Palliative Care for Advanced Cancer in Medicare Advantage and Physician Networks

Xin Hu; Changchuan Jiang; Youngmin Kwon; Qinjin Fan; Kewei Sylvia Shi; Zhiyuan Zheng; Jingxuan Zhao; Craig Evan Pollack; Amanda L Blackford; Joan L Warren; Xuesong Han; K Robin Yabroff

doi:10.1001/jamanetworkopen.2026.36865

. 2026 Sep 30;9(9):e2636865. doi: 10.1001/jamanetworkopen.2026.36865

Palliative Care for Advanced Cancer in Medicare Advantage and Physician Networks

Xin Hu ^1,^2,^3,^✉, Changchuan Jiang ⁴, Youngmin Kwon ⁵, Qinjin Fan ⁶, Kewei Sylvia Shi ⁶, Zhiyuan Zheng ⁶, Jingxuan Zhao ⁶, Craig Evan Pollack ^7,⁸, Amanda L Blackford ⁹, Joan L Warren ⁶, Xuesong Han ⁶, K Robin Yabroff ⁶

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 Surveillance, Prevention, & Health Services Research, American Cancer Society, Atlanta, Georgia

⁷Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland

⁸School of Nursing, Johns Hopkins University, Baltimore, Maryland

⁹Division of Quantitative Sciences, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, Maryland

Accepted for Publication: August 7, 2026.

Published: September 30, 2026. doi:10.1001/jamanetworkopen.2026.36865

^✉

Corresponding Author: Xin Hu, PhD, Department of Radiation Oncology, School of Medicine, Emory University, 1365 Clifton Rd NE, C3005, Atlanta, GA 30322 (xin.hu@emory.edu).

Author Contributions: Dr 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, Shi, Zhao, Yabroff.

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

Drafting of the manuscript: Hu, Shi.

Critical review of the manuscript for important intellectual content: Hu, Jiang, Kwon, Fan, Zheng, Zhao, Pollack, Blackford, Warren, Han, Yabroff.

Statistical analysis: Hu, Shi.

Obtained funding: Hu.

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

Supervision: Blackford.

Conflict of Interest Disclosures: Dr Hu reported receiving grants from the National Institutes of Health (NIH), American Cancer Society, Donaghue Foundation, Pfizer, Johnson & Johnson, and St Jude Children’s Research Hospital outside the submitted work. Dr Jiang reported receiving grants from the American Cancer Society, Conquer Cancer Foundation, and Robert Winn Foundation and honoraria from Johnson & Johnson for serving on the advisory board outside the submitted work. Dr Kwon reported receiving grants from the National Cancer Institute (NCI) and Agency for Healthcare Research and Quality outside the submitted work. Dr Shi reported receiving a fellowship from PhRMA Foundation outside the submitted work. Dr Pollack reported having stock ownership in Gilead Pharmaceuticals outside the submitted work and being an unpaid board member of the nonprofit American Association of Service Coordinators. Ms Blackford reported receiving grants from the NCI during the conduct of the study. Dr Yabroff reported receiving grants from the NIH and Blood Cancer United outside the submitted work and honoraria from the American Society of Clinical Oncology. No other disclosures were reported.

Funding/Support: This research was partially supported by awards K12AR084234 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases, NIH (Dr Hu) and RSG-25-1512174-01-HOPS from the American Cancer Society (Dr Hu). 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 (CDC) National Program of Cancer Registries under cooperative agreement NU58DP007156; the National Cancer Institute 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.

Role of the Funder/Sponsor: The funders 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, NCI, or CDC or their contractors and subcontractors. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the American Cancer Society.

Meeting Presentation: This study was presented in part at a podium session during the 2026 annual meeting of AcademyHealth; May 30, 2026; Seattle, Washington, and at a poster session at the annual meeting of the American Society of Clinical Oncology; June 1, 2026; Chicago, Illinois.

Data Sharing Statement: See Supplement 2.

Additional Contributions: We thank the NCI; the Office of Research, Development and Information of the Center for Medicare & Medicaid Services; and Information Management Services, Inc and acknowledge the SEER Program tumor registries in the creation of the SEER-Medicare database.

^✉

Corresponding author.

PMCID: PMC13628318 PMID: 42814462

This cohort study compares receipt of palliative care, systemic therapy, and hospice enrollment between Medicare Advantage and traditional Medicare beneficiaries with advanced cancer and explores whether differences are associated with oncology network composition.

Key Points

Question

Do palliative care (PC), systemic therapy, and hospice utilization differ between Medicare Advantage (MA) and traditional Medicare (TM) beneficiaries with advanced cancer, and is physician network composition associated with these differences?

Findings

In this cohort study of 135 402 Medicare beneficiaries, MA enrollees had higher PC utilization than TM beneficiaries, but this difference was largely attenuated after matching patients to the same treating oncologists. In contrast, lower systemic therapy and higher hospice utilization among MA beneficiaries persisted after matching.

Meaning

The findings suggest that differences in PC between MA and TM beneficiaries are associated with physician network composition, highlighting the importance of monitoring network adequacy and access to PC.

Abstract

Importance

Palliative care (PC) is guideline recommended to improve symptoms and quality of life concurrent with cancer-directed therapy for patients with advanced cancer, yet utilization remains low among Medicare beneficiaries. While Medicare Advantage (MA) enrollment has surpassed traditional Medicare (TM), how PC and other health care utilization differs between MA and TM for beneficiaries with advanced cancer is unknown.

Objective

To compare PC, systemic therapy, and hospice utilization among patients with advanced cancer covered by MA vs TM and evaluate whether differences are associated with physician network composition.

Design, Setting, and Participants

This retrospective cohort study used linked Surveillance, Epidemiology, and End Results (SEER) and Medicare data for beneficiaries aged 66 years or older diagnosed with distant-stage breast, colorectal, lung, pancreatic, or prostate cancer from January 1, 2016, to December 31, 2021, who had continuous MA or TM enrollment and at least 2 months’ survival. Beneficiaries were followed up from diagnosis until the earlier of death or December 31, 2021. Each beneficiary was assigned a treating oncologist; MA beneficiaries within each plan were matched 1:1 to TM beneficiaries based on treating oncologists. Data were analyzed from November 18, 2025, to July 26, 2026.

Exposure

Medicare plan type.

Main Outcomes and Measures

Time to first PC, systemic therapy, and hospice utilization. Multivariable models estimated the associations between MA enrollment and outcomes before and after matching on treating oncologist, adjusting for sociodemographic and clinical characteristics.

Results

Among 135 402 eligible beneficiaries, 32.3% were enrolled in MA and 67.7% in TM; mean (SD) age was 76.2 (6.7) years, and 53.3% were male. Medicare Advantage beneficiaries had higher 6-month cumulative incidence of PC compared with TM beneficiaries (13.3% vs 9.3%; adjusted hazard ratio [AHR], 1.39 [95% CI, 1.35-1.44]; P < .001), particularly those in a health maintenance organization (HMO) plan (AHR, 1.66 [95% CI, 1.60-1.73]; P < .001). After matching, there were no overall differences between MA and TM in PC receipt (AHR, 1.05 [95% CI, 0.98-1.12]; P = .19), and the AHR was attenuated for HMO plans vs TM (AHR, 1.16 [95% CI, 1.08-1.24]; P < .001). Medicare Advantage beneficiaries had lower systemic therapy receipt (adjusted probability difference, −5.38 percentage points [pp]; 95% CI, −6.02 to −4.74 pp; P < .001) and higher hospice enrollment (3.42 pp; 95% CI, 2.82-4.03 pp; P < .001) than TM beneficiaries; differences persisted after matching (systemic therapy: −5.70 pp [95% CI, −6.89 to −4.52 pp]; P < .001; hospice enrollment: 1.68 pp [95% CI, 0.56-2.80 pp]; P = .003).

Conclusions and Relevance

In this retrospective cohort study of Medicare beneficiaries with advanced cancer, MA enrollment was associated with increased PC utilization, primarily attributable to differential oncologist network composition. Persistent differences in systemic therapy use and hospice enrollment after matching by treating oncologists suggest that MA plan–level features beyond physician networks and patient characteristics may be associated with these outcomes.

Introduction

Palliative care (PC) is associated with improved quality of life through symptom management, communication about goals and preferences, and psychosocial support for individuals with serious illness, including advanced cancers.¹ Clinical guidelines from the American Society of Clinical Oncology and the National Comprehensive Cancer Network (NCCN) recommend early integration of specialty outpatient PC concurrent with cancer-directed therapy—primarily systemic therapy for patients with advanced cancers.^2,3,4,5 However, early PC remains underused: among traditional Medicare beneficiaries with common distant-stage cancers, utilization reached only about 10% by 2019,⁶ with persistent disparities across sociodemographic groups.^7,8,9,10,11

Medicare covers nearly all adults aged 65 years or older and is the largest payer for cancer care in the US, largely because cancer incidence increases with age. Historically, most beneficiaries received coverage through traditional Medicare (TM), but enrollment in the privately operated alternative, Medicare Advantage (MA), increased substantially in the past decade and has now surpassed half of Medicare beneficiaries.¹² Yet, evidence on PC utilization among beneficiaries with advanced cancer has focused on TM,^6,8,9 and little is known about whether PC utilization differs in MA.

Medicare Advantage plans, paid by a risk-adjusted capitated amount per member, are designed to reduce costs through managed care principles, including utilization management and contracted physician or health care organizations, which can shape access to and patterns of cancer care.¹³ Moreover, hospice remains a carve-out benefit (ie, hospice services are covered by TM even for MA enrollees), giving MA plans financial incentives to facilitate early hospice enrollment.¹⁴ Studies found that MA enrollment was associated with lower likelihood of receiving systemic therapy, higher hospice enrollment, and worse survival than TM.^{14,15,16,17,18,19,20} However, to our knowledge, PC has not been evaluated as a distinct component of cancer care delivery in studies of MA vs TM, and how plan-level features contribute to differences in PC remains unknown.

