Effects of different palliative care models on decedents with kidney failure receiving maintenance dialysis: a nationwide population-based retrospective observational study in Taiwan

Abstract

Objectives

Patients with kidney failure receiving maintenance dialysis are a particularly important population and carry a heavy disease burden. However, evidence related to palliative care for patients with kidney failure receiving maintenance dialysis remains scarce, especially in regard to palliative care consultation services and palliative home care. This study aimed to evaluate the effects of different palliative care models on aggressive treatment among patients with kidney failure receiving maintenance dialysis during the end of life.

Design

A population-based retrospective observational study.

Setting

This study used a population database maintained by Taiwan’s Ministry of Health and Welfare in combination with Taiwan’s National Health Research Insurance Database.

Participants

We enrolled all decedents who were patients with kidney failure receiving maintenance dialysis from the period 1 January 2017 to 31 December 2017 in Taiwan.

Main exposure measure

Hospice care during the 1-year period before death.

Main outcome measures

Eight aggressive treatments within 30 days before death, more than one emergency department visit, more than one admission, a longer than 14-day admission, admission to an intensive care unit, death in hospital, endotracheal tube use, ventilator use and need for cardiopulmonary resuscitation.

Results

A total of 10 083 patients were enrolled, including 1786 (17.7%) patients with kidney failure who received palliative care 1 year before death. Compared with patients without palliative care, patients with palliative care had significantly less aggressive treatments within 30 days before death (Estimates: −0.09, CI: −0.10 to −0.08). Patients with inpatient palliative care, palliative home care or a mixed model experienced significantly lower treatment aggressiveness within 30 days before death.

Conclusions

Palliative care, particularly use of a mixed care model, inpatient palliative care and palliative home care in patients with kidney failure receiving dialysis, could all significantly reduce the aggressiveness of treatment within 30 days before death.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study used population-based data to evaluate the effects of different palliative care models on aggressive treatment among patients with kidney failure receiving maintenance dialysis during the end of life.

• The population-based study design involving a large number of patients elevated the scope of this study.

• Laboratory data such as blood tests and patients’ and their families’ preference of treatment or preferences for place of death were not recorded in the database.

Introduction

Populations are rapidly ageing while life expectancy has increased worldwide in recent decades due to improvements in both public health and medical healthcare, thus the ageing population is expected to continue growing.¹ The nature of ageing requires a consistently shifting approach to healthcare provided by healthcare professionals, patients and families.² However, despite the overall increase in lifespan being seen in worldwide populations, patients with chronic diseases still suffer from both physical symptoms and the psychological burden of requiring care for their disabilities and multimorbidity.³

Patients with kidney failure receiving maintenance dialysis are a particularly important population and carry a heavy disease burden. In recent years, efforts such as multimodal kidney replacement therapy which aim at providing better treatment for these patients have been undertaken.^{4 5} Despite advances in kidney replacement therapy, including kidney transplantation, hemodialysis and peritoneal dialysis, many patients experiencing kidney failure still suffer from various types of physical stressors, including infection,⁶ anaemia, conscious disturbance,⁷ heart failure,⁸ pain,⁹ frailty and disability,¹⁰ psychological stress^{11 12} and even low quality of life.¹³ Moreover, the mortality rate is particularly high among patients with kidney failure¹⁴ and sudden death often occurs.¹⁵ The possibility of sudden death is a risk factor for complicated grief being experienced by caretakers, and therefore palliative care staff are trained to help provide psychosocial support for caretakers who are coping with grief.^{16 17} Patients with kidney failure often require invasive treatment during the end of life (EOL), including the use of an endotracheal tube (ETT), ventilator support, cardiopulmonary resuscitation (CPR) and even admission to the intensive care unit (ICU) prior to death,¹⁸ resulting in poor experiences for patients and their caregivers during the EOL period.^{19 20} Such aggressive treatments, along with admission to the ICU, are associated with higher medical costs, lower psychological health in caregivers and lower self-esteem for patients.^{21 22}

