How does patient complexity differ between people with and without cancer who access specialist community palliative care? A comparative analysis study in Western Sydney specialist community palliative care

Abstract

Introduction

This study describes patient characteristics on entry and episode outcomes among patients admitted to an Australian community-based specialist palliative care service, comparing patients admitted with a cancer diagnosis with those admitted with a non-cancer diagnosis.

Methods

We conducted a comparative analysis of patients who were discharged from Specialist Community Palliative Care Service and received at least one episode of care between 1 July 2017 and 28 February 2023. We considered outcomes for socio-demographic characteristics, diagnoses and comorbidity burden, palliative care needs on admission to the service, episode duration and place of death.

Results

A total of 4116 distinct patients were included in the study. A quarter (26.0%) of the cohort were aged <65 years and almost half were culturally or linguistically diverse. Most (78.6%) were admitted with a cancer diagnosis, predominantly originating in the lung (23.1%, n=747), digestive system (21.3%, n=689) or genitourinary system (12.9%, n=418). Cardiovascular conditions were most common among patients admitted with non-cancer diagnosis (19.3%), followed by respiratory conditions (16.9%) and genitourinary conditions (13.9%). Dementia was present for 10.3% of patients with a non-cancer diagnosis on admission. Patients with a non-cancer diagnosis had more comorbidities (median=8 vs 5), they were more commonly frail and took more medications (median=7 vs 6) in comparison to patients with malignancy. On admission, patients with a non-cancer diagnosis had significantly lower functional scores for activities of daily living and were less likely to enter the service in a stable phase. However, patients with a non-cancer diagnosis were more likely to die at home and die in their preferred place compared with patients with cancer (p=<0.001).

Conclusions

Cancer and non-cancer cohorts of patients admitted to community palliative care services have different admission characteristics and service needs. Both groups have complex requirements, and further investment in palliative care should target optimal community-based models that positively influence early entry, patient-centred care and death in place of choice.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Palliative care in Australia has historically focused on patients with cancer, who comprise 70% of admissions, while evidence suggests patients with a non-cancer diagnosis may have different care needs and timing of access.

WHAT THIS STUDY ADDS

• Patients with a non-cancer diagnosis accessing community palliative care in Western Sydney presented with greater complexity (more comorbidities, frailty and medications) and lower functional status than patients with cancer, yet were more likely to achieve death at home and in their preferred location.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings highlight the need for differentiated service models that address the distinct admission characteristics and care needs of patients with cancer and those with a non-cancer diagnosis, as well as targeted investment to reduce barriers to early palliative care access for patients with a non-cancer diagnosis.

Introduction

Palliative care services are specialised clinical services for patients and their families facing a life-limiting illness, which focus on improved quality of life and reduced physical and emotional stress. Early initiation of palliative care in patients’ disease progression can significantly improve symptoms over the course of care, improving quality of life and reducing healthcare costs.1,4

Between July 2015 and June 2022, there was a 29% increase in palliative care-related hospitalisations in Australia, increasing at almost three times the rate of all hospitalisations (10% increase) over the same period.⁵ With demand expected to double by 2050,⁵ the Australian federal government has invested significantly in workforce training, person-centred care quality and rural and regional support.⁶

Most palliative care is delivered in community settings,² aligning with the 70% of Australians that express a preference to die at home (70%).⁵ This preference is more achievable when services are provided outside hospitals. Community palliative care services in Australia are delivered through a combination of inpatient, outpatient and home-based care.⁵ The services are delivered by a multidisciplinary team of healthcare professionals, including doctors, nurses, social workers and pharmacists, who work together to address the physical, psychosocial and spiritual needs of patients.

The delivery of community palliative care services is guided by the National Palliative Care Strategy 2018. The strategy provides a shared direction and authorising environment for the continuous improvement of palliative care services throughout Australia.⁵ Evidence suggests community settings are superior or comparable to hospital settings in delivering effective care,² and they may be more cost-effective.⁷ However, as the need for community palliative care services continues to grow, to deliver high quality of care and remain cost-effective in community settings, palliative care services must evolve to meet the needs of the changing population.