Several mechanisms may drive the differences in PC between MA and TM. Because specialty PC is billed through evaluation and management codes²¹ that typically have lower cost than systemic therapies, it may be less influenced by prior authorization. Palliative care access may instead be especially sensitive to network composition, as referrals are commonly initiated by treating oncologists⁷ and timely access depends on whether PC specialists are available within the same health system.⁶ Narrow physician networks could therefore limit PC access if plans exclude oncology systems that house established specialty PC programs, and prior work has shown that many MA networks exclude National Cancer Institute (NCI)–designated comprehensive cancer centers, which typically have integrated PC programs.²²

Network breadth is not the only structural feature that may shape PC access. Even given similar network size, integrated delivery models, which are more common among some health maintenance organization (HMO) plans, may facilitate PC through centralized care management and coordinated referral processes. The association between MA enrollment and PC use may therefore be heterogenous, with higher utilization concentrated in more integrated plans. These differences may be particularly consequential for patients with advanced-stage cancers, as PC is closely intertwined with systemic therapy decisions and hospice transitions.^23,24

Accordingly, this study addressed 2 questions. First, do PC utilization, systemic therapy receipt, and hospice enrollment differ between MA and TM beneficiaries with distant-stage cancers, and do these differences vary across MA plan types? Second, to what extent are observed differences explained by oncologist networks? We hypothesized that higher PC utilization among MA beneficiaries would be concentrated among more integrated plan types and differences in PC would be substantially explained by oncologist network composition. In contrast, we anticipated lower systemic therapy and higher hospice enrollment among MA beneficiaries and that these differences would be less attributable to oncologist networks.

Methods

Study Design and Sample

In this retrospective cohort study, we used 2024 Surveillance, Epidemiology, and End Results (SEER)–Medicare linked data and included Medicare beneficiaries newly diagnosed with cancer from January 1, 2016, to December 31, 2021; fee-for-service claims through December 31, 2022; and MA encounter data through December 31, 2021.²⁵ SEER data include tumor characteristics, patient demographics, and neighborhood-level measures from regions covering approximately 35% of the US population.²⁶ Medicare fee-for-services claims and MA encounter data capture health care utilization for TM and MA beneficiaries, respectively. We also used the Medicare Data on Provider Practice and Specialty (MD-PPAS)²⁷ to obtain physician information. This study was approved by the Emory University institutional review board under expedited review, with an informed consent waiver for secondary data analysis. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

We identified patients with de novo distant-stage breast cancer, non–small cell lung cancer (NSCLC), small cell lung cancer (SCLC), colorectal cancer, prostate cancer, or pancreas cancer from January 1, 2016, to December 31, 2021, based on International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) and histology codes (eTable 1 in Supplement 1).²⁸ For patients with multiple distant-stage diagnoses, we selected the first qualifying diagnosis. Inclusion criteria were diagnosis at or later than age 66 years, at least 2 months’ survival after diagnosis, and continuous enrollment in Medicare Parts A and B from 12 months before cancer diagnosis through death or end of follow-up (December 31, 2021). Beneficiaries who switched between TM and MA during follow-up or were diagnosed at autopsy or on the death certificate were excluded (eFigure 1 in Supplement 1). We also defined 3 subcohorts: decedents who died by December 31, 2021, to evaluate hospice enrollment; beneficiaries with additional Medicare Part D coverage to evaluate systemic therapy receipt (2 outcomes intertwined with PC and known to differ between TM and MA^{14,15,16,17,18,19}); and beneficiaries surviving 6 months or more to address prognostic selection bias.

To assess the extent to which oncologist networks contributed to differences in PC and related outcomes between MA and TM, we used a shared-physician design. Prior studies have implemented such designs by matching patients to common physicians and incorporating physician fixed effects.^29,30 Following our team’s previous approach,^6,31 our primary analysis implemented physician-based matching. We first assigned each beneficiary a treating oncologist, defined as the National Provider Identifier (NPI) that appeared most frequently on outpatient or carrier claims with cancer diagnosis codes from 90 days before through 180 days after diagnosis, death, or end of follow-up. Because a nontrivial share of MA claims lacked individual NPIs, we used the MD-PPAS file to recover individual NPIs and reduce missingness by approximately 10% (eTable 2 and eMethods 1 in Supplement 1). For each MA plan, we performed 1:1 exact matching of MA to TM beneficiaries by treating oncologist using a greedy algorithm, which selects the control unit that best matches the treated unit without replacement; thus, a TM beneficiary could be matched to more than 1 MA plan because we conducted matching separately for each MA plan. We did not match additional patient characteristics because it would substantially reduce successful matches for MA contracts with a small number of beneficiaries. We therefore adjusted for patient covariates (age, sex, race and ethnicity, metropolitan residence, marital status, dual eligibility, Yost Index quintile, cancer site, year of diagnosis, and NCI Comorbidity Index) in regressions using both the unmatched and matched samples. Matching was conducted separately for the primary cohort and each subcohort.

Measures

Time to First Palliative Care Visit

Our primary outcome was time to first outpatient PC visit, identified by claims with corresponding diagnosis codes or physician specialty codes^8,32 through 6 months after cancer diagnosis or until hospice admission or death, whichever occurred first. A recent validation study demonstrated that PC diagnosis codes with specialty codes have 78.1% sensitivity and nearly 100% specificity for identifying outpatient PC.³³

Systemic Therapy Use and Hospice Enrollment Among Decedents

Systemic therapy receipt was defined as at least 1 claim containing a qualifying antineoplastic drug or infusion procedure code from diagnosis to death: International Classification of Diseases, Ninth Revision or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision procedure code, Healthcare Common Procedure Coding System code, or a generic or brand name oral agent (eTable 3 in Supplement 1).⁶ The eligible drug list was reviewed by oncology clinician members of the study team (C.J.) and restricted to therapies with US Food and Drug Administration–approved cancer indications and NCCN guideline recommendations. Hospice enrollment was identified by any hospice claim.

Study Covariates

Patient characteristics included age, sex, race and ethnicity, metropolitan residency, census-tract socioeconomic status (Yost Index³⁴), dual Medicaid-Medicare eligibility, cancer type, diagnosis year, and NCI Comorbidity Index score³⁵ measured 12 months before diagnosis. Race and ethnicity were included because they capture structural and social factors that are associated with both differential enrollment into Medicare Advantage and unequal access to palliative care; they were abstracted from medical records by cancer registrars at the time of diagnosis as Hispanic, non-Hispanic Black, non-Hispanic White, non-Hispanic other (American Indian or Alaska Native, Asian or Other Pacific Islander, and other unspecified race), or unknown. To mitigate biases in comorbidity burden between MA and TM due to coding intensity,^36,37 we excluded MA claims flagged as medical record review without procedure codes when constructing comorbidity measures for MA beneficiaries. We also measured MA plan types, including HMOs, local preferred provider organizations (PPOs), regional PPOs, point of service (POS), and others.

Statistical Analysis

We compared sample characteristics between MA and TM in the full and matched cohorts and assessed differences between matched and unmatched MA beneficiaries. We estimated the cumulative incidence of first PC encounter by MA vs TM status in both the unmatched cohort and the matched cohort. We then conducted multivariable regression analyses using the unmatched and matched samples; each set of the model included a binary indicator for MA (vs TM) and patient sociodemographic and clinical characteristics to account for potential patient selection into MA plans. We intentionally did not use instrumental variable analysis designed to address unobserved confounding; rather, we tested a specific, observable mechanism (ie, oncologist network as captured through treating oncologist matching) hypothesized to be associated with differences between MA and TM. Because TM beneficiaries could be matched to multiple MA beneficiaries, we clustered SEs at the beneficiary level in the matched cohort analysis.

Cox proportional hazards regression models accounting for death as a competing risk were used for time to first PC visit. Logistic regression with marginal effects was used for systemic therapy receipt and hospice enrollment, with estimates interpreted as adjusted probability differences.³⁸ In addition to the comparison of MA and TM overall, we examined more granular MA plan types given that certain plan types, such as HMOs, have more selective contracting and limited network breadth.³⁹

We performed several sensitivity analyses. First, we repeated analyses among beneficiaries with 6 or more months of survival and those who were diagnosed and died outside the COVID-19 disruption period (March 1 to December 31, 2020). Second, we stratified the analyses by cancer sites, considering different prognoses that may affect study outcomes. Third, because earlier MA encounter data completeness varied across MA contracts^40,41,42 and our own investigation identified nonnegligible and nonrandom plan-level missingness of individual NPIs (eFigure 2 in Supplement 1), we repeated analyses restricting to MA contracts with high data completeness and no contracts with 100% missing individual NPIs, for which we expected more complete capture of PC visits. Fourth, because MA beneficiaries still showed higher mean comorbidity indices than TM beneficiaries after excluding reviewed medical records, we repeated regression analyses without the NCI Comorbidity Index. Fifth, to address concerns about sample representativeness after matching, we implemented an alternative shared-physician design that compared differences between MA and TM in outcomes with and without controlling for oncologist fixed effects (eMethods 2 in Supplement 1).

Data analyses were performed from November 18, 2025, to July 26, 2026, using SAS, version 9.4 (SAS Institute Inc) and Stata, version 18.0 (StataCorp LLC). Statistical significance was evaluated at a P < .05 threshold using 2-sided tests.

Results

Among 135 402 Medicare beneficiaries with advanced cancer, 43 774 (32.3%) were enrolled in MA and 91 628 (67.7%) in TM. Mean (SD) age was 76.2 (6.7) years; 46.7% were female, and 53.3% were male. A total of 9.5% were Black; 7.3%, Hispanic; 78.2%, White; 4.8%, non-Hispanic other race; and 0.2%, unknown race and ethnicity. Medicare Advantage beneficiaries vs TM beneficiaries were more likely to be racial and ethnic minority individuals (eg, Black: 13.2% vs 7.7%; Hispanic: 11.0% vs 5.6%; P < .001 for both) and to reside in metropolitan areas (91.9% vs 83.7%; P < .001) and areas with lower socioeconomic status (Yost Index quintile 1, indicating lowest socioeconomic status: 14.9% vs 12.4%; P < .001) (Table). After matching by treating oncologists, 23 033 MA observations and 23 033 TM observations among 17 606 unique TM beneficiaries were included. Matched MA patients were less likely than unmatched MA patients to enroll in HMO plans (49.1% vs 76.7%; P < .001), with most unmatched beneficiaries enrolled in the Kaiser Foundation Health Plan (7684 of 7969 [96.4%]) (eTable 4 in Supplement 1).

Table. Sample Characteristics by Medicare Advantage and Traditional Medicare Before and After 1:1 Exact Matching by Treating Oncologist.