The Quality Indicators for End-of-Life Care are tools used to measure the quality of care for patients during their end stages of life. Developed through empirical research and consensus, these indicators can serve as tools for either quality monitoring or as a means of continually improving care. Earle et al identified 19 indicators related to invasive medical interventions and palliative care, which were then ranked in order of importance following expert consensus through a modified Delphi method, after feedback from patients and family members during focus group discussions had been heard. These indicators can be used to measure poor quality EOL care, including: (1) chemotherapy during one’s final month of life; (2) more than one emergency room visit during the final month of life; (3) more than one hospitalisation during the final month of life; (4) hospitalisation for more than 14 days during the final month of life; (5) need for the ICU in the final month of life; (6) no use of palliative care in the year prior to death, or only using it during the final 3 days of life and (7) dying in the hospital.²³ Tang et al conducted a literature review and found that indicators of poor quality EOL care for patients with terminal cancer included: use of endotracheal intubation in the last month of life, use of mechanical ventilation in the last month of life and use of CPR in the last month of life.²⁴ A previous report has highlighted the importance of improving the quality of EOL care, with a particular emphasis on ICUs and the EOL care received in these units. The main focus in the report included: (1) patient-centred and family-centred decision making, (2) communication, (3) continuity of care, (4) emotional and practical support, (5) symptom management and comfort care, (6) spiritual support and (7) emotional and organisational support for the clinical physicians in the ICU.²⁵

Palliative care is specialised medical care for people living with a serious illness.²⁶ This type of care can improve the quality of life for both patients and families during EOL.²⁷ Multiple studies have shown that the use of palliative care among non-cancer patients has increased in recent years.^{28 29} Additionally, previous studies have shown that inpatient palliative care and palliative care service in the community can relieve symptom burden and reduce costs. For decedents who experienced kidney failure, inpatient palliative care has been proven to shorten hospital length of stay and reduce hospitalisation costs. Also, patients with kidney failure receiving inpatient palliative care were more likely to enter a hospice, while having a lower likelihood of rehospitalisation.³⁰ Inpatient palliative care was also related to a greater likelihood of dying at home, resulting in reduced costs.³¹ Another study revealed that dialysis patients in a renal supportive care clinic that focused on symptom control and patient-centred care showed an improvement in symptom burden.³² However, from a previous study in the USA, only 13.5% of patients receiving maintenance hemodialysis used hospice service, and even among those patients who withdrew from dialysis, more than 50% of them did not use hospice service.³¹

Multiple palliative care services have been initiated in Taiwan, including inpatient care in palliative care units, palliative home care service and palliative care consultation service. Inpatient palliative care is similar to hospice units in the USA,³³ where palliative care is provided in a special palliative care unit in the hospital. Palliative care consultation service is similar to inpatient palliative care consultation service in the USA,³⁴ where high-quality palliative care is provided to inpatients in need. Palliative home care is much like routine palliative home care in the USA.³⁵ A previous study has shown that palliative care consultation service during hospitalisation for non-cancer patients can improve patients’ and families’ awareness of disease, while also increasing the number of Do Not Resuscitate (DNR) consent orders.³⁶ Taiwan has made all three of these palliative care services available through its National Health Insurance (NHI) programme. In 2009, the service covered by NHI was expanded to include non-cancer terminally ill patients. The amendments to the ‘Hospice Palliative Care Act’ in 2011 and 2013, and the further announcement and execution of the ‘Patient Right to Autonomy Act’ in 2016 and 2019, have helped palliative care in Taiwan become more popular, more widely well-known and more accessible. In 2021, Taiwan was ranked third in the world for delivering EOL care and quality of death.³⁷ A previous study also showed that palliative service covered by the NHI programme led to reduced medical costs for patients experiencing kidney failure.³⁸

However, despite advances in palliative care, Taiwan has the highest incidence and prevalence of kidney failure worldwide per capita, with a previous study revealing that the annual incidence and prevalence of dialysis increased steadily from the year 2010 to 2018.³⁹ In 2010, there were 68 962 patients who received dialysis treatment. However, in 2019, the number had increased to 92 624 patients, showing a yearly growth rate of more than 2%.⁴⁰ Another study revealed that the number of patients receiving dialysis in Taiwan was the highest in Asia.⁴¹

Previous studies have shown that although palliative care among patients with kidney failure is related to lower medical costs and lower medical utilisation, evidence related to the effects of palliative care on aggressive treatment during EOL among this population remains scarce, particularly when concerning different palliative care models. Also, most studies regarding this issue were performed in Western countries; however, from a recent review, incidence rates of treated kidney failure rose substantially and predominantly in East and Southeast Asia,⁴² with most patients experiencing kidney failure who are first enrolled in a hemodialysis programme living in Asia.⁴³ Additionally, there are international differences in the prevalence of dialysis outcomes due to variations in patient population and different socioeconomic contexts. On the other hand, palliative care development and its execution are largely influenced by policies and funding, as well as cultural views and beliefs among stakeholders in each country.⁴⁴ Thus, it is important to evaluate the impact of different palliative care approaches for patients with kidney failure receiving maintenance dialysis, particularly among the Asian population. In this population-based observational study, we aimed to evaluate different palliative care approaches, including inpatient palliative care, palliative care consultation service, palliative home care and mixed type service, while also analysing their effects on various aggressive treatments during EOL among decedents who had experienced kidney failure while receiving dialysis treatment in Taiwan.