Palliative care patients experience a range of severe symptoms over the course of their illness trajectory.⁸ These symptoms are dynamic and are shaped by disease progression, medications, treatment side effects and exacerbation of concurrent conditions. Seventy per cent of patients admitted to palliative care in Australia have a cancer diagnosis.⁹ This is a result of the long-standing close relationship between oncology and palliative care, where it has been largely the oncology community that has driven development within the palliative care sector.¹⁰ Historically, palliative care has focused on terminally ill patients with cancer; however, there is good evidence for positive outcomes in palliative care among patients with a non-cancer diagnosis and the incidence of patients with a non-cancer diagnosis requiring palliative care is increasing.¹¹ There is some evidence from European settings that the complexity of care needs, timing of admission and integration with medical specialists among patients with a non-cancer diagnosis accessing palliative care differs from patients with cancer.¹² If this difference exists in the Australian setting, investment in improvements to meet future palliative care needs should address the current gaps in practice that do not serve the needs of patients with a non-cancer diagnosis entering the service.

Even though there is a growing preference for care at home in Australia, and a likely considerable expansion of palliative care delivered in the community setting, there is little published data on the epidemiology of patients receiving specialist community palliative care, in particular the variation in the presentation of patients with cancer and patients with a non-cancer diagnosis within this cohort. The objective of this study is to describe the characteristics of patients entering specialist community palliative care services in an urban setting and to differentiate between patients with cancer and those with a non-cancer diagnosis in terms of their episode of care and place of death. Understanding these patients and how they vary will add to the evidence base supporting advanced care planning, integration between palliative services and specialist clinical care and the delivery of palliative care for patients with cancer and those with a non-cancer diagnosis in community palliative care settings in Australia.

Methods

Design

We conducted a comparison study of patients discharged from Silverchain Specialist Community Palliative Care Service in Western Sydney between 1 July 2017 and 28 February 2023. Patients were included if they received at least one visit from Silverchain within the study period. In the context of community palliative care, an episode is defined as the period between a patient’s first visit and their death, or discharge from the service. Discharge may occur when the patient’s palliative care needs change, are met, they die or other circumstances make the service no longer the most appropriate provider (eg, admission to hospice). Patients may have multiple episodes of care. We used administrative data to examine the characteristics of patients. A manual record review was also performed on a sample of patient records to extract comorbidity and medication data. This study was conducted and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.¹³

Setting

Specialist palliative care is available to Australians when symptoms cannot be effectively managed in the generalist setting (eg, general practitioners). Specialist services consist of multidisciplinary teams, with staff trained specifically in palliative care provision. Silverchain is a not-for-profit community health and aged care provider that delivers specialist and generalist palliative care services in four Australian states, including New South Wales (NSW) up to July 2024. In NSW, specialist palliative care services comprised multidisciplinary teams including doctors, nurses, social workers and allied health staff. The service commenced on 1 July 2017 and provided care 24 hours, 7 days a week to 700–900 patients in their home annually. Palliative care was delivered within the Western Sydney Local Health District, which includes five public hospitals and a catchment of >1.1 million people. Western Sydney is one of the fastest growing areas in NSW and one of the more culturally diverse communities. Nearly half (46.8%) of the community are born outside of Australia. First Nation Australians comprise approximately 2% of the population.^{14 15}

Consumer engagement

The Silverchain Consumer Advisory Group (CAG) serves as a driver of the research agenda at Silverchain, ensuring that research priorities align with consumer needs and perspectives. Prior to initiating this study, the CAG was consulted regarding the importance and relevance of palliative care research, particularly research aimed at understanding outcomes across different patient groups and populations.

The CAG affirmed the significance of this research area and supported the need to better understand palliative care outcomes to inform service delivery and improve patient experiences. This consumer input was instrumental in establishing the research priority and confirming the value of investigating outcome variations across diverse patient cohorts.

The CAG was not directly involved in the design, conduct or analysis phases of this specific study, but their foundational role in identifying palliative care outcomes research as a priority area ensured that this work addresses questions of direct relevance to consumers and the broader public. Plans for dissemination of results to the CAG and broader consumer networks will be developed following study publication to ensure findings reach relevant stakeholders and inform future service improvements.

