Discussion

The results of this study demonstrate an overall increased trend toward death at home or in hospice care between 1999–2020, with an average annual percent change (AAPC) of 6.55%. However, subgroup analysis reveals significant disparities in the location of death when comparing different population groups, regions, and age groups with CCA. Notably, there are decreased odds of dying at home among American Indian, Black, and Asian subgroups when compared to the White individuals, with Blacks demonstrating the lowest likelihood of dying at home. Blacks were the only racial group that had a lower overall number of deaths at home or in a hospice facility than other facilities. At the same time, Blacks showed the highest AAPC of deaths at home or hospice facilities between 1999 and 2020. In terms of geographical variations for CCA deaths in the U.S., the Southern region and those in the Northeast and Midwest regions exhibited the lowest likelihood of dying at home or hospice facilities. In contrast, those in the Western region demonstrated the highest odds of dying at home or hospice facilities. The Northeastern region showed equivalent rates of deaths either at home or in a hospice facility. Interestingly, individuals aged 85 years or older were the least likely to die at home or hospice. Of all the age groups, only this group had a higher number of deaths outside home or hospice facility. The lowest AAPC was seen amongst those aged 45–54, followed by those under 45. These findings emphasize the need for more personalized care to address the cultural, regional, and age-related factors that influence end-of-life care preferences and practices, particularly among racial minorities, extremities of age, and those pertaining to certain geographical areas.

Our findings both corroborate existing literature and reveal novel insights into end-of-life care, highlighting important trends and disparities when assessing sociodemographic and geographic attributes. The growing number of patients with CCA opting for hospice care at the time of death, compared to inpatient settings or nursing facilities, is a pattern consistent with a broader trend. A similar trend has been reported in England, where home and hospice deaths from cancer have risen since 2005 (Gao et al. Reference Gao, Ho and Verne2013). Likewise, Korea has experienced a notable increase in home deaths for cancer patients following the introduction of insurance coverage for home-based hospice care (Yun et al. Reference Yun, Jang and Park2023). This shift toward hospice and home deaths, along with a decline in inpatient hospital deaths, may reflect several factors. First, the ubiquitous emphasis on advance care planning has allowed patients to express their end-of-life preferences earlier, resulting in care that better aligns with their wishes – often favoring home or hospice deaths surrounded by loved ones (Khan et al. Reference Khan, Gomes and Higginson2014). Improvements in palliative care practices and the decreasing stigma around death in the United States may also contribute to this trend.

In patients with CCA, we observed that American Indian, Asian, and Black individuals were less likely to die at home or in a hospice facility compared to their white counterparts. Although limited research specifically addresses how racial disparities affect the place of death in CCA, a study by Chino et al. found that Black, American Indian, and nonwhite Hispanic cancer patients were significantly less likely to utilize hospice care than white patients (Chino et al. Reference Chino, Kamal and Leblanc2018). Together, these findings suggest potential barriers to home or hospice deaths among minority groups. This may stem from reduced family support, financial or environmental challenges in accessing hospice care, or implicit bias affecting provider referrals. Furthermore, Black patients often present with CCA at a more advanced stage than white patients (Munir et al. Reference Munir, Woldesenbet and Endo2023). This leads to accelerated health deterioration and higher rates of hospital deaths due to limited time and worsening prognosis. While our study reflects a general increase in hospice and home deaths, significant racial disparities persist.

There were notable regional differences in hospice utilization across the U.S. We observed that patients with CCA in the Northeast and Midwest were less likely to die at home or in a hospice facility compared to those in the South. Despite the Northeast’s leadership in legislated palliative care access – such as New York’s Palliative Care Information Act, which mandates that patients be informed about palliative care options – this did not correlate with higher hospice utilization in the region. In contrast, other areas, particularly the South, may lag in legal mandates and available services. However, given that CCA is an aggressive cancer with a poor prognosis, the Northeast’s higher density of advanced medical centers may contribute to these findings, explaining why there were increased rates of in-hospital deaths. Our results could reflect a higher prevalence of advanced treatments in these regions, inadvertently leading to more hospital deaths. For example, a study by Uhlig et al. in the Annals of Surgical Oncology found that the Middle Atlantic region, which includes states in the Northeast, had the highest surgery rates for intrahepatic CCA, indicating a more aggressive surgical approach (Uhlig et al. Reference Uhlig, Sellers and Cha2019). This could lead to increased complications and risk of subsequent hospital deaths. Similarly, our findings align with a study done by Connor et al., which also demonstrated higher hospice use in the South compared to the Northeast and Midwest. In another study by Virnig et al., rates of hospice use were over 10-fold higher in Florida than in Maine. In comparison, the West has even greater rates of hospice use compared to the South. The Western regions are known to have fewer hospital beds per capita (Hansen et al. Reference Hansen, Tolle and Martin2002). The higher percentage of deaths occurring at home in the West correlates with a lower availability of in-hospital beds. Physician perspectives also play a crucial role in shaping patient awareness and acceptance of hospice care. Data on geographic differences in physician attitudes toward hospice are limited, but further research that explores how these perspectives vary across regions and influence hospice utilization in patients with CCA would be valuable.

The results of the current study show an important trend in the place of death among different age groups. We noticed a lower AAPC of hospice use or home death amongst younger patients compared to their older counterparts. Prior research indicates that younger patients are more likely to die in the hospital secondary to more aggressive treatments that often require hospital-based care (Falchook et al. Reference Falchook, Dusetzina and Tian2017). This is likely because younger patients may be better surgical candidates and have fewer comorbid conditions. In contrast, we noted a significant increase in the rates of home or hospice death amongst those aged 55–74. Research by Cagle et al. found that older age is a significant predictor of hospice utilization among cancer patients (Cagle et al. Reference Cagle, Lee and Ornstein2020). Additionally, the literature suggests that older patients with primary liver cancer, including CCA, have a shorter time to hospice enrollment (Fukui et al. Reference Fukui, Golabi and Otgonsuren2018). However, our findings show lower hospice utilization among patients over 85 years of age compared to those in late middle age. This discrepancy may reflect higher rates of postoperative complications in elderly patients undergoing major hepatectomy or interventional radiology procedures for CCA (de la Fuente Sg Bk and Scarborough Reference de la Fuente and Scarborough2013; Maeda et al. Reference Maeda, Kuriyama and Ito2022), which can lead to prolonged hospital stays and increased mortality. These patients were also likely sicker, more debilitated, and had fewer resources for hospice. Further investigation into subcategorizing age groups based on the aggressiveness of treatment received could provide additional insights.

The odds of hospice or home death compared to hospital death among males versus females with CCA were similar. The observed lack of difference in gender and place of death is consistent with other studies that examine hospice use across various cancer diagnoses (Virnig et al. Reference Virnig, Marshall Mcbean and Kind2002). Although a study by Mojtahedi et al. suggested that females with extrahepatic CCA may be more likely to receive palliative care services compared to males, specific rates of hospice utilization by gender were not provided in this study (Mojtahedi et al. Reference Mojtahedi, Yoo and Callahan2021).

We identified some limitations with our study, mostly pertaining to the dataset. Firstly, the WONDER dataset relies on death reporting by healthcare professionals, and some cases may have been missed. Secondly, the underlying cause of death reported in the WONDER dataset reflects the cause of death listed on the death certificate, which may not always be accurate and is open to interpretation. This may lead to an under-identification of patients who had CCA as the cause of death. Of note, Meino et al. reported that patients with cancer as an underlying cause of death tend to have an accurate death certificate (Mieno et al. Reference Mieno, Tanaka and Arai2016); these observations may not be accurate when applied to a larger database. Additionally, we expect the differences in cause of death reporting to remain constant over the years, allowing us to compare the groups adequately. The location of death reported may not always be accurate, such as when patients are being palliatively treated at home or in a hospice facility and are transferred to a medical facility right before their death. Furthermore, some patients may be treated palliatively in a medical facility, especially if prolonged survival is not anticipated. Moreover, hospice facilities as places of death were not introduced into the WONDER dataset until 2003. The CDC also suppresses values that involve less than ten individuals. To circumvent this, for trend graphs and AAPC, we requested data only divided by place of death and each of the variables, race, gender, age groups, and census regions. We did not employ this technique for logistic regression in order to maintain the same data across univariate and multivariate analyses. Finally, the WONDER dataset lacks the granularity to report additional patient or disease factors such as socioeconomic status, access to care, insurance status, malignancy stage, or treatment received. Further studies are needed to assess the effect of the aforementioned factors on place of death for patients with CCA.

The strengths of our study are several; our analysis not only highlights the disparities in the location of death for patients with CCA, but it also demonstrates trends in palliative care utilization and variation across different sociodemographic factors. The inequality that exists between non-White and White populations in place of death from CCA is strongly evidenced in our study. We hope that these results will positively impact the discussions around and management of end-of-life care in racial minorities with CCA. Education about the goals and benefits of hospice care with non-White populations will allow enhanced decision-making that aligns with the patient’s goals of care. More research is warranted to discover and address the barriers to hospice care that exist among non-White populations in the United States. Additional findings demonstrating the disparities between regions and age groups show a significant disproportion in end-of-life care in the United States, regardless of race. The healthcare costs associated with these disparities are striking. Many studies have shown that the cost of dying in the hospital is dramatically more costly than dying at home. Hoverman et al. observed that when analyzing patients with cancer, dying in hospital is twice as costly as dying at home, with an average cost in the final month of life of $20,113 vs. $10,803, respectively (Hoverman et al. Reference Hoverman, Mann and Phu2020). The emotional and physical costs of dying in the hospital can be alleviated by improving access to home hospice care. Wright et al. demonstrated that cancer patients who die in the hospital setting experience more physical and emotional distress than those who die at home (Wright et al. Reference Wright, Keating and Balboni2010). Dying at home impacts multiple aspects of a patient’s final moments of life. Our study has demonstrated a clear discrepancy in place of dying among specific races, regions, and age groups in patients with CCA and has important implications for public health stakeholders in addressing barriers to and potential improvement in the utilization of home hospice care for those with CCA.