Methods

Data source and study group

This population-based retrospective study used a population database maintained by Taiwan’s Ministry of Health and Welfare in conjunction with Taiwan’s National Health Insurance Research Database (NHIRD). The NHI’s claims database is called the NHIRD. We enrolled all decedents from 1 January 2017 to 31 December 2017 in Taiwan who had been given a diagnosis of chronic kidney disease from clinics, emergency departments (EDs) and admission file records 1 year prior to death (ICD 10: N18). Because the main purpose of the study was to evaluate the impact of different palliative care approaches for patients with kidney failure receiving maintenance dialysis, we excluded certain participants, including those with: (1) no dialysis treatment (including hemodialysis and peritoneal dialysis 1 year before death); (2) no maintenance dialysis treatment, meaning dialysis less than twice 1 year prior to death or an interval period between the two dialysis sessions of more than 3 months. We included this particular exclusion criteria because multiple epidemiological studies have defined maintenance dialysis as receiving dialysis for more than 90 days^{45 46} and (3) an age at death of under 18 years. After the initial analysis, a total of 10 083 subjects were included (figure 1).

Figure 1.

Flow chart of study. CKD, chronic kidney disease; NHIRD, National Health Insurance Research Database.

Independent variables

We analysed all types of palliative care use among decedents during the period 1 year prior to death. Palliative care utilisation was categorised into four groups: inpatient palliative care, palliative care consultation service, palliative home care and mixed service. Patients receiving inpatient palliative care would be admitted to a palliative care unit in a hospital and cared for by professional palliative healthcare professionals. Patients using palliative care consultation service were admitted to his/her clinical ward and received shared care from both the original clinical care team and the palliative share care team. Patients receiving palliative home care were cared for at home by palliative home care team members. And finally, patients using mixed services received at least two different types of palliative care services.

Dependent variables

The outcome variables were aggressive treatments at EOL. The indicators used in this study regarding the poor quality of EOL care were adapted from Earle et al²³ and Tang et al²⁴ on the basis of the data available from the NHI claims database and the synthesised evidence taken from the futility of life-sustaining treatments for patients with kidney failure. Specifically, we analysed eight aggressive treatments: more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to an ICU within 30 days before death, death in hospital, use of an ETT within 30 days before death, ventilator use within 30 days before death and the need for CPR within 30 days before death. The Patient Right to Autonomy Act was launched in 2019 but had been officially announced in 2016. Because the act focuses on choices surrounding life-sustaining treatment and artificial nutrition, we wanted to see how those treatments were used among patients with kidney failure before the act was launched, so we therefore added blood transfusion (BT) within 30 days before death, and nasogastric tube (NG)/percutaneous endoscopic gastrostomy/total parenteral nutrition (TPN) within 30 days before death as additional treatments to better understand the choices which patients experiencing kidney failure have.

Moreover, we calculated the aggressiveness of a two-treatment combination during the EOL by considering composite aggressive treatments, using a method adapted from work performed by Chang et al.⁴⁷ Composite indicators have the ability to condense the quality of care into a singular value, rendering them useful when evaluating, ranking and selecting healthcare providers, making them useful as an alternative to assessing performance based only on numerous individual indicators.⁴⁸ Opting for composite indicators over individual ones can also lead to heightened reliability, as the amalgamation of several individual indicators can result in a larger underlying number of observations.⁴⁹ We used the overall percentage method, in which the composite score was calculated as the total number of processes of care delivered to all patients divided by the total number of eligible care processes,⁵⁰ in order to create the composite aggressive treatments. The first composite aggressive treatment consisted of eight aggressive treatments modified from work undertaken by Earle et al and Tang et al, and included more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to ICU within 30 days before death, death in hospital, use of an ETT within 30 days before death, ventilator use within 30 days before death and requiring CPR within 30 days before death. We calculated the first aggressiveness rate as (number of aggressive treatments received/8)×100%. Because we wanted to understand the outcome specifically on medical service usage, we calculated the second treatment aggressiveness. The second was adapted from previous work performed by Earle et al,²³ where we deleted chemotherapy use and ultimately included five aggressive treatments, consisting of: more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to ICU within 30 days before death and death in hospital. We calculated the second aggressiveness rate as (number of aggressive treatments received/5)×100%. We considered them to be good composites to use considering the multiple EOL care outcomes and concerns.

Control variables

We took several related confounding variables into consideration, including age at death, gender, low income, living in a rural area and comorbidities based on prior empirical research.^{51 52} Data regarding comorbidities were calculated using Charlson’s Comorbidity Index (CCI),⁵³ a widely used comorbidity index in acute kidney injury and chronic kidney disease outcome studies.^{54 55} This index is the sum of weighted scores based on the presence or absence of 17 different medical conditions. The low-income qualification was regulated by Article 4 of the Public Assistance Act of Taiwan under the following conditions: (1) individuals are approved by their local municipality’s competent authority via application, (2) the average divided monthly income among each person in the household falls below the lowest living index and (3) total household assets do not exceed the specified amount announced by both the central and municipality competent authorities in the year of application.⁵⁶

Statistical analysis

Data were expressed as numbers with percentages and as an average with SD for categorised and continuous variables, respectively. The Pearson’s χ² test and the t-test were used to compare categorical and continuous variables, respectively. For aggressive treatments as binary variables, multivariable logistic regression was used to analyse the relationship between aggressive treatments at EOL and different palliative care models, adjusted for all of the above-mentioned confounders. For aggressiveness rates as continuous variables, multivariable linear regression was used to analyse the relationship between treatment aggressiveness and different palliative care models, adjusted for all of the above-mentioned confounders. A two-tailed p value <0.05 was considered statistically significant. Statistical analyses were performed using SAS V.9.4 (Statistical Analysis Software 9.4, SAS Institute).

Patient and public involvement

Patients nor the public were involved in the design, conduct or reporting of our research.

Results

Table 1 shows the distribution of the general characteristics of all participants. A total of 10 083 patients were enrolled, with 1786 (17.7%) patients with kidney failure receiving palliative care 1 year prior to death, and 8297 (82.3%) patients with kidney failure not receiving palliative care prior to death. The average age at death was 73.6 years. Between the palliative care group and the non-palliative care group, the distributions of age, low income, living in a rural area and CCI were statistically different. In the palliative care group, most patients (69%) received palliative care consultation service before death, while only 2.8% received palliative home care; 22.3% of patients received a mixed model, with the majority among them receiving inpatient palliative care with palliative care consultation service. Furthermore, 53.6% of patients received palliative care service in their last month of life.

Table 1.

Distribution of characteristics between palliative care group and non-palliative care group

Variables	Total	With PC	Without PC	P value
	n (%)/mean (SD)	n (%)/mean (SD)	n (%)/mean (SD)
Total	10 083 (100.0)	1786 (100.0)	8297 (100.0)	−
Gender
Male	5307 (52.6)	913 (51.1)	4394 (53.0)	0.166
Female	4776 (47.4)	873 (48.9)	3903 (47.0)
Age	73.6 (15.2)	73.5 (14.5)	72.7 (12.0)	0.013
Low income
Yes	395 (3.9)	41 (2.3)	354 (4.3)	<0.001
No	9688 (96.1)	1745 (97.7)	7943 (95.7)
Living in rural areas
Yes	3243 (32.2)	498 (27.9)	2745 (33.1)	<0.001
No	6840 (67.8)	1288 (72.1)	5552 (66.9)
CCI	3.9 (3.4)	5.0 (3.0)	3.6 (2.4)	<0.001
PC model
None	8297 (82.3)	0 (0.0)	8297 (100.0)	−
IPC only	106 (1.1)	106 (5.9)	−
PCCS only	1232 (12.2)	1232 (69.0)	−
PHC only	50 (0.5)	50 (2.8)	−
Mixed model	398 (3.9)	398 (22.3)	−
IPC+PCCS+PHC	−	0 (0.0)	−
IPC+PCCS	−	284 (71.4)	−
IPC+PHC	−	22 (5.5)	−
PCCS+PHC	−	92 (23.1)	−

CCI, Charlson’s Comorbidity Index; IPC, inpatient palliative care; PC, palliative care; PCCS, palliative care consultation service; PHC, palliative home care.

