive screening for myocardial ischemia and therapeutic and preventive interventions should be performed.In-hospital deaths because of senility/cachexia and malignancy should be shifted to home care as far as possible.CVD, cardiovascular disease; SCD, sudden cardiac death between place of death and cause of death among PD patients.Our findings suggest that PD patients can be enabled to spend their end of life at home, which is important given the expected increase in the number of elderly PD patients and the subsequent shortage of inpatient beds.

Conclusions
In the present study, the in-hospital death rate in PD was lower than that previously reported in HD and equal to that of the general population in Japan.Based on the place and cause of death in PD patients, to further increase the out-of-hospital death rate, which many PD patients would have preferred, it may be important to prevent infectious diseases, strengthen aggressive screening and therapeutic and preventive interventions for patients at risk for cardiovascular disease, promote ACP throughout the country, and strengthen collaboration with home physicians regarding PD.

Table 1
Comparison of patient characteristics by the place of death

Table 2
Cause of death

Table 3
Comparison of patient characteristics by the place of death limited to sudden cardiac death