1 BACKGROUND

Patients with cancer have high care needs toward the end of life. Patients dying with cancer tend to have multiple comorbidities,¹ and are more likely to experience severe pain at the end-of-life stage compared to other end-of-life trajectories.² They also experience aggressive interventions during the end-of-life phase,^{3, 4} with almost a quarter of decedents who died of cancer in Ontario, Canada experiencing at least one aggressive intervention (i.e., receiving chemotherapy in the last 14 days of life, having more than one emergency room visit or hospitalization in the last 30 days of life).³ A retrospective cohort study of cancer decedents from 7 countries reported that between 44% to 60% of decedents experienced at least one hospitalization during the last 30 days of life and 5% to 13% of decedents received chemotherapy during this period as well.⁵

Research has been conducted on preferred settings of death for individuals with cancer. A qualitative study of patients dying with cancer reported that patients desire a comfortable end-of-life experience with limited suffering, maintaining their independence and autonomy, and, ultimately, dying in their own homes.⁶ A survey of bereaved caregivers in Ontario also reported that patients with terminal illnesses (e.g., cancer) overwhelmingly preferred to die at home as well.⁷ Similarly, a systematic review of 210 studies from over 33 countries identified that most individuals prefer to die at home instead of an institution and that most individuals' preferences remained static as they approached death.⁸ Despite this, 53% of cancer decedents die in a hospital setting in Ontario,¹ thereby contradicting the established benefits of homecare during this phase⁹ and patient preferences in spending their final days at home.

Palliative care in Ontario—as provided by physicians and nurse practitioners-- is delivered in the hospital (both in a consultative model and a dedicated palliative care unit), in complex continuing care/sub-acute care (in the form of a palliative care unit), at outpatient clinics, in nursing homes, and in the home. Almost half of decedents receive at least 1 palliative care service in the last year of life, with the majority of it being delivered in acute care and outpatient settings.¹⁰ Hospice in Ontario refers only to dedicated facilities and there are only approximately 271 hospice beds in the entire province.¹¹ Unfortunately, health services information for hospice facilities is not captured through health administrative datasets in Ontario, Canada. Nevertheless, a 2014 cross-country comparison study estimated that 16% to 30% of decedents received hospice palliative care services in Canada, compared to 23% in England, 12% in Germany, and 41% in the United States.¹² Few studies have explored which settings decedents spend their end-of-life stage in.^13-16 Of these studies, only two limited their study cohorts to cancer decedents, of which one was specific to hematologic cancers.^{13, 15} Cheung et al. and Andersen et al. both reported that cancer decedents spend the majority of their final days at home. However, they only studied the number of days spent at home in the last 6 months of life; neglecting to differentiate between days at home with home care and days at home without any care, nor reporting on days spent in other settings.

To our knowledge, no studies have described the different settings of care experienced by cancer decedents during the last 100 days of life using population-level health administrative data. Therefore, the objective of this study was to describe the different settings of care experienced in the last 100 days of life by decedents whose primary cause of death was cancer and how settings of care change as the decedent approached death. The last 100 days of life is a timeframe of interest as healthcare utilization and cost increase significantly during this timeframe for decedents with a history of cancer.¹⁷ In addition, a systematic review of retrospective studies using linked-health administrative datasets demonstrated that the use of various healthcare resources increased toward the end-of-life phase for patients dying with cancer.^{5, 18, 19}

2 METHODS

3 RESULTS

From January 1, 2013, to December 31, 2017, there were 475,108 decedents aged 19 years and older after 8655 excluding decedents ineligible for OHIP during the last year of life, decedents aged <19 years or > 105 years at death, decedents who did not have healthcare contact within the last 5 years of life, and decedents with non-Ontario residence status. After excluding non-cancer decedents, females with prostate cancer as their cause of death and missing cause of death information or record in the Ontario Cancer Registry, 125,755 decedents remained in the study cohort (Figure 1). The number of excluded females being recorded as having prostate cancer was suppressed in Figure 1 to prevent reidentification due to small cell issues. The mean age of decedents was 73.0 (SD 12.9), 46.1% were female, 14.1% resided in rural regions, and 86.2% had at least one other prevalent condition in addition to cancer before the last 100 days of life. Of the total cohort, 23.6% died of lung cancer, 7.2% breast cancer, 7.2% colorectal cancer, 6.6% pancreatic cancer, 5.2% prostate cancer, and 50.3% other cancer types, consisting of various cancers with small proportions preventing functional grouping (Table 1). Decedents experienced 1.7 emergency department visits and 1.2 inpatient hospital admissions in the last 100 days of life respectively.

In the last 100 days of life, decedents spent a mean of 1.7 days in emergency rooms, 12.6 days in inpatient hospitals, 4.4 days in palliative care units, 2.5 days in complex continuing care or rehabilitation, 4.7 days in long-term care, 4.6 days at home with outpatient care, 21.2 days at home with home care, and 48.3 days at home without any care (Figure 2). Of note, the median days were less than the means for all settings, except for days at home without healthcare (Supplementary File S4), suggesting a right skew to the data.

Results were stratified by cancer type, age category, and neighborhood income quintile (Figure 3). Lung cancer and pancreatic cancer decedents spent an average of 52.1 and 52.6 days at home without care, while prostate cancer and breast cancer decedents spent an average of 41.6 and 45.1 days at home without care. Decedents spent between an average of 3.8 (pancreatic) and 5.2 (breast) days in a palliative care unit, between 9.9 (breast) and 14.2 (other) days in an inpatient hospital, and between 3.2 (pancreatic) and 8.0 (prostate) days in long-term care. As age increased, decedents spent fewer days at home without care (19–44 = 47.3 days; 95 + = 37.7 days) and in hospital inpatient settings (19–44 = 17.6 days; 95 + = 7.2 days). Days spent in long-term care increased as decedents' age increased, with decedents aged 95+ spending 24.9 days in long-term care. Differences in places of care according to neighborhood income quintile were marginal; however, there was a general pattern that as income increased, days in institutions decreased and days at home increased. Please see Supplementary File S5 for results stratified by both cancer type and age.

Places of care were analyzed weekly across the last 14 weeks of life (i.e., approximately the last 100 days) to measure changes in setting as decedents approached death (Figure 4). Absolute and relative differences were produced for each setting to measure change between 14 weeks away from death, until the last week before death. Overall, the average number of weekly days spent at home decreased while days in institutions increased as decedents approached death. Days at home without care decreased by 3.5 days, representing a 76% reduction, while days in hospital inpatient settings increased by 1.5 days, representing a 371% increase, and days at home with home care increased by 0.9 days, representing a 1242% increase. Please refer to Supplementary File S6 for results stratified by cancer type.

Transitions between home, hospital, long-term care, and palliative care units, were captured from the last 4 months of life (Table 2). Between 4 months to 3 months before death, most decedents remained in their original location, while some moved from home to hospital (n = 6317; 5%) and hospital to home (n = 3040; 2%). Most decedents remained in their original setting as they approached death, while the percentage of decedents moving from home to hospital continued to increase in the last month (n = 22,612; 18%).

4 DISCUSSION

In this retrospective study describing the different settings of care experienced in the last 100 days of life in Ontario, Canada, cancer decedents spent almost half of their final days at home without receiving any care, 25 days in institutions, and 21 days at home with home care services, including physician home services. Further, decedents increasingly spent more days in institutions as they approached death, resulting in fewer days spent at home consequently.

Cancer decedents of less aggressive cancers, like breast, colorectal, and prostate, spent more days in institutions compared to decedents with more aggressive cancers like pancreatic and lung cancers. These findings may reflect the differences in treatment on survival rates between less and more aggressive cancers. For instance, people with breast, colorectal, and prostate cancers, have significantly higher five- and 10-year survival rates than people with lung or pancreatic cancers,³⁴ demonstrating the differences in treatment effectiveness between cancers. In turn, people dying from less aggressive cancers may spend more time in institutions as they may receive more treatment. This may be further compounded by aggressive cancers, such as lung and pancreatic cancers, rapidly progressing which may make it difficult for physicians to coordinate homecare or palliative care. Nevertheless, further statistical analyses would be required to verify that differences between cancer types are not attributed to random variation. Regardless of cancer type, decedents spent more days in institutions as they approached death, complementing existing literature.¹⁴ Further, our findings demonstrate that almost one in five decedents transitioned from home to hospital in the last month of life, which reinforces previous findings as well.^{35, 36}

Despite transitioning to hospital settings, decedents spent little time in palliative care units that provide end-of-life care to referred patients. This may be attributed to there only being 31 palliative care units in Ontario and few in rural areas.³⁷ As a result, rural decedents may experience difficulties in accessing palliative care units, as described in the literature.³⁸ Further, limited time spent in palliative care units may be due to the units having limited financial support and inadequate human resources,³⁷ which may prevent higher uptake.

Our findings highlight the role of homecare services for people dying from cancer in Ontario, Canada. While there was an increase in days at home with home care toward death, most cancer decedents spent their final days either in an institution or at home without any care. Palliative home care resources are stretched, and many people lack the social supports to achieve a home death, even when resources are available. Increasing access to homecare services at the end-of-life can have positive outcomes, such as decreased acute care usage^{39, 40} and providing patients with a dignified end-of-life experience by maintaining patient autonomy and integrity,⁴¹ while also aligning with patient preferences and values regarding their preferred location of death.^{4, 6} Therefore, our findings, coupled with the literature, emphasize the importance of improving end-of-life care delivery to keep patients in the community and out of institutions during this important life stage.

5 CONCLUSION

In the last 100 days of life, people dying of cancer in Ontario, Canada, spend 25 days in institutions and nearly 50 days at home without any care, and only 21 days at home with home care services, including physician home visits. As individuals approach death, they increasingly spend more time in institutions instead of in the community, contradicting established patient preferences. It is unlikely or desirable to avoid all institutional admissions during the end-of-life. At a population level, increasing days spent in acute care indicate that we can expect a certain number of transitions to acute care, despite the receipt of homecare. This realization may prompt decision-makers to shift resources to where needed most instead of a blanket approach in increasing services based on a diagnosis; for example, the often-recommended bolstering of homecare or physician home visits to address the deterioration during the end-of-life stage.

AUTHOR CONTRIBUTIONS

Abe Hafid: Conceptualization (equal); investigation (equal); methodology (equal); project administration (lead); visualization (lead); writing – original draft (lead); writing – review and editing (equal). Michelle Howard: Conceptualization (equal); funding acquisition (lead); investigation (equal); methodology (equal); writing – review and editing (equal). Colleen Webber: Conceptualization (equal); investigation (equal); methodology (equal); writing – review and editing (equal). Ana Gayowsky: Formal analysis (equal); methodology (equal); writing – review and editing (equal). Mary Scott: Methodology (equal); writing – review and editing (equal). Aaron Jones: Methodology (equal); writing – review and editing (equal). Amy T Hsu: Funding acquisition (equal); methodology (equal); writing – review and editing (equal). Peter Tanuseputro: Funding acquisition (equal); methodology (equal); writing – review and editing (equal). James Downar: Funding acquisition (equal); methodology (equal); writing – review and editing (equal). Katrin Conen: Methodology (equal); writing – review and editing (equal). Doug Manuel: Funding acquisition (equal); methodology (equal); writing – review and editing (equal). Sarina R Isenberg: Conceptualization (equal); funding acquisition (equal); investigation (equal); methodology (equal); writing – review and editing (equal).

FUNDING INFORMATION

This study was funded by a grant from the Canadian Institutes of Health Research project #159771. This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

ETHICS

The use of data in this project was authorized under section 45 of Ontario's Personal Health Information Protection Act, which does not require review by a Research Ethics Board.

CONFLICT OF INTEREST

All authors declare that they have no competing interests.