For the outcomes of all participants, the distribution of more than one ED visit within 30 days before death, more than one hospital admission within 30 days before death, a longer than 14-day hospitalisation within 30 days before death, admission to an ICU within 30 days before death, ETT insertion within 30 days before death, ventilator use within 30 days before death, need for CPR within 30 days before death and BT within 30 days before death were all significantly different between the palliative care group and the non-palliative care group (table 2). The average interval period between final dialysis and death was 30.0 days for the palliative care group, compared with 17.5 days for the non-palliative care group.

Table 2.

Outcomes between palliative care group and non-palliative care group

Variables	Total	With PC	Without PC	P value
	n (%)/mean (SD)	n (%)/mean (SD)	n (%)/mean (SD)
More than one ED visit within 30 days before death
Yes	5798 (57.5)	896 (50.2)	4902 (59.1)	<0.001
No	4285 (42.5)	890 (49.8)	3395 (40.9)
More than one admission within 30 days before death
Yes	5911 (58.6)	926 (51.8)	4985 (60.1)	<0.001
No	4172 (41.4)	860 (48.2)	3312 (39.9)
More than 14 days of hospitalisation within 30 days before death
Yes	1908 (18.9)	407 (22.8)	1501 (18.1)	<0.001
No	8175 (81.1)	1379 (77.2)	6796 (81.9)
Admission to ICU within 30 days before death
Yes	3426 (34.0)	493 (27.6)	2933 (35.4)	<0.001
No	6657 (66.0)	1293 (72.4)	5364 (64.6)
Death in hospital
Yes	4721 (46.8)	845 (47.3)	3876 (46.7)	0.666
No	5362 (53.2)	941 (52.7)	4421 (53.3)
ETT within 30 days before death
Yes	2689 (26.7)	230 (12.9)	2459 (29.6)	<0.001
No	7394 (73.3)	1556 (87.1)	5838 (70.4)
Ventilator use within 30 days before death
Yes	3483 (34.5)	404 (22.6)	3079 (37.1)	<0.001
No	6600 (65.5)	1382 (77.4)	5218 (62.9)
CPR within 30 days before death
Yes	2073 (20.6)	109 (6.1)	1964 (23.7)	<0.001
No	8010 (79.4)	1677 (93.9)	6333 (76.3)
Aggressiveness rate* (%)	37.2 (26.7)	30.2 (25.7)	38.7 (26.6)	<0.001
Aggressiveness rate† (%)	43.2 (29.8)	39.9 (30.7)	43.9 (29.5)	<0.001
BT within 30 days before death
Yes	3635 (36.1)	702 (39.3)	2933 (35.4)	0.002
No	6448 (63.9)	1084 (60.7)	5364 (64.6)
Use of NG, gastrostomy or TPN within 30 days before death
Yes	4893 (48.5)	855 (47.9)	4038 (48.7)	0.559
No	5190 (51.5)	931 (52.1)	4259 (51.3)

*Consisted of eight aggressive treatments, including more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to ICU within 30 days before death, death in hospital, endotracheal tube insertion within 30 days before death, ventilator use and CPR within 30 days before death.

†Consisted of five aggressive treatments, including more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to ICU within 30 days before death and death in hospital.

BT, blood transfusion; CPR, cardiopulmonary resuscitation; ED, emergency department; ETT, endotracheal tube; ICU, intensive care unit; NG, nasogastric tube; PC, palliative care; TPN, total parenteral nutrition.

Table 3 reveals the relationship between aggressive treatments and aggressiveness rates prior to death with palliative care among patients with kidney failure through multivariable logistic and linear regression, respectively. Palliative care prior to death significantly reduced the odds of more than one ED visit within 30 days before death, more than one admission within 30 days before death, more than 14 days of hospitalisation within 30 days before death, admission to an ICU within 30 days before death, ETT within 30 days before death and need for CPR within 30 days before death when compared with no palliative care. The two treatment aggressiveness measures were significantly reduced through palliative care (Aggressiveness 1, Estimates: −0.09, p<0.001; Aggressiveness 2, Estimates: −0.06, p<0.001).

Table 3.

