Article Text

other Versions

Association of advance care planning with place of death and utilisation of life-sustaining treatments in deceased patients at Taipei City Hospital in Taiwan
  1. Yung-Feng Yen1,2,3,4,5,
  2. Ya-Ling Lee5,6,7,
  3. Hsiao-Yun Hu2,4,5,
  4. Wen-Jung Sun8,9,10,11,
  5. Ming-Chung Ko3,5,12,
  6. Shen-Shong Chang2,13,14,
  7. Chu-Chieh Chen3,
  8. Sheng-Jean Huang15,16 and
  9. Dachen Chu2,3,5,15
  1. 1Section of Infectious Diseases, Taipei City Hospital Yangming Branch, Taipei, Taiwan
  2. 2Institute of Public Health, National Yang-Ming University, Taipei, Taiwan
  3. 3Department of Health Care Management, National Taipei University of Nursing and Health Sciences, Taipei, Taiwan
  4. 4Department of Education and Research, Taipei City Hospital, Taipei, Taiwan
  5. 5University of Taipei, Taipei, Taiwan
  6. 6Department of Dentistry, Taipei City Hospital, Taipei, Taiwan
  7. 7Department of Dentistry, School of Dentistry, National Yang-Ming University, Taipei, Taiwan
  8. 8Department of Community Medicine, Taipei City Hospital, Taipei, Taiwan
  9. 9Center of Research and Development in Community Based Palliative Care, Taipei City Hospital, Taipei, Taiwan
  10. 10Holistic Mental Health Center, Taipei City Hospital, Taipei, Taiwan
  11. 11Community Medicine Department & Family Medicine Division, Taipei City Hospital Zhongxing Branch, Taipei, Taiwan
  12. 12Department of Urology, Taipei City Hospital, Taipei, Taiwan
  13. 13Division of Gastroenterology, Department of Internal Medicine, Taipei City Hospital Yangming Branch, Taipei, Taiwan
  14. 14School of Medicine, National Yang-Ming University, Taipei, Taiwan
  15. 15Department of Neurosurgery, Taipei City Hospital, Taipei, Taiwan
  16. 16Department of Surgery, Medical College, National Taiwan University Hospital, Taipei, Taiwan
  1. Correspondence to Dr Yung-Feng Yen, Section of Infectious Diseases, Taipei City Hospital Yangming Branch, Taipei 10341, Taiwan; yfyen1{at}gmail.com

Abstract

Objective Evidence is mixed regarding the impact of advance care planning (ACP) on place of death. This cohort study investigated the effect of ACP programmes on place of death and utilisation of life-sustaining treatments for patients during end-of-life (EOL) care.

Methods This prospective cohort study identified deceased patients between 2015 and 2016 at Taipei City Hospital. ACP was determined by patients’ medical records and defined as a process to discuss patients’ preferences with respect to EOL treatments and place of death. Place of death included hospital or home death. Stepwise logistic regression determined the association of ACP with place of death and utilisation of life-sustaining treatments during EOL care.

Results Of the 3196 deceased patients, the overall mean age was 78.6 years, and 46.5% of the subjects had an ACP communication with healthcare providers before death. During the study follow-up period, 166 individuals died at home, including 98 (6.59%) patients with ACP and 68 (3.98%) patients without ACP. After adjusting for sociodemographic factors and comorbidities, patients with ACP were more likely to die at home during EOL care (adjusted OR (AOR)=1.71, 95% CI 1.24 to 2.35). Moreover, patients with ACP were less likely to receive cardiopulmonary resuscitation (AOR 0.36, 95% CI 0.25 to 0.51) as well as intubation and mechanical ventilation support (AOR 0.54, 95% CI 0.44 to 0.67) during the last 3 months of life.

Conclusion Patients with ACP were more likely to die at home and less likely to receive life-sustaining treatments during EOL care.

  • end of life care
  • hospice care

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Advance care planning (ACP) is the process of discussing patient preferences concerning place of death and goals of care for patients who may lose the capacity for communication ability in the future.1 During the ACP communication with healthcare providers, patients could complete advance directives with respect to their preferred place of death and end-of-life (EOL) treatments. ACP has been promoted for decades in the USA2 and the UK.3 A previous randomised trial found that ACP, when implemented on hospitalised patients over the age of 80 years, could improve patients’ EOL care and reduce family stress, anxiety and depression.4

ACP is less common in East Asia.5 Taiwan has been a leader in Asia regarding the promotion of ACP and palliative care,6 and launched the Patient Self-Determination Act in 2015.7 Medical institutions in Taiwan, by law, are required to provide ACP consultation for terminally ill patients.

