Medicare Plan Switching and Hospice Care Among Decedents With Advanced Cancer

Xin Hu; Changchuan Jiang; Youngmin Kwon; Fangli Geng; Qinjin Fan; Kewei Sylvia Shi; Zhiyuan Zheng; Jingxuan Zhao; Joan L Warren; K Robin Yabroff; Xuesong Han

doi:10.1001/jamanetworkopen.2026.0755

. 2026 Mar 24;9(3):e260755. doi: 10.1001/jamanetworkopen.2026.0755

Medicare Plan Switching and Hospice Care Among Decedents With Advanced Cancer

Xin Hu ^1,^2,^3,^✉, Changchuan Jiang ⁴, Youngmin Kwon ⁵, Fangli Geng ⁶, Qinjin Fan ⁷, Kewei Sylvia Shi ⁷, Zhiyuan Zheng ⁷, Jingxuan Zhao ⁷, Joan L Warren ⁷, K Robin Yabroff ⁷, Xuesong Han ⁷

¹Department of Radiation Oncology, School of Medicine, Emory University, Atlanta, Georgia

²Winship Cancer Institute, Emory University, Atlanta, Georgia

³Department of Health Policy and Management, Rollins School of Public Health, Emory University, Atlanta, Georgia

⁴Department of Internal Medicine, UT Southwestern Medical Center, Dallas, Texas

⁵Department of Health Policy, Vanderbilt University Medical Center, Nashville, Tennessee

⁶Department of Health Services, Policy and Practice, Brown University School of Public Health, Providence, Rhode Island

⁷American Cancer Society, Atlanta, Georgia

Accepted for Publication: January 2, 2026.

Published: March 24, 2026. doi:10.1001/jamanetworkopen.2026.0755

^✉

Corresponding Author: Xin Hu, PhD, Assistant Professor, Department of Radiation Oncology, Emory University School of Medicine, 1365 Clifton Rd, Ste A1320, Atlanta, GA 30322 (xin.hu@emory.edu).

Author Contributions: Xin Hu had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Hu, Geng, Zhao, Yabroff.

Acquisition, analysis, or interpretation of data: Hu, Jiang, Kwon, Geng, Fan, Shi, Zheng, Warren, Yabroff, Han.

Drafting of the manuscript: Hu.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Hu.

Obtained funding: Hu.

Administrative, technical, or material support: Hu, Kwon, Shi, Warren.

Supervision: Kwon.

Conflict of Interest Disclosures: Dr Hu reported grants from the National Institutes of Health (NIH), Pfizer, Johnson & Johnson, PhRMA Foundation, St Jude Children's Research Hospital, and the Patrick and Catherine Weldon Donaghue Medical Research Foundation outside the submitted work. Dr. Jiang received honoraria from MJH Science. Dr Kwon reported grants from the Agency for Healthcare Research and Quality outside the submitted work. Dr Yabroff reported honoraria from the National Comprehensive Cancer Network and the American Society of Clinical Oncology and grants from the National Cancer Institute and Blood Cancer United outside the submitted work. Dr Geng reports grant support from the National Institute on Aging, the National Bureau of Economic Research, and the Rhode Island Parent Information Network. No other disclosures were reported.

Funding/Support: The collection of cancer incidence data used in this study was supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; Centers for Disease Control and Prevention’s (CDC’s) National Program of Cancer Registries, under cooperative agreement 1NU58DP007156; the National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) program under contract HHSN261201800032I awarded to the University of California, San Francisco; contract HHSN261201800015I awarded to the University of Southern California; and contract HHSN261201800009I awarded to the Public Health Institute. Research reported in this publication was partially supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health under award number K12AR084234 (Hu).

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The ideas and opinions expressed herein are those of the authors and do not necessarily reflect the opinions of the State of California, Department of Public Health, the National Cancer Institute, or the Centers for Disease Control and Prevention or their contractors and subcontractors. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Meeting Presentations: This paper was presented at the 2025 Meeting of the American Society of Clinical Oncology; May 31, 2025; Chicago, IL; the 2025 AcademyHealth Annual Research Meeting; June 7, 2025; Minneapolis, MN; and 2025 Georgia Clinical & Translational Science Alliance Health Services Research Day; May 6, 2025; Atlanta, GA.

Data Sharing Statement: See Supplement 2.

Additional Contributions: The authors acknowledge the efforts of the National Cancer Institute; the Office of Research, Development, and Information, Centers for Medicare and Medicaid Services; Information Management Services, Inc; and the Surveillance, Epidemiology, and End Results (SEER) program tumor registries in the creation of the SEER Medicare database.

^✉

Corresponding author.

PMCID: PMC13014202 PMID: 41874509

This cohort study reports on differences in hospice use between Medicare decedents with advanced cancers who were continuously enrolled in Medicare Advantage, traditional Medicare, or who switch plans in the last year of life.

Key Points

Question

What are the hospice utilization implications for Medicare decedents with advanced cancers continuously enrolled in Medicare Advantage (MA), traditional Medicare (TM) or who switch plans in the last year of life?

Findings

In this cohort study among 196 536 beneficiaries who died in 2010-2020, continuous MA enrollees had 6.93 percentage points higher probability of hospice use and 5.13 days longer stay than those with continuous TM. Continuous MA enrollees more often received hospice care at home, while those who switched from MA to TM more often received hospice care in nursing homes.

Meaning

Results of this study suggest greater hospice use among MA decedents, which may improve end-of-life care; those who switched from MA to TM may represent a subgroup with greater nursing home needs and access barriers under MA.

Abstract

Importance

Hospice is central to end-of-life (EOL) care for patients with advanced cancers and is an excluded benefit under Medicare Advantage (MA), with coverage instead provided by traditional Medicare (TM). With growing MA penetration, more beneficiaries also switch between MA and TM for financial protection and physician access considerations, although less is known about how different Medicare programs and plan switching behaviors affect EOL care for patients with advanced cancers.

Objective

To evaluate hospice utilization and places of hospice care by Medicare plan switching patterns.

Design, Setting, and Participants

This retrospective cohort study used Surveillance, Epidemiology, and End Results (SEER) Medicare data to identify beneficiaries aged ≥66 years diagnosed with distant-stage female breast, colorectal, lung, pancreatic, or prostate cancers from 2010 to 2019 who died by 2020. Beneficiaries were followed for up to 1 year before death. Data were analyzed from August 1, 2024, to December 14, 2025.

Exposure

Plan switching patterns classified as continuous MA, continuous TM, MA to TM, TM to MA, and other (ie, multiple switches).

Main Outcomes and Measures

Main outcomes were hospice enrollment in the last year of life and within 3 days of death, total hospice length of stay, and place of last hospice stay (home, nursing home, hospice facility, inpatient facility, or other) using multivariable regressions.

Results

The sample included 196 536 decedents (46.5% female, 49.2% aged 66-74 years). Plan switching was infrequent (1.5% TM to MA; 1.8% MA to TM). Those who switched plans were more likely to be members of racial and ethnic minority groups and dual Medicare-Medicaid enrollees. Hospice enrollment was highest for those with continuous MA (74.8%), followed by those who switched from TM to MA (69.0%), those with continuous TM (68.5%), and those who switched from MA to TM (66.4%). Continuous MA beneficiaries had longer hospice stays than continuous TM beneficiaries (48.3 vs 43.8 days). Compared with continuous TM, continuous MA beneficiaries were more likely to receive hospice at home (1.93 percentage points [pp]; 95% CI, 1.40-2.45 pp; P < .001), while those who switched from MA to TM were more likely to receive hospice in nursing homes (2.45 pp; 95% CI, 1.26-3.63 pp; P < .001), particularly among dual Medicare-Medicaid enrollees (6.01 pp; 95% CI, 2.80 to 9.21 pp; P < .001).

Conclusions and Relevance

In this cohort study of Medicare decedents with advanced cancers, continuous MA enrollees were most likely to receive hospice at home, while those who switched from MA to TM more frequently received hospice care in nursing homes. Plan switching near the EOL may reflect access barriers, highlighting the importance of addressing care coordination to improve EOL care.