Study size

The study included the episodes of all patients with at least one episode of care in Western Sydney Specialist Community Palliative Care delivered by Silverchain. Post hoc power analysis confirmed the sample provided >99% power to detect clinically meaningful differences of 5–10 percentage points between groups for proportion outcomes (α=0.05, two-tailed), and >95% power to detect small-to-moderate effect sizes (Cohen’s d ≥0.2) for continuous variables. A manual record review for detailed comorbidity and medication data was conducted on a pragmatic subsample of 474 patients (11.5% of the total cohort), determined by available research resources and the labour-intensive nature of clinical record extraction. Records for clients admitted to the service between 1 March 2022 and 28 February 2023 were randomly selected using record IDs in STATA V.16. This subsample size provided adequate representation to characterise clinical complexity and validate key variables not reliably captured in administrative data alone.

Data and covariates

Data for eligible patients were de-identified and extracted from electronic care records and clinical notes. We defined age as age in years at time of admission and analysed as continuous and as a categorical variable as <16, 16–64 and 65 years and over. All socio-demographic data and preference of death were self-reported at the time of admission and analysed categorically. Primary diagnoses were defined using the diagnosis cited as the reason for referral at the time of admission and refer to the condition that requires the highest resource needs, defined by the attending clinician. Comorbidities are defined as any diagnosis within the clinical record that was not listed as the primary diagnosis. We analysed both patient characteristics and episode characteristics. Patient characteristics are analysed at the patient level, while episode characteristics are analysed at the episode level.^{16 17} Palliative Care Outcomes Collaboration (PCOC) is an Australian national programme established in 2005 to systematically improve palliative care by using standardised tools to measure and benchmark patient and carer outcomes. We used PCOC data for phase of care and analysed these in line with PCOC guidelines.¹⁷ The Australian-modified Karnofsky Performance Score (AKPS) is a measure of the patient’s overall performance status or ability to perform their activities of daily living. It is a single score between 10 and 100 assigned by a clinician based on observations of a patient’s ability to perform common tasks related to activity, work and self-care. A score of 100 signifies normal physical abilities with no evidence of disease.^{18 19} Decreasing numbers indicate a reduced performance status. We used AKPS as a performance measure for dimensions of activity, work and self-care and the Resource Utilisation Groups-Activities of Daily Living tool (RUG-ADL)¹⁸ as a measure of functional dependency as the sum of domain scores for four domains (bed mobility, toileting, transfer and eating). We analysed disadvantage using Australian national Socio-Economic Indexes for Areas quintiles.²⁰ Details of how these data are collected and the categories constructed are available in online supplemental material A. Missing data were considered missing at random, classified as unknown and presented in tables.

Statistical analysis

The study describes the characteristics of cancer and non-cancer palliative case cohorts. Categorical variables are presented as frequencies and percentages. Continuous variables are presented as means and SD or medians with IQRs. Comparisons between groups for categorical variables were made using Pearson’s χ² test, Fisher’s exact test and post hoc residual analysis. Variables with continuous data used t-tests for those with a normal distribution or Kruskal-Wallis analysis of variance for data that were not normally distributed. A p value of <0.05 was considered statistically significant. Data analysis was performed using STATA V.18 and R V.4.02.²¹

Bias

This descriptive, comparative study aimed at characterising patients accessing specialist community palliative care services rather than establishing causal relationships; therefore, traditional approaches to bias control such as multivariable adjustment were not the primary methodological focus. Selection bias was minimised by including all patients discharged from the service during the study period who met inclusion criteria, creating a complete census of the target population. Information bias was addressed by supplementing administrative data extraction with manual record review for a subset of patients (n=474) to validate comorbidity and medication data. Reporting bias was minimised by using objective, routinely collected administrative data extracted from standardised clinical documentation. While unmeasured confounding may influence observed differences between cancer and non-cancer patient groups, extensive confounder control was not essential as the primary aim was to describe and compare patient characteristics rather than establish causal relationships.

Results

Socio-demographic characteristics

There were 4116 patients discharged from the service during the study period, among whom there were a similar number of males compared with females. Most patients were 65 years or older (74.0%, n=3052), with a small cohort of paediatric patients (<1%), and most had a life-limiting primary diagnosis of cancer (78.6%, n=3236). Patients had a diverse ethnic profile, reflective of the geographic catchment area, 16% needed an interpreter for palliative care services. Socio-demographic characteristics of patients are presented in table 1.

Table 1. Socio-demographic characteristics of patients, 1 July 2017–28 February 2023, n=4116.

Characteristic	Overall, n=4116*	Cancer, n=3236*	Non-cancer, n=880*	P value†
Mean age (years) at episode start date	75 (65, 83)	73 (63, 82)	80 (71, 87)	<0.001
Age group (years)				<0.001
<16	7 (0.2%)	3 (0.1%)	4 (0.5%)
16–64	1057 (25.7%)	918 (28.4%)	139 (15.8%)
65+	3052 (74.1%)	2315 (71.5%)	737 (83.8%)
Gender				0.11
Female	1932 (46.9%)	1498 (46.3%)	434 (49.3%)
Male	2184 (53.1%)	1738 (53.7%)	446 (50.7%)
First Nations	76 (2.1%)	52 (1.9%)	24 (3.2%)	0.027
Unknown	569	445	124
CALD	1864 (47.3%)	1443 (46.5%)	421 (50.5%)	0.040
Unknown	178	132	46
SEIFA				0.12
Lowest disadvantage	1069 (26.0%)	824 (25.5%)	245 (27.8%)
Low	494 (12.0%)	409 (12.6%)	85 (9.7%)
Middle	1120 (27.2%)	877 (27.1%)	243 (27.6%)
High	375 (9.1%)	289 (8.9%)	86 (9.8%)
Highest disadvantage	1058 (25.7%)	837 (25.9%)	221 (25.1%)
Marital status				<0.001
Married/De facto	2087 (61.3%)	1701 (63.0%)	386 (55.0%)
Widow/Widower	787 (23.1%)	556 (20.6%)	231 (32.9%)
Divorced/Separated	307 (9.0%)	269 (10.0%)	38 (5.4%)
Never married	222 (6.5%)	175 (6.5%)	47 (6.7%)
Unknown	713	535	178

Western Sydney palliative care episodes discharged between 1 July 2017 and 28 February 2023, among 4116 individuals with at least one episode of palliative care.

SEIFA quintiles based on the National Scale. Based on patients’ first episode of care.

^*Median (IQR); n (%).

^†Wilcoxon rank-sum test; Fisher’s exact test; Pearson’s χ2 test.

CALD, culturally and linguistically diverse; SEIFA, Socio-Economic Indexes for Areas: Index of Relative Socio-Economic Disadvantage.

Patients with a primary cancer diagnosis were younger than patients with a non-cancer diagnosis (median 73 years vs 80 years), less commonly identified as First Nations (p=0.027) and less culturally and linguistically diverse (table 1).

Primary diagnoses on referral

The most common primary diagnostic grouping associated with palliative care referrals was cancer (n=3236). Among patients with cancer, cancers of the lung (23.1%, n=747) were the most common, followed by digestive (including colon) (21.3%, n=689) and genitourinary (12.9%, n=418). Non-cancer diagnoses were the primary reason for referral for more than one in five patients (21.2%, n=880). Cardiovascular conditions were the most common (19.3%, n=170) among this group, followed by respiratory conditions (16.9%, n=149) and genitourinary conditions (13.9%, n=122). Over 10% (10.3%, n=93) of patients with a non-cancer diagnosis were referred to palliative care for dementia. Details of all recorded primary diagnoses for cancer and non-cancer admissions are available in online supplemental tables 1a and 1b.

Comorbidities and medications

Among the 474 patients included in the manual record review, there was a median of five comorbidities, including primary diagnosis. Patients with a non-cancer diagnosis had a higher number of comorbidities (median=8) compared with patients with cancer (p<0.001) (online supplemental table 2). Frailty indicators were recorded for nearly half (46.0%) of patients. A higher proportion of those with a non-cancer diagnosis presented with frailty indicators compared with those with cancer (60.5% vs 39.4%, p<0.001) (online supplemental table 2). Most patients, 71.2% (n=338), experienced polypharmacy of five or more medications. Active prescribed medications per patient on admission to community palliative care did not differ significantly between patients with cancer and patients with a non-cancer diagnosis, although patients with cancer were more commonly prescribed opioids (p=0.02) and patients without cancer were more commonly prescribed benzodiazepines (p<0.001). Medicinal cannabis use was rare (1.6%, n=6), and only observed among patients with a cancer diagnosis.

Among the cohort, hypertension (41.1%, n=195), diabetes (26.4%, n=125), cardiac arrhythmia (22.2%, n=105) and chronic pulmonary disease (17.9%, n=85) were most frequently identified as comorbidities. A comorbidity of dementia was recorded within 5.3% (n=25) of the manual record review. Additionally, 12.0% (n=57) had a mental health-related diagnosis (depression, anxiety, psychosis or post-traumatic stress disorder). Details of all recorded comorbidities are available in online supplemental table 3.

Comorbidities (excluding primary diagnosis) retrieved from the manual record review where we saw significant differences between patients with cancer and patients with a non-cancer diagnosis are presented in online supplemental table. In all cases where there were significant differences, the larger comorbidity burden was among the non-cancer group.

Episodes of care

Among the 4116 distinct patients in the cohort, 4013 had one episode of palliative care during the study period. There were 103 patients who had multiple episodes, 101 had two distinct episodes of community palliative care and two individuals had three distinct episodes. Among these, 24 were patients with a non-cancer diagnosis (23.3%) and 79 were patients with cancer (76.7%). There were 4221 distinct episodes of care during the study period.

Characteristics of patients on admission

On admission to the service, almost all patients were recorded as either being in a stable (50.8%, n=2145) or deteriorating (44.0%, n=1856) palliative care phase (table 2). Patients with cancer were more commonly admitted during a stable phase, while patients with a non-cancer diagnosis were more commonly admitted during deteriorating, unstable or terminal phases (table 2).

Table 2. Phase and function of patients on admission to palliative care (n=4221).

Characteristic	Overall, n=4221*	Cancer, n=3317*	Non-cancer, n=904*	P value†
Phase at admission				<0.001
Stable	2145 (50.8%)	1784 (53.8%)	361 (39.9%)
Deteriorating	1856 (44.0%)	1403 (42.3%)	453 (50.1%)
Unstable	111 (2.6%)	82 (2.5%)	29 (3.2%)
Terminal	109 (2.6%)	48 (1.4%)	61 (6.7%)
Bedfast >50% of time	1185 (28.1%)	717 (21.6%)	468 (51.8%)	<0.001
RUG-ADL fully functional	2019 (48.5%)	1811 (55.2%)	208 (23.5%)	<0.001
Unknown	54	34	20

Western Sydney palliative care episodes discharged between 1 July 2017 and 28 February 2023, among 4116 patients.

^*N (%).

^†Pearson’s χ2 test.

RUG-ADL, Resource Utilisation Groups-Activities of Daily Living tool.

More than half of patients with a non-cancer diagnosis were admitted with an Australia-modified Karnofsky Performance Status (AKPS) score of ≤40, indicating that they were bedfast >50% of the time. This was significantly lower for patients with cancer, among whom only 21.6% were admitted with an AKPS score of ≤40 (table 2). The distribution of AKPS scores among patients with cancer and patients with a non-cancer diagnosis is presented in figure 1. On admission to palliative care services, patients with a non-cancer diagnosis were far less functional than patients with cancer, almost a quarter of whom were admitted while able to carry out normal activity (figure 2).

Figure 1. Resource Utilisation Groups-Activities of Daily Living scores for toileting, eating, mobility and transfers on admission to palliative care (n=4221).

Figure 2. Australia-modified Karnofsky Performance Status (AKPS) scale scores on entry for patients with cancer and patients with a non-cancer diagnosis.

Functionality on the RUG-ADL scale among four domains, for eating, toileting, mobility and transfers was higher among patients with cancer. More than half (55.2%) of patients with cancer were admitted with a total RUG-ADL score of 4, indicating full functionality across all four domains. This was significantly lower among patients with a non-cancer diagnosis, among whom only 23.5% were admitted with a total RUG-ADL score of 4 (table 2). Among the four RUG-ADL domains assessed, patients with a non-cancer diagnosis were admitted with higher scores, indicating less independence when compared with patients with cancer (figure 1).

Episode outcomes

The median episode duration for all patients was 49 days (IQR 23–113 days). Patients with a non-cancer diagnosis had a shorter median episode length (43 days vs 51 days) and higher variability in episode length (table 3).

Table 3. Characteristics of palliative care episodes, outcomes among cancer and non-cancer episodes (n=4221).

Characteristic	Overall, n=4221*	Cancer, n=3317*	Non-cancer, n=904*	P value†
Discharge outcome				0.2
Discharge due to death	3447 (94.8%)	2685 (94.7%)	762 (94.9%)
Died within 7 days of discharge	30 (0.8%)	28 (1.0%)	2 (0.2%)
Subsequent admission to community palliative care	105 (2.9%)	81 (2.9%)	24 (3.0%)
Lived >7 days following discharge	55 (1.5%)	40 (1.4%)	15 (1.9%)
Unknown	584	483	101
Place of death				<0.001
Home	1790 (51.9%)	1285 (47.9%)	505 (66.3%)
Other	1657 (48.1%)	1400 (52.1%)	257 (33.7%)
Context of separation unknown/did not die during episode	774	632	142
Died in preferred place				<0.001
Yes	1685 (68.4%)	1260 (66.5%)	425 (75.1%)
No	777 (31.6%)	636 (33.5%)	141 (24.9%)
Died—preference unknown	985	789	196
Context of separation unknown/did not die during episode	774	632	142
Episode length (days)	49 (23, 113)	51 (25, 111)	43 (17, 129)	0.002

Western Sydney palliative care episodes discharged between 1 July 2017 and 28 February 2023, among 4116 patients.

^*N (%); median (IQR).

^†Pearson’s χ2 test; Wilcoxon rank-sum test.

Most episodes concluded with the patient dying (94.8%). Among the episodes where the patient did not die, 2.9% went on to have an additional episode of palliative care. The characteristics of the episode separation were similar for patients with cancer and patients with a non-cancer diagnosis.

Among episodes that ended with a death, roughly half (51.9%) of patients died at home. This was higher among episodes with patients with a non-cancer diagnosis, where 66.3% of patients died at home, compared with only 47.9% of patients with cancer. Among those that died, where their preference of place of death was known, 68.4% died in their place of preference. This was higher for the non-cancer episodes, where 75.1% of patients died in their place of preference, while 66.5% of patients with cancer died in their place of preference.

Discussion

The term ‘complex need’ is frequently used; however, there is no standard definition of complexity in palliative care, nor a distinct set of needs that are understood as ‘complex’. Our results add to the limited research that exists around the complexity of patients receiving community palliative care.²² Just over a quarter of these patients were aged under 65 years and almost half were culturally and linguistically diverse. There was considerable polypharmacy in the cohort (median=7), and almost half of the population were not admitted in a stable phase. On admission, most patients had a cancer diagnosis (78.6%); however, among the patients admitted with a diagnosis not related to cancer, diagnoses were diverse, reflecting a high level of variability of palliative care needs of the population. Most palliative care episodes ended in death (94.8%) and most people died in their preferred place (68.4%).

We saw important differences in the palliative care needs between patients referred with a cancer diagnosis and those referred with a non-cancer diagnosis. Non-cancer referrals were older, more culturally and linguistically diverse and more commonly widowed. Their clinical picture on admission was more complex than cancer referrals, they had more comorbidities (median=8 vs 5), were more commonly frail and took more medications (median=7 vs 6). Patients with a non-cancer diagnosis entered the service with significantly poorer function, were less likely to be admitted in a stable phase and more commonly had care needs associated with being bedfast. Although non-cancer referrals had shorter episodes (43 days vs 51 days), their outcomes in relation to dying at home and death in their preferred place were more favourable than cancer referrals (p≤0.001).

Our findings highlight delayed access to palliative care among patients, with almost half admitted outside of a stable phase. The benefits of early entry to palliative care are widely recognised.^{4 23 24} Early entry is associated with improved quality of life and symptom intensity among patients.²⁴ Patients with a non-cancer diagnosis in our population were particularly less able to benefit from palliative care services to the extent they might have if admitted earlier in their disease trajectory. Enabling early access to palliative care among individuals suffering from diseases not related to cancer should be a focus of initiatives to improve palliative care.

Over 10% of our cohort had dementia. Dementia is currently the leading cause of death in Australia for women and the second leading cause of death overall. It is the largest cause of disability and death in people aged over 65 years. Over 400 000 Australians have a diagnosis of dementia, and this is predicted to more than double by 2058.²⁵ Despite it being a common cause of death, many people do not realise that dementia is a terminal illness, nor that it can benefit from palliative care.²⁶ There is a pressing need for dementia-specific palliative care with respect to early access and multidisciplinary treatment.

A quarter of our cohort was aged <65 years, which is similar to evidence that home-based palliative care services are more common among younger groups, possibly reflecting increased support in the home at younger ages.²⁷ Our findings contrasting patients with cancer and those with a non-cancer diagnosis are consistent with international evidence, indicating patients referred to palliative care with a non-cancer diagnosis present later, have more complex comorbidities and have significantly different needs compared with those referred with a cancer diagnosis.¹² We did not assess symptoms in our cohort; however, one study identified similar symptom clusters between patients with cancer and those with a non-cancer diagnosis.²⁸ There is a little evidence in the Australian setting describing patient comorbidities at entry and episode outcomes among specialist community palliative care patients, differentiated by patient group.

The findings strongly support the need to examine current policies, funding mechanisms and models of care internationally to ensure they meet the needs of the diverse population of patients who would benefit from palliative care. This may include investment in earlier identification pathways, targeted public health campaigns, improved outcome measures and the co-design and piloting of diagnosis-specific, needs-based models of care. Although patients with a non-cancer diagnosis represent a smaller proportion of those receiving community palliative care, this cohort is growing significantly and service models should be tailored to their specific needs. One clear example is patients with dementia. Dementia is the most rapidly increasing primary diagnosis of patient’s accessing palliative care in Australia, doubling from 8.3% in 2014 to 16.8% in 2024.²⁹

These findings are in line with other research on palliative care in patients with a non-cancer diagnosis. Similar to our study, an observational study of non-cancer palliative care patients in Northern Italy found patients with a non-cancer diagnosis enter palliative care with high symptom burden.³⁰ In a population-based matched cohort study of palliative and healthcare outcomes among adults with terminal non-cancer illness, palliative care was associated with reduced emergency attendances and hospital admissions and was more likely to die at home, rather than in hospital.³¹

However, traditional models of palliative care were designed for people with cancer and do not suit the needs of people with dementia. These patients usually live with their illness for years and experience gradual functional and cognitive decline over this period. Underscored is the requirement for multidisciplinary care teams and effective care planning of resource allocation. This is initiated with prognostication of illness trajectories using tools that are sensitive to the nuanced requirements among non-cancer referrals. Most tools currently used to assess patients’ palliative care needs on entry to services were developed specifically for patients with cancer.³² In addition, there is currently no consensus on referral criteria, which often fail to capture the complexity of need.²²

Our study does have some limitations. Our findings pertain to a cohort accessing community specialised palliative care in Western Sydney Local Health District and may not be generalisable to all palliative care patients. The data in this study were limited to what was captured within routine health service administrative records. Missing data, particularly for preferred place of death and occurrence of death, resulted in an incomplete profile. We report all missing data as unknown to enable readers to interpret this. Our study may underestimate the morbidity burden among patients because the service collected only the comorbidities considered by experienced clinicians to be relevant to care and outcomes. Modified morbidity index scores were used due to the absence of International Classification of Diseases codes; therefore, comparison of our results with other studies is limited.

Despite the importance of palliative care, timely access and choice in setting remains a challenge, particularly among patients referred to care with non-cancer diagnoses. Our results support the call to invest in understanding and addressing the barriers to early access to community palliative care among all patients who could benefit from improvements in quality of life and symptom control. This work is vital to ensure optimised service design and an ability to offer best care and dying in place of choice for all Australians.

The number of terminally ill patients with non-cancer diagnoses is increasing, requiring access to community palliative care to manage symptoms and improve quality of life. Yet our study demonstrates that these patients are accessing care later in their disease trajectory than patients with cancer and with increased complexities such as comorbidities and medication burden. Australian services need to adapt to the changing aetiology of patients accessing palliative care services, which will require integration with broader teams such as gerontology and neurology. Co-design of models should occur with multidisciplinary providers, care navigators, advocacy groups, clients and carers. New models should be needs-based, proactive and support appropriate data collection and outcomes of care in all patients.

Data availability statement

Data are available on reasonable request.