Place of death is an indicator to evaluate the quality of EOL care.8A previous study reported that terminally ill patients who died at home were more likely to have a higher quality of life during EOL care compared with those who died in a hospital or intensive care unit.9 Two prior meta-analyses showed that factors associated with an increased likelihood of home death included multidisciplinary home palliative care,10 preference for home death,10 11 cancer,10 early referral to palliative care,10 not living alone,10 11 having a caregiver10 and extended family support.11 However, patients with prior hospital admission were less likely to die at home during EOL care.10 11 Since dying at home is more natural and has a higher quality of EOL care than dying at a hospital,9 12 the majority of patients would prefer to die at home.13 However, many patients nearing the end of their lives do not die in their preferred location.14 Moreover, the promotion of home death for patients and family has been challenged because of the easy access to well established palliative care services in hospitals.15

ACP discusses patients’ preferred place of death and enables patients to make plans regarding their favoured location of death during EOL care. However, studies determining the impact of ACP on patients’ place of death have had inconsistent findings. Three prior retrospective cohort studies showed that ACP was significantly associated with greater odds of dying at home,16–18 but another randomised study found no significant association between ACP and place of death.19 These previous studies of the association between ACP and place of death, however, enrolled only those with heart failure19 or cancer patients,17 or had a limited sample size (n<110),17 19 and were inadequately controlled for potential confounders such as patients’ functional status.16–19

Since 2015, Taipei City Hospital (TCH) has launched a large-scale palliative care programme to promote ACP for hospitalised patients.20 When patients are admitted to TCH, their palliative care needs are evaluated using the Palliative Care Screening Tool (PCST).21 If patients’ PCST score is ≥4 at the time of hospitalisation, they are informed about ACP meeting to discuss their preferred place of death and goals of treatment during EOL care. Through the ACP communication with healthcare providers, patients can consider and document their preferences regarding the place of death and life-sustaining treatments during EOL care.

Understanding the impact of ACP on the place of death and the utilisation of life-sustaining treatments during EOL care is imperative to guide future healthcare policy. Thus, this interventional cohort study aimed to investigate the effect of ACP programme on the place of death and utilisation of life-sustaining treatments in the patients in Taipei, Taiwan.

Methods

Background information and study subjects

TCH is a 4700-bed healthcare organisation in Northern Taiwan. Since 2015 it has initiated a palliative care programme to provide ACP for the patients and to identify those needing palliative care.20To promote ACP and improve patients’ palliative care, TCH held a series of palliative training programmes for all healthcare providers from 2015.22 These palliative training programmes included: (1) training in communication skills with patients during the ACP discussion; (2) education regarding palliative care for patients with terminal illness and (3) education on respecting patients’ decisions associated with the place of death and goal of care during EOL care.

When patients are admitted to TCH, their palliative care needs are evaluated using the PCST.21 The checklists of PCST include four categories: (A) primary diseases associated with palliative care, (B) secondary comorbidities, (C) functional status score according to the Eastern Cooperative Oncology Group (ECOG) Performance Status, and (D) frequency of disease exacerbations.21 If a patient’s PCST score is four points or higher, ACP meetings are informed for the patients to discuss their preferred place of death and goal of treatment during EOL care. ACP meetings are also informed for patients nearing the end of their lives.

Research ethics

This cohort study comprised all patients aged ≥18 years who were admitted to TCH from 2015 to 2016. The TCH dataset was linked to the Taiwanese death certificate database before being accessed by the researchers. All information that allows a specific individual patient to be identified was encrypted.

Outcome variables

The primary outcomes were place of death and life-sustaining treatments during EOL care in patients. Place of death in decedent patients was classified as home or hospital death, which was confirmed by examining the Taiwanese death certificate database.23 Life-sustaining treatments in the last 3 months of life included cardiopulmonary resuscitation as well as intubation and mechanical ventilation support,24 which were determined by patients’ medical records.

Main explanatory variable

The main explanatory variable was ACP communication with healthcare providers during patients’ hospitalisations, which was recorded by the physicians. ACP was defined as a process to discuss patients’ preferences with respect to place of death and the goals of EOL care when patients were in critical condition and lost their decision-making capacity.25

Controlling variables

Covariates identified in previous studies10 11 26 as factors associated with place of death or life-sustaining treatments were assessed in our analyses; these included the individuals’ sociodemographic (age and gender), comorbidities, ECOG Performance Status and frequency of underlying disease exacerbations. Comorbidity was determined by patients’ medical records and included heart failure, cancer, liver cirrhosis, end-stage renal disease and dementia. Heart failure statuses included normal, mild (New York Heart Association (NYHA) class I and II), moderate (NYHA class III) and severe (NYHA class IV).27 ECOG functional status consisted of five categories, ranging from grade 0 ‘fully active’ to grade 4 ‘completely disabled’.28 When patients were admitted to TCH, the frequency of underlying disease exacerbations was evaluated using the following questions: ‘Did the patient visit the Emergency Department more than once, for the same diagnosis, in the last 30 days?’ and ‘Was the patient hospitalised more than once for the same diagnosis in the last 30 days?’

Statistical analysis

First, the demographic data of the study subjects were analysed. Continuous data are presented as the mean (SD), and the two-sample t-test was used for comparisons between patients with and without ACP communication with healthcare providers. Categorical data were analysed using the Pearson χ2 test, where appropriate.

We assessed the crude associations of ACP and other covariates with the outcome (place of death and life-sustaining treatments during the last 3 months of life) by computing the ORs and corresponding 95% CIs. We calculated the attributable proportion and population attributable fraction of the outcome (place of death and life-sustaining treatments during the last 3 months of life) due to ACP. The attributable proportion was defined as the theoretical proportion of the outcomes of interest (home death and life-sustaining treatments during the last 3 months of life) in patients with ACP when the explanatory variable (ACP) was implemented. The attributable proportion was calculated using the following equation:

Attributable proportion among patients with ACP = |(OR-1)|/OR where OR indicates the OR of ACP associated with the outcome (home death and life-sustaining treatments during the last 3 months of life).

The population attributable fraction was defined as the theoretical proportion of the outcome of interest (home death and life-sustaining treatments during the last 3 months of life) in the entire population when the explanatory variable (ACP) was implemented. The population attributable fraction was calculated using the following equation:

Population attributable fraction = (proportion of patients with ACP) x (attributable proportion among patients with ACP).

A stepwise logistic regression was used to estimate the association of ACP with home death and the utilisation of life-sustaining treatments after adjusting for potential confounders.29 30 A variable with p<0.05 was defined as a significant factor associated with the outcomes in the multivariate analysis. Adjusted ORs (AOR) with 95% CIs were presented to indicate the direction and strength of these associations.

To evaluate the robustness of the main findings, a propensity score analysis was conducted to evaluate the associations of ACP with home death and the utilisation of life-sustaining treatments during EOL care. Logistic regression was used to calculate the probability of patients receiving ACP communication with healthcare providers according to their age, sex, ECOG Performance Status, CCI and frequency of underlying disease exacerbations. A stepwise logistic regression was used to estimate the association of ACP with home death and the utilisation of life-sustaining treatments after adjusting for patients’ propensity score. To evaluate the interaction of ACP with age and sex, subgroup analyses were conducted after stratifying patients by age and sex. All data analyses and management were performed using SAS V.9.4 software package (SAS Institute).

Results

Participant selection

This cohort study included 3210 decedent patients at TCH from 2015 to 2016. After excluding those younger than 18 years (n=6), those with unknown diagnosis (n=7), and those with unknown places of death (n=1), the remaining 3196 subjects were included in the analysis. The overall mean (SD) age was 78.6 (14.1) years and 56.9% of the subjects were male. Of all study subjects, 1487 (46.5%) patients had an ACP communication with healthcare providers before death. Moreover, of all patients with an ACP, the mean (SD) time to death was 44.2 (71.7) days from the ACP communication with healthcare providers.

Characteristics of deceased patients with and without ACP

Table 1 shows the characteristics of patients with and without ACP. As compared with individuals without ACP communication with healthcare providers, patients with ACP communication with healthcare providers had a higher CCI and PCST. There was no significant difference regarding the age, gender, and ECOG performance status between the groups.

Table 1

Characteristics of deceased patients, by advance care planning

Place of death and life-sustaining treatments in patients

During the study follow-up period, 166 individuals died at home, including 98 (6.59%) patients with ACP and 68 (3.98%) patients without ACP (table 1). In terms of life-sustaining treatments, 520 patients received life-sustaining treatments during the last 3 months of life, including 171 (11.49%) patients with ACP and 349 (20.42%) patients without ACP. Moreover, 183 patients received cardiopulmonary resuscitation during the last 3 months of life, including 45 (3.02%) patients with ACP and 138 (8.07) patients without ACP. Furthermore, 457 subjects received intubation and mechanical ventilation support during the last 3 months of life, including 154 (10.35%) patients with ACP and 303 (17.73%) patients without ACP.

Association of ACP with home death and life-sustaining treatments in patients

Table 2 shows the univariate of factors associated with home death and life-sustaining treatments in patients. Patients with ACP were more likely to die at home (OR 1.70, 95% CI 1.24 to 2.34) but were less likely to receive cardiopulmonary resuscitation (AOR 0.36, 95% CI 0.25 to 0.50) as well as intubation and mechanical ventilation support (AOR 0.54, 95% CI 0.44 to 0.66) during the last 3 months of life compared with those without ACP. The population attributable fractions of home death, cardiopulmonary resuscitation, and intubation and mechanical ventilation support among patients with ACP were 30.49%, 29.04% and 19.33%, respectively (online supplemental table 1).

Table 2

Univariate analyses of factors associated with home death and life-sustaining treatments among deceased patients

Multivariate analysis showed that, after controlling for the sociodemographic factors, ECOG Performance Status and CCI, patients with ACP had a higher likelihood of dying at home during EOL care (AOR 1.71, 95% CI 1.24 to 2.35) (table 3). Moreover, patients with ACP had a lower likelihood of receiving cardiopulmonary resuscitation (AOR 0.36, 95% CI 0.25 to 0.51) as well as intubation and mechanical ventilation support (AOR 0.54, 95% CI 0.44 to 0.67) during the last 3 months of life compared with those without ACP.

Table 3

Multivariate analyses of factors associated with home death and life-sustaining treatments among deceased patients

Propensity score analysis for the association of ACP with home death and the utilisation of life-sustaining treatments

The propensity score analysis showed that patients with ACP were more likely to die at home during EOL care (AOR 1.65, 95% CI 1.20 to 2.29). Moreover, patients with ACP were less likely to receive cardiopulmonary resuscitation (AOR 0.36, 95% CI 0.25 to 0.51) and intubation and mechanical ventilation support during the last 3 months of life (AOR 0.55, 95% CI 0.44 to 0.68).

Subgroup analysis for the association of ACP with place of death and life-sustaining treatments

Figure 1 shows the results of subgroup analyses of the association between ACP with place of death and life-sustaining treatments after stratifying participants by age and sex. ACP was significantly associated with a higher likelihood of dying at home during EOL care in patients aged 65 years or older and male subjects. Moreover, ACP was significantly associated with a lower likelihood of receiving cardiopulmonary resuscitation as well as intubation and mechanical ventilation support during EOL care in all patient subgroups.

Figure 1

Sensitivity analysis for the associations of advance care planning with home death and life-sustaining treatments after adjusting for patient characteristics. AOR, adjusted OR.

Discussion

In this cohort study of 3196 deceased patients, the overall rate of ACP communication with healthcare providers was 46.5%. After adjusting for demographic factors, ECOG performance status and CCI, patients with ACP communication with healthcare providers were more likely to die at home during EOL care compared with those without ACP communication with healthcare providers. Moreover, patients with ACP were less likely to receive cardiopulmonary resuscitation as well as intubation and mechanical ventilation support during the last 3 months of life compared with those without ACP.

Our study revealed robust associations of ACP with home death and life-sustaining treatments after stratifying patients by age and sex. ACP communication with healthcare providers was significantly associated with home death in patients aged 65 years or older and in male subjects. Moreover, ACP significantly lowered the likelihood of receiving cardiopulmonary resuscitation as well as intubation and mechanical ventilation support in the last 3 months of life in all patient subgroups.

This study found that the percentage of home deaths among deceased patients at TCH was 5.2%, which was lower than that of deceased patients with cancer in Taiwan (32.4%)31 and deceased patients in Japan (13.0%)32 and Norway (14.2%).33 Taiwan government, since 1995, has launched a universal, national health insurance programme to provide affordable healthcare services for all residents.34 Consequently, easy access to hospital services under a national health insurance program34 may account for the lower proportions of dying at home in Taiwanese patients during EOL care. Since dying at home is associated with a good quality of dying,9 our study suggests that it is important to proactively promote home death for patients during EOL care.

ACP is intended to respect patients’ wishes regarding the place of death and the goals of care, and it has been promoted for many years in Western countries.2 3However, the effect of ACP on the place of death during EOL care has been unclear. A UK study using the National Survey of Bereaved People found that decedents with ACP had significantly greater odds of dying at home than in hospital (OR 6.25; 99% CI 5.56 to 7.14).16 Another study involving 107 patients with cancer at home visit nursing agencies in Japan showed that cancer patients with ACP were more likely to die at home than those without ACP (OR 41.76; 99% CI 5.87 to 297.07).17 However, a randomised study recruiting 97 individuals in Singapore found that ACP was not significantly associated with the place of death in patients with heart failure.19 The present cohort study found that patients with ACP were more likely to die at home during EOL care. When TCH initiated a large-scale palliative care programme to provide ACP meetings for the patients in 2015, healthcare providers proactively discussed patients’ preferred place of death during the ACP communication with the patients. If patients preferred to die at home, social support and palliative care home visits were offered for the patients during their EOL care. Since home is most patients’ preferred location of death and is associated with good quality of dying,9 13 the findings of our study suggest that it is important to provide ACP consultation for patients to discuss their preferred place of death during EOL care.