Characteristic	Beneficiaries before matching, No. (%)				Observations after matching
Characteristic	Total (N = 135 402)	MA (n = 43 774 [32.3%])	TM (n = 91 628 [67.7%])	P value	MA (n = 23 033 [50.0%])	TM (n = 23 033 [50.0%])	P value
Year of diagnosis
2016	21 931 (16.2)	5748 (13.1)	16 183 (17.7)	<.001	2936 (12.8)	2936 (12.8)	>.99
2017	22 243 (16.4)	6146 (14.0)	16 097 (17.6)		3090 (13.4)	3090 (13.4)
2018	23 006 (17.0)	6864 (15.7)	16 142 (17.6)		3623 (15.7)	3623 (15.7)
2019	24 377 (18.0)	8182 (18.7)	16 195 (17.7)		4336 (18.8)	4336 (18.8)
2020	22 691 (16.8)	8225 (18.8)	14 466 (15.8)		4493 (19.5)	4493 (19.5)
2021	21 154 (15.6)	8609 (19.7)	12 545 (13.7)		4555 (19.8)	4555 (19.8)
Age at diagnosis, mean (SD), y	76.2 (6.7)	76.3 (6.6)	76.2 (6.7)	<.001	75.7 (6.4)	75.6 (6.5)	<.001
Age group at diagnosis, y
66-74	61 390 (45.3)	19 641 (44.9)	41 749 (45.6)	<.001	10 977 (47.7)	11 240 (48.8)	.03
75-84	55 028 (40.6)	18 190 (41.6)	36 838 (40.2)		9414 (40.9)	9146 (39.7)
≥85	18 984 (14.0)	5943 (13.6)	13 041 (14.2)		2642 (11.5)	2647 (11.5)
Sex
Female	63 220 (46.7)	20 378 (46.6)	42 842 (46.8)	.48	11 089 (48.1)	10 969 (47.6)	.26
Male	72 182 (53.3)	23 396 (53.4)	48 786 (53.2)	.48	11 944 (51.9)	12 064 (52.4)	.26
Race and ethnicity
Hispanic	9936 (7.3)	4820 (11.0)	5116 (5.6)	<.001	1995 (8.7)	1264 (5.5)	<.001
Non-Hispanic Black	12 814 (9.5)	5787 (13.2)	7027 (7.7)		2858 (12.4)	1954 (8.5)
Non-Hispanic White	105 950 (78.2)	30 491 (69.7)	75 459 (82.4)		17 183 (74.6)	18 781 (81.5)
Non-Hispanic other^a	6452 (4.8)	2599 (5.9)	3853 (4.2)		965 (4.2)	983 (4.3)
Unknown	250 (0.2)	77 (0.2)	173 (0.2)		32 (0.1)	51 (0.2)
Metropolitan residence status^b
Nonmetropolitan	18 491 (13.7)	3535 (8.1)	14 956 (16.3)	<.001	2489 (10.8)	3236 (14.1)	<.001
Metropolitan	11 6911 (86.3)	40 239 (91.9)	76 672 (83.7)	<.001	20 544 (89.2)	19 797 (86.0)	<.001
Dual eligibility^c
No	118 443 (87.5)	37 940 (86.7)	80 503 (87.9)	<.001	20 026 (86.9)	20 321 (88.2)	<.001
Yes	16 959 (12.5)	5834 (13.3)	11 125 (12.1)	<.001	3007 (13.1)	2712 (11.8)	<.001
Marital status
Married	55 218 (40.8)	18 786 (42.9)	36 432 (39.8)	<.001	9385 (40.8)	9528 (41.4)	.40
Not married	46 648 (34.5)	15 806 (36.1)	30 842 (33.7)		7917 (34.4)	7834 (34.0)
Unknown	33 536 (24.8)	9182 (21.0)	24 354 (26.6)		5731 (24.9)	5671 (24.6)
Yost Index quintile^d
1 (Lowest SES)	17 899 (13.2)	6543 (14.9)	11 356 (12.4)	<.001	3502 (15.2)	3004 (13.0)	<.001
2	20 134 (14.9)	6706 (15.3)	13 428 (14.7)		3660 (15.9)	3510 (15.2)
3	23 091 (17.1)	7837 (17.9)	15 254 (16.6)		4112 (17.9)	4194 (18.2)
4	27 019 (20.0)	9325 (21.3)	17 694 (19.3)		4794 (20.8)	4626 (20.1)
5 (Highest SES)	33 038 (24.4)	10 076 (23.0)	22 962 (25.1)		4655 (20.2)	5434 (23.6)
Unknown	14 221 (10.5)	3287 (7.5)	10 934 (11.9)		2310 (10.0)	2265 (9.8)
Cancer site
Breast	9237 (6.8)	2815 (6.4)	6422 (7.0)	<.001	1391 (6.0)	1674 (7.3)	<.001
Colorectal	18 861 (13.9)	5908 (13.5)	12 953 (14.1)		3063 (13.3)	3327 (14.4)
NSCLC	59 976 (44.3)	19 567 (44.7)	40 409 (44.1)		10 708 (46.5)	10 243 (44.5)
Pancreas	16 837 (12.4)	5509 (12.6)	11 328 (12.4)		3028 (13.2)	2889 (12.5)
Prostate	17 520 (12.9)	5852 (13.4)	11 668 (12.7)		2255 (9.8)	2416 (10.5)
SCLC	12 971 (9.6)	4123 (9.4)	8848 (9.7)		2588 (11.2)	2484 (10.8)
NCI Comorbidity Index, mean (SD)^e	0.54 (0.60)	0.58 (0.60)	0.53 (0.60)	<.001	0.56 (0.59)	0.58 (0.58)	<.001
MA plan type
HMO	27 217 (20.1)	27 217 (62.2)	NA	NA	11 315 (49.1)	NA	NA
Local PPO	11 705 (8.6)	11 705 (26.7)	NA		8490 (36.9)	NA
Regional PPO	1989 (1.5)	1989 (4.5)	NA		1455 (6.3)	NA
POS	1623 (1.2)	1623 (3.7)	NA		1004 (4.4)	NA
Other or unknown	1240 (0.9)	1240 (2.8)	NA		769 (3.3)	NA

Open in a new tab

Abbreviations: HMO, health maintenance organization; MA, Medicare Advantage; NA, not applicable; NCI, National Cancer Institute; NSCLC, non–small cell lung cancer; POS, point of service; PPO, preferred provider organization; SCLC, small cell lung cancer; SES, socioeconomic status; TM, traditional Medicare.

^{^a}

Other includes American Indian or Alaska Native, Asian or Other Pacific Islander, and other unspecified race.

^{^b}

Based on 2013 Rural-Urban Commuting Area Codes.

^{^c}

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

^{^d}

The Yost Index is a composite SES score for census tracts based on median household income, median house value, median rent, percentage of residents below 150% of the federal poverty level, educational level, percentage of residents who are working class, and percentage of residents who are unemployed.

^{^e}

Total score range, 0 to 5.4, with higher scores indicating higher comorbidity burden.

Palliative Care

Prior to matching, the 6-month cumulative incidence of PC was 13.3% and 9.3% among MA and TM beneficiaries, respectively (Figure 1A). In adjusted analyses, MA enrollment was associated with a higher rate of PC receipt (adjusted hazard ratio [AHR], 1.39; 95% CI, 1.35-1.44; P < .001). Across MA plan types, the difference between MA and TM was most pronounced in HMOs (AHR, 1.66; 95% CI, 1.60-1.73; P < .001), but there were no differences for local PPO, regional PPO, POS, or other plan types (Figure 2 and eTables 5 and 6 in Supplement 1).

Two-panel cumulative incidence plots over 0 to 6 months for M A and T M. Two stacked panels labeled A and B, each a line chart with vertical axis labeled Cumulative incidence and horizontal axis labeled Time from diagnosis, mo. In both panels, the horizontal axis runs from 0 to 6, and the vertical axis has tick marks at 0, 0 point 05, 0 point 10, and 0 point 15, with light horizontal gridlines. Panel A title at upper left reads Overall sample. Two solid lines begin near time 0 and rise over time: an orange line labeled M A near the upper right end, and a dark teal line labeled T M near the mid right end. At 6 months, the orange M A line ends near 0 point 13 to 0 point 14, while the dark teal T M line ends near 0 point 09 to 0 point 10. A numbers-at-risk table beneath Panel A lists two rows labeled M A and T M aligned under time points 0 through 6. Values for M A: 43,774; 43,178; 41,327; 33,496; 28,641; 25,197; 22,475. Values for T M: 91,628; 90,699; 88,256; 74,591; 65,707; 59,262; 54,021. Panel B title at upper left reads Matched and unmatched cohorts. Four lines appear: two orange lines for M A and two dark teal lines for T M. Solid lines correspond to matched cohorts, labeled near the right as M A, matched and T M, matched; dashed lines correspond to not matched cohorts, labeled near the right as M A, not matched and T M, not matched. By 6 months, dashed orange M A, not matched ends near 0 point 15 to 0 point 16; solid orange M A, matched ends near 0 point 11; solid dark teal T M, matched ends near 0 point 10 to 0 point 11; dashed dark teal T M, not matched ends near 0 point 09. A numbers-at-risk table beneath Panel B lists four rows aligned under time points 0 through 6: M A, matched 23,033; 22,786; 22,057; 18,342; 15,864; 14,011; 12,517. M A, not matched 20,741; 20,392; 19,270; 15,154; 12,777; 11,186; 9,958. T M, matched 23,033; 22,791; 22,089; 19,233; 17,038; 15,351; 13,993. T M, not matched 74,022; 73,264; 71,310; 59,811; 52,587; 47,412; 43,206. — MA indicates Medicare Advantage; TM, traditional Medicare.