Introduction

Hospice is a specialized, supportive, comfort-focused care model for people with serious illness near the end of life (EOL), such as those with advanced cancers. It prioritizes relief of pain and other symptoms, psychosocial and spiritual support for patients and families, and alignment of care with patient goals.^1,2 Early hospice enrollment has been associated with improved quality of life,³ reduced unnecessary and burdensome care,⁴ and a greater likelihood of dying in the preferred setting, often at home.^5,6 Despite these benefits, nearly one-third of patients with advanced cancer do not enroll in hospice before death.⁷

Medicare is the largest payer for both cancer treatment and hospice in the United States.^8,9 A growing share of Medicare beneficiaries are enrolled in Medicare Advantage (MA), privately managed alternatives to traditional Medicare (TM).¹⁰ While MA plans manage most Medicare-covered services, hospice has historically been carved out of MA.^11,12 In other words, when MA beneficiaries elect hospice, responsibility shifts to TM for hospice-related services, while MA retains responsibilities to manage or coordinate other services (although minimal) near the EOL. This carve-out arrangement gives MA plans financial incentives to encourage early hospice enrollment to offset high costs of intensive cancer treatments but also creates misaligned accountability for care coordination during hospice stay. Consistent with this dynamic, MA decedents are more likely to enroll in hospice than those in TM.¹¹

At the same time, switching between MA and TM has become increasingly common.¹³ Importantly, patients newly diagnosed with cancer are more likely to switch from MA to TM than the general Medicare population.¹⁴ Although the underlying reasons for plan switching in patients with cancer are not fully understood, prior studies suggest potential challenges accessing oncologists and other specialists, as well as utilization management practices that contribute to unmet care needs and subsequent plan switching.^15,16,17 For example, limited MA coverage of National Cancer Institute (NCI)–designated cancer centers¹⁶ may prompt patients to switch plans to access clinical trials and novel therapies, potentially prolonging active treatment and delaying hospice transition; narrower skilled nursing facility (SNF) networks and prior authorization under MA may prompt patients to move to TM for broader access after hospitalizations related to cancer treatment or complications, which can affect downstream hospice care, including greater reliance on certain care settings such as nursing homes. Conversely, continuous MA enrollment may facilitate earlier more coordinated hospice referral, potentially explaining higher enrollment and longer stays.¹¹

Despite these clinical implications, little is known about whether beneficiaries who switch between MA and TM during the last year of life experience different hospice utilization. This evidence gap is especially salient amid recent efforts to integrate hospice into MA through the Value-Based Insurance Design model, a model discontinued by the Centers for Medicare and Medicaid Services in 2024.¹⁸ Under either arrangement, MA plans retain incentives to manage utilization near the EOL, yet it remains unclear how these incentives intersect with EOL care quality, particularly for individuals who switch plans close to death. Using the population-based Surveillance, Epidemiology, and End Results (SEER) Medicare data, we examined hospice enrollment, timing of enrollment, and place of hospice care among beneficiaries with distant-stage cancers by plan switching patterns during the last year of life.

Methods

Study Design and Sample

We conducted a retrospective cohort study of Medicare beneficiaries newly diagnosed with distant-stage female breast cancer, colorectal cancer, non–small cell lung cancer (NSCLC), small cell lung cancer (SCLC), pancreatic cancer, or prostate cancer from 2010 to 2019 who died by 2020 using the linked SEER Medicare data.¹⁹ SEER registries cover approximately 48% of the US population,^20,21 and provide information on patients’ cancer history, demographic characteristics, and neighborhood-level characteristics. Medicare administrative data, including Medicare Beneficiary Summary File and claim files, capture monthly enrollment information (eg, Medicare program type) and health care utilization (eg, hospice enrollment).

We used International Classification of Diseases for Oncology, Third Edition (ICD-O-3) codes and histology codes to identify patients with the aforementioned 6 types of cancer²² (eTable 1 in Supplement 1). We selected the first distant-stage diagnosis among individuals aged 66 years and older, with known diagnosis month and not diagnosed on autopsy or death certificate. Patients had to have 3 months or greater survival for sufficient time to measure plan switching. We defined the index date as 1 year prior to death or date of diagnosis for those who survived less than 1 year; patients had to have continuous parts A/B coverage with or without part C (ie, managed care) from the index date to death (eFigure 1 in Supplement 1). This study was deemed exempt under Common Rule (45 CFR 46, subpart A) by the Morehouse School of Medicine Institutional Review Board; informed consent was not required. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Measures

Medicare Plan Switching

We used Medicare monthly enrollment indicators for up to 1 year before death to classify beneficiaries’ plan switching patterns: (1) continuous TM, (2) continuous MA, (3) switching from MA to TM, (4) switching from TM to MA, and (5) other switching patterns (ie, multiple switches). Of note, for patients who survived less than 12 months (eg, 3-month survival), plan switching patterns were assessed within the window from the date of cancer diagnosis to month of death (ie, 3-month window). Because Medicare open enrollment occurs annually,^13,23 we conducted sensitivity analyses restricted to patients surviving 12 months or longer to allow plan switching pattern classification within a 12-month window. eFigure 2 in Supplement 1 illustrates plan switching patterns for patients with varying length of survival.

Hospice Utilization and Place of Last Hospice Care

Outcomes included hospice enrollment in the last year of life, late hospice enrollment (first hospice claim within 3 days of death among hospice users), total hospice length of stay, and site of last hospice service (home, nursing home, hospice facility, inpatient facility, or other [long-term care hospital; inpatient psychiatric facility; place not otherwise specified; hospice residential facility]). Hospice enrollment was extracted from the hospice fee-for-service claim file. We also evaluated late hospice enrollment and total hospice length of stay as potential quality indicators.^24,25 The site of hospice service was identified using place-specific Healthcare Common Procedure Coding System codes (Q5001-Q5010; eTable 2 in Supplement 1) among beneficiaries with any hospice enrollment.²⁶ Notably, as beneficiaries may transition to different sites of care during a hospice stay, we identified the Healthcare Common Procedure Coding System code shown in the last line of the last hospice claim.

Study Covariates

Patient baseline socioeconomic characteristics abstracted from medical records by cancer registrars at the time of diagnosis included age at diagnosis, sex, race and ethnicity (abstracted from medical records by cancer registrars at the time of diagnosis as Hispanic, non-Hispanic Black, non-Hispanic White, and other or unknown; the other group includes American Indian or Alaska Native, Asian or Other Pacific Islander, and other unspecified), and marital status. Census tract of residence at cancer diagnosis was used to identify metropolitan residency and quintiles of socioeconomic status (Yost Socioeconomic Index²⁷). Clinical characteristics of cancer type and diagnosis year were available from SEER; dual Medicaid-Medicare eligibility was classified into never dual, gain dual, lose dual, always dually enrolled, and other patterns using the monthly dual eligibility code from the Medicare Beneficiary Summary File.²⁸ We also measured county-level MA penetration rate as a proxy for market structure, which may confound plan enrollment and switching, as well as hospice enrollment.¹²

Statistical Analysis

Sample characteristics by Medicare plan switching patterns were compared using χ² tests. We described the distribution of hospice utilization and place of last hospice care by plan switching patterns, and conducted multivariable regression models to examine the differences in hospice utilization and place of hospice care by plan switching patterns, adjusted for all patient characteristics. Logistic models were used for any hospice utilization and late hospice enrollment, Poisson distribution was used for total hospice length of stay, and a multinomial model was used for place of last hospice care. Postestimation margins were used to estimate the adjusted differences in the probability of hospice use and a specific place of hospice care (eg, home vs all other places of hospice care) and adjusted differences in days for hospice length of stay.

We conducted sensitivity analyses restricted to patients surviving 12 or more months to address selection bias driven by prognosis and bias from underestimation of plan switching, given that the open enrollment period for plan switching occurs once per year.^13,23 We also stratified the analysis by dual Medicare and Medicaid enrollment, given programmatic differences in the flexibility of plan switching for those with and without Medicaid coverage and the potential that Medicaid coverage of nursing home room and board may be associated with place of hospice care.^29,30 Further, to address potential confounding of comorbidities, we examined a subgroup diagnosed in 2016 to 2019 for whom TM claims and MA encounter data were available to measure claim-based NCI comorbidity index in the year prior to diagnosis.³¹ Regressions with and without adjusting for comorbidity were conducted. Last, considering SNF utilization management and network restrictions under MA may prompt plan switching among beneficiaries with postacute care needs and affect choice of hospice site, we used a 2016-2019 subcohort during which SNF data files are available to describe switching patterns by SNF use after cancer diagnosis and estimated associations between plan switching and hospice outcomes with and without adjusting for SNF stay. Data analyses were performed using SAS version 9.4 (SAS Institute) and Stata software version 18.0 (StataCorp). Statistical significance was evaluated at the P < .05 threshold using 2-sided tests.

Results

Among a total of 196 536 patients included in the analysis (eFigure 1 in Supplement 1), 46.5% were female, 53.5% were male, and 49.2% were aged 66 to 74 years. There were 10 873 (5.5%) with female breast cancer, 29 162 (14.8%) with colorectal cancer, 95 557 (48.6%) with NSCLC, 21 357 (10.9%) with SCLC, 21 564 (11.0%) with pancreatic cancer, and 18 023 (9.2%) with prostate cancer. Medicare plan switching was relatively infrequent, with 1.5% (2861) switching from TM to MA and 1.8% (3534) switching from MA to TM. Beneficiaries who switched from MA to TM were more likely to be younger (66-74 years: 53.8% vs 49.0%), from racial or ethnic minority groups (non-Hispanic Black: 16.2% vs 9.0%), and reside in areas with lower socioeconomic status (ie, Yost quintile 1: 20.1% vs 16.2%) than beneficiaries with continuous TM; similar differences were seen for beneficiaries who switched from TM to MA compared with beneficiaries with continuous TM. In particular, there was a higher proportion of always dually enrolled beneficiaries among those who switched from TM to MA (18.3%) and from MA to TM (22.2%) than those with continuous TM (12.2%) or MA (12.5%) (Table 1).

Table 1. Sample Characteristics by Medicare Plan Switching Patterns.

Characteristic	Continuous		TM to MA (2861 [1.5%])	MA to TM (3534 [1.8%])	Other (623 [0.3%])	Total (196 536 [100.0%])	P value
Characteristic	TM (128 095 [65.2%])	MA (61 423 [31.3%])	TM to MA (2861 [1.5%])	MA to TM (3534 [1.8%])	Other (623 [0.3%])	Total (196 536 [100.0%])	P value
Cancer site, No. (%)
Breast	7176 (5.6)	3278 (5.3)	171 (6.0)	189 (5.3)	59 (9.5)	10873 (5.5)	<.001
Colorectal	19 126 (14.9)	8925 (14.5)	482 (16.8)	530 (15.0)	99 (15.9)	29 162 (14.8)
NSCL	62 219 (48.6)	30 110 (49.0)	1281 (44.8)	1687 (47.7)	260 (41.7)	95 557 (48.6)
Pancreas	14 152 (11.0)	6663 (10.8)	260 (9.1)	432 (12.2)	57 (9.1)	21 564 (11.0)
Prostate	11 387 (8.9)	5890 (9.6)	355 (12.4)	317 (9.0)	74 (11.9)	18 023 (9.2)
SCL	14 035 (11.0)	6557 (10.7)	312 (10.9)	379 (10.7)	74 (11.9)	21 357 (10.9)
Age group at diagnosis, No. (%)
66-74 y	62 730 (49.0)	29 951 (48.8)	1694 (59.2)	1900 (53.8)	370 (59.4)	96 645 (49.2)	<.001
75-84 y	48 965 (38.2)	24 513 (39.9)	921 (32.2)	1341 (37.9)	200 (32.1)	75 940 (38.6)
≥85 y	16 400 (12.8)	6959 (11.3)	246 (8.6)	293 (8.3)	53 (8.5)	23 951 (12.2)
Sex, No. (%)
Male	68 775 (53.7)	32 401 (52.8)	1674 (58.5)	1866 (52.8)	343 (55.1)	105 059 (53.5)	<.001
Female	59 320 (46.3)	29 022 (47.2)	1187 (41.5)	1668 (47.2)	280 (44.9)	91 477 (46.5)	<.001
Race and ethnicity, No. (%)^a
Hispanic	8175 (6.4)	6554 (10.7)	363 (12.7)	439 (12.4)	126 (20.2)	15 657 (8.0)	<.001
Non–Hispanic Black	11 476 (9.0)	8292 (13.5)	554 (19.4)	572 (16.2)	166 (26.6)	21 060 (10.7)
Non–Hispanic White	102 549 (80.1)	43 336 (70.6)	1761 (61.6)	2185 (61.8)	266 (42.7)	150 097 (76.4)
Other/unknown^b	5895 (4.6)	3241 (5.3)	183 (6.4)	338 (9.6)	65 (10.5)	9722 (5.0)
Metropolitan residency, No. (%)^c
Nonmetropolitan	19 508 (15.2)	5254 (8.6)	315 (11.0)	314 (8.9)	47 (7.5)	25 438 (12.9)	<.001
Metropolitan	108 587 (84.8)	56 169 (91.4)	2546 (89.0)	3220 (91.1)	576 (92.5)	171 098 (87.1)	<.001
Marital status, No. (%)
Married	42 033 (32.8)	20 171 (32.8)	916 (32.0)	828 (23.4)	149 (23.9)	64 097 (32.6)	<.001
Not married	35 470 (27.7)	16 633 (27.1)	837 (29.3)	862 (24.4)	169 (27.1)	53 971 (27.5)
Unknown	50 592 (39.5)	24 619 (40.1)	1108 (38.7)	1844 (52.2)	305 (49.0)	78 468 (39.9)
Yost quintile, No. (%)^d
First (lowest SES)	20 702 (16.2)	11 141 (18.1)	685 (23.9)	710 (20.1)	186 (29.9)	33 424 (17.0)	<.001
Second	21 967 (17.1)	10 332 (16.8)	506 (17.7)	613 (17.3)	119 (19.1)	33 537 (17.1)
Third	23 690 (18.5)	11 872 (19.3)	554 (19.4)	660 (18.7)	110 (17.7)	36 886 (18.8)
Fourth	27 353 (21.4)	13 645 (22.2)	559 (19.5)	730 (20.7)	94 (15.1)	42 381 (21.6)
Fifth (highest SES)	33 631 (26.3)	14 029 (22.8)	538 (18.8)	786 (22.2)	>103 (>16.5)	49 089 (25.0)
Unknown	752 (0.6)	404 (0.7)	19 (0.7)	35 (1.0)	<11 (<1.8)	1219 (0.6)
Dual Medicare-Medicaid eligibility, No. (%)^e
Never dual	101 767 (79.4)	49 090 (79.9)	1975 (69.0)	2090 (59.1)	263 (42.2)	155 185 (79.0)	<.001
Gain dual	2615 (2.0)	1551 (2.5)	122 (4.3)	259 (7.3)	44 (7.1)	4591 (2.3)
Lose dual	6785 (5.3)	2305 (3.8)	186 (6.5)	295 (8.3)	102 (16.4)	9673 (4.9)
Always dual	15 646 (12.2)	7689 (12.5)	523 (18.3)	785 (22.2)	191 (30.7)	24 834 (12.6)
Other	1282 (1.0)	788 (1.3)	55 (1.9)	105 (3.0)	23 (3.7)	2253 (1.1)

Open in a new tab

Abbreviations: MA, Medicare Advantage; NSCL, non–small cell lung; SCL, small cell lung; SES, socioeconomic status; TM, traditional Medicare.

^{^a}

Race and ethnicity information was abstracted from medical records by cancer registrars at the time of diagnosis.

^{^b}

Other group includes American Indian/Alaska Native, Asian/Pacific Islander, and other unspecified. The unknown category has been reported with other to avoid reporting small numbers in accordance with the National Cancer Institute data use agreement.

^{^c}

Based on 2013 Rural Urban Continuum Code. Codes 1-3 were classified as metropolitan and codes 4-9 were classified as nonmetropolitan.

^{^d}

The Yost index is a composite SES score for census tracts based on median household income, median house value, median rent, percentage below 150% of the federal poverty line, Education Index, percentage working class, and percentage unemployed. Lower quintiles represent the lower SES groups.

^{^e}

With state buy-in (Medicaid) from cancer diagnosis to the end of follow-up.

Hospice Utilization by Medicare Plan Switching Patterns

Hospice enrollment in the last year of life was highest among beneficiaries with continuous MA (74.8%), followed by those who switched from TM to MA (69.0%), those with continuous TM (68.5%), and those who switched from MA to TM (66.4%; P < .001) (Figure 1). Such difference persisted in multivariable logistic regression, where beneficiaries with continuous MA had 6.93 percentage points (pp; 95% CI, 6.50-7.37 pp) higher likelihood of hospice enrollment than those with continuous TM coverage. Similarly, beneficiaries who switched from TM to MA and from MA to TM had 2.92 pp (95% CI, 1.30-4.54 pp) and 2.77 pp (95% CI, 1.32-4.22 pp) higher likelihood of hospice enrollment than those with continuous TM. In addition, beneficiaries diagnosed with pancreatic cancers had considerably higher hospice enrollment than those with breast cancers (11.80 pp; 95% CI, 10.72-12.88 pp). Compared with non-Hispanic White beneficiaries, non-Hispanic Black and Hispanic beneficiaries had −8.56 pp (95% CI, –9.30 to −7.82 pp) and −2.62 pp (95% CI, –3.42 to −1.83) lower likelihood of hospice enrollment (Table 2).

Figure 1. — MA indicates Medicare Advantage; TM, traditional Medicare.

Table 2. Medicare Plan Switching Patterns and Patient Characteristics Associated With Hospice Enrollment^a.

Characteristic	Any hospice in the last year of life		Hospice in the last 3 d of life		Hospice length of stay
Characteristic	Adjusted percentage point difference (95% CI)	P value	Adjusted percentage point difference (95% CI)	P value	Adjusted differences in days (95% CI)	P value
Total No.	196 536		138 417		138 417
Medicare plan switching
Full TM	[Reference]	NA	[Reference]	NA	[Reference]	NA
TM to MA	2.92 (1.30 to 4.54)	<.001	0.04 (−1.60 to 1.69)	.96	0.28 (−0.02 to 0.58)	.06
MA to TM	2.77 (1.32 to 4.22)	<.001	0.44 (−1.10 to 1.99)	.57	−0.63 (−0.91 to −0.36)	<.001
Full MA	6.93 (6.50 to 7.37)	<.001	−1.39 (−1.83 to −0.96)	<.001	5.13 (5.05 to 5.22)	<.001
Other	2.65 (−0.73 to 6.03)	.13	−0.03 (−3.65 to 3.59)	.99	−3.41 (−4.02 to −2.80)	<.001
Cancer site
Breast	[Reference]	NA	[Reference]	NA	[Reference]	NA
Colorectal	6.53 (5.47 to 7.58)	<.001	−5.28 (−6.36 to −4.20)	<.001	4.01 (3.84 to 4.19)	<.001
NSCL	3.04 (2.06 to 4.02)	<.001	−2.82 (−3.83 to −1.80)	<.001	0.88 (0.72 to 1.04)	<.001
Pancreas	11.80 (10.72 to 12.88)	<.001	−6.32 (−7.42 to −5.21)	<.001	−7.08 (−7.26 to −6.90)	<.001
Prostate	0.38 (−0.83 to 1.59)	.54	−3.00 (−4.24 to −1.76)	<.001	1.78 (1.57 to 1.98)	<.001
SCL	5.97 (4.87 to 7.08)	<.001	−2.45 (−3.59 to −1.31)	<.001	−8.21 (−8.39 to −8.04)	<.001
Age group at diagnosis
66-74 y	[Reference]	NA	[Reference]	NA	[Reference]	NA
75-84 y	4.65 (4.23 to 5.08)	<.001	−2.92 (−3.35 to −2.49)	<.001	9.28 (9.21 to 9.36)	<.001
≥85 y	8.79 (8.18 to 9.40)	<.001	−6.11 (−6.68 to −5.53)	<.001	24.81 (24.68 to 24.94)	<.001
Sex
Male	[Reference]	NA	[Reference]	NA	[Reference]	NA
Female	6.00 (5.56 to 6.44)	<.001	−3.00 (−3.43 to −2.57)	<.001	6.38 (6.30 to 6.45)	<.001
Race and ethnicity^a
Hispanic	−2.62 (−3.42 to −1.83)	<.001	−0.41 (−1.20 to 0.38)	.31	−1.70 (−1.84 to −1.56)	<.001
Non–Hispanic-Black	−8.56 (−9.30 to −7.82)	<.001	−0.07 (−0.79 to 0.65)	.84	−1.71 (−1.84 to −1.58)	<.001
Non–Hispanic-White	[Reference]	NA	[Reference]	NA	[Reference]	NA
Other^b	−10.45 (−11.48 to −9.42)	<.001	0.84 (−0.25 to 1.92)	.13	−2.19 (−2.37 to −2.01)	<.001
Unknown	−3.43 (−10.41 to 3.55)	.34	−1.68 (−8.59 to 5.23)	.63	−0.62 (−1.87 to 0.62)	.33
Metropolitan residency^c
Nonmetropolitan	[Reference]	NA	[Reference]	NA	[Reference]	NA
Metropolitan	0.70 (−0.01 to 1.41)	.06	0.14 (−0.55 to 0.82)	.70	1.13 (1.01 to 1.25)	<.001
Marital status
Married	[Reference]	NA	[Reference]	NA	[Reference]	NA
Not married	0.70 (0.16 to 1.24)	.01	−1.52 (−2.05 to −0.99)	<.001	4.85 (4.75 to 4.94)	<.001
Unknown	−0.23 (−1.45 to 0.99)	.71	−1.38 (−2.59 to −0.16)	.03	5.26 (5.05 to 5.47)	<.001
Yost quintile^d
First (lowest SES)	[Reference]	NA	[Reference]	NA	[Reference]	NA
Second	0.34 (−0.37 to 1.04)	.35	0.32 (−0.37 to 1.00)	.36	−1.87 (−1.99 to −1.74)	<.001
Third	1.25 (0.54 to 1.96)	.001	0.47 (−0.22 to 1.16)	.18	−2.42 (−2.55 to −2.29)	<.001
Fourth	2.30 (1.58 to 3.01)	<.001	0.51 (−0.19 to 1.21)	.15	−2.54 (−2.67 to −2.41)	<.001
Fifth (highest SES)	1.83 (1.10 to 2.55)	<.001	1.17 (0.45 to 1.89)	.001	−4.42 (−4.55 to −4.29)	<.001
Unknown	0.37 (−2.10 to 2.85)	.77	0.97 (−1.68 to 3.61)	.48	0.87 (0.36 to 1.38)	.001
Dual Medicare-Medicaid eligibility^e
Never dual	[Reference]	NA	[Reference]	NA	[Reference]	NA
Gain dual	1.69 (0.42 to 2.96)	.01	−3.92 (−5.08 to −2.76)	<.001	14.36 (14.09 to 14.62)	<.001
Lose dual	−7.14 (−8.20 to −6.08)	<.001	0.30 (−0.84 to 1.43)	.61	7.65 (7.45 to 7.85)	<.001
Always dual	−3.09 (−3.75 to −2.44)	<.001	−2.03 (−2.66 to −1.40)	<.001	8.75 (8.62 to 8.88)	<.001
Other	−1.08 (−2.93 to 0.78)	.26	−2.14 (−3.96 to −0.32)	.02	11.73 (11.36 to 12.11)	<.001
MA penetration rate	4.15 (2.20 to 6.10)	<.001	2.42 (0.43 to 4.42)	.02	−12.33 (−12.69 to −11.98)	<.001

Open in a new tab

Abbreviations: NA, not applicable; other acronym expansions in Table 1 footnote.

^{^a}

Footnote a in Table 1 gives race and ethnicity information.

^{^b}

Footnote b in Table 1 gives other group information.

^{^c}

Footnote c in Table 1 gives Rural Urban Continuum Code information.

^{^d}

Footnote d in Table 1 gives Yost index information.

^{^e}

Footnote e in Table 1 gives eligibility information.

Among hospice users, enrollment within 3 days of death was lower among continuous MA enrollees (15.5%) compared with other groups: 17.1% with continuous TM, 18.0% among those who switched from TM to MA, and 17.6% among those who switched from MA to TM. In adjusted analysis, hospice enrollment within 3 days of death was lower among beneficiaries with continuous MA (−1.39 pp; 95% CI, –1.83 to −0.96 pp) compared with beneficiaries with continuous TM, but similar among those who switched from TM to MA (0.04 pp; 95% CI, –1.60 to 1.69 pp) and those who switched from MA to TM (0.44 pp; 95% CI, –1.10 to 1.99 pp).

Among hospice users, the average total hospice length of stay was longer among beneficiaries with continuous MA (48.3 days) than those with continuous TM (43.8 days), those who switched from TM to MA (42.7 days), and those who switched from MA to TM (42.1 days). Adjusted regression analysis showed persistent positive differences in hospice length of stay (5.13 days; 95% CI, 5.05-5.22 days) for beneficiaries with continuous MA than for those with continuous TM.

Place of Hospice Care by Medicare Plan Switching Patterns

The last hospice service was received at home by 70.4% of hospice enrollees, followed by 10.8% at hospice facilities, 9.6% at nursing homes, and 7.6% at inpatient facilities (Figure 2). Compared with continuous TM, continuous MA beneficiaries had 1.93 pp (95% CI, 1.40-2.45 pp) higher likelihood of receiving their last hospice service at home. In addition, beneficiaries who were older (≥85 years vs 66-74 years: 2.43 pp; 95% CI, 1.70-3.16 pp), female (1.88 pp; 95% CI, 1.37-2.39 pp), and Hispanic (vs non-Hispanic White: 4.89 pp; 95% CI, 3.96-5.81 pp) were more likely to receive home hospice, while non-Hispanic Black beneficiaries (−1.95 pp; 95% CI, –2.80 to −1.10 pp) and those with newly gained dual coverage (−31.98 pp; 95% CI, –33.69 to −30.27 pp) were less likely to receive home hospice (Table 3).

Table 3. Medicare Plan Switching Patterns and Patient Characteristics Associated With Place of Last Hospice Stay^a.

Characteristic	Home hospice		Nursing home hospice
Characteristic	Adjusted probability difference (95% CI)	P value	Adjusted probability difference (95% CI)	P value
Total No.	138 417	NA	138 417	NA
Medicare plan switching
Full TM	[Reference]	NA	[Reference]	NA
TM to MA	1.75 (−0.16 to 3.65)	.07	−0.77 (−1.93 to 0.38)	.19
MA to TM	−3.26 (−5.08 to −1.45)	<.001	2.45 (1.26 to 3.63)	<.001
Full MA	1.93 (1.40 to 2.45)	<.001	−1.83 (−2.16 to −1.51)	<.001
Other	1.93 (−2.14 to 6.01)	.35	0.21 (−2.15 to 2.58)	.86
Cancer site
Breast	[Reference]	NA	[Reference]	NA
Colorectal	3.97 (2.76 to 5.18)	<.001	−1.91 (−2.69 to −1.13)	<.001
NSCL	3.57 (2.46 to 4.69)	<.001	−2.60 (−3.32 to −1.88)	<.001
Pancreas	7.29 (6.04 to 8.54)	<.001	−4.98 (−5.78 to −4.19)	<.001
Prostate	0.92 (−0.51 to 2.34)	.21	0.27 (−0.68 to 1.22)	.58
SCL	5.62 (4.36 to 6.87)	<.001	−3.41 (−4.21 to −2.60)	<.001
Age group at diagnosis
66-74 y	[Reference]	NA	[Reference]	NA
75-84 y	0.99 (0.49 to 1.50)	<.001	1.60 (1.29 to 1.92)	<.001
≥85 y	2.43 (1.70 to 3.16)	<.001	3.63 (3.14 to 4.13)	<.001
Sex
Male	[Reference]	NA	[Reference]	NA
Female	1.88 (1.37 to 2.39)	<.001	−0.62 (−0.95 to −0.29)	<.001
Race and ethnicity^a
Hispanic	4.89 (3.96 to 5.81)	<.001	−4.55 (−4.98 to −4.13)	<.001
Non–Hispanic-Black	−1.95 (−2.80 to −1.10)	<.001	−0.73 (−1.23 to −0.23)	.004
Non–Hispanic-White	[Reference]	NA	[Reference]	NA
Other^b	0.03 (−1.32 to 1.37)	.97	−1.30 (−1.99 to −0.62)	<.001
Unknown	−3.44 (−13.06 to 6.18)	.48	−5.15 (−8.70 to −1.59)	.005
Metropolitan residency^c
Nonmetropolitan	[Reference]	NA	[Reference]	NA
Metropolitan	−3.72 (−4.49 to −2.95)	<.001	−0.25 (−0.76 to 0.25)	.33
Marital status
Married	[Reference]	NA	[Reference]	NA
Not married	−5.83 (−6.49 to −5.17)	<.001	4.08 (3.67 to 4.48)	<.001
Unknown	−1.48 (−2.93 to −0.02)	.05	3.11 (2.27 to 3.94)	<.001
Yost quintile^d
First (lowest SES)	[Reference]	NA	[Reference]	NA
Second	0.31 (−0.49 to 1.12)	.45	0.49 (−0.01 to 0.98)	.06
Third	−0.62 (−1.43 to 0.19)	.14	0.20 (−0.31 to 0.70)	.45
Fourth	−0.48 (−1.31 to 0.35)	.26	−0.03 (−0.55 to 0.48)	.90
Fifth (highest SES)	0.18 (−0.67 to 1.03)	.69	−0.77 (−1.29 to −0.24)	.004
Unknown	−5.63 (−8.81 to −2.45)	.001	1.36 (−0.56 to 3.29)	.16
Dual Medicare Medicaid eligibility^e
Never dual	[Reference]	NA	[Reference]	NA
Gain dual	−31.98 (−33.69 to −30.27)	<.001	33.76 (32.10 to 35.43)	<.001
Lose dual	−18.72 (−20.25 to −17.19)	<.001	18.60 (17.26 to 19.94)	<.001
Always dual	−15.55 (−16.40 to −14.70)	<.001	16.54 (15.81 to 17.27)	<.001
Other	−21.12 (−23.60 to −18.64)	<.001	23.23 (20.98 to 25.47)	<.001
MA penetration rate	−8.64 (−11.06 to −6.22)	<.001	0.59 (−0.94 to 2.12)	.45

Open in a new tab

Abbreviations: MA, Medicare Advantage; NA, not applicable; NSCL, non–small cell lung; SCL, small cell lung; Ref, reference; SES, socioeconomic status; TM, traditional Medicare.

^{^a}

Race and ethnicity information was abstracted from medical records by cancer registrars at the time of diagnosis.

^{^b}

Other group includes American Indian/Alaska Native, Asian/Pacific Islander, and other unspecified.

^{^c}

Based on 2013 Rural Urban Continuum Code. Codes 1-3 were classified as metropolitan and codes 4-9 were classified as nonmetropolitan.

^{^d}

The Yost index is a composite SES score for census tracts based on median household income, median house value, median rent, percentage below 150% of federal poverty line, Education Index, percentage working class, and percentage unemployed. Lower quintiles represent the lower SES groups.

^{^e}

With state buy-in (Medicaid) from cancer diagnosis to the end of follow-up.

In contrast, those who switched from MA to TM had 2.45 pp (95% CI, 1.26-3.63 pp) higher likelihood and those with continuous MA had −1.83 pp (95% CI, –2.16 to −1.51 pp) lower likelihood of receiving last hospice service at nursing homes than beneficiaries with continuous TM, after adjusting for patient characteristics. Racial and ethnic minority groups, including Hispanic (−4.55 pp; 95% CI, –4.98 to −4.13 pp), non-Hispanic Black (−0.73 pp; 95% CI, –1.23 to −0.23 pp), and non-Hispanic other (−1.30 pp; 95% CI, –1.99 to −0.62 pp) beneficiaries were less likely to receive hospice at nursing homes. Dual Medicare and Medicaid coverage had highest adjusted probability difference values with nursing home use, where those gaining dual coverage was associated with 33.76 pp (95% CI, 32.10-35.43 pp) higher likelihood of hospice use at a nursing home compared with those without dual coverage (Table 3).

Sensitivity Analyses

Sensitivity analyses among those who survived 12 months showed a lower percentage of beneficiaries with pancreatic cancer, NSCLC, and SCLC than the primary sample (eTable 3 in Supplement 1), but consistent statistical significance and direction of the adjusted probability differences between plan switching patterns and hospice utilization and place of hospice care (eTable 4 in Supplement 1). When stratified by any dual Medicare and Medicaid coverage, those with continuous MA coverage consistently had a higher likelihood of hospice utilization than those with continuous TM coverage, although the magnitude was higher among dual than nondual beneficiaries (9.15 pp [95% CI, 8.10-10.20 pp] and 6.32 pp [95% CI, 5.81-6.84 pp]) (eTables 5 and 6 in Supplement 1). Similar patterns were observed for hospice use at home and in a nursing home: the increased likelihood of home hospice among beneficiaries with continuous MA (vs continuous TM) was higher among dual (2.71 pp; 95% CI, 1.36-4.06 pp) than nondual (1.63 pp; 95% CI, 1.05-2.21 pp) beneficiaries; and the increased likelihood of hospice at nursing homes among MA-to-TM beneficiaries was higher among dual (6.01 pp; 95% CI, 2.80-9.21 pp) than nondual (1.27 pp; 95% CI, 0.16-2.37 pp) beneficiaries (eTables 5 and 6 in Supplement 1). Regression results with and without adjusting for NCI comorbidity index were similar and consistent with the primary analysis (eTable 7 in Supplement 1). Among the subcohort with SNF data available, we found a higher rate of MA-to-TM switching (3.0%) among those with an SNF stay after cancer diagnosis than those without (1.5%; eTable 8 in Supplement 1); and the adjusted probability differences in nursing home hospice use between MA to TM and continuous TM became smaller (2.38 pp; 95%- CI, 0.66-4.10 pp) after adjusting for SNF stay than without adjustments (3.54 pp; 95% CI, 1.62-5.45). (eTable 9 in Supplement 1).

Discussion

In this large, population-based cohort of Medicare beneficiaries diagnosed with distant-stage cancers, plan switching between MA and TM in the last year of life was relatively infrequent. However, we found distinct patterns of hospice utilization and place of last hospice service across plan switching patterns.

Our findings extend prior research showing that MA enrollees are more likely to use hospice than TM enrollees.^11,32,33 Importantly, we found longer hospice stays and lower rates of late hospice enrollment among continuous MA enrollees, suggesting that higher overall hospice use among MA beneficiaries was driven by earlier transition into hospice.¹² This aligns with the hypothesis that MA plans proactively encourage hospice enrollment, whether through financial incentives (eg, utilization management, provider networks) to avoid costly hospitalizations, care management approaches that facilitate EOL transitions, or better integration of supportive care services.^12,34 Our study cannot directly test these mechanisms; further investigation to identify which aspects of plan benefit design may support timely, preference-concordant EOL care across both MA and TM is warranted.

We also examined how transitions between programs are associated with the setting of EOL care. Continuous MA enrollees were more likely to receive hospice care at home, consistent with prior evidence that MA decedents more often enter hospice from community settings.¹² While patient preferences for place of EOL care are highly contingent on clinical and social context, home remains the most commonly preferred place of EOL care for patients and families, particularly when symptom burden is manageable and informal caregiving support is available.³⁵ In contrast, those who switched from MA to TM were more likely to receive hospice care in nursing homes. Although MA is not responsible for administering and managing hospice care, several MA features may explain this pattern. First, MA plans maintain narrower SNF networks and stricter prior authorization and length-of-stay for SNF stay limits than TM.³⁶ Patients with advanced cancer who require postacute or rehabilitative care facing access barriers under MA may switch to TM for broader SNF coverage. The SNF facility may become the care setting where patients ultimately receive their hospice services. Our sensitivity analyses confirmed this potential mechanism (ie, mediation): (1) the MA-to-TM switching rate was higher among those with an SNF stay after cancer diagnosis than those without and (2) the adjusted probability differences in nursing home hospice use between MA to TM and continuous TM became smaller after adjusting for SNF stay.

Another potential explanation for the higher use of nursing home hospice among those who switched from MA to TM relates to the integration of Medicare and Medicaid in financing of nursing home care. Although MA plans manage Medicare-covered services, custodial long-term nursing home care is covered by Medicaid.^37,38 Many dual-eligible beneficiaries in MA must still navigate 2 separate programs, unless they are in a fully integrated plan such as a Fully Integrated Dual-Eligible Special Needs Plan,²⁹ which could create administrative complexity, limited overlaps between MA networks and Medicaid-covered nursing homes, and fewer facility choices.^36,39,40 For patients with advanced cancer and substantial nursing home needs, these barriers may prompt switching to TM, which simplifies coordination with Medicaid-covered long-term care. Once in nursing homes under TM, transitions into hospice at the same facility may follow more directly. Consistent with this, prior research shows that dual Medicare-Medicaid enrollees are more likely to disenroll from MA than non-dual enrollees.⁴¹ We similarly found switching was more common among dual enrollees, a group characterized by both greater clinical complexity and higher likelihood of needing intensive nursing care.^29,42,43,44 Moreover, the associations between MA-to-TM switching and nursing home hospice use were larger among dually covered than non–dually covered beneficiaries. These findings suggest that structural barriers to long-term care access in MA, combined with Medicaid’s coverage of room and board for nursing home care, may drive plan switching and subsequent receipt of hospice in nursing homes.

These results highlight important policy implications related to structural constraints from benefit design and plan switching rules that may influence where hospice care is ultimately received. Although home remains the most preferred place of death, population aging, multimorbidity, and declines in family caregiving capacity are likely to increase demand for nursing home–based hospice,^45,46 particularly among individuals residing in long-term care facilities who may even view the nursing home as their “home” at the EOL.^35,47 In this context, higher nursing home hospice use among those who switched from MA to TM may reflect structural mismatches between care needs, benefit design, and beneficiaries’ capacity to navigate coverage options. For example, dual Medicare-Medicaid enrollees benefit from Special Enrollment Periods that allow plan changes once per quarter and, beginning in January 2025, will permit monthly switches.⁴⁸ In contrast, non–dually covered beneficiaries face a narrower annual enrollment period.¹³ Their choices are further constrained by financial barriers to obtaining supplemental coverage under TM. Because TM lacks an annual out-of-pocket maximum, many beneficiaries rely on Medigap policies to offset cost-sharing.⁴⁹ Yet, in most states, Medigap insurance providers may deny coverage or charge prohibitively high premiums based on preexisting conditions.^50,51 For patients newly diagnosed with cancer, these barriers can effectively lock them into MA plans even when they are facing narrow networks or other unmet needs.

Limitations

Our study has limitations. First, causal relationships between plan switching and hospice utilization or place of hospice cannot be inferred from this observational study. Instead, those who switch from MA to TM may represent distinct subgroups with underlying nursing care needs. Second, our outcome measures such as late hospice enrollment and total length of hospice stay may not capture all dimensions of hospice quality. We also could not capture several unmeasured confounders, such as patients’ or family members’ underlying preference for place of hospice care (eg, home, nursing home), which may be influenced by programmatic constraints, caregiver support, and other factors. Last, findings were restricted to older Medicare beneficiaries diagnosed with 6 common solid tumors within SEER regions and may not be generalizable to other patient populations.

Conclusions

This retrospective cohort study of Medicare decedents diagnosed with distant-stage cancers in 2010 to 2019 found that hospice utilization patterns at the EOL varied substantially by Medicare plan type and plan switching behavior. Continuous MA enrollment was associated with higher hospice use and more frequently home hospice, while MA-to-TM switching was associated with greater nursing home hospice use, particularly among dually covered beneficiaries. These findings suggest that insurance program structures and enrollment policies influence where and how patients receive hospice care at the EOL. Future research to understand potential care coordination gaps, access to patient-centered hospice care settings, and barriers to plan switching for medically vulnerable beneficiaries is critical to identify targeted intervention to promote equitable and patient-centered EOL care.

Supplement 1.

eFigure 1. CONSORT Chart for creation of patient cohort

eFigure 2. Examples of Medicare Plan Switching Patterns

eTable 1. Cancer Identification

eTable 2. Place Specific Healthcare Common Procedure Coding System for Hospice Services

eTable 3. Sample Characteristics of Medicare Beneficiaries Who Survived At Least 12 Months

eTable 4. Medicare Plan Switching and Patient Characteristics Associated with Hospice Enrollment and Place of Last Hospice Stay Among Beneficiaries Survived At Least 12 Months

eTable 5. Medicare Plan Switching and Patient Characteristics Associated with Hospice Enrollment and Place of Last Hospice Stay Among Beneficiaries With Dual Medicaid and Medicare Enrollment

eTable 6. Medicare Plan Switching and Patient Characteristics Associated with Hospice Enrollment and Place of Last Hospice Stay Among Beneficiaries Without Dual Medicaid and Medicare Enrollment

eTable 7. Association Between Medicare Plan Switching and Hospice Enrollment and Place of Last Hospice Stay With and Without Adjusting for NCI Comorbidity Index

eTable 8. Medicare Plan Switching Stratified by Skilled Nursing Facility Stay Post Cancer Diagnosis

eTable 9. Medicare Plan Switching and Patient Characteristics Associated with Nursing Home Hospice Enrollment With and Without Adjusting for Skilled Nursing Facility Stay Post Cancer Diagnosis

jamanetwopen-e260755-s001.pdf^{(614.3KB, pdf)}

Supplement 2.

Data Sharing Statement

jamanetwopen-e260755-s002.pdf^{(13.1KB, pdf)}

References

1.Currow DC, Agar MR, Phillips JL. Role of hospice care at the end of life for people with cancer. J Clin Oncol. 2020;38(9):937-943. doi: 10.1200/JCO.18.02235 [DOI] [PubMed] [Google Scholar]
2.Wagner V, Marks A. What is hospice? JAMA. 2024;332(14):1218. doi: 10.1001/jama.2024.13144 [DOI] [PubMed] [Google Scholar]
3.Mulville AK, Widick NN, Makani NS. Timely referral to hospice care for oncology patients: a retrospective review. Am J Hosp Palliat Care. 2019;36(6):466-471. doi: 10.1177/1049909118820494 [DOI] [PubMed] [Google Scholar]
4.Kelley AS, Deb P, Du Q, Aldridge Carlson MD, Morrison RS. Hospice enrollment saves money for Medicare and improves care quality across a number of different lengths-of-stay. Health Aff (Millwood). 2013;32(3):552-561. doi: 10.1377/hlthaff.2012.0851 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Lysaght S, Ersek M. Settings of care within hospice: new options and questions about dying “at home”. J Hosp Palliat Nurs. 2013;15(3):171-176. doi: 10.1097/NJH.0b013e3182765a17 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Committee on Approaching Death . Addressing key end of life issues, Institute of Medicine. Dying in America: improving quality and honoring individual preferences near the end of life. National Academies Press (US); 2015. Accessed August 6, 2025. https://www.ncbi.nlm.nih.gov/books/NBK285681/ [PubMed]
7.Cagle JG, Lee J, Ornstein KA, Guralnik JM. Hospice utilization in the United States: a prospective cohort study comparing cancer and noncancer deaths. J Am Geriatr Soc. 2020;68(4):783-793. doi: 10.1111/jgs.16294 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Park J, Look KA. Health care expenditure burden of cancer care in the United States. Inquiry. 2019;56:46958019880696. doi: 10.1177/0046958019880696 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Fine PG. Hospice underutilization in the U.S.: the misalignment of regulatory policy and clinical reality. J Pain Symptom Manage. 2018;56(5):808-815. doi: 10.1016/j.jpainsymman.2018.08.005 [DOI] [PubMed] [Google Scholar]
10.Freed M, Biniek JF, Damico A, Published TN. Medicare Advantage in 2024: enrollment update and key trends. KFF. August 8, 2024. Accessed May 4, 2025. https://www.kff.org/medicare/issue-brief/medicare-advantage-in-2024-enrollment-update-and-key-trends/
11.Nicholas LH, Fischer SM, Arbaje AI, Perraillon MC, Jones CD, Polsky D. Medicare-covered services near the end of life in Medicare Advantage vs traditional Medicare. JAMA Health Forum. 2024;5(7):e241777. doi: 10.1001/jamahealthforum.2024.1777 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Ankuda CK, Belanger E, Bunker J, et al. Comparison of the pathway to hospice enrollment between Medicare Advantage and traditional Medicare. JAMA Health Forum. 2023;4(2):e225457. doi: 10.1001/jamahealthforum.2022.5457 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Xu L, Welch WP, Sheingold S, De Lew N, Sommers BD. Medicare switching: patterns of enrollment growth in Medicare Advantage, 2006-22. Health Aff (Millwood). 2023;42(9):1203-1211. doi: 10.1377/hlthaff.2023.00224 [DOI] [PubMed] [Google Scholar]
14.Parsons HM, Greenwald SJ, Jarosek S, et al. Switching between Medicare Advantage and traditional Medicare for individuals newly diagnosed with cancer 2015-2019. J Natl Cancer Inst. 2025;117(6):1218-1227. doi: 10.1093/jnci/djaf036 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Meyers DJ, Belanger E, Joyce N, McHugh J, Rahman M, Mor V. Analysis of drivers of disenrollment and plan switching among Medicare Advantage beneficiaries. JAMA Intern Med. 2019;179(4):524-532. doi: 10.1001/jamainternmed.2018.7639 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Jacobson G, Rae M, Neuman T, Orgera K, Published CB. Medicare Advantage: how robust are plans’ physician networks? KFF. October 5, 2017. Accessed May 4, 2025. https://www.kff.org/medicare/report/medicare-advantage-how-robust-are-plans-physician-networks/
17.Zhao Y, Diggs K, Chen ZX, Hohmann N, Kwon WS, Westrick SC. Qualitative exploration of factors influencing the plan selection process by Medicare beneficiaries. J Manag Care Spec Pharm. 2021;27(3):339-353. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Ankuda CK, Morrison RS, Aldridge MD. The 2021 Medicare Advantage hospice carve-in. JAMA. 2021;326(23):2367-2368. doi: 10.1001/jama.2021.21161 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Hankey BF, Ries LA, Edwards BK. The Surveillance, Epidemiology, and End Results program: a national resource. Cancer Epidemiol Biomarkers Prev. 1999;8(12):1117-1121. [PubMed] [Google Scholar]
20.Enewold L, Parsons H, Zhao L, et al. Updated overview of the SEER-Medicare data: enhanced content and applications. J Natl Cancer Inst Monogr. 2020;2020(55):3-13. doi: 10.1093/jncimonographs/lgz029 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.National Cancer Institute. What is a cancer registry? SEER. Accessed September 24, 2025. https://seer.cancer.gov/registries/cancer_registry/index.html
22.National Cancer Institute. Site Recode ICD-O-3/WHO 2008 definition. . SEER data reporting tools. Accessed August 1, 2024. https://seer.cancer.gov/siterecode/icdo3_dwhoheme/index.html
23.Cottrill A, Biniek JF, Neuman T, Cubanski J, Freed M. What to know about the Medicare open enrollment period and Medicare coverage options. KFF. September 26, 2024. Accessed August 17, 2025. https://www.kff.org/medicare/issue-brief/what-to-know-about-the-medicare-open-enrollment-period-and-medicare-coverage-options/
24.O’Connor NR, Hu R, Harris PS, Ache K, Casarett DJ. Hospice admissions for cancer in the final days of life: independent predictors and implications for quality measures. J Clin Oncol. 2014;32(28):3184-3189. doi: 10.1200/JCO.2014.55.8817 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Wang SY, Hall J, Pollack CE, et al. Trends in end-of-life cancer care in the Medicare program. J Geriatr Oncol. 2016;7(2):116-125. doi: 10.1016/j.jgo.2015.11.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Jarosek SL, Shippee TP, Virnig BA. Place of death of individuals with terminal cancer: new insights from Medicare Hospice Place-of-Service codes. J Am Geriatr Soc. 2016;64(9):1815-1822. doi: 10.1111/jgs.14269 [DOI] [PubMed] [Google Scholar]
27.Yost K, Perkins C, Cohen R, Morris C, Wright W. Socioeconomic status and breast cancer incidence in California for different race/ethnic groups. Cancer Causes Control. 2001;12(8):703-711. [DOI] [PubMed] [Google Scholar]
28.ResDAC. Medicare-Medicaid Dual Eligibility Code—January. Accessed August 17, 2025. https://resdac.org/cms-data/variables/medicare-medicaid-dual-eligibility-code-january
29.Roberts ET, Johnston KJ, Figueroa JF. Integrating Medicare and Medicaid coverage for dual eligibles—recommendations for reform. JAMA. 2023;330(5):409-410. doi: 10.1001/jama.2023.8879 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.National Council on Aging. Does Medicaid pay for nursing home care? April 29, 2025. Accessed September 24, 2025. https://www.ncoa.org/article/does-medicaid-pay-for-nursing-homes-a-comprehensive-guide/
31.Klabunde CN, Legler JM, Warren JL, Baldwin LM, Schrag D. A refined comorbidity measurement algorithm for claims-based studies of breast, prostate, colorectal, and lung cancer patients. Ann Epidemiol. 2007;17(8):584-590. doi: 10.1016/j.annepidem.2007.03.011 [DOI] [PubMed] [Google Scholar]
32.Stevenson DG, Ayanian JZ, Zaslavsky AM, Newhouse JP, Landon BE. Service use at the end-of-life in Medicare Advantage versus traditional Medicare. Med Care. 2013;51(10):931-937. doi: 10.1097/MLR.0b013e3182a50278 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Park S, Teno JM, White L, Coe NB. Effects of Medicare Advantage on patterns of end-of-life care among Medicare decedents. Health Serv Res. 2022;57(4):863-871. doi: 10.1111/1475-6773.13953 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Gadbois EA, Tyler DA, Shield RR, et al. Medicare Advantage control of postacute costs: perspectives from stakeholders. Am J Manag Care. 2018;24(12):e386-e392. [PMC free article] [PubMed] [Google Scholar]
35.Pinto S, Lopes S, de Sousa AB, Delalibera M, Gomes B. Patient and family preferences about place of end-of-life care and death: an umbrella review. J Pain Symptom Manage. 2024;67(5):e439-e452. doi: 10.1016/j.jpainsymman.2024.01.014 [DOI] [PubMed] [Google Scholar]
36.Meyers DJ, Mor V, Rahman M. Medicare Advantage enrollees more likely to enter lower-quality nursing homes compared to fee-for-service enrollees. Health Aff (Millwood). 2018;37(1):78-85. doi: 10.1377/hlthaff.2017.0714 [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Centers for Medicare & Medicaid Services. Beneficiaries dually eligible for Medicare & Medicaid. Centers for Medicare & Medicaid Services, Medicare Learning Network; 2024. Accessed August 17, 2025. https://www.cms.gov/files/document/beneficiaries-dually-eligible-medicare-medicaid.pdf
38.Data Book: beneficiaries dually eligible for Medicare and Medicaid. Medicare Payment Advisory Commission and Medicaid and CHIP Payment and Access Commission; 2023. Accessed August 17, 2025. https://www.macpac.gov/wp-content/uploads/2023/02/Feb23_MedPAC_MACPAC_DualsDataBook-WEB-508.pdf
39.Peña MT, Mohamed M, Fuglesten Biniek J, Burns A, Cubanski J, Neuman T. The landscape of Medicare and Medicaid coverage arrangements for dual-eligible individuals across states. KFF. October 24, 2024. Accessed September 24, 2025. https://www.kff.org/medicare/the-landscape-of-medicare-and-medicaid-coverage-arrangements-for-dual-eligible-individuals-across-states/
40.Jung HY, Li Q, Rahman M, Mor V. Medicare Advantage enrollees’ use of nursing homes: trends and nursing home characteristics. Am J Manag Care. 2018;24(8):e249-e256. [PMC free article] [PubMed] [Google Scholar]
41.Unuigbe A, Cintina I, Koenig L. Beneficiary switching between traditional Medicare and Medicare Advantage between 2016 and 2020. JAMA Health Forum. 2022;3(12):e224896. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Chatterjee P, Liao JM, Wang E, Feffer D, Navathe AS. Characteristics, utilization, and concentration of outpatient care for dual-eligible Medicare beneficiaries. Am J Manag Care. 2022;28(10):e370-e377. doi: 10.37765/ajmc.2022.89189 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Maciejewski ML, Hammill BG. Measuring the burden of multimorbidity among Medicare beneficiaries via condition counts and cumulative duration. Health Serv Res. 2019;54(2):484-491. doi: 10.1111/1475-6773.13124 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Rahman M, Tyler D, Thomas KS, Grabowski DC, Mor V. Higher Medicare SNF care utilization by dual-eligible beneficiaries: can Medicaid long-term care policies be the answer? Health Serv Res. 2015;50(1):161-179. doi: 10.1111/1475-6773.12204 [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Favreault MM, Johnson RW. Projections of Risk of Needing Long-Term Services and Supports at Ages 65 and Older. US Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation; 2021. [PubMed] [Google Scholar]
46.Sørstrøm AK, Kymre IG, Ludvigsen MS. Nursing care to patients who have the home as the preferred place of death: a scoping review. BMC Health Serv Res. 2024;24(1):1302. doi: 10.1186/s12913-024-11757-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Sopcheck J, Tappen RM. Nursing home residents, family members, and staff members views about a good death and preferred place of death for a nursing home resident. Omega (Westport). 2025;302228251323422. Published online February 20, 2025. doi: 10.1177/00302228251323422 [DOI] [PubMed] [Google Scholar]
48.Cottrill A, Biniek JF, Neuman T, Cubanski J, Freed M. What to know about the Medicare open enrollment period and Medicare coverage options. KFF. September 26, 2024. Accessed August 17, 2025. https://www.kff.org/medicare/issue-brief/what-to-know-about-the-medicare-open-enrollment-period-and-medicare-coverage-options/
49.Roberts ET. The unintended cost of high cost sharing in Medicare-assessing consequences for patients and options for policy. JAMA Health Forum. 2021;2(12):e213624. doi: 10.1001/jamahealthforum.2021.3624 [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Kwon Y, Jazowski SA, Hu X, et al. Medigap protection and plan switching among Medicare Advantage enrollees with cancer. JAMA Health Forum. 2025;6(6):e252018. doi: 10.1001/jamahealthforum.2025.2018 [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Freed M, Ochieng N, Cubanski J, Neuman TN. Medigap may be elusive for Medicare Beneficiaries with pre-existing conditions. KFF. October 18, 2024. Accessed August 17, 2025. https://www.kff.org/medicare/issue-brief/medigap-may-be-elusive-for-medicare-beneficiaries-with-pre-existing-conditions/

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials