Abstract
Background:
Adolescents and young adults (AYAs) with life-limiting illnesses face unique challenges and often receive late or no palliative care (PC). This study examines the correlation between PC involvement and the intensity of end-of-life care among AYAs with nonmalignant life-limiting illnesses.
Design:
A retrospective cohort study analyzing population-based health care data from 2010 to 2018.
Setting/Subjects:
The study population included AYAs aged 15–39 who died in Ontario, Canada, from nonmalignant life-limiting illnesses during the study period (n = 2313).
Measurements:
PC involvement was defined as at least one encounter with a PC provider. End-of-life (EOL) care intensity was measured using rates of emergency department visits, hospitalizations, intensive care unit admissions, and mechanical ventilation in the last 30 days of life.
Results:
Of the 2313 AYAs studied, 37.5% had at least one PC encounter during their lifetime. Specialist PC delivered ≥90 days before death was associated with lower intensity of EOL care, including fewer intensive care unit deaths (17% vs. 34% versus 31%, p < 0.0001) and emergency department visits (17% vs. 27% versus 21%, p = 0.0091) when compared to generalist PC and no PC, respectively.
Conclusions:
AYAs with nonmalignant illnesses received high EOL care intensity and had a high percentage of death in acute care settings. Specialist PC involvement was associated with improved EOL care outcomes compared with generalist and no PC.
Key Message
AYAs with nonmalignant illnesses received high EOL care intensity and were more likely to die in acute care settings. Specialist PC involvement can improve the EOL care outcomes for this age group.
Introduction
Adolescents and young adults (AYAs) with life-limiting illnesses represent a unique and often under-served population within health care systems. 1 These individuals face complex medical, emotional, and psychosocial challenges as they navigate their conditions.2,3 Life-limiting illnesses encompass a wide range of diseases, including oncological, neuromuscular, cardiovascular, and metabolic diseases, each with diverse trajectories and health care needs.4,5 Managing these conditions requires a multifaceted approach that addresses not only the physical symptoms but also the emotional and social impacts on AYA patients and their families.6,7
Palliative care (PC) is an important component of the management of life-limiting illnesses. 8 PC aims to improve the quality of life for patients and their families through the prevention and relief of suffering by means of early identification, assessment, and treatment of pain and other physical, psychosocial, and spiritual problems. 9 For AYAs, PC can provide tailored support that respects their developmental stage and individual preferences, thereby enhancing their overall well-being and potentially altering the trajectory of their care.10,11 Despite the recognized benefits of PC, its integration into the care of AYAs with non-malignant life-limiting illnesses remains inconsistent. 12
There is growing evidence that PC involvement can reduce the intensity of end-of-life (EOL) care, characterized by metrics such as hospital admissions and intensive care unit (ICU) stays.13–15 Intensity of care at the EOL is an important indicator of the quality of care, as high-intensity interventions are often associated with increased patient suffering, reduced quality of life, and higher health care expenditures without necessarily improving outcomes. 16 The correlation between PC involvement and the intensity of EOL care in AYAs with nonmalignant illnesses is an area that warrants further investigation. 17 Understanding this relationship can inform health care providers and policymakers about the role of PC for this vulnerable population.18,19 This study aims to explore the association between PC involvement and the intensity of EOL care among AYAs with nonmalignant life-limiting illnesses. By analyzing a cohort of patients with varying levels of PC engagement, we seek to provide insights into how PC can impact health care utilization and outcomes in the final stages of life.
Methods
Study design
We conducted a retrospective cohort study to examine the intensity of EOL care among AYAs who died in Ontario with nonmalignant illnesses. We used individually linked health administrative data that are routinely collected through the delivery of services covered by the provincial’s single-payer health care system. These data are held at the ICES, formerly known as the
Population and setting
All AYAs (defined as individuals aged 15 to 39 years old) 20 who died between January 1, 2010, and December 31, 2018, in Ontario, Canada’s most populous province (15.6 million). This time frame was chosen based on the availability and completeness of linked administrative health data at the ICES at the time of data collection in 2024.
Only AYA decedents who had nonmalignant life-limiting illnesses listed as the cause of death on their death certificates were included in our study. Exclusion criteria included nonmedical causes of death such as assault/homicide, suicide, and accidents, or deaths due to acute causes (i.e., infections, pregnancy, and puerperium-related). We also excluded individuals without valid birthdates or sex, those who were not Ontario residents at the index date, and those for whom the cause of death was missing.
Outcomes
The primary outcome was the involvement of PC physicians in the care of AYAs. PC physicians’ involvement was categorized according to the timing of involvement into (1) at least one PC encounter during the lifetime in any setting (inpatient, outpatient, community) among the study cohort, (2) involvement in the last 90 days of life, and (3) involvement in the last 30 days of life. We divided the type of PC received into generalist and specialist PCs based on previously validated definitions (see below). Secondary outcomes included the intensity of EOL care in the last 30 days of life, measured by emergency department (ED) visits, hospitalizations, or ICU admissions. 21
Variables
Supplementary Appendix A has a list of the databases used for data extraction. Examined variables included age at death, sex, type of nonmalignant illness (e.g., cardiovascular, neuromuscular, gastrointestinal, metabolic), rural residence, socioeconomic status, and involvement of a PC physician. Cause of death was captured from the Office of the Registrar General-Deaths Database. We used the Registered Persons Database and the Postal Code Conversion File (PCCF) to collect data on demographics, residential postal code, neighborhood level, and sociodemographic characteristics (e.g., rural/urban, neighborhood income quantile). The Ontario Health Insurance Plan (OHIP) Claims Database was used to determine PC involvement. We identified hospital and ICU admissions from the Discharge Abstract Database.
Life-limiting illness was defined as any advanced disease where death is expected within two years. 22 We used the International Classification of Diseases, 10th Revision (ICD-10) codes to identify the nonmalignant life-limiting illnesses as causes of death.23,24 All codes used are listed in Supplementary Appendix B.
Involvement of a PC physician was defined using previously validated algorithms based on billing codes. 25 PC physician involvement was classified into generalist (those whose PC billing comprised less than 10% of their total billings in the previous year) and specialist (those who used PC billing codes for 10% or more of their annual billing codes). 25 Supplementary Appendix C has a list of the PC billing codes used to determine PC involvement.
Location of death was divided into “hospital,” “community,” “palliative care unit,” and “others.” From hospital deaths, we identified those who died in an ICU. “Others” included long-term care (LTC), nursing homes, and complex continuing care. “Community” included home, hospice, and respite care. We estimated home deaths using the Home Care Database and the OHIP billing code for death pronouncement at home. Other decedents who died in the community were identified indirectly as they did not fall into any of the listed locations.
Analyses
Descriptive statistics were used to summarize the baseline characteristics of the cohort. Categorical variables were presented as frequencies and percentages to provide a clear depiction of the distribution within each category. Normally distributed continuous variables were reported as means with standard deviations (SD), providing a measure of central tendency and dispersion, respectively. For non-normally distributed continuous variables, medians and interquartile ranges were used to better represent the central tendency and variability.
A chi-square test was employed to assess the associations between categorical variables, with the results indicating the number and percentage of occurrences within each category. For continuous variables, we conducted one-way analysis of variance to compare means across different groups. For nonnormally distributed continuous variables, the Wilcoxon rank-sum test was employed when comparing between two groups, and the Kruskal–Wallis test when comparing more than two groups. Statistical significance was defined as a p value of <0.05.
Results
There were 2313 AYAs who died from nonmalignant life-limiting illnesses between January 2010 and December 2018 (Table 1). Overall, 862 individuals (37%) were female, and 36% were aged 35–39 years. Approximately two thirds had at least one medical comorbidity. Twenty-eight percent of the cohort were from the lowest income quintile, and the majority (87%) lived in urban areas. The most common cause of death was cardiovascular conditions (33%), followed by neuromuscular diseases (29%).
Demographic Characteristics
Q1: lowest, Q2: low, Q3: middle, Q4: high, Q5: highest.
We examined the demographic characteristics of the cohort based on the type of nonmalignant illnesses (Table 2). Deaths were more common among males, except in respiratory illnesses (52% were females). The highest percentage of deaths among the youngest age group (15–19 years) was from neuromuscular diseases (46%). Similarly, cardiovascular illnesses were the most common cause of death among the oldest age group (35–39 years), accounting for 37% of all deaths within this age group.
Distribution of the Demographic Characteristics across Different Types of Nonmalignant Illnesses
Ranges presented to prevent back-calculation of small cell counts (≤5).
Q1: lowest, Q2: low, Q3: middle, Q4: high, Q5: highest.
Table 3 examines the intensity level of EOL care. In the entire cohort, 21% had ED visits, 41% were hospitalized, 32% were admitted to an ICU, and 30% were mechanically ventilated within the last 30 days of life. The intensity of EOL care varied between the different subgroups. ED visits were most common among those with renal disease (26%); hospitalization was most common among those with gastrointestinal disease (61%); and ICU admission was most common among those with respiratory illnesses (53%). Overall, the mean number of days spent in hospital during the last year of life was 23 days (SD: 44.87), and the mean number of days spent in an ICU within the last year of life was 7 days (SD: 19.09). Inpatient hospital units were the most common place of death (60%), and approximately half of these deaths happened in an ICU (30% of the entire cohort). The highest proportion of hospital and ICU deaths was among decedents with respiratory illnesses (85% and 56%, respectively). Approximately 38% of the study population had at least one encounter with a PC physician, with the proportion ranging from 22% (cardiovascular) to 82% (respiratory).
Intensity of EOL Care
Ranges presented to prevent back-calculation of small cell counts (≤5).
ED, Emergency Department.
ICU, Intensive Care Unit.
Comparing outcomes between patients with and without PC involvement, those with PC were more likely to die at home (19% vs. 5%, p < 0.0001) (Table 4). The parameters of EOL care intensity were slightly higher among the group who had at least one PC encounter; however, the results were not statistically significant. Our results showed that patients who had at least one encounter with a PC physician were more likely to have ED visits (23% vs. 21%, p = 0.266), ICU admission (33% vs. 31%, p = 0.486), and mechanical ventilation (32% vs. 28%, p = 0.0627).
Correlation between PC Involvement and the Intensity of EOL Care in the Total Cohort
Ranges presented to prevent back-calculation of small cell counts (≤5).
ICU, Intensive Care Unit.
ED, Emergency Department.
The cohort was then categorized based on PC provider model in the last 90 days and 30 days of life (Table 5). The proportion seen by a PC physician in the last 90 days of life was 29% (n = 663). We divided decedents according to the PC care received in the last 90 days of life: specialist PC (n = 272, 12%), generalist PC (n = 391, 17%), and no PC (n = 1650, 71%). A higher proportion of patients who received specialist PC died at home (33% vs. 15% vs. 6%, p < 0.0001), and a lower proportion died in hospitals (48% vs. 75% vs. 59%) compared to those with generalist PC and no PC, respectively. The specialist PC group also had the lowest prevalence of ICU death (17% vs. 34% and 31%, p < 0.0001), ED visits (17% vs. 27% and 21%, p < 0.0091), and ICU admissions (24% vs. 36% and 32%, p < 0.0026) when compared with generalist and no PC groups, respectively.
PC Provider Models in the Last 90 Days and 30 Days of Life and Its Correlation with EOL Care
ICU, Intensive Care Unit.
ED, Emergency Department.
The proportion seen by a PC physician in the last 30 days of life was 26% (n = 600). Similar to the findings in the last 90 days of life, 15% (n = 353) received generalist PC, and 11% (n = 247) were under specialist PC. Similarly, patients with generalist PC were more likely to die in the hospital (76% vs. 47% and 59%, p < 0.0001) and in the ICU (32% vs. 15% and 31%, p < 0.0001) and less likely to die in the community (17% vs. 39% and 38%, p < 0.0001) or in a PC unit (3% vs. 11% and 0.4%, p < 0.0001) when compared with specialist PC and no PC, respectively. Specialist PC involvement was correlated with less intense EOL care, including ED visits (17% vs. 27% and 21%, p < 0.0085) and ICU admissions (23% vs. 35% and 32%, p < 0.0061) when compared to generalist PC and no PC, respectively.
Discussion
In this cohort study of AYAs who died of nonmalignant illnesses, we found that only 38% were seen by a PC physician prior to death, with the lowest involvement seen among people with cardiovascular illness and the highest among those with respiratory illness. The majority died in the hospital, although only half of the hospital decedents died in the ICU. Those who were seen by a PC physician were more likely to die at home but also more likely to be admitted to hospital in the final days of life.
The AYA population with nonmalignant conditions represents a unique demographic in PC research that has been less studied compared to the cancer population. 1 Our study contributes novel insights into their demographic characteristics, comorbid conditions, health care utilization patterns, and PC. In our cohort, we identified a higher representation in the low- and lowest-income quintiles compared with the general Ontario population during the same period. This may raise the need for equitable access to PC services among this population. 26
Our findings are similar to those of Kassam et al., 15 who found that 43% of AYAs dying of cancer within the same health care system and relatively comparable study period were seen by a PC physician before EOL. The lower prevalence of PC integration in nonmalignant illnesses compared to cancer patients has been reported across different age groups.27,28 This highlights significant gaps in PC integration and can be attributed to a lack of PC awareness, difficult prognostication in nonmalignant illnesses, insufficient AYA-designed PC training among health care providers, and societal misconceptions about PC being solely for cancer patients or those imminently dying.29,30
In our study, PC involvement was correlated with higher intensity of EOL care; however, only hospitalization rates were statistically significant. This is different than some other studies that showed a correlation between PC and lower intensity of EOL care.14,31 These studies were among cancer patients, which may represent different care needs and disease trajectories. Higher hospitalization among patients who received PC might suggest that PC involvement facilitated more appropriate utilization of the health care services, ensuring patients received necessary interventions in a timely manner. PC involvement may also be more impactful in some conditions than others; patients dying of respiratory illness had the highest prevalence of PC involvement but also the highest proportion who died in an ICU. We could not explore the causes of hospitalization in our cohort and whether this was for avoidable reasons. Also, we could not explore the patients’ goals of care in this study, and whether this had an impact on the intensity of EOL care received. The higher likelihood of home deaths among PC recipients aligns with previous findings that home is the preferred place of death among patients in general, and PC can enhance the alignment of care with patients’ goals and preferences. 32
We found significant differences between specialist and generalist PC involvement and the intensity of EOL care. Specialist PC was linked to lower EOL care intensity, while generalist PC showed similar or higher intensity compared to no PC involvement.
This suggests specialist PCs may be more effective in managing complex symptoms and reducing aggressive interventions. Furthermore, this may reflect better access to community or hospice care instead of hospital admissions for symptom control.
The correlation between specialist PC involvement and the intensity of EOL care has been studied among AYA decedents with cancer in Ontario. 33 Similarly, Jewitt et al. showed that specialist PC for AYA patients with cancer was associated with lower intensity of EOL care, but unlike our results, specialist PC was more commonly delivered to cancer patients compared to generalist PC. 33
Our findings underscore the need for improved integration of PC in the care of AYAs with non-malignant life-limiting illnesses. Policymakers and health care providers should prioritize the development and implementation of PC programs tailored to this population. Strategies might include enhancing AYA-specific PC training for health care providers, increasing awareness about the benefits of PC among AYA patients and their caregivers, and ensuring equitable access to specialist PC services for AYA patients with nonmalignant illnesses.
Advancing research in this area should focus on longitudinal assessments of PC outcomes and patient-reported experiences. Prospective cohort studies can provide deeper insights into the trajectories of illness, caregiving dynamics, and impacts on family members, informing evidence-based interventions and policy recommendations. Furthermore, comparative effectiveness research should explore the impact of different PC models, interdisciplinary team approaches, and innovative care delivery strategies on patient satisfaction and EOL care outcomes in the AYA population.
Limitations of the Study
This study has several limitations. As a retrospective cohort study, it is subject to biases related to the accuracy and completeness of historical data. The reliance on administrative databases and death certificates means the accuracy of the recorded causes of death and the details of health care utilization are dependent on the quality of the documentation.
The sole use of PC physicians billing codes to detect PC involvement may result in missed or inaccurate data. For example, there were instances where patients died in a PC unit without a corresponding PC billing code from a physician. This discrepancy could be due to administrative and coding issues, multidisciplinary care dynamics, or late initiation of PC, highlighting potential gaps in the accurate capture of PC services.
The study’s generalizability is another limitation, as it used data from Ontario, Canada, which may not apply to other regions with different health care systems. Unmeasured confounding variables such as illness severity, patient and family care preferences, and social determinants of health could influence outcomes.
We acknowledge that the perception and availability of PCs in Canada may vary between age groups within the AYA population due to different health care settings in pediatric (for those aged 15–18 years) versus adult (for those aged >18 years) locations. These factors were not explored in this study. However, we aim to cover this part in our future research.
Conclusion
To our knowledge, this is the first population-based study to explore the intensity of EOL care among AYA patients with nonmalignant illnesses, the prevalence of PC involvement in this population and the correlation of PC involvement with EOL outcomes. Our findings highlight the limited involvement of PC and the high intensity of EOL care in this population, and the potential importance of involving specialist (as opposed to generalist) PC providers in their care. These findings advocate for policy and practice changes aimed at ensuring timely and appropriate PC access for this vulnerable population.
Footnotes
Authors’ Contributions
M.A., S.G.F., S.Y., P.T., and J.D. contributed to the conception and design of the work. S.Y. contributed to the acquisition of the data. M.A., S.G.F., S.Y., and J.D. contributed to data analysis. All authors contributed to interpretation of the data. M.A. drafted the article. All authors revised the article critically for important intellectual content, gave final approval of the version to be published, and agreed to be accountable for all aspects of the work.
Ethical Considerations
The use of the data in this project is authorized under section 45 of Ontario’s Personal Health Information Protection Act (PHIPA) and does not require review by a Research Ethics Board.
Data Sharing Statement
While legal data sharing agreements between ICES and data providers (e.g., health care organizations and government) prohibit ICES from making the dataset publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at 
(email:
Disclaimer
This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health (MOH) and the Ministry of Long-Term Care (MLTC). This document used data adapted from the Statistics Canada Postal CodeOM Conversion File, which is based on data licensed from Canada Post Corporation, and/or data adapted from the Ontario Ministry of Health PCCF, which contains data copied under license from ©Canada Post Corporation and Statistics Canada. Parts of this material are based on data and/or information compiled and provided by the Canadian Institute for Health Information and the Ontario MOH. The analyses, results, conclusions, opinions and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement by ICES, the Ontario MOH or MLTC, its partners, or the Province of Ontario is intended or should be inferred.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
This study was supported by ICES, which is funded by a grant from the Ontario government. This study was also supported by the PanCanadian Palliative Care Research Collaborative (PCPCRC). This work has been funded (in part) by a contribution from Health Canada, Health Care Policy and Strategies Program. The views expressed herein do not necessarily represent the views of Health Canada or the other funding organizations.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