Dot and whisker plot of A H R for palliative care receipt by Medicare plan type. Left side table with two columns labeled Variable and A H R, with 95 percent C I. Two grouped sections appear: Medicare program and Medicare plan type. Under Medicare program, T M has two rows, Matched and Not matched, each listed as 1 Reference. M A has Matched 1.05 with 95 percent C I 0.98 to 1.12, and Not matched 1.39 with 95 percent C I 1.35 to 1.44. Under Medicare plan type, T M again has Matched 1 Reference and Not matched 1 Reference. H M O has Matched 1.16 with 95 percent C I 1.08 to 1.24, and Not matched 1.66 with 95 percent C I 1.60 to 1.73. Local P P O has Matched 0.90 with 95 percent C I 0.83 to 0.98, and Not matched 0.95 with 95 percent C I 0.89 to 1.01. Regional P P O P has Matched 0.97 with 95 percent C I 0.82 to 1.15, and Not matched 1.02 with 95 percent C I 0.88 to 1.18. P O S has Matched 1.19 with 95 percent C I 1.00 to 1.43, and Not matched 1.15 with 95 percent C I 0.99 to 1.33. Other or unknown has Matched 0.98 with 95 percent C I 0.77 to 1.25, and Not matched 0.94 with 95 percent C I 0.77 to 1.15. Right side forest plot aligns with the table rows. A vertical dotted reference line at A H R equals 1.00 divides headings at top: Less likely to receive P C on the left and More likely to receive P C on the right. Each row has a square marker with a horizontal whisker for the 95 percent C I; matched markers are dark teal squares and not matched markers are orange squares. Horizontal axis label at bottom reads A H R, with 95 percent C I, with tick labels 0.75, 1.00, 1.25, 1.50, and 2.00. — Exact 1:1 matching of Medicare Advantage (MA) beneficiaries under each contract to traditional Medicare (TM) beneficiaries treated by the same oncologist was done. The sample size without matching was 135 402 and with matching was 46 066. The regression model adjusted for patient age, sex, race and ethnicity, metropolitan residence, marital status, dual eligibility, Yost Index quintile, cancer site, year of diagnosis, and National Cancer Institute Comorbidity Index. SEs were clustered to the beneficiary level in the matched cohort because some TM beneficiaries were matched to multiple MA beneficiaries. Whiskers represent 95% CIs. AHR indicates adjusted hazard ratio; HMO, health maintenance organization; POS, point of service; PPO, preferred provider organization.

^aOther plan types included medical savings account, private fee-for-service plan, Section 1876 Cost Plan, or Program of All-Inclusive Care for the Elderly.

In the matched cohort, the 6-month cumulative incidence of PC was 11.0% and 10.4% among MA and TM beneficiaries, respectively (Figure 1B). In adjusted analyses, there was no overall difference in PC receipt between MA and TM (AHR, 1.05; 95% CI, 0.98-1.12; P = .19). The AHR was attenuated for HMO plans (AHR, 1.16; 95% CI, 1.08-1.24; P < .001) (Figure 2 and eTable 6 in Supplement 1).

PC, Systemic Therapy, and Hospice

Among the decedent subcohort with Part D coverage (n = 76 845), we similarly observed differences in PC receipt without matching (AHR, 1.34; 95% CI, 1.28-1.39; P < .001) and no differences after matching (AHR, 0.98; 95% CI, 0.91-1.07; P = .70) (eFigure 3 in Supplement 1). Systemic therapy receipt from diagnosis to death was lower in MA than in TM (66.3% vs 71.6%), especially in HMO (64.6%) and POS (64.7%) plans (Figure 3). Among all decedents (n = 100 481), hospice enrollment was higher in MA than in TM (74.1% vs 71.9%) (Figure 3). Prior to matching, MA beneficiaries were less likely than TM beneficiaries to receive systemic therapy (adjusted probability difference, −5.38 percentage points [pp]; 95% CI, −6.02 to −4.74 pp; P < .001) and more likely to enroll in hospice (adjusted probability difference, 3.42 pp; 95% CI, 2.82-4.03 pp; P < .001). Differences in systemic therapy receipt were largest for HMO (−7.32 pp; 95% CI, −8.09 to −6.54 pp; P < .001) and POS (−6.78 pp; 95% CI, −9.44 to −4.11 pp; P < .001) plans compared with TM. Differences in hospice enrollment were largest for HMO (4.09 pp; 95% CI, 3.40-4.78 pp; P < .001) and local PPO (2.94 pp; 95% CI, 1.94-3.94 pp; P < .001) plans compared with TM (Figure 4 and eTables 7 and 8 in Supplement 1). After matching, differences in adjusted probability of systemic therapy for MA vs TM persisted (−5.70 pp; 95% CI, −6.89 to −4.52 pp; P < .001). Differences in hospice enrollment for MA vs TM were smaller (1.68 pp; 95% CI, 0.56-2.80 pp; P = .003) (Figure 4 and eTable 7 in Supplement 1). When stratified by plan type, differences in adjusted probability of hospice enrollment were also attenuated for HMO (1.49 pp; 95% CI, 0.15-2.84 pp; P = .03) and local PPO (2.23 pp; 95% CI, 0.76-3.69 pp; P = .003) plans compared with TM (Figure 4 eTable 8 in Supplement 1).

Two-panel bar charts of beneficiaries percent by systemic therapy and hospice, MA and TM. Two side-by-side panels labeled A at upper left and B at upper left. Panel A title text reads MA vs TM. Vertical axis label reads Beneficiaries, percent, with tick marks at 0, 20, 40, 60, and 80. Horizontal axis label reads Systemic therapy use and hospice enrollment. Two category labels on the horizontal axis read Systemic therapy and Hospice. For Systemic therapy, two adjacent bars appear: a pale orange bar labeled in the legend as M A at about 65 percent, and a light blue bar labeled T M at about 71 percent. For Hospice, two adjacent bars appear: pale orange M A at about 74 percent and light blue T M at about 72 percent. A boxed legend centered below the panel contains two color swatches labeled M A in pale orange and T M in light blue. Panel B title text reads MA type vs TM. Vertical axis label reads Beneficiaries, percent, with tick marks at 40, 60, and 80. Horizontal axis label reads Systemic therapy use and hospice enrollment. Two category labels on the horizontal axis read Systemic therapy and Hospice. Under Systemic therapy, six narrow bars appear left to right with colors matching the legend: H M O in dark gray at about 64 percent; Local P P O in dark orange at about 70 percent; Regional P P O in medium tan at about 65 percent; P O S in light tan at about 64 percent; Other with a superscript a in pale orange at about 66 percent; and T M in light blue at about 71 percent. Under Hospice, six narrow bars appear left to right: H M O about 73 percent; Local P P O about 74 percent; Regional P P O about 72 percent; P O S about 73 percent; Other a about 70 percent; and T M about 71 percent. A wide legend below panel B lists H M O, Local P P O, Regional P P O, P O S, Other a, and T M with their corresponding color swatches. — HMO indicates health maintenance organization; MA, Medicare Advantage; POS, point of service; PPO, preferred provider organization; TM, traditional Medicare.

^aOther plan types included medical savings account, private fee-for-service plan, Section 1876 Cost Plan, or Program of All-Inclusive Care for the Elderly.

Two-panel forest plots of adjusted probability differences for systemic therapy and hospice. Two vertically stacked panels labeled A and B, each a dot-and-whisker forest plot with a left text table and a right horizontal axis. Panel A title at upper left: Systemic therapy receipt. Left table header: Variable; Adjusted probability difference, p p, with 95 percent C I. Section header Medicare program with T M listed as 1 reference, and M A with two rows: Matched minus five point seven zero with 95 percent C I minus six point eight nine to minus four point five two; Not matched minus five point three eight with 95 percent C I minus six point zero two to minus four point seven four. Section header Medicare plan type with T M as 1 reference, then H M O: Matched minus seven point seven six with 95 percent C I minus nine point two six to minus six point two six; Not matched minus seven point three two with 95 percent C I minus eight point zero nine to minus six point five four. Local P P O: Matched minus two point five four with 95 percent C I minus four point one six to minus one point one two; Not matched minus one point two seven with 95 percent C I minus two point three two to minus zero point two two. Regional P P O: Matched minus seven point one one with 95 percent C I minus ten point two zero to minus four point zero two; Not matched minus four point five one with 95 percent C I minus six point seven six to minus two point two five. P O S: Matched minus ten point four one with 95 percent C I minus fourteen point six zero to minus six point two three; Not matched minus six point seven eight with 95 percent C I minus nine point four four to minus four point one one. Other or unknown with footnote marker: Matched minus two point three zero with 95 percent C I minus six point four four to one point eight four; Not matched minus three point three three with 95 percent C I minus six point two five to minus zero point four zero. To the right, a dotted vertical reference line at zero with headings above reading Receipt less likely on the left and Receipt more likely on the right. Each estimate appears as a square with a horizontal whisker; teal squares correspond to matched rows and orange squares to not matched rows. Horizontal axis label: Adjusted probability difference, p p, with 95 percent C I, with tick marks at minus fifteen, minus ten, minus five, zero, and five. Panel B title at left: Hospice enrollment. Left table structured similarly with Medicare program and Medicare plan type. Medicare program: T M 1 reference; M A Matched one point six eight with 95 percent C I zero point five six to two point eight zero; Not matched three point four two with 95 percent C I two point eight two to four point zero three. Medicare plan type: T M 1 reference; H M O Matched one point four nine with 95 percent C I zero point one five to two point eight four; Not matched four point zero nine with 95 percent C I three point four zero to four point seven eight. Local P P O Matched two point two three with 95 percent C I zero point seven six to three point six nine; Not matched two point nine four with 95 percent C I one point nine four to three point nine four. Regional P P O Matched minus zero point zero seven with 95 percent C I minus two point nine zero to two point seven seven; Not matched one point zero seven with 95 percent C I minus one point one two to three point two seven. P O S Matched three point three four with 95 percent C I minus zero point zero nine to six point seven seven; Not matched two point one one with 95 percent C I minus zero point four two to four point six four. Other or unknown with footnote marker: Matched zero point five zero with 95 percent C I minus three point three six to four point three seven; Not matched minus one point nine eight with 95 percent C I minus four point eight eight to zero point nine three. Right side includes a dotted vertical line at zero with headings Enrollment less likely and Enrollment more likely. Axis label repeats Adjusted probability difference, p p, with 95 percent C I, with tick marks at minus five, zero, five, and ten. Teal and orange squares with whiskers align to the listed rows. — Exact 1:1 matching of Medicare Advantage (MA) beneficiaries under each contract to traditional Medicare (TM) beneficiaries treated by the same oncologists was done. The sample size without matching was 135 402 and with matching was 46 066. The regression model adjusted for patient age, sex, race and ethnicity, metropolitan residence, marital status, dual eligibility, Yost Index quintile, cancer site, year of diagnosis, and National Cancer Institute Comorbidity Index. SEs were clustered to the beneficiary level in the matched cohort because some TM beneficiaries were matched to multiple MA beneficiaries. Whiskers represent 95% CIs. HMO indicates health maintenance organization; POS, point of service; PPO, preferred provider organization.

^aOther plan types included medical savings account, private fee-for-service plan, Section 1876 Cost Plan, or Program of All-Inclusive Care for the Elderly.

Sensitivity Analyses

Sensitivity analyses among beneficiaries who survived 6 or more months and those who were diagnosed and died outside the COVID-19 disruption period showed similarly higher PC utilization among MA beneficiaries than among TM beneficiaries (eFigures 4 and 5 in Supplement 1), with attenuated differences or no differences after matching (eTables 9 and 10 in Supplement 1). Stratification by cancer sites showed higher PC use among MA beneficiaries than among TM beneficiaries across all cancer sites, with attenuated differences or no differences after matching (eFigure 6 in Supplement 1). Similar patterns were found for systemic therapy and hospice enrollment across cancer sites (eFigure 7 in Supplement 1). Restricting MA beneficiaries to contracts with high data completeness and no 100% missing individual NPIs showed consistent, although smaller, differences between MA and TM (AHR, 1.12; 95% CI, 1.07-1.17; P < .001) (eTable 11 in Supplement 1). This suggests that data completeness is unlikely to explain the observed differences because more complete data capture would be expected to increase rather than decrease the AHR. Results excluding the NCI Comorbidity Index from regressions were also similar (eTable 12 in Supplement 1). The alternative shared-physician design showed consistent results that differences in PC receipt between MA and TM were attenuated after controlling for oncologist fixed effects (HR, 1.08; 95% CI, 1.03-1.13) compared with models without oncologist fixed effects (HR, 1.38; 95% CI, 1.33-1.43). Patterns for systemic therapy and hospice enrollment also remained similar to those in the primary analysis (eTables 13 and 14 in Supplement 1).

Discussion

In this population-based study of Medicare beneficiaries diagnosed with common distant-stage cancers from 2016 to 2021, PC utilization within 6 months of diagnosis was 39% higher in MA (particularly in HMO plans) than in TM, a magnitude comparable to socioeconomic disparities in PC use (ie, the Yost Index) (eTable 5 in Supplement 1). This difference was largely eliminated after matching by treating oncologists. Conversely, systemic therapy receipt remained substantially lower in MA, and higher hospice enrollment differences were attenuated but persisted after matching by oncologists.

These findings suggest that specialty PC is particularly sensitive to the organizational structure of the care delivery system. Access to PC often depends on whether PC clinicians are available within the same organization where oncology care is delivered,⁶ and PC capacity is unevenly distributed and frequently concentrated in academic centers and larger organizations.^43,44 Consistent with this, the higher PC utilization observed among MA beneficiaries was concentrated in HMO plans, many of which, including the Kaiser Foundation Health Plan, feature more selective physician contracting and integrated delivery systems with embedded PC programs that may facilitate PC referrals.^45,46 As anticipated, unmatched MA beneficiaries were disproportionately enrolled in integrated HMO plans, which reflect closed networks in which oncologists rarely treat TM beneficiaries.⁴⁷ These findings highlight that organizational structure may independently influence access to PC in addition to network inclusion. As vertical integration between insurers and health care organizations continues to expand,⁴⁸ future analyses that jointly consider delivery system structure and MA network may clarify which features are associated with PC access and downstream outcomes. Higher PC utilization measured in claims, however, cannot be interpreted as more appropriate or goal-concordant referrals and also warrants future research based on patient-reported outcomes.

The persistently lower systemic therapy receipt among MA beneficiaries after oncologist matching, which removed shared oncologist confounding, suggests that utilization management (eg, prior authorization and formulary rules)^49,50 may affect initiation or continuation of therapy even within the same oncology practice. Unobserved patient characteristics, such as performance status and disease burden, may also contribute to differences in systemic therapy receipt. The partial persistence of hospice enrollment differences after matching suggests a more multifactorial pathway shaped by patient preferences and caregiver capacity not fully driven by treating oncologists.⁵¹ Higher hospice enrollment was only observed in HMO and local PPO plans, not in other MA plans (compared with TM), with attenuation in such differences in the matched cohort, which was potentially attributable to the tighter care management and more structured referral pathways that may facilitate earlier hospice transitions by HMO and local PPO plans. This combination of higher specialty PC use alongside lower systemic therapy receipt and higher hospice enrollment also raises an important alternative interpretation. In particular, the financial incentives created by hospice carve-out benefits may have facilitated earlier hospice election with systemic therapy consequently forgone. It is also possible that the observed differences reflect divergent patient preferences or clinical trajectories that are not fully captured in claims. Future research to assess other patient-centered outcomes, such as alignment of goal-concordant care with patients’ symptom burden, prognosis, and preferences, is critically needed.

Our findings have important policy implications as MA enrollment continues expanding. Although PC utilization was higher among individuals with MA HMO plans compared with TM, the absolute utilization remained low (approximately 10% by 6 months). Because specialty PC capacity is concentrated in a limited number of organizations, this already underused service is particularly vulnerable to structural barriers: plans that exclude organizations with established PC programs may substantially constrain access for their enrollees, even as average utilization under MA appears to be favorable. Current monitoring of MA network adequacy covers 29 specialties, including oncology,⁵² but does not explicitly consider PC specialist availability, which is a gap that our findings suggest warrants attention. Moreover, the attenuated PC differences combined with persistent systemic therapy differences after matching suggests that evaluations of MA performance in oncology should consider network design, utilization management, and other plan features with potentially divergent effects on different aspects of cancer care.

Limitations

Our study has limitations. First, given the observational nature, residual confounding was possible because claim-based measures do not capture symptom severity, functional status, patient preferences, or informal caregiver support, which influence PC referrals. Second, our systemic therapy and hospice measures captured utilization but not appropriateness or goal concordance. Lower systemic therapy receipt and differences in hospice enrollment may reflect unmeasured plan-level utilization management, care coordination, patient preferences, or differences in access and affordability. Third, recovering missing NPIs using the MD-PPAS file may have excluded predominantly MA-serving physicians, as MD-PPAS is derived from fee-for-service claims.⁵³ The encryption of individual NPIs in SEER-Medicare data prevents linkage to external sources. However, these predominantly MA-serving physicians likely fall outside the shared-physician population that our matched analysis targeted. Fourth, we excluded beneficiaries who switched plans to ensure that exposure to a physician network and other plan-level structures remained consistent. Beneficiaries who switch plans may represent a distinct group with unmet needs and warrant future research. Fifth, the study cohort included beneficiaries in SEER regions and with selected cancers. Results may not generalize to other populations or to more recent years as MA enrollment and delivery models continue to evolve.

Conclusions

In this retrospective cohort study of Medicare beneficiaries diagnosed with the most common distant-stage cancers, specialty PC use was higher among MA beneficiaries, particularly beneficiaries in HMO plans, than among TM beneficiaries. This difference largely disappeared after matching beneficiaries by treating oncologists, suggesting that oncologist networks may be key mechanisms affecting specialty PC receipt. In contrast, differences in systemic therapy receipt and hospice enrollment remained even after matching by treating oncologists, suggesting that plan-level policies beyond network composition may contribute to differences in treatment intensity and end-of-life transitions. As MA enrollment continues to grow, efforts to evaluate and refine MA plan design should consider network adequacy alongside other plan features that shape different components of cancer care delivery.

Supplement 1.

eTable 1. Cancer Identification Codes

eFigure 1. CONSORT Chart for Creation of Patient Cohort

eTable 2. Percentage of Claims With Missing Individual NPI by File Type

eMethods 1. Recovering Missing Individual NPI Using Medicare Data on Provider Practice and Specialty File

eFigure 2. Missingness of Provider Identifiers on Medicare Advantage Plan Level

eTable 3. Systemic Therapy Identification Codes

eMethods 2. Alternative Shared-Physician Design

eTable 4. Sample Characteristics Among Traditional Medicare and Medicare Advantage Beneficiaries Who Were Matched and Not Matched Based on Treating Oncologists

eTable 5. Cox Proportional Hazards Regression Model for Palliative Care Comparison Between Medicare Advantage and Traditional Medicare Before and After Matching Based on Treating Oncologists

eTable 6. Cox Proportional Hazards Regression Model for Palliative Care Comparison Between Medicare Advantage Plan Types and Traditional Medicare Before and After Matching Based on Treating Oncologists

eFigure 3. Cumulative Incidence of Palliative Care Up to 6 Months Before and After Matching Based on Treating Oncologists Among Subsample With Part D Coverage and Died by 2021

eTable 7. Logistic Regression for Systemic Therapy Receipt and Hospice Enrollment Comparison Between Medicare Advantage and Traditional Medicare Before and After Matching Based on Treating Oncologists

eTable 8. Logistic Regression for Systemic Therapy Receipt and Hospice Enrollment Comparison Between Medicare Advantage Plan Types and Traditional Medicare Before and After Matching Based on Treating Oncologists

eFigure 4. Cumulative Incidence of Palliative Care up to 6 Months Before and After Matching Based on Treating Oncologists Among Beneficiaries With 6 or More Months of Survival

eFigure 5. Cumulative Incidence of Palliative Care up to 6 Months Before and After Matching Based on Treating Oncologists Among Beneficiaries Who Were Diagnosed and Died Outside of COVID-19 Disruption Period

eTable 9. Cox Proportional Hazards Regression Model for Palliative Care Receipt up to 6 Months Before and After Matching Based on Treating Oncologists Among Beneficiaries With ≥6 Months’ Survival

eTable 10. Cox Proportional Hazards Regression Model for Palliative Care Receipt up to 6 Months Before and After Matching Based on Treating Oncologists Among Beneficiaries Diagnosed and Died Outside COVID-19 Disruption Period

eFigure 6. Cox Proportional Hazards Regression Model Comparing Palliative Care Receipt up to 6 Months Between Medicare Advantage and Traditional Medicare Before and After Matching Based on Treating Oncologists by Cancer Site

eFigure 7. Systemic Therapy Receipt and Hospice Enrollment Comparison Between Medicare Advantage and Traditional Medicare Before and After Matching Based on Treating Oncologists by Cancer Site

eTable 11. Cox Proportional Hazards Regression Model for Palliative Care Comparison Between Medicare Advantage and Traditional Medicare Among Plans With High Data Completeness and Without 100% Missing Provider IDs in Outpatient and Carrier Files

eTable 12. Cox Proportional Hazards Regression Model for Palliative Care Comparison Between Medicare Advantage and Traditional Medicare With and Without Controlling for NCI Comorbidity Index

eTable 13. Sample Characteristics With and Without Propensity Score Weighting for Beneficiary-Level Characteristics

eTable 14. Palliative Care, Systemic Therapy, and Hospice Enrollment Comparison Between Medicare Advantage and Traditional Medicare With and Without Controlling for Treating Oncologist Fixed Effects Using Propensity Score Weighted Sample

jamanetwopen-e2636865-s001.pdf^{(894KB, pdf)}

Supplement 2.

Data Sharing Statement

jamanetwopen-e2636865-s002.pdf^{(14KB, pdf)}

References

1.World Health Organization. Integrating palliative care and symptom relief into primary health care: a WHO guide for planners, implementers and managers. Accessed January 5, 2024. https://qualityhealthservices.who.int/quality-toolkit/qt-catalog-item/integrating-palliative-care-and-symptom-relief-into-primary-health-care-a-who-guide-for-planners-implementers-and-managers
2.Dans M, Kutner JS, Agarwal R, et al. NCCN guidelines insights: palliative care, version 2.2021. J Natl Compr Canc Netw. 2021;19(7):780-788. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Sanders JJ, Temin S, Ghoshal A, et al. Palliative care for patients with cancer: ASCO guideline update. J Clin Oncol. 2024;42(19):2336-2357. doi: 10.1200/JCO.24.00542 [DOI] [PubMed] [Google Scholar]
4.Susman E. NCCN produces its first guidelines on palliative care. Oncol Times. 2001;23(5):25-26. doi: 10.1097/01.COT.0000285846.31466.1e [DOI] [Google Scholar]
5.Smith TJ, Temin S, Alesi ER, et al. American Society of Clinical Oncology provisional clinical opinion: the integration of palliative care into standard oncology care. J Clin Oncol. 2012;30(8):880-887. doi: 10.1200/JCO.2011.38.5161 [DOI] [PubMed] [Google Scholar]
6.Hu X, Kwon Y, Jiang C, et al. Trend and provider- and organizational-level factors associated with early palliative care billing among patients diagnosed with distant-stage cancers in 2010-2019 in the United States. J Clin Oncol. 2025;43(15):1789-1799. doi: 10.1200/JCO-24-01935 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Cavallo J. Understanding oncologists’ perceptions about palliative care and the barriers preventing its use. The ASCO Post. May 25, 2022. Accessed October 20, 2023. https://ascopost.com/issues/may-25-2022/understanding-oncologists-perceptions-about-palliative-care-and-the-barriers-preventing-its-use/
8.Huo J, Hong YR, Turner K, et al. Timing, costs, and survival outcome of specialty palliative care in Medicare beneficiaries with metastatic non-small-cell lung cancer. JCO Oncol Pract. 2020;16(12):e1532-e1542. doi: 10.1200/OP.20.00298 [DOI] [PubMed] [Google Scholar]
9.Ferrario A, Zhang F, Ross-Degnan D, Wharam JF, Twaddle ML, Wagner AK. Use of palliative care among commercially insured patients with metastatic cancer between 2001 and 2016. JCO Oncol Pract. 2022;18(5):e677-e687. doi: 10.1200/OP.21.00516 [DOI] [PubMed] [Google Scholar]
10.Islam JY, Deveaux A, Previs RA, Akinyemiju T. Racial and ethnic disparities in palliative care utilization among gynecological cancer patients. Gynecol Oncol. 2021;160(2):469-476. doi: 10.1016/j.ygyno.2020.11.031 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Islam JY, Braithwaite D, Zhang D, Guo Y, Tailor TD, Akinyemiju T. Racial and ethnic inequities of palliative care use among advanced non-small cell lung cancer patients in the US. Cancer Med. 2023;12(7):8567-8580. doi: 10.1002/cam4.5538 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Freed M, Fuglesten Biniek J, Damico A, Ochieng N, Neuman T. Medicare Advantage in 2026: enrollment update and key trends. KFF. June 5, 2026. Accessed August 24, 2026. https://www.kff.org/medicare/medicare-advantage-enrollment-update-and-key-trends/
13.Xu L, Welch WP, Ruhter J, et al. Medicare Advantage overview: a primer on enrollment and spending: issue brief. Office of the Assistant Secretary for Planning and Evaluation (ASPE). May 2023. Accessed February 10, 2026. https://www.ncbi.nlm.nih.gov/books/NBK604796/ [PubMed]
14.Hu X, Jiang C, Kwon Y, et al. Medicare plan switching and hospice care among decedents with advanced cancer. JAMA Netw Open. 2026;9(3):e260755. doi: 10.1001/jamanetworkopen.2026.0755 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Bradley CJ, Liang R, Lindrooth RC, Sabik LM, Perraillon MC. High-cost cancer drug use in Medicare Advantage and traditional Medicare. JAMA Health Forum. 2025;6(1):e244868. doi: 10.1001/jamahealthforum.2024.4868 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.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]
17.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]
18.Waheed MT, Sullivan KM, Hernandez M, et al. Comparison of cancer care delivery and outcomes in Medicare Advantage with traditional Medicare. Ann Surg. Published online June 26, 2025. doi: 10.1097/SLA.0000000000006819 [DOI] [PubMed] [Google Scholar]
19.Jung J, Carlin C, Feldman R, Song G, Mitchell A. Use of low-value cancer treatments in Medicare Advantage versus traditional Medicare. J Clin Oncol. 2025;43(20):2245-2254. doi: 10.1200/JCO-24-01907 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Raman HS, Greenwald S, Banach E, Abel GA, Lindvall C, Odejide OO. End-of-life care for older adults with blood cancers with Medicare Advantage versus Medicare fee-for-service insurance. JCO Oncol Pract. 2026;22(2):285-293. doi: 10.1200/OP-25-00106 [DOI] [PubMed] [Google Scholar]
21.Hu X, Jiao B, Pan X, Nipp R, Jiang C. Illuminating the value of palliative care in cancer: a path to incentivizing high-value cancer care. Ann Intern Med. 2024;177(12):1722-1724. doi: 10.7326/ANNALS-24-00702 [DOI] [PubMed] [Google Scholar]
22.Jacobson G, Rae M, Neuman T, Orgera K, Boccuti C. 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/
23.Hua M, Guo L, Ing C, Lackraj D, Wang S, Morrison RS. Specialist palliative care use and end-of-life care in patients with metastatic cancer. J Pain Symptom Manage. 2024;67(5):357-365.e15. doi: 10.1016/j.jpainsymman.2024.01.029 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Kavalieratos D, Corbelli J, Zhang D, et al. Association between palliative care and patient and caregiver outcomes: a systematic review and meta-analysis. JAMA. 2016;316(20):2104-2114. doi: 10.1001/jama.2016.16840 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.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]
26.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] [PMC free article] [PubMed] [Google Scholar]
27.White DP, Enewold L, Geiger AM, Banks R, Warren JL. Comparison of physician data in two data files available for cancer health services research. J Natl Cancer Inst Monogr. 2020;2020(55):66-71. doi: 10.1093/jncimonographs/lgz031 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.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
29.Cohen KR, Vabson B, Podulka J, et al. Medicare risk arrangement and use and outcomes among physician groups. JAMA Netw Open. 2025;8(1):e2456074. doi: 10.1001/jamanetworkopen.2024.56074 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Ahammer A, Schober T. Exploring variations in health-care expenditures—what is the role of practice styles? Health Econ. 2020;29(6):683-699. doi: 10.1002/hec.4011 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Hu X, Lipscomb J, Jiang C, Graetz I. Vertical integration of oncologists and cancer outcomes and costs in metastatic castration-resistant prostate cancer. J Natl Cancer Inst. 2023;115(3):268-278. doi: 10.1093/jnci/djac233 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Huo J, Hong YR, Turner K, Bian J, Grewal R, Wilkie DJ. Utilization pattern and service settings of palliative care for patients with metastatic non-small cell lung cancer. Cancer. 2019;125(24):4481-4489. doi: 10.1002/cncr.32478 [DOI] [PubMed] [Google Scholar]
33.Hua M, Yang Z, Guo L, Cassel JB, Morrison RS, Li G. Validation of a claims-based algorithm for specialist palliative care delivery in metastatic cancer. J Pain Symptom Manage. 2026;71(3):385-393.e2. [DOI] [PMC free article] [PubMed]
34.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: 10.1023/A:1011240019516 [DOI] [PubMed] [Google Scholar]
35.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]
36.Curto VE, Politzer E, Anderson TS, et al. Coding intensity variation in Medicare Advantage. Health Aff Sch. 2025;3(1):qxae176. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Meyers DJ, Werner RM. Studying Medicare Advantage with existing data-pitfalls, challenges, and opportunities. JAMA Netw Open. 2025;8(7):e2522833. doi: 10.1001/jamanetworkopen.2025.22833 [DOI] [PubMed] [Google Scholar]
38.Williams R. Using the margins command to estimate and interpret adjusted predictions and marginal effects. Stata J. 2012;12(2):308-331. doi: 10.1177/1536867X1201200209 [DOI] [Google Scholar]
39.Sen AP, Meiselbach MK, Anderson KE, Miller BJ, Polsky D. Physician network breadth and plan quality ratings in Medicare Advantage. JAMA Health Forum. 2021;2(7):e211816. doi: 10.1001/jamahealthforum.2021.1816 [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Jung J, Carlin C, Feldman R. Measuring resource use in Medicare Advantage using encounter data. Health Serv Res. 2022;57(1):172-181. doi: 10.1111/1475-6773.13879 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Jung J, Carlin C, Feldman R, Tran L. Implementation of resource use measures in Medicare Advantage. Health Serv Res. 2022;57(4):957-962. doi: 10.1111/1475-6773.13970 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Johnston KJ, Hendricks M, Dabas M, et al. Nationwide availability of and enrollment in Medicare and Medicaid dual-eligible special needs plans with exclusively aligned enrollment. JAMA Health Forum. 2024;5(10):e243546. doi: 10.1001/jamahealthforum.2024.3546 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Lamb BW, Brown KF, Nagpal K, Vincent C, Green JSA, Sevdalis N. Quality of care management decisions by multidisciplinary cancer teams: a systematic review. Ann Surg Oncol. 2011;18(8):2116-2125. doi: 10.1245/s10434-011-1675-6 [DOI] [PubMed] [Google Scholar]
44.Selby P, Popescu R, Lawler M, Butcher H, Costa A. The value and future developments of multidisciplinary team cancer care. Am Soc Clin Oncol Educ Book. 2019;39(39):332-340. doi: 10.1200/EDBK_236857 [DOI] [PubMed] [Google Scholar]
45.Nguyen HQ, Cuyegkeng T, Phung TO, et al. Integration of a palliative care intervention into community practice for lung cancer: a study protocol and lessons learned with implementation. J Palliat Med. 2017;20(12):1327-1337. doi: 10.1089/jpm.2017.0143 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Khang PS, Wang SE, Liu IA, et al. Impact of inpatient palliative care on quality of end-of-life care and downstream acute and postacute care utilization. J Palliat Med. 2018;21(7):913-923. doi: 10.1089/jpm.2017.0275 [DOI] [PubMed] [Google Scholar]
47.Gitterman DP, Weiner BJ, Domino ME, McKethan AN, Enthoven AC. The rise and fall of a Kaiser Permanente expansion region. Milbank Q. 2003;81(4):567-601. doi: 10.1046/j.0887-378X.2003.00295.x [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Bejarano G, Ryan A, Trivedi A, Offiaeli K, Meyers DJ. Medicare Advantage plan and health system vertical integration, 2011-2020. JAMA Netw Open. 2024;7(7):e2423733. doi: 10.1001/jamanetworkopen.2024.23733 [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Medicare and Medicaid Programs; Patient Protection and Affordable Care Act; Advancing Interoperability and Improving Prior Authorization Processes for Medicare Advantage Organizations, Medicaid Managed Care Plans, State Medicaid Agencies, Children’s Health Insurance Program (CHIP) Agencies and CHIP Managed Care Entities, Issuers of Qualified Health Plans on the Federally-Facilitated Exchanges, Merit-Based Incentive Payment System (MIPS) Eligible Clinicians, and Eligible Hospitals and Critical Access Hospitals in the Medicare Promoting Interoperability Program. 42 CFR §422, 431, 435, 438, 440, and 457 and 45 CFR §156 (2024). Accessed May 10, 2025. https://www.federalregister.gov/documents/2024/02/08/2024-00895/medicare-and-medicaid-programs-patient-protection-and-affordable-care-act-advancing-interoperability
50.Jiang C, Tang Y, Shih YCT, Yabroff R, Han X, Hu X. Medicare Advantage premium, utilization management, and costs for generic abiraterone and imatinib. JCO Oncol Pract. 2025;21(10)(suppl):84. doi: 10.1200/OP.2025.21.10_suppl.84 [DOI] [Google Scholar]
51.Wright AA, Mack JW, Kritek PA, et al. Influence of patients’ preferences and treatment site on cancer patients’ end-of-life care. Cancer. 2010;116(19):4656-4663. doi: 10.1002/cncr.25217 [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Centers for Medicare & Medicaid Services. Medicare-Medicaid Plan (MMP) application & annual requirements. Accessed April 29, 2025. https://www.cms.gov/medicare/medicaid-coordination/plans/mmp-application-annual-requirements
53.Carlin CS, Feldman R, Jung J. Provider attribution in Medicare: challenges and solutions. Health Serv Res. 2026;61(1):e70062. doi: 10.1111/1475-6773.70062 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Supplement 1.

eTable 1. Cancer Identification Codes

eFigure 1. CONSORT Chart for Creation of Patient Cohort

eTable 2. Percentage of Claims With Missing Individual NPI by File Type

eMethods 1. Recovering Missing Individual NPI Using Medicare Data on Provider Practice and Specialty File

eFigure 2. Missingness of Provider Identifiers on Medicare Advantage Plan Level

eTable 3. Systemic Therapy Identification Codes

eMethods 2. Alternative Shared-Physician Design

eTable 4. Sample Characteristics Among Traditional Medicare and Medicare Advantage Beneficiaries Who Were Matched and Not Matched Based on Treating Oncologists

eTable 5. Cox Proportional Hazards Regression Model for Palliative Care Comparison Between Medicare Advantage and Traditional Medicare Before and After Matching Based on Treating Oncologists

eFigure 3. Cumulative Incidence of Palliative Care Up to 6 Months Before and After Matching Based on Treating Oncologists Among Subsample With Part D Coverage and Died by 2021

eFigure 4. Cumulative Incidence of Palliative Care up to 6 Months Before and After Matching Based on Treating Oncologists Among Beneficiaries With 6 or More Months of Survival

eFigure 7. Systemic Therapy Receipt and Hospice Enrollment Comparison Between Medicare Advantage and Traditional Medicare Before and After Matching Based on Treating Oncologists by Cancer Site

eTable 12. Cox Proportional Hazards Regression Model for Palliative Care Comparison Between Medicare Advantage and Traditional Medicare With and Without Controlling for NCI Comorbidity Index

eTable 13. Sample Characteristics With and Without Propensity Score Weighting for Beneficiary-Level Characteristics

jamanetwopen-e2636865-s001.pdf^{(894KB, pdf)}

Supplement 2.

Data Sharing Statement

jamanetwopen-e2636865-s002.pdf^{(14KB, pdf)}

[zoi260990r1] 1.World Health Organization. Integrating palliative care and symptom relief into primary health care: a WHO guide for planners, implementers and managers. Accessed January 5, 2024. https://qualityhealthservices.who.int/quality-toolkit/qt-catalog-item/integrating-palliative-care-and-symptom-relief-into-primary-health-care-a-who-guide-for-planners-implementers-and-managers

[zoi260990r2] 2.Dans M, Kutner JS, Agarwal R, et al. NCCN guidelines insights: palliative care, version 2.2021. J Natl Compr Canc Netw. 2021;19(7):780-788. [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r3] 3.Sanders JJ, Temin S, Ghoshal A, et al. Palliative care for patients with cancer: ASCO guideline update. J Clin Oncol. 2024;42(19):2336-2357. doi: 10.1200/JCO.24.00542 [DOI] [PubMed] [Google Scholar]

[zoi260990r4] 4.Susman E. NCCN produces its first guidelines on palliative care. Oncol Times. 2001;23(5):25-26. doi: 10.1097/01.COT.0000285846.31466.1e [DOI] [Google Scholar]

[zoi260990r5] 5.Smith TJ, Temin S, Alesi ER, et al. American Society of Clinical Oncology provisional clinical opinion: the integration of palliative care into standard oncology care. J Clin Oncol. 2012;30(8):880-887. doi: 10.1200/JCO.2011.38.5161 [DOI] [PubMed] [Google Scholar]

[zoi260990r6] 6.Hu X, Kwon Y, Jiang C, et al. Trend and provider- and organizational-level factors associated with early palliative care billing among patients diagnosed with distant-stage cancers in 2010-2019 in the United States. J Clin Oncol. 2025;43(15):1789-1799. doi: 10.1200/JCO-24-01935 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r7] 7.Cavallo J. Understanding oncologists’ perceptions about palliative care and the barriers preventing its use. The ASCO Post. May 25, 2022. Accessed October 20, 2023. https://ascopost.com/issues/may-25-2022/understanding-oncologists-perceptions-about-palliative-care-and-the-barriers-preventing-its-use/

[zoi260990r8] 8.Huo J, Hong YR, Turner K, et al. Timing, costs, and survival outcome of specialty palliative care in Medicare beneficiaries with metastatic non-small-cell lung cancer. JCO Oncol Pract. 2020;16(12):e1532-e1542. doi: 10.1200/OP.20.00298 [DOI] [PubMed] [Google Scholar]

[zoi260990r9] 9.Ferrario A, Zhang F, Ross-Degnan D, Wharam JF, Twaddle ML, Wagner AK. Use of palliative care among commercially insured patients with metastatic cancer between 2001 and 2016. JCO Oncol Pract. 2022;18(5):e677-e687. doi: 10.1200/OP.21.00516 [DOI] [PubMed] [Google Scholar]

[zoi260990r10] 10.Islam JY, Deveaux A, Previs RA, Akinyemiju T. Racial and ethnic disparities in palliative care utilization among gynecological cancer patients. Gynecol Oncol. 2021;160(2):469-476. doi: 10.1016/j.ygyno.2020.11.031 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r11] 11.Islam JY, Braithwaite D, Zhang D, Guo Y, Tailor TD, Akinyemiju T. Racial and ethnic inequities of palliative care use among advanced non-small cell lung cancer patients in the US. Cancer Med. 2023;12(7):8567-8580. doi: 10.1002/cam4.5538 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r12] 12.Freed M, Fuglesten Biniek J, Damico A, Ochieng N, Neuman T. Medicare Advantage in 2026: enrollment update and key trends. KFF. June 5, 2026. Accessed August 24, 2026. https://www.kff.org/medicare/medicare-advantage-enrollment-update-and-key-trends/

[zoi260990r13] 13.Xu L, Welch WP, Ruhter J, et al. Medicare Advantage overview: a primer on enrollment and spending: issue brief. Office of the Assistant Secretary for Planning and Evaluation (ASPE). May 2023. Accessed February 10, 2026. https://www.ncbi.nlm.nih.gov/books/NBK604796/ [PubMed]

[zoi260990r14] 14.Hu X, Jiang C, Kwon Y, et al. Medicare plan switching and hospice care among decedents with advanced cancer. JAMA Netw Open. 2026;9(3):e260755. doi: 10.1001/jamanetworkopen.2026.0755 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r15] 15.Bradley CJ, Liang R, Lindrooth RC, Sabik LM, Perraillon MC. High-cost cancer drug use in Medicare Advantage and traditional Medicare. JAMA Health Forum. 2025;6(1):e244868. doi: 10.1001/jamahealthforum.2024.4868 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r16] 16.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]

[zoi260990r17] 17.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]

[zoi260990r18] 18.Waheed MT, Sullivan KM, Hernandez M, et al. Comparison of cancer care delivery and outcomes in Medicare Advantage with traditional Medicare. Ann Surg. Published online June 26, 2025. doi: 10.1097/SLA.0000000000006819 [DOI] [PubMed] [Google Scholar]

[zoi260990r19] 19.Jung J, Carlin C, Feldman R, Song G, Mitchell A. Use of low-value cancer treatments in Medicare Advantage versus traditional Medicare. J Clin Oncol. 2025;43(20):2245-2254. doi: 10.1200/JCO-24-01907 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r20] 20.Raman HS, Greenwald S, Banach E, Abel GA, Lindvall C, Odejide OO. End-of-life care for older adults with blood cancers with Medicare Advantage versus Medicare fee-for-service insurance. JCO Oncol Pract. 2026;22(2):285-293. doi: 10.1200/OP-25-00106 [DOI] [PubMed] [Google Scholar]

[zoi260990r21] 21.Hu X, Jiao B, Pan X, Nipp R, Jiang C. Illuminating the value of palliative care in cancer: a path to incentivizing high-value cancer care. Ann Intern Med. 2024;177(12):1722-1724. doi: 10.7326/ANNALS-24-00702 [DOI] [PubMed] [Google Scholar]

[zoi260990r22] 22.Jacobson G, Rae M, Neuman T, Orgera K, Boccuti C. 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/

[zoi260990r23] 23.Hua M, Guo L, Ing C, Lackraj D, Wang S, Morrison RS. Specialist palliative care use and end-of-life care in patients with metastatic cancer. J Pain Symptom Manage. 2024;67(5):357-365.e15. doi: 10.1016/j.jpainsymman.2024.01.029 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r24] 24.Kavalieratos D, Corbelli J, Zhang D, et al. Association between palliative care and patient and caregiver outcomes: a systematic review and meta-analysis. JAMA. 2016;316(20):2104-2114. doi: 10.1001/jama.2016.16840 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r25] 25.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]

[zoi260990r26] 26.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] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r27] 27.White DP, Enewold L, Geiger AM, Banks R, Warren JL. Comparison of physician data in two data files available for cancer health services research. J Natl Cancer Inst Monogr. 2020;2020(55):66-71. doi: 10.1093/jncimonographs/lgz031 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r28] 28.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

[zoi260990r29] 29.Cohen KR, Vabson B, Podulka J, et al. Medicare risk arrangement and use and outcomes among physician groups. JAMA Netw Open. 2025;8(1):e2456074. doi: 10.1001/jamanetworkopen.2024.56074 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r30] 30.Ahammer A, Schober T. Exploring variations in health-care expenditures—what is the role of practice styles? Health Econ. 2020;29(6):683-699. doi: 10.1002/hec.4011 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r31] 31.Hu X, Lipscomb J, Jiang C, Graetz I. Vertical integration of oncologists and cancer outcomes and costs in metastatic castration-resistant prostate cancer. J Natl Cancer Inst. 2023;115(3):268-278. doi: 10.1093/jnci/djac233 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r32] 32.Huo J, Hong YR, Turner K, Bian J, Grewal R, Wilkie DJ. Utilization pattern and service settings of palliative care for patients with metastatic non-small cell lung cancer. Cancer. 2019;125(24):4481-4489. doi: 10.1002/cncr.32478 [DOI] [PubMed] [Google Scholar]

[zoi260990r33] 33.Hua M, Yang Z, Guo L, Cassel JB, Morrison RS, Li G. Validation of a claims-based algorithm for specialist palliative care delivery in metastatic cancer. J Pain Symptom Manage. 2026;71(3):385-393.e2. [DOI] [PMC free article] [PubMed]

[zoi260990r34] 34.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: 10.1023/A:1011240019516 [DOI] [PubMed] [Google Scholar]

[zoi260990r35] 35.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]

[zoi260990r36] 36.Curto VE, Politzer E, Anderson TS, et al. Coding intensity variation in Medicare Advantage. Health Aff Sch. 2025;3(1):qxae176. [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r37] 37.Meyers DJ, Werner RM. Studying Medicare Advantage with existing data-pitfalls, challenges, and opportunities. JAMA Netw Open. 2025;8(7):e2522833. doi: 10.1001/jamanetworkopen.2025.22833 [DOI] [PubMed] [Google Scholar]

[zoi260990r38] 38.Williams R. Using the margins command to estimate and interpret adjusted predictions and marginal effects. Stata J. 2012;12(2):308-331. doi: 10.1177/1536867X1201200209 [DOI] [Google Scholar]

[zoi260990r39] 39.Sen AP, Meiselbach MK, Anderson KE, Miller BJ, Polsky D. Physician network breadth and plan quality ratings in Medicare Advantage. JAMA Health Forum. 2021;2(7):e211816. doi: 10.1001/jamahealthforum.2021.1816 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r40] 40.Jung J, Carlin C, Feldman R. Measuring resource use in Medicare Advantage using encounter data. Health Serv Res. 2022;57(1):172-181. doi: 10.1111/1475-6773.13879 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r41] 41.Jung J, Carlin C, Feldman R, Tran L. Implementation of resource use measures in Medicare Advantage. Health Serv Res. 2022;57(4):957-962. doi: 10.1111/1475-6773.13970 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r42] 42.Johnston KJ, Hendricks M, Dabas M, et al. Nationwide availability of and enrollment in Medicare and Medicaid dual-eligible special needs plans with exclusively aligned enrollment. JAMA Health Forum. 2024;5(10):e243546. doi: 10.1001/jamahealthforum.2024.3546 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r43] 43.Lamb BW, Brown KF, Nagpal K, Vincent C, Green JSA, Sevdalis N. Quality of care management decisions by multidisciplinary cancer teams: a systematic review. Ann Surg Oncol. 2011;18(8):2116-2125. doi: 10.1245/s10434-011-1675-6 [DOI] [PubMed] [Google Scholar]

[zoi260990r44] 44.Selby P, Popescu R, Lawler M, Butcher H, Costa A. The value and future developments of multidisciplinary team cancer care. Am Soc Clin Oncol Educ Book. 2019;39(39):332-340. doi: 10.1200/EDBK_236857 [DOI] [PubMed] [Google Scholar]

[zoi260990r45] 45.Nguyen HQ, Cuyegkeng T, Phung TO, et al. Integration of a palliative care intervention into community practice for lung cancer: a study protocol and lessons learned with implementation. J Palliat Med. 2017;20(12):1327-1337. doi: 10.1089/jpm.2017.0143 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r46] 46.Khang PS, Wang SE, Liu IA, et al. Impact of inpatient palliative care on quality of end-of-life care and downstream acute and postacute care utilization. J Palliat Med. 2018;21(7):913-923. doi: 10.1089/jpm.2017.0275 [DOI] [PubMed] [Google Scholar]

[zoi260990r47] 47.Gitterman DP, Weiner BJ, Domino ME, McKethan AN, Enthoven AC. The rise and fall of a Kaiser Permanente expansion region. Milbank Q. 2003;81(4):567-601. doi: 10.1046/j.0887-378X.2003.00295.x [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r48] 48.Bejarano G, Ryan A, Trivedi A, Offiaeli K, Meyers DJ. Medicare Advantage plan and health system vertical integration, 2011-2020. JAMA Netw Open. 2024;7(7):e2423733. doi: 10.1001/jamanetworkopen.2024.23733 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r49] 49.Medicare and Medicaid Programs; Patient Protection and Affordable Care Act; Advancing Interoperability and Improving Prior Authorization Processes for Medicare Advantage Organizations, Medicaid Managed Care Plans, State Medicaid Agencies, Children’s Health Insurance Program (CHIP) Agencies and CHIP Managed Care Entities, Issuers of Qualified Health Plans on the Federally-Facilitated Exchanges, Merit-Based Incentive Payment System (MIPS) Eligible Clinicians, and Eligible Hospitals and Critical Access Hospitals in the Medicare Promoting Interoperability Program. 42 CFR §422, 431, 435, 438, 440, and 457 and 45 CFR §156 (2024). Accessed May 10, 2025. https://www.federalregister.gov/documents/2024/02/08/2024-00895/medicare-and-medicaid-programs-patient-protection-and-affordable-care-act-advancing-interoperability

[zoi260990r50] 50.Jiang C, Tang Y, Shih YCT, Yabroff R, Han X, Hu X. Medicare Advantage premium, utilization management, and costs for generic abiraterone and imatinib. JCO Oncol Pract. 2025;21(10)(suppl):84. doi: 10.1200/OP.2025.21.10_suppl.84 [DOI] [Google Scholar]

[zoi260990r51] 51.Wright AA, Mack JW, Kritek PA, et al. Influence of patients’ preferences and treatment site on cancer patients’ end-of-life care. Cancer. 2010;116(19):4656-4663. doi: 10.1002/cncr.25217 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260990r52] 52.Centers for Medicare & Medicaid Services. Medicare-Medicaid Plan (MMP) application & annual requirements. Accessed April 29, 2025. https://www.cms.gov/medicare/medicaid-coordination/plans/mmp-application-annual-requirements

[zoi260990r53] 53.Carlin CS, Feldman R, Jung J. Provider attribution in Medicare: challenges and solutions. Health Serv Res. 2026;61(1):e70062. doi: 10.1111/1475-6773.70062 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Palliative Care for Advanced Cancer in Medicare Advantage and Physician Networks

Xin Hu, PhD

Changchuan Jiang, MD

Youngmin Kwon, PhD

Qinjin Fan, PhD

Kewei Sylvia Shi, PhD

Zhiyuan Zheng, PhD

Jingxuan Zhao, PhD

Craig Evan Pollack, MD, MHS

Amanda L Blackford, ScM

Joan L Warren, PhD

Xuesong Han, PhD

K Robin Yabroff, PhD

Key Points

Question

Findings

Meaning

Abstract

Importance

Objective

Design, Setting, and Participants

Exposure

Main Outcomes and Measures

Results

Conclusions and Relevance

Introduction

Methods

Study Design and Sample

Measures

Time to First Palliative Care Visit

Systemic Therapy Use and Hospice Enrollment Among Decedents

Study Covariates

Statistical Analysis

Results

Table. Sample Characteristics by Medicare Advantage and Traditional Medicare Before and After 1:1 Exact Matching by Treating Oncologist.

Palliative Care

Figure 1. Cumulative Incidence Curves of Palliative Care Receipt Up to 6 Months After Cancer Diagnosis.

Figure 2. Dot-and-Whisker Plot of Adjusted Differences in Palliative Care (PC) Receipt Up to 6 Months Before and After Matching Based on Treating Oncologists.

PC, Systemic Therapy, and Hospice

Figure 3. Bar Graphs Showing Proportions of Decedents With Systemic Therapy Receipt and Hospice Enrollment.

Figure 4. Dot-and-Whisker Plots of Adjusted Differences in Systemic Therapy Receipt and Hospice Enrollment Among Decedents Before and After Matching Based on Treating Oncologists.

Sensitivity Analyses

Discussion

Limitations

Conclusions

References

Associated Data

Supplementary Materials

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases