Introduction
Every year approximately 87,000 new cancers are diagnosed among adolescents and young adults (AYAs) (15–39 years) in the United States (US) (National Cancer Institute 2020, 2022). There are nearly 9,180 cancer-related deaths among AYAs every year, making it one of the leading causes of disease-related mortality (National Cancer Institute 2020, 2022). AYAs have unique physical and psychological needs, and there is a higher prevalence of psychological distress and complex pain among AYAs with cancer (Devlin et al. Reference Devlin, Murphy and Yeung2019; Ellis et al. Reference Ellis, Lin and Walsh2009; Kazak et al. Reference Kazak, DeRosa and Schwartz2010). Palliative care – which aims to relieve symptoms and psychological distress – is associated with improved quality of life, lower health-care costs, and increased survival with early integration in standard oncology care. (Bakitas et al. Reference Bakitas, Lyons and Hegel2009, Reference Bakitas, Tosteson and Li2015; Yadav et al. Reference Yadav, Heller and Schaefer2020) Palliative care has been proven to have benefits and promising results in AYAs with advanced cancer when integrated early into their treatment (Abdelaal et al. Reference Abdelaal, Mosher and Gupta2021; Sansom-Daly et al. Reference Sansom-Daly, Wakefield and Patterson2020). However, there have been relatively few studies investigating its availability and utilization (Donovan et al. Reference Donovan, Knight and Quinn2015).
AYAs require focused palliative care services to fulfill their complex needs; for example, AYA cancer patients have developmental, psychological, and social needs that are specific to their age group, which may not be fully understood or addressed in general oncology settings (Cheng and Wangmo Reference Cheng and Wangmo2020; Clark and Fasciano Reference Clark and Fasciano2015). Clinical guidelines encourage integrating palliative care with standard oncology care for AYAs with cancer across the continuum of care, starting early on at the stage of diagnosis (National Comprehensive Cancer Network 2022). Despite the clinical recommendation, the integration of palliative care in the cancer care of AYAs remains suboptimal to date (Maciasz et al. Reference Maciasz, Arnold and Chu2013). Evidence has shown that patients diagnosed with cancer are not routinely referred to palliative care due to various barriers, including limited resources, and supportive care is prioritized for those with the highest need as defined by the oncology team (Abdelaal et al. Reference Abdelaal, Mosher and Gupta2021; Wolfe and Rosenberg Reference Wolfe and Rosenberg2013). Specifically, many AYA patients receive care in adult cancer centers, where the focus is primarily on treatment and cure, rather than on addressing the unique needs of this age group (e.g., psychosocial needs) (Cheng and Wangmo Reference Cheng and Wangmo2020; Clark and Fasciano Reference Clark and Fasciano2015; Linebarger et al. Reference Linebarger, Ajayi and Jones2014). Moreover, AYA cancer patients and their families may have limited knowledge and awareness of palliative care and its benefits, which can make it difficult for them to access palliative care or communicate in care coordination (Huo et al. Reference Huo, Hong and Grewal2019a; Ivey and Johnston Reference Ivey and Johnston2022; Linebarger et al. Reference Linebarger, Ajayi and Jones2014; Mallon et al. Reference Mallon, Slater and Hasson2021).
Previous studies have primarily focused on palliative care delivery and utilization among the older adult population (Cheng and Wangmo Reference Cheng and Wangmo2020; Cohen-Gogo et al. Reference Cohen-Gogo, Marioni and Laurent2011; Roeland et al. Reference Roeland, Triplett and Matsuno2016; Ruck et al. Reference Ruck, Canner and Smith2018). A recent passing of the Childhood Cancer Survivorship, Treatment, Access, and Research (STAR) Act of 2018 highlights the need for AYA survivorship treatment and research and encourages the development of targeted interventions to improve quality of life and reduce the cancer burden for AYA cancer survivors and their families (Congress.Gov 2018). Population-based evidence is needed to better inform health systems’ planning and implementation of palliative care programs for AYA patients. However, no known studies have been conducted to estimate palliative care use among the AYA cancer patient population. To address this gap, the current study aimed to (1) estimate the national prevalence of inpatient palliative care use and (2) determine predictors of inpatient palliative care use among AYAs with cancer using nationally representative data of hospital discharge records in the US.
Methods
Data and study population
We analyzed data from the 2016–2019 National Inpatient Sample (NIS), which provides information on hospitalizations across all payers in the US. The NIS is maintained by the Agency for Health-care Research and Quality (AHRQ) as part of the Health-care Cost and Utilization Project (HCUP) (Agency for Healthcare Research and Quality 2021). The HCUP database can be accessed through the HCUP Central Distributor (https://www.hcup-us.ahrq.gov/tech_assist/centdist.jsp). De-identified NIS data were delivered for analysis following completion of a data–user agreement with AHRQ (S.Y., Z.X., and Y.-R.H. completed the agreement and had full access to all of the data in the study). The NIS provides a cross-sectional representative sample of discharges from US hospitals and is the largest inpatient database in the US. The data contain information about patient demographic and hospital characteristics related to inpatient admissions (Agency for Healthcare Research and Quality 2021).
A total of 10,979 hospitalizations were included in the analysis. We included patient admissions that met these criteria of (1) age between 15 and 39 years (National Cancer Institute 2020, 2022), (2) primary diagnosis of cancer, and (3) highest likelihood of in-hospital mortality. Cases with a primary diagnosis of cancer were identified using the International Classification of Disease, Tenth Revision (ICD-10) codes C00–C96. Given the unknown pattern of palliative care use among AYA cancer patients, we focused on those with high-risk mortality, defined by All-Patient Refined Diagnosis-Related Group (APR-DRG) Risk of Mortality rating of 3–4 (Baram et al. Reference Baram, Daroowalla and Garcia2008), likely to use palliative care. The APR-DRG rating calculates disease-specific mortality risk incorporating comorbidity conditions. With every one-unit increase in APR-DRG Risk of Mortality, there is a 3 times increase in in-hospital mortality (Baram et al. Reference Baram, Daroowalla and Garcia2008). This approach is consistent with previous studies that have used mortality rating to indicate palliative care prioritization among pediatric and adult populations (Cheng and Wangmo Reference Cheng and Wangmo2020; Mulvey et al. Reference Mulvey, Smith and Gourin2016; Ruck et al. Reference Ruck, Canner and Smith2018). As the NIS data is de-identified, it does not constitute research involving human subjects; therefore, this study was exempted from review by the University of Florida Institutional Review Board. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology guideline (von Elm et al. Reference von Elm, Altman and Egger2014).
Outcome variable: inpatient palliative care
The primary outcome of this study was inpatient palliative care utilization, which was identified by the ICD-10 code Z51.5 (Ruck et al. Reference Ruck, Canner and Smith2018). This was coded as a binary variable indicating whether having palliative care encounter during hospitalization.
Independent variables: patient and hospital characteristics
Patients’ age in years (15–24 and 25–39), sex (male and female), race and ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic/Latinx, other [including Asian, Native Hawaiian, Pacific Islander, Native American, and multiple races]), median household income quartiles based on patient’s ZIP code (0–25th percentile [lowest quartile], 26th–50th percentile, 51st–75th percentile, and 76th–100th percentile [highest quartile]), health insurance type (defined as primary payer; private, public, self-pay, or no charge), length of stay (0–1, 2–4, and 5+ days), patient location (metropolitan, micropolitan, and rural), and elective admission (yes and no) were used as patient-level covariates. Hospital-level covariates included hospital region (Northeast, Midwest, South, and West), hospital location (rural and urban), hospital teaching status (teaching and non-teaching), and hospital size based on HCUP hospital bed size categorized based on hospital location and teaching status (categorized as small, medium, and large). For example, a large hospital size refers to a hospital with more than 100 beds in a rural area, 200 beds in an urban non-teaching hospital, and 500 beds in an urban teaching hospital (Agency for Healthcare Research and Quality 2006). We accounted for these patient- and hospital-level characteristics in the adjusted analysis to control for their direct or indirect effect on outcomes.
Statistical analysis
We constructed 2 sequential multivariable regression models to examine patient and hospital characteristics associated with palliative care utilization. Both unadjusted and adjusted odds ratios (OR) with 95% confidence interval (CI) were calculated. The first model included patient characteristics including age, sex, race and ethnicity, zip code–level income, insurance type, patient location, admission type, and length of stay (Model 1). Model 2 then adjusted for hospital characteristics (hospital region, location, teaching status, and bed size) in addition to the patient characteristics used in Model 1. Given a concern of potential residual confounding with sex-specific cancer types (e.g., female breast), we did not include cancer type in the multivariable model and only reported the results from the bivariate analysis (for most prevalent cancer types). We tested interactions of race and ethnicity, household income, insurance type, patient location, and hospital region by year; however, we did not find any significant interaction. All analyses were conducted in accordance with NIS national discharge-level estimates (Agency for Healthcare Research and Quality 2018) and the HCUP formal data use agreement. We applied recommended survey weights to account for patient discharge weights, hospital-level clusters, and survey strata (Agency for Healthcare Research and Quality 2018). All analyses were conducted using SAS Version 9.4 (SAS Institute). We determined statistical significance at a 2-sided p-value of less than 0.05.
Results
Sample characteristics
Of all the hospitalizations recorded in the NIS data between 2016 and 2019, 10,979 (equating to 54,895 national estimates) met our inclusion criteria (AYA patients with cancer and higher in-hospital mortality risk). Of the study sample, the majority (78%) were 25 to 39 years old, male (52%), and non-Hispanic White (50%). The most common cancer types identified were brain cancer (15.5%), leukemia (11.5%), and colorectal cancer (7.2%) (Table 1).
Demographic characteristics of the sample population by inpatient palliative care, 2016 to 2019
a Estimates were weighted to represent national discharge level.
b Includes Asian, Pacific Islander, Native American, and multiple races.
c Quartile classification of the estimated median household income of residents in the patient’s ZIP code.
d Based on Rural–Urban Continuum Code (RUCC) and Urban Influence Code (UIC) of the Economic Research Service of the US Department of Agriculture.
e NIS bed size categories are determined based on hospital beds and are specific to the hospital’s location and teaching status. For more details, see https://www.hcup-us.ahrq.gov/db/vars/hosp_bedsize/nisnote.jsp. For example, a large hospital size refers to a hospital with more than 100 beds in a rural area, 200 beds in an urban non-teaching hospital, and 500 beds in an urban teaching hospital.
Inpatient palliative care utilization patterns
Overall, 19.9% of the hospitalizations involved palliative care services use (Table 1). AYA cancer patients aged 25–39 years old (20.9%) were more likely to use palliative care than those aged 15–24 (16.1%) (p < 0.001). Compared with non-Hispanic White (18.6%), patients of race or ethnicity other than White had more palliative care utilization (non-Hispanic Black, 21.9%, Hispanic, 21.2%, and other, 20.4%) (p = 0.045). Female AYA patients (22.0%) had a higher palliative care use than male patients (17.8%) (p < 0.001). AYA cancer patients who were covered with public insurance (22.6% vs. 17.6% with private), with emergent admission (22.7% vs. 9.1% in elective admission), and in the Northeast region (23.1% vs. 18.2% in West) had higher utilization of palliative care services. By cancer type (Figure 1), palliative care utilization rate was more commonly used among female breast cancer patients (27.7%), followed by lung (25.1%), other cancer types (24.4%), colorectal (22.9%), and multiple cancers (17.7%).
Weighted prevalence of inpatient palliative care use among AYA cancer patients with high-risk mortality, 2016–2019.
Predictors of inpatient palliative care utilization
The results of multivariable logistic regression models are shown in Table 2. When adjusted for patient characteristics (Model 1), older age (25–39 years; OR 1.30, 95% CI 1.14–1.48) and female sex (OR 1.26, 95% CI 1.14–1.40) were significantly associated with palliative care utilization. Palliative care use was significantly less likely in patients with elective admissions (OR 0.34, 95% CI 0.28–0.41) than emergent admission type and covered under private insurance (reversed OR 0.81, 95% CI 0.72–0.91) or self-pay payor status (OR 0.69, 95% CI 0.52–0.90) than covered under Medicare.
Multivariate logistic regression analysis of variables associated with inpatient palliative care
OR, odds ratio; Ref, reference.
a Model 1 adjusted for patient age, race/ethnicity, sex, household income, insurance type, location, admission type, length of stay, and year.
b Mode 2 adjusted for patient characteristics and admission type included in Model 1 plus hospital characteristics.
After controlling for all patient- and hospital-related characteristics in the multivariable logistic regression model (Model 2), the independent predictors of palliative care use were age between 25 and 39 (OR 1.31, 95% CI 1.15–1.49), Hispanic /Latinx ethnicity (vs. non-Hispanic White; OR 1.16, 95% CI 1.01–1.34), female (vs. male; OR 1.27, 95% CI 1.14–1.41) sex, elective admission (vs. emergent admission; OR 0.34, 95% CI 0.28–0.41), length of stay of 2 to 4 days (vs. 0–1 day; OR 0.79, 95% CI 0.63–0.99), public insurance (vs. private insurance; OR 1.23, 95% CI 1.10–1.38), geographic region South (vs. Northeast; OR 0.78, 95% CI 0.66–0.94) and Western region (vs. Northeast; OR 0.72, 95% CI 0.60–0.86), and large hospital (vs. small hospital; OR 0.83, 95% CI 0.72–0.96).
Discussion
Using a nationally representative sample of US hospital admissions, our study provides a population estimate of palliative care use in AYAs with cancer and a high risk of mortality. Overall, 19.9% of hospitalizations in the study cohort had a palliative care encounter. Significant predictors of palliative care included age, sex, race/ethnicity, type of insurance, length of stay, admission type, hospital region, and hospital size. A higher likelihood of palliative care use was observed for patients in the age group 25 to 39 years compared to patients 18 to 24 years. This finding suggests there are differences in palliative care practice between the age groups. Two previous population-based studies examined palliative care utilization in pediatric and adult populations with advanced cancer. Both studies found a lower prevalence of palliative care utilization among younger age groups and a relatively higher prevalence in the older age groups within their study cohorts (Cheng and Wangmo Reference Cheng and Wangmo2020; Mulvey et al. Reference Mulvey, Smith and Gourin2016). A possible explanation for these differences in referral practices could be the limited number of specialized AYA palliative care practitioners relative to adult palliative care practitioners, resulting in lower access to palliative care services forAYAs (Feudtner et al. Reference Feudtner, Womer and Augustin2013). Palliative care services are important for younger patients and their families as they need support to cope with the psycho–social–emotional needs (i.e., development continuum) by the advanced disease and during the end-of-life stage (Cheng and Wangmo Reference Cheng and Wangmo2020; Clark and Fasciano Reference Clark and Fasciano2015; Linebarger et al. Reference Linebarger, Ajayi and Jones2014). To meet the growing palliative care demands for AYA population, public health efforts should be focused on providing training to clinicians for providing specialist AYA palliative care services. Further research is required to explore the reasons for lower palliative care utilization in the younger age groups.
Our study found a positive association between female sex and inpatient palliative care use. This finding is consistent with prior literature on the adult cancer population (Ruck et al. Reference Ruck, Canner and Smith2018). Evidence across health-care utilization literature shows a greater consumption of health-care services by women. Previous studies on costs associated with palliative care use also suggest that female sex is associated with lower hospital daily costs (Cheng and Wangmo Reference Cheng and Wangmo2020). A possible explanation could be better knowledge of palliative care among women than men (Huo et al. Reference Huo, Hong and Grewal2019a). Studies suggest gender-based preference for palliative care services. According to a recent study, women were 3 times more likely than men to prefer palliative care services. These gender differences in awareness and preferences related to palliative care services may lead to disparities in cancer end-of-life care; therefore, interventions should be undertaken to promote awareness and utilization of palliative care among men (Saeed et al. Reference Saeed, Hoerger and Norton2018).
We also found a higher prevalence of palliative care use among non-Hispanic Black and Hispanic AYAs than non-Hispanic White AYAs. However, after controlling for both patient- and hospital-related characteristics, these differences in the use of inpatient palliative care services were no longer significant for non-Hispanic Black AYAs. Previous studies of advanced cancer patients found that Black AYAs and Hispanic AYAs were more likely than White AYAs to use inpatient palliative care services and suggested that inpatient palliative care services are more accessible and equitable for these groups, compared to other settings (Griggs Reference Griggs2020; Sharma et al. Reference Sharma, Cameron and Chmiel2015). It is important to consider that inpatient settings may not be the most appropriate or preferred setting for all patients, especially for AYA cancer patients who may wish to receive care in an outpatient setting or at home (Muni et al. Reference Muni, Engelberg and Treece2011; Sharma et al. Reference Sharma, Cameron and Chmiel2015). Additionally, while inpatient palliative care may be more accessible, it may not necessarily be the most effective or efficient way to deliver palliative care. Inpatient settings may also not be able to address all of the unique needs of AYA cancer patients, such as their developmental, psychological, and social needs (Cheng and Wangmo Reference Cheng and Wangmo2020; Clark and Fasciano Reference Clark and Fasciano2015; Linebarger et al. Reference Linebarger, Ajayi and Jones2014). There is also the possibility that the physician might disagree with families of critically ill non-White patients regarding end-of-life care, which may explain these differences (Muni et al. Reference Muni, Engelberg and Treece2011; Sharma et al. Reference Sharma, Cameron and Chmiel2015). Overall, more research is needed to investigate the underlying causes of disparities or access barriers across races and ethnicities to ensure that all AYA cancer patients have access to adequate palliative care as part of their cancer treatment.
Although we did not observe significant racial and ethnic disparities in palliative care use, the overall proportion of utilization use in our AYA population was less than 20%. This is in the lowest end of the utilization range of older cancer populations (10%–70% by cancer type) (Huo et al. Reference Huo, Hong and Turner2019b; Rubens et al. Reference Rubens, Ramamoorthy and Saxena2019; Ruck et al. Reference Ruck, Canner and Smith2018). The role of health-care providers in evaluating each patient individually and referring them for palliative care can be further strengthened to increase the uptake of palliative care services (Griggs Reference Griggs2020). Our analysis also revealed hospital characteristics associated with palliative care use among AYA cancer patients. Hospitals located in the South region or large bed sizes had lower utilization, contrasting with studies on inpatient palliative care use among the adult population (Lee et al. Reference Lee, Gani and Canner2021; Singh et al. Reference Singh, Peters and Tirschwell2017; The Center to Advance Palliative Care 2020). For example, a recent report from the Center to Advance Palliative Care shows that the vast majority (>95%) of large hospitals with 300 or more beds have palliative care teams (The Center to Advance Palliative Care 2020). However, this study found that AYA patients in large hospitals were less likely to receive palliative care. It would be worthwhile to investigate further geographic variations and the availability of palliative care teams in hospital settings for AYA patients in future studies.
Our analysis has some limitations. First, our dataset is limited to inpatient palliative care services. However, there is evidence that the majority of palliative care services are provided in the inpatient setting (Roeland et al. Reference Roeland, Triplett and Matsuno2016). A further study is needed to estimate how many palliative care consultations and home health or hospice referrals are provided in the outpatient setting to this population. Second, we used ICD-10 code Z51.5 for identifying patients who received palliative care services, which is subject to limitation of incomplete or inaccurate administrative coding. However, it has been used in previous studies utilizing the national inpatient sample dataset (Ruck et al. Reference Ruck, Canner and Smith2018; Singh et al. Reference Singh, Peters and Tirschwell2017). Third, the dataset did not provide any information about the cancer stage; we tried to overcome this limitation by using the APR-DRG Risk of Mortality score to identify patients at high risk of mortality. This approach has been previously used to eliminate the different cancer stages among cancer patients (Cheng and Wangmo Reference Cheng and Wangmo2020; Ruck et al. Reference Ruck, Canner and Smith2018). Despite these limitations, to our knowledge, this is the first study to assess national estimates of inpatient palliative care utilization among AYAs in recent years. Our findings could inform efforts in health policy and clinical guidelines to improve the integration of palliative care into standard cancer care in AYAs.
Conclusion
In this nationally representative sample population, less than 20% AYAs with cancer and a high risk of mortality received inpatient palliative care services. Predictors of palliative care included age, sex, race, type of insurance, length of stay, admission type, hospital region, and hospital size. These findings address an important gap in palliative care service use among AYAs with cancer and have implications for resource and personnel allocation and strategies for ensuring timely access to palliative care services. There is a need for future research to explain the barriers to the use of palliative care services and to develop policy and clinical guidelines for the wider adoption of palliative care services for AYA patients with cancer.
Acknowledgments
The authors thank the Health-care Cost and Utilization Project (HCUP) and their Data Partners (52 state orgnizations) for preparing and supplying the data for analysis.
Funding
There was no external funder for this study.
Conflicts of interest
The authors declare no conflict of interest
Open access
