Place of death outcomes of home ventilation patients in Singapore: a retrospective cohort study using a linked database

Introduction

Background

Patients under the care of Tan Tock Seng Hospital’s (TTSH’s) Home Ventilation and Respiratory Support Service (HVRSS) in Singapore mostly have chronic respiratory failure from progressive medical conditions needing long-term home ventilation support. The median survival for HVRSS patients ranged from 1.8 to 6.7 years, among those with motor neuron disease (MND) like amyotrophic lateral sclerosis (ALS), complex intensive care unit stays, spinal cord injury (SCI), and other neuromuscular and chest wall diseases (NMCWs) conditions (1). There is some international evidence on home death rates of deceased patients receiving home ventilation. An Italian study found that 46.4% of those studied died at home (2), and a UK-based study of 1,210 home ventilation patients found that 20.0% of those studied died at home, although a large proportion in the study had an unknown place of death (POD) (3). Specifically, in a review of end-of-life management for those with ALS who constitute the majority of HVRSS patients, there is also considerable variation in home death rates worldwide, ranging from 36% (France) to 65% (US), and even up to 85% (Italy) (4). Singapore, as a multi-cultural city-state in Southeast Asia, with its population made up of four main ethnicities (Chinese, Malay, Indian, and Eurasian), values and beliefs regarding preferences for POD may differ from the West. As the available studies concerning these home death outcomes are among cohorts in Western populations, these may not be representative of the cultural context in Singapore.

Regarding preferences, high variability (21–100%) was observed in a systematic review that examined concordance between actual POD and preferred POD (PPOD) for cancer and non-cancer patients at end-of-life (5). To the best of our knowledge, there does not appear to be published evidence on POD concordance with preferences among patients requiring long-term home ventilation support globally, which is an important consideration for end-of-life care for those with chronic, progressive, complex medical conditions. Especially so, as concordance with POD preferences is increasingly being recognised as an important end-of-life care quality indicator (6).

In Singapore, end-of-life preferences like cardiopulmonary resuscitation in the event of a cardiac arrest, type of medical intervention during deterioration, preferred place of terminal care, and PPOD can be documented in a process of Advance Care Planning (ACP). Importantly, ACP allows individuals to share their values and how these values shape end-of-life preferences. The prevalence of ACP uptake among HVRSS patients is lower compared to elsewhere (44% vs. 76% in the US) (7,8).

Rationale and knowledge gap

The survival of HVRSS patients, although a heterogenous group, is predictably limited. End-of-life care of these patients is complicated by specialised respiratory care needs requiring frequent multi-disciplinary inputs and highly technical medical equipment, and communication difficulties, which likely impact end-of-life outcomes. Beyond clinical needs, the heterogeneity of HVRSS patients with differential clinical trajectories under the care of the programme and the interaction of clinical care with local cultural values may also affect end-of-life outcomes. Despite these considerations, information regarding the end-of-life circumstances of HVRSS patients is lacking. Specifically, there is no available information currently on different POD rates, and importantly, how concordant preferences are with actual POD for HVRSS patients. Understanding the current state of POD circumstances would be a critical first step towards any future planning of improvement initiatives seeking to improve end-of-life care for HVRSS patients.

Objectives

We aimed to examine the distribution of deceased HVRSS patients by PODs, over a 10-year period (e.g., home, hospital, nursing home, hospice, etc), and explore clinical factors which may be associated with specific death sites. Additionally, the secondary aim of this study was to understand the extent to which end-of-life preferences like PPOD is in agreement with actual POD among deceased HVRSS patients. These are pertinent information that can facilitate planning of high-quality end-of-life care for HVRSS patients in the future. We present this article in accordance with the STROBE reporting checklist (available at https://apm.amegroups.com/article/view/10.21037/apm-25-94/rc).