The present cohort study found that, compared with patients without ACP, patients with ACP had 64% and 45% lower rates of receiving cardiopulmonary resuscitation as well as intubation and mechanical ventilation support, respectively, in the last 3 months of life. ACP communication with healthcare providers may account for the lower rate of receiving life-sustaining treatments in patients with ACP during EOL care. While healthcare providers had an ACP meeting with patients at TCH, patients’ goals of treatment during EOL care was discussed and emphasised. Through the ACP communication with healthcare providers, patients could consider and document their treatment preferences regarding EOL care. A previous report showed that patients with ACP have stable preferences for future treatments35 and may reduce their likelihood of receiving unwanted aggressive treatments during EOL care. Since patients’ wishes with respect to EOL care were more likely to be followed if they have an ACP communication with healthcare providers, the findings of our study suggest that it is imperative to promote ACP for patients nearing the end of their lives.

This study found that patients with ACP had 1.71-fold increased likelihood of dying at home during EOL care than those without ACP. However, the population attributable fraction of home death due to ACP was only 30.49%. Since home death was associated with a higher quality of life in patients during EOL care,9 our study suggests that it is imperative to proactively promote ACP and discuss the place of death for terminally ill patients.

Our study was the first large-scale palliative programme to provide ACP for patients in Eastern Asia. Although palliative programmes to promote ACP are less common in Eastern Asia,5 our study demonstrates the feasibility of implementing an ACP programme to discuss patient preferences concerning the place of death and goals of care.

Nonetheless, several limitations should be considered in interpreting our findings. First, there may be important factors (eg, patients’ religion) associated with the decision of the preferred place of death or receiving life-sustaining treatments, which were not collected in this study. Second, patients’ decisions regarding the preference of place of death and life-sustaining treatments during the ACP communication with healthcare providers was not available in this study. This study used an ACP as the proxy for patients’ preference of place of death and life-sustaining treatments during EOL care, which would lead to an underestimation regarding the association of patients’ preference of place of death and life-sustaining treatments with each outcome. Third, none of the 3210 deceased patients died in a nursing home, which precluded this study from determining the impact of ACP on nursing home death. Finally, the external validity of our findings may be a concern because almost all of our study subjects were deceased patients at TCH in Taiwan. The generalisability of our results to other hospital settings and non-Asian ethnic groups thus requires further verification.

Conclusion

This population-based cohort study found that the overall rate of ACP communication with healthcare providers was 46.5% in decedents in Taiwan. After adjusting for demographics, ECOG performance status and comorbidities, patients with ACP were more likely to die at home and were less likely to receive life-sustaining treatments during EOL care. Since the home is most patients’ preferred location of death and is associated with good quality of dying, the findings of our study suggest that it is important to provide ACP consultation for patients to discuss their preferred place of death and goals of care during EOL care.

Acknowledgments

The authors are grateful to the members of the Research Office for Health Data, Department of Education and Research, Taipei City Hospital, Taiwan for their valuable contributions in data management and statistical analysis.

References

Footnotes

  • Y-FY, Y-LL and H-YH contributed equally.

  • Contributors Y-FY, Y-LL, H-YH, W-JS, M-CK, S-SC, C-CC, S-JH and DC substantially contributed to the conception and design of the study, data analysis, data interpretation and the drafting of the manuscript. Y-FY, Y-LL, H-YH, W-JS, M-CK, S-JH and DC substantially contributed to data acquisition and interpretation of the results. Y-FY, Y-LL, H-YH, W-JS, M-CK, S-SC, C-CC, S-JH and DC all approved the final version of the manuscript.

  • Funding This study was supported by a grant from the Department of Health, Taipei City Government, Taiwan.

  • Disclaimer The study sponsor was not involved in the study design, the collection, analysis, or interpretation of the data, the writing of this report, or the decision to submit it for publication.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval This study was approved by the Institutional Review Board of TCH (TCHIRB-10502121-E).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available on reasonable request. This prospective cohort study analysed the palliative data at Taipei City Hospital. All information that allows a specific individual patient to be identified was encrypted. After encryption of the data, we collected patients’ demographics, comorbidities, place of death and life-sustaining treatments during EOL care.