Relationship between aggressive treatments and level of treatment aggressiveness before death with palliative care among patients with kidney failure by multivariable regression

	OR/estimate	95% CI	P value	C/R2	Likelihood ratio/F value	P value
More than one ED visit within 30 days before death	0.68	0.61 to 0.75	<0.001	0.547	68.13	<0.001
More than one admission within 30 days before death	0.62	0.56 to 0.69	<0.001	0.586	209.65	<0.001
More than 14 days of hospitalisation within 30 days before death	1.19	1.05 to 1.35	0.009	0.573	96.29	<0.001
Admission to ICU within 30 days before death	0.70	0.62 to 0.78	<0.001	0.561	97.69	<0.001
Death in hospital	0.93	0.83 to 1.03	0.152	0.600	308.29	<0.001
ETT within 30 days before death	0.38	0.33 to 0.44	<0.001	0.652	628.76	<0.001
Ventilator use within 30 days before death	0.83	0.69 to 1.01	0.059	0.598	261.75	<0.001
CPR within 30 days before death	0.23	0.18 to 0.28	<0.001	0.669	621.19	<0.001
Aggressiveness rate†	−0.09	−0.10 to 0.08	<0.001	0.027	46.26	<0.0001
Aggressiveness rate*	−0.06	−0.07 to 0.04	<0.001	0.013	22.67	<0.001
BT within 30 days before death	1.09	0.98 to 1.21	0.127	0.550	69.3917	<0.001
Use of NG, gastrostomy or TPN within 30 days before death	0.91	0.82 to 1.02	0.094	0.554	94.89	<0.001

Adjusted for age, gender, low income, living in rural areas and CCI.

*Consisted of five aggressive treatments, including more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to ICU within 30 days before death and death in hospital.

†Consisted of eight aggressive treatments, including more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to ICU within 30 days before death, death in hospital, ETT within 30 days before death, ventilator use within 30 days before death and CPR within 30 days before death.

BT, blood transfusion; CCI, Charlson’s Comorbidity Index; CPR, cardiopulmonary resuscitation; ED, emergency department; ETT, endotracheal tube; ICU, intensive care unit; NG, nasogastric tube; TPN, total parenteral nutrition.

In multivariable logistic regression, for aggressive treatments before death, patients receiving the mixed palliative care model had a reduced odds of experiencing more than one ED visit within 30 days prior to death, more than one admission within 30 days before death, more than 14 days of hospitalisation within 30 days before death, admission to an ICU within 30 days before death, in-hospital death, ETT within 30 days before death, need for CPR within 30 days before death and NG/percutaneous endoscopic gastrostomy/TPN within 30 days before death when compared with patients not receiving palliative care, while the patients receiving only palliative care consultation service benefited less (table 4). Palliative care consultation service was related to significantly increased odds of hospitalisation for more than 14 days, in-hospital death and need for BT. The goodness of fit in the multivariable regression model is shown in online supplemental table 1.

Table 4.

Relationship between aggressive treatments and level of treatment aggressiveness before death with palliative care among patients with kidney failure by multivariable regression

	PC model (reference: without PC)
	Only IPC		Only PCCS		Only PHC		Mixed model*
	OR/ estimate	P value	OR/ estimate	P value	OR/ estimate	P value	OR/ estimate	P value
More than one ED visit within 30 days before death	0.78	0.197	0.70	<0.001	0.68	0.172	0.59	<0.001
More than one admission within 30 days before death	0.11	<0.001	1.12	0.096	0.38	0.001	0.14	<0.001
More than 14 days of hospitalisation within 30 days before death	0.20	0.001	1.82	<0.001	0.34	0.041	0.18	<0.001
Admission to ICU within 30 days before death	0.34	<0.001	0.83	0.005	0.21	0.001	0.51	<0.001
Death in hospital	0.01	<0.001	2.00	<0.001	0.20	<0.001	0.07	<0.001
ETT within 30 days before death	0.16	<0.001	0.48	<0.001	<0.001	0.945	0.19	<0.001
Ventilator use within 30 days before death	0.55	0.158	0.93	0.527	0.180	0.085	0.72	0.087
CPR within 30 days before death	0.04	0.001	0.29	<0.001	<0.001	0.951	0.10	<0.001
Aggressiveness rate†	−0.25	<0.001	−0.02	0.004	−0.23	<0.001	−0.23	<0.001
Aggressiveness rate‡	−0.27	<0.001	0.03	<0.001	−0.20	<0.001	−0.26	<0.001
BT within 30 days before death	0.83	0.385	1.20	0.005	0.32	0.003	0.98	0.863
Use of NG, gastrostomy or TPN within 30 days before death	0.92	0.689	1.00	0.982	0.37	0.002	0.77	0.012

Adjusted for age, gender, low income, living in rural areas and CCI.

*Mixed: at least two PC services.

†Consisted of eight aggressive treatments, including more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to ICU within 30 days before death, death in hospital, ETT within 30 days before death, ventilator use within 30 days before death and CPR within 30 days before death.

‡Consisted of five aggressive treatments, including more than one ED visit within 30 days before death, more than one admission within 30 days before death, a longer than 14-day admission within 30 days before death, admission to ICU within 30 days before death and death in hospital.

BT, blood transfusion; CCI, Charlson’s Comorbidity Index; CPR, cardiopulmonary resuscitation; ED, emergency department; ETT, endotracheal tube; ICU, intensive care unit; IPC, inpatient palliative care; NG, nasogastric tube; PC, palliative care; PCCS, palliative care consultation service; PHC, palliative home care; TPN, total parenteral nutrition.

Discussion

Summary of the main findings

This population-based observational study explored 10 083 decedents who had experienced kidney failure while receiving maintenance dialysis in 2017, with the main findings of this study showing that among all decedents with kidney failure, only 17.7% of patients had received palliative care prior to death. Most patients (69%) received palliative care consultation service only, while 22.3% received a mixed palliative care model. The interval between the start of palliative care and death was 57.7 days on average. Compared with patients who had received no palliative care, patients who had received palliative care during EOL experienced a significant reduction in aggressive treatments. When compared with patients who had not received palliative care, patients who had received inpatient palliative care, palliative home care, palliative care consultation service and a mixed model experienced significantly reduced levels of treatment aggressiveness during EOL. However, the patients who had received palliative care consultation service benefited less when compared with the other palliative care models.

Patients with kidney failure and treatment involving palliative care

In our study, only a small portion of patients with kidney failure who were in dialysis received palliative care before death, and this result is consistent with previous studies. Couchoud et al followed up for 5 years on 51 834 patients with kidney failure who were in dialysis and discovered that only 3.6% had a palliative care-related hospitalisation. Even for patients who withdrew from dialysis, only 10% required a palliative care-related hospitalisation.⁵⁷ Murray et al explored Medicare/Medicaid Services in the USA and found that 13.5% of decedents with kidney failure used a hospice, while of those who withdrew from dialysis, 41.9% used a hospice.³¹ Another study conducted in a medical centre in the USA for adult patients on incident hemodialysis revealed that 34% of patients who withdrew from dialysis received palliative care services.⁵⁸ The reason why patients with kidney failure receiving dialysis and approaching EOL seldom receive palliative care could be multifactorial. Previous research has demonstrated that patients with kidney failure who are receiving dialysis have little choice regarding disease treatments and have less chance of expressing preference.⁵⁹ A national survey of palliative care physicians in the UK showed that the majority of specialist palliative care (SPC) services accept patients with kidney failure, but limited numbers are referred to the services available.⁶⁰ This illustrates that implementing the optimum model of care for such physiologically and psychologically affected patients is still warranted, which is consistent with prior research.⁶¹ Closer collaboration, together with better communication and education between renal and SPC services, could be the key to addressing this issue. Future studies are also warranted in order to better explore the association between referral timings and clinical outcomes for different palliative care models.

Different palliative care models for patients with kidney failure

From our results, most palliative care, such as inpatient palliative care, palliative home care and mixed model care during EOL can improve the quality indicators of EOL by significantly reducing the use of aggressive treatments prior to death. This finding is consistent with previous studies. Hsu et al discovered that palliative care consultation was associated with a higher odds of change in code status for DNR, as well as release from intensive care in patients with kidney failure who had experienced CPR.⁶² Chettiar et al also found that inpatient palliative care was associated with a 21% shorter length of hospital stay among decedents with kidney failure. Additionally, palliative care was also associated with a lower likelihood of rehospitalisation within 30 days post-discharge among non-decedents.³⁰ However, our analysis of the different palliative care models showed that palliative care consultation service was related to an increase in hospitalisations for more than 14 days, in-hospital death and BTs. These results were partly consistent with previous studies. Chen et al found that palliative care consultation service among patients who went through withdrawal of dialysis was associated with longer hemodialysis duration, hospitalisation within 30 days of death and death in hospital.⁵⁸

The reasons for palliative care consultation service having a relatively higher treatment aggressiveness rate could be as follows: first, the symptom burden among patients with kidney failure who received palliative care consultation service was higher. According to our results, patients receiving palliative care consultation service had a significantly increased likelihood of having a longer than 14-day admission within 30 days before death, as well as a greater risk of in-hospital death. This could be due to the increased care needs and heavy symptom burden, such as pain, among this population, and thus the palliative care consultation service team was sought.⁶³ Second, the time interval between first palliative care and death was relatively short. According to our results, this interval was 57.7 days, equivalent to approximately 2 months. This short interval period is unlikely to have had any notable effect on the level of treatment aggressiveness before death, particularly in this group of patients, which typically displays many symptoms.

In contrast, inpatient palliative care, palliative home care and the mixed care model all decreased treatment aggressiveness during EOL. Since most patients receiving mixed model care had received palliative care consultation service, we suggest that patients with kidney failure receiving dialysis not only receive this service, but should also be referred to inpatient palliative care or palliative home care, according to the care needs, wishes and preferences of each patient. In Taiwan, most terminal non-cancer patients are referred to a palliative care consultation service first if their clinical condition meets the criteria published by NHI, and then if the nephrologist, palliative care physician and patient or family agree that there is further care needed, the patient would be transferred to either inpatient palliative care or palliative home care service.³⁶ However, achieving this outcome necessitates both interdisciplinary collaboration and effective communication among patients and healthcare professionals, encompassing a shared comprehension of the concept of ‘terminal’ illness, the advantages of palliative care provision and the preferences of patients. To accomplish this goal, early consultation is needed. Any future studies should focus on the barriers surrounding early referral to a palliative care team among patients with kidney failure having a heavy care burden.

Strengths and limitations

This study has several strengths. First, it is the first known study to compare different palliative care models among patients with kidney failure receiving maintenance dialysis during EOL. We found that palliative care, particularly the mixed care model, as well as inpatient palliative care and palliative home care for patients with kidney failure receiving dialysis during EOL could significantly reduce the need for aggressive treatment during EOL. Second, this investigation is the first study to discuss palliative care and aggressiveness of treatment during EOL among patients with kidney failure in Asia, a region in which the number of patients with kidney failure is growing rapidly. Third, this population-based study design involved a large number of patients, which in turn elevated the scope of our study.

Our study also has some limitations. First, there were no laboratory data such as blood tests which could be retrieved from the NHIRD. Thus, there may have been some important lab data related to morbidity, but these were not included as confounding factors in multivariable analysis. However, we did include CCI in our final analysis so as to to minimise this limitation. Second, though this study was population-based and had a large sample size, we only analysed a Taiwanese population, so the generalisability of these findings is limited. However, as noted above, Asia is an area with the fastest growth rate of hemodialysis patients, so we believe that our results are still representative.⁴³ Third, the patients’ and their families’ preference of treatment and preference for place of death were not recorded in the NHIRD.⁶⁴ Future research is still warranted in order to investigate patients’ preferences on these matters and their relation to aggressive treatments during EOL.

Implications and suggestions

Our results show that most decedents who had undergone dialysis received palliative care consultation service as the sole palliative care resource. However, patients who had received this palliative care consultation service benefited less when compared with the other palliative care models. This may suggest that the process of involving palliative care consultation service in patients with kidney failure should be re-examined and re-organised, with a particular emphasis being placed on collaboration between nephrologists and the palliative care team. This could be achieved through regular comprehensive prognosis assessments of all dialysis patients using established tools involving further timely referrals and a curriculum for interprofessional clinicians to exchange thoughts and ideas.⁶⁵ Meanwhile, for patients under palliative care consultation service, regular meetings should be held to discuss if future inpatient palliative care or palliative home care is needed. Moreover, to better understand patients’ preferences and thoughts, advance care planning plays an important role. Previous qualitative research has demonstrated that patients experiencing kidney failure actually wish to have high-quality conversations with their physicians regarding further treatments.⁶⁶ Further efforts should be undertaken to establish how best to initiate palliative care consultation service early and link the service to other palliative care models, such as inpatient palliative care and palliative home care, while also determining how to implement early advance care planning for patients experiencing kidney failure.

Conclusions

This population-based observational study analysed quality of care during EOL among more than 10 000 decedents with kidney failure who were receiving dialysis, with the results showing that palliative care, particularly the mixed care model, inpatient palliative care and palliative home care, could significantly reduce aggressive treatment during EOL. Health policy makers should incorporate the palliative care approach into the clinical pathway of patients with kidney failure, while also facilitating cooperation between nephrologists and palliative care teams in order to improve each patient’s quality of life.

Data availability statement

Data are available upon reasonable request. The datasets used and analysed during the current study are not publicly available but are available from the corresponding author on reasonable request with the permission of the Ministry of Health and Welfare, Taiwan.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of National Taiwan University, Taiwan (Approval No. 201801056RIND). The Institutional Review Board of National Taiwan University approved this study and deemed that informed consent should be waived owing to the retrospective nature of the study.