Methods

Design and study population

A retrospective cohort study was conducted. The HVRSS clinical team was based at TTSH. The inclusion criteria of the population studied encompassed the following reflected in their medical documentation:

• Patients enrolled in HVRSS between 2009 and 2019;
• Decedent HVRSS patients who were deceased by 2019;
• Patients who were discharged home with at least one HVRSS home or outpatient clinic visit after programme enrolment.

Outcomes, data, and sources

The primary outcome of the study was the POD of study subjects. This was obtained from an administrative dataset for POD and other death-related information, like date and causes of death (CODs), for deaths in Singapore. The secondary outcome of the study was the PPOD of study subjects available for those who had completed ACP with any TTSH staff. Besides PPOD, available ACP information includes other formally documented end-of-life preferences [(I) cardiopulmonary resuscitation (to proceed or not during cardiac arrest); (II) medical intervention (comfort measures, limited intervention, or full treatment during deterioration of condition); and (III) place of terminal care]. Additionally, we collected other information like socio-demographic factors and clinical information like main HVRSS diagnosis, pre-existing comorbid chronic health conditions, HVRSS enrollment duration, tracheostomy status, type of ventilation support, and feeding type at death, was also collected.

After relevant study outcomes and other variables required were prepared, the data sources used were linked, and a de-identified dataset was assembled for final analysis. In doing so, we were able to determine our primary outcome of POD distribution, explore clinical factors associated with specific PODs, and check agreement between POD and PPOD information among study subjects with completed ACP using the aforementioned linked de-identified dataset. For example, study subjects who died at home and indicated in their ACP that home was PPOD were considered a concordant case. For study subjects whose preference involved more than one POD option, like “home or hospital”, and the actual POD was either home or in hospital, this was deemed as agreement. Study subjects who expressed no preference or did not discuss PPOD during ACPs were excluded from the determination of levels of agreement between POD and PPOD.

Statistical analysis

The sample size was based on actual decedent numbers derived from the HVRSS decedent list, which fulfilled the inclusion criteria. We computed descriptive statistics of socio-demographic and clinical characteristics [means and standard deviations (SDs) for continuous variables, numbers and percentages for categorical variables], for our study population and across death-site subgroups. Thereafter, we used logistic regression to explore clinical factors that may be associated with specific death sites. Lastly, we determined the concordance for POD and PPOD among study subjects with valid PPOD through an agreement analysis. Our POD and PPOD agreement analysis included percentage agreement (not chance corrected) and Gwet’s AC1 (chance corrected), less affected by prevalence and marginal probability, while measuring reliability.

Ethical consideration

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the National Healthcare Group Domain Specific Review Board in Singapore (ethics approval DSRB#: 2019/00854). In addition, the study team obtained applicable data access approvals to carry out the data access and linkages to enable necessary statistical analysis. Data access and linkage procedures performed in this study adhered to data security governance requirements of the organisations involved. Given the retrospective nature of the study, the requirement for informed consent was waived; however, informed consent was obtained from all patients and their caregivers participating in the prospective component.

Results

Death-related, socio-demographic, and clinical information

We included 118 study subjects for analysis, and the inclusion process for study subjects has been previously outlined (8). Among 118 study subjects, we were able to establish that 52.5% (n=62) died in hospitals, 44.1% (n=52) had home deaths, and 3.4% (n=4) died elsewhere (Figure 1). Notably, no patient died in an inpatient hospice setting. Further, POD outcomes were similar between ACP status groups (P=0.73). The top 2 CODs recorded were chronic neurological disorder (n=47, 39.8%) and others (n=45, 38.1%), the category that usually captures pneumonia deaths (Figure 2). Cancer-related deaths constituted 9.3% (n=11), whilst end-stage organ failure (chronic obstructive pulmonary disease, end-stage renal failure, chronic heart failure: n=12, 10.2%) and stroke (n=2, 1.7%) made up the rest of COD types recorded for study subjects. One study subject had an unknown COD.

Figure 1 POD of study participants by ACP status. ACP, Advance Care Planning; POD, place of death.

Figure 2 COD of study participants. CHF, chronic heart failure; COD, cause of death; COPD, chronic obstructive pulmonary disease; ESRF, end-stage renal failure.

We show socio-demographic information of our study subjects overall and broken down into those with home and hospital deaths in Table 1. The mean overall HVRSS enrolment age was 64.2 (SD, 16.1) and 66.0 (SD, 16.6) years at death. Overall clinical characteristics of the study population, and by home or hospital POD, are presented in Table 2. The most common main HVRSS diagnosis was MND (n=56, 47.5%), and the mean HVRSS enrolment duration was 1.8 (SD, 1.8) years.

Home- vs. hospital-death comparison

We examined clinical factors (diagnosis, ventilation status at death, and feeding type at death) in a logistic regression analysis comparing study subjects who died at home vs. those who died in hospital. The analysis was adjusted for socio-demographic variables such as gender, ethnicity, marital status, living arrangements of the primary familial caregiver, and age at death. Within our study population, we found that the following clinical features were associated with a higher likelihood to die at home:

• Diagnosed with MND [odds ratio (OR) =5.6; 95% confidence interval (CI): 1.2–25.0; P=0.02], compared to those diagnosed with central, restrictive, obstructive, complex, or rare pathologies, or obesity syndrome, as their main HVRSS diagnosis;
• On invasive ventilation at the time of death (OR =14.9; 95% CI: 3.8–58.3; P<0.001), compared to those on non-invasive ventilation;
• Receiving oral feeding at the time of death (OR =3.8; 95% CI: 1.0–14.0; P=0.044), compared to those fed via nasogastric tube, nasojejunal tube, percutaneous endoscopic gastrostomy, or percutaneous endoscopic jejunostomy.

Actual and preference POD agreement

For end-of-life preferences, 52 study subjects (44.1%) had completed ACP prior to death. The details of available PPOD choices of study subjects mapped to their actual POD data are tabulated in Table 3. The majority of study subjects with ACP indicated that they wanted to die at home (n=38, 73.1%). Four study subjects had combination choices. Three study subjects either had no preference or did not discuss their PPOD during ACP, and these subjects were not included in the POD-PPOD agreement analysis (n=49). In our agreement analysis, the percentage agreement was 57.1% (95% CI: 43–72%; P<0.001), and the Gwet’s AC1 coefficient obtained was 0.49 (95% CI: 0.31–0.66; P<0.001), indicating a moderate level of agreement between the actual and PPOD.

Discussion

Key findings

This is the first study of death outcomes of a multi-ethnic, nationally representative, patient population with chronic respiratory failure, cared for by the only adult home ventilation service in Singapore. Key study findings include establishing rates of different death sites, the prevalence of different PPOD, and the level of agreement between actual and PPOD among deceased HVRSS patients. The different POD and PPOD rates of HVRSS decedents found in this study may better inform evidence-based policy and practice of end-of-life care models that support HVRSS patients, particularly at their chosen POD. Clinicians involved in the care of patients receiving long-term home mechanical ventilation, and such patients can be better informed during end-of-life conversations, as we now know the different rates of various POD options of a decedent cohort enrolled over a decade. Knowing that the prevalent PPOD choice is home may help clinicians involved in HVRSS patient-care manage end-of-life matters while mindful of patients’ emotional well-being, an integral aspect of patient-centred care. We determined the current agreement between actual and PPOD among HVRSS decedents, which sets a necessary baseline for patients who receive long-term home ventilation support in Singapore. This is a critical first step towards any development and planning of any end-of-life care improvement initiatives for such patients, and an important metric for future evaluation of potential programmes.

Strengths and limitations

The key strength of this study is that it represents the first examination of POD outcomes of Singapore’s sole adult home ventilation service. Additionally, the population studied is a nationally representative sample, a multi-ethnic group who received home ventilation support, and our findings illustrate the cultural context in Singapore among these patients. Furthermore, our primary POD outcome was data extracted from an administrative data source, ensuring data completeness.

This study is not without weakness. Although the study results are nationally representative of Singapore’s context, we are still limited by a relatively small sample size, which therefore could restrict generalisability. Our small sample size imposes limitations on statistical power, particularly for subgroup analyses undertaken, which should be considered. Additionally, there could potentially be some incomplete data capture, like national-level ACP information outside of TTSH sources; however, we envision the incompleteness of data to be small. The management of home ventilation patients is highly supported by the TTSH HVRSS team, so we think that the likelihood of available end-of-life preferences captured in ACP recorded outside of TTSH sources to be small. Third, we did not consider informal POD preferences that may have been recorded in medical case notes, but not formalised into an ACP. Also, we were not able to verify if other end-of-life preferences, besides PPOD, were concordant with actual end-of-life circumstances, for example, place of terminal care, cardiopulmonary resuscitation, or medical intervention. These directions would have unduly stretched the scope of the research beyond feasibility.

Comparison with similar research

Like a related study on home death among those who died from chronic diseases in Singapore (9), our results may indicate that certain patient-specific characteristics of HVRSS patients potentially allow home deaths to be more feasible, which are type of diagnosis, ventilation, and feeding at end-of-life, should that be the patient’s choice. Particularly for those on non-invasive ventilation who can be escalated to invasive ventilation when they return to hospital during a medical crisis, can likewise increase the risk of dying in hospital. In addition, being on non-invasive ventilation may indicate that their disease course is less advanced compared to those on invasive ventilation and that the patient and family has had less time to come to terms with choosing to die at home and not go back to hospital for care escalation. Conversely, more HVRSS decedents on invasive ventilation likely passed on at home because they were at the height of possible ventilatory support, so even if care was escalated to a hospital setting, it would likely be a futile endeavour without further ventilatory progression possible. Due to the advanced trajectory of their disease state, HVRSS decedents on invasive ventilation may have experienced repeated hospitalisation and a longer period of home care. This may enable a better-informed choice to choose home as their preferred place of terminal care to be more comfortable, with a trial of home treatment during exacerbations, and in the event for a home death as their PPOD. Such sentiments are corroborated elsewhere (8).

This study found that there was a strong preference for home deaths (73.1%) among home ventilation patients who expressed a POD preference. This level is similar to 77% among a general population (10), but much higher than the 40% amongst a population with a life-limiting or advanced illness in Singapore (11). There have been efforts towards a deeper understanding of the preferences of home ventilation patients (12), although preferences about end-of-life have been unexamined. This study represents the first examination of end-of-life preferences concerning POD in this patient population. With this information about POD preference, healthcare providers and policymakers can be equipped to make better, appropriate funding and resource allocation decisions to deliver patient-centred care for home ventilation patients at the end-of-life.

The agreement between preferred and actual POD for home ventilation patients was slightly higher compared to the previous study mentioned in a Singapore population with a life-limiting or advanced illness (11). This may not be the best comparison, but we are limited by the lack of studies in this area, as the agreement between actual and PPOD for home ventilation patients is understudied, like end-of-life preferences. As we speculate that some of our patients passed on due to pneumonia, respiratory or urinary tract infections secondary to their primary diagnoses, leading to the need for long-term home ventilation support, it would be beneficial for the end-of-life management of the death-causing condition to account for PPOD, as this may impact POD agreement. Nonetheless, our results may be considered an important metric of concern by healthcare providers delivering care to home ventilation patients at the end-of-life as part of quality monitoring of patient-centered care outcomes (6).

Explanations of findings

This study depicted the current state of death outcomes for HVRSS decedents and was able to identify some potential care gaps. Firstly, despite home care support, only approximate half of study participants were able to die at home, even though the majority of them had home as their PPOD. This potentially demonstrates that the needs of HVRSS decedents may not be adequately managed by the HVRSS clinical team. This may be in terms of longer service hours, as the homecare provided by the HVRSS clinical team is not 24-hour due to manpower constraints, or additional support like specialist palliative measures during the terminal phase (13). In a study of the impact of palliative care among end-stage organ disease patients in Singapore, those with end-stage respiratory failure were least likely to be referred for palliative care, despite them having the high inpatient hospital admissions and emergency department visits (14). This may indicate that there are barriers for patients with respiratory failure at the end-of-life to access palliative care, which is a care gap that should be addressed. Secondly, dying at an inpatient hospice setting was a preference for some patients; however, this was not an actual death site for any HVRSS decedent in the study. This may be a reflection of the care capacity of inpatient hospices to care for ventilator-assisted patients historically in Singapore, and so the introduction of the Inpatient Hospice Palliative Care Service in 2020 may help better meet the end-of-life needs for home ventilation patients (15). Other factors, such as family preference superseding a patient’s preference when they lose mental capacity, or changes in their informal POD preferences, may have also contributed to this finding. Thirdly, the earlier-mentioned UK-based study reported that only 14% of their 1,210-cohort died in a hospice (3), and a UK-wide qualitative study of MND patients on home ventilation and family members found that many are reluctant to have end-of-life discussions, and do not understand the dying process and what their available options are surrounding end-of-life decision-making (16). These findings demonstrate that more needs to be done towards demystifying and clarifying end-of-life care options and decision-making for patients with chronic respiratory failure who need long-term home ventilation support.

The burden of care on caregivers for patients on home ventilation is high, and this can be a strong influence on healthcare-seeking patterns for home ventilation patients and their families. For patients on invasive ventilation, patients and family may feel that care escalation or curative treatment in hospital to prolong life is financially challenging without much benefit, and dying in the comfort of home may be better than risking hospital admission in the hope of improving patients’ current condition. Furthermore, the significant dyspnoea of the dying process on non-invasive ventilation can present high symptom burden without additional support both for patients and family, so the choice to return to hospital for relief is understandable. All this knowledge may help clinicians and HVRSS patients, and families to be better informed during end-of-life POD conversations.

Implications and actions needed

Enhanced death outcomes may potentially alleviate avoidable distress and suffering during the dying process for an especially vulnerable patient group due to their high degree of medical complexity, disability, and care requirements. Moving forward, a better understanding of potential facilitators and barriers of specific PPOD, like a hospice death, and reasons for non-concordance between preferred and actual POD to facilitate better concordance levels may lead to improved POD outcomes for the home ventilation patient group.

Conclusions

In this first study of POD outcomes among home ventilation patients in Singapore, we found that the rates of home ventilation patients across different PODs were 52.5% in-hospital, 44.1% home deaths, and 3.4% at other locations. We also found that the majority of these patients with a formal POD preference wanted a home death, and that concordance between preferred and actual POD was 57.1%. These findings may inform policy and practice of end-of-life care models that support HVRSS patients, particularly at their chosen POD.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://apm.amegroups.com/article/view/10.21037/apm-25-94/rc

Data Sharing Statement: Available at https://apm.amegroups.com/article/view/10.21037/apm-25-94/dss

Peer Review File: Available at https://apm.amegroups.com/article/view/10.21037/apm-25-94/prf

Funding: This work was supported by the Palliative Care Centre for Excellence in Research and Education (PalC) Research Grant 2020 (No. PalC-RG-20/P004).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://apm.amegroups.com/article/view/10.21037/apm-25-94/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the National Healthcare Group Domain Specific Review Board in Singapore (ethics approval DSRB#: 2019/00854). Any subject identifiers used in the study were not disclosed to anyone outside the relevant research study members or trusted third parties, and hence cannot be used by others to identify individuals, resulting in the improbability of re-identification. Given the retrospective nature of the study, the requirement for informed consent was waived; however, informed consent was obtained from all patients and their caregivers participating in the prospective component.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Tan GP, Soon LHY, Ni B, et al. The pattern of use and survival outcomes of a dedicated adult Home Ventilation and Respiratory Support Service in Singapore: a 7-year retrospective observational cohort study. J Thorac Dis 2019;11:795-804. [Crossref] [PubMed]
2. Vitacca M, Grassi M, Barbano L, et al. Last 3 months of life in home-ventilated patients: the family perception. Eur Respir J 2010;35:1064-71. [Crossref] [PubMed]
3. Schwarz EI, Mackie M, Weston N, et al. Time-to-death in chronic respiratory failure on home mechanical ventilation: A cohort study. Respir Med 2020;162:105877. [Crossref] [PubMed]
4. Connolly S, Galvin M, Hardiman O. End-of-life management in patients with amyotrophic lateral sclerosis. Lancet Neurol 2015;14:435-42. [Crossref] [PubMed]
5. García-Sanjuán S, Fernández-Alcántara M, Clement-Carbonell V, et al. Levels and Determinants of Place-Of-Death Congruence in Palliative Patients: A Systematic Review. Front Psychol 2021;12:807869. [Crossref] [PubMed]
6. Finkelstein EA, Bhadelia A, Goh C, et al. Cross Country Comparison of Expert Assessments of the Quality of Death and Dying 2021. J Pain Symptom Manage 2022;63:e419-29. [Crossref] [PubMed]
7. Moss AH, Oppenheimer EA, Casey P, et al. Patients with amyotrophic lateral sclerosis receiving long-term mechanical ventilation. Advance care planning and outcomes. Chest 1996;110:249-55. [Crossref] [PubMed]
8. Pereira MJ, Chieh PP, Molina JAD, et al. Home ventilation patients: associations of having advance care plans with end-of-life healthcare utilisation. J Thorac Dis 2025;17:5626-38. [Crossref] [PubMed]
9. Tan WS, Bajpai R, Low CK, et al. Individual, clinical and system factors associated with the place of death: A linked national database study. PLoS One 2019;14:e0215566. [Crossref] [PubMed]
10. Blackbox Research. Lien Foundation Survey on Death Attitudes: Doctors & Nurses Survey. Available online: https://lienfoundation.org/wp-content/uploads/2024/11/Lien-Foundation-Doctors-and-Nurses-Report_18-June_14_for-uploading-on-website_0.pdf
11. Tan WS, Bajpai R, Low CK, et al. Using Routinely Collected Data to Ascertain Concordance With Advance Care Planning Preferences. J Pain Symptom Manage 2018;56:659-666.e2. [Crossref] [PubMed]
12. Masefield S, Vitacca M, Dreher M, et al. Attitudes and preferences of home mechanical ventilation users from four European countries: an ERS/ELF survey. ERJ Open Res 2017;3:00015-2017. [Crossref] [PubMed]
13. Neudert C, Oliver D, Wasner M, et al. The course of the terminal phase in patients with amyotrophic lateral sclerosis. J Neurol 2001;248:612-6. [Crossref] [PubMed]
14. Hum A, Yap CW, Koh MYH. End-stage organ disease-Healthcare utilisation: Impact of palliative medicine. BMJ Support Palliat Care 2024;14:e2657-63. [Crossref] [PubMed]
15. Ministry of Health Singapore. New Inpatient Palliative Care Service from 1 April 2020. 2020. Available online: https://www.moh.gov.sg/newsroom/new-inpatient-palliative-care-service-from-1-april-2020/
16. Wilson E, Palmer J, Armstrong A, et al. End of life decision making when home mechanical ventilation is used to sustain breathing in Motor Neurone Disease: patient and family perspectives. BMC Palliat Care 2024;23:115. [Crossref] [PubMed]