Leveraging CMS Medicare Data for Oncology RWE: Benefits, Limitations, and Opportunities, and Insights into Enrollment Patterns
Author(s)
Keri Calkins, PhD1, Michael Barna, MA1, Jelena Zurovac, MS, PhD2, Alex Bohl, PhD3, Shiven BHARDWAJ, MA, MASc4.
1Mathematica, Ann Arbor, MI, USA, 2Mathematica, Washington DC, DC, USA, 3Mathematica, Cambridge, MA, USA, 4CHOICE Institute at the University of Washington, Seattle, WA, USA.
1Mathematica, Ann Arbor, MI, USA, 2Mathematica, Washington DC, DC, USA, 3Mathematica, Cambridge, MA, USA, 4CHOICE Institute at the University of Washington, Seattle, WA, USA.
Presentation Documents
OBJECTIVES: CMS’s data for all Medicare fee-for-service (FFS) and Medicare Advantage (MA) beneficiaries offer valuable opportunities for real world evidence (RWE) studies, including oncology. We analyze enrollment changes (churn) and discuss the advantages and limitations of these data. Studying churn informs dataset selection, protocol development, and study design.
METHODS: Using Medicare data, we created Sankey diagrams showing the number of FFS beneficiaries who transitioned to MA, who died, or into only Medicare Part A or B between 2017, 2020, and 2022. We analyzed all Medicare beneficiaries and those with lung cancer, breast cancer, and multiple myeloma.
RESULTS: Among 31.5 million FFS beneficiaries in 2017, 23.3 million retained FFS in 2020, 3.6 million transitioned to MA, and 4.4 million died. Among 120,831 FFS beneficiaries with breast cancer (80,198 with lung cancer) in 2017, 75,854 (20,835) retained FFS, 26,008 (54,248) died, and 18,720 (5,052) enrolled in MA by 2022. Few beneficiaries who transitioned to MA by 2020 switched back to FFS by 2022.
CONCLUSIONS: CMS’ Medicare data are underused in oncology RWE but offer important advantages over closed claims datasets by comprehensively capturing enrollment, mortality, demographics, comorbidities, and healthcare utilization for nearly all U.S. adults aged ≥65 years, while improving power and reducing selection bias by including MA beneficiaries and those switching between FFS and MA. This is particularly important for oncology research where survival is a key outcome and cancer treatment costs may prompt switching from MA to FFS. Disadvantages are MA data lags and limited staging and histology information. CMS data are well suited for research questions not requiring recent data and where lines of therapy can approximate disease severity, such as multiple myeloma.
METHODS: Using Medicare data, we created Sankey diagrams showing the number of FFS beneficiaries who transitioned to MA, who died, or into only Medicare Part A or B between 2017, 2020, and 2022. We analyzed all Medicare beneficiaries and those with lung cancer, breast cancer, and multiple myeloma.
RESULTS: Among 31.5 million FFS beneficiaries in 2017, 23.3 million retained FFS in 2020, 3.6 million transitioned to MA, and 4.4 million died. Among 120,831 FFS beneficiaries with breast cancer (80,198 with lung cancer) in 2017, 75,854 (20,835) retained FFS, 26,008 (54,248) died, and 18,720 (5,052) enrolled in MA by 2022. Few beneficiaries who transitioned to MA by 2020 switched back to FFS by 2022.
CONCLUSIONS: CMS’ Medicare data are underused in oncology RWE but offer important advantages over closed claims datasets by comprehensively capturing enrollment, mortality, demographics, comorbidities, and healthcare utilization for nearly all U.S. adults aged ≥65 years, while improving power and reducing selection bias by including MA beneficiaries and those switching between FFS and MA. This is particularly important for oncology research where survival is a key outcome and cancer treatment costs may prompt switching from MA to FFS. Disadvantages are MA data lags and limited staging and histology information. CMS data are well suited for research questions not requiring recent data and where lines of therapy can approximate disease severity, such as multiple myeloma.
Conference/Value in Health Info
2025-05, ISPOR 2025, Montréal, Quebec, CA
Value in Health, Volume 28, Issue S1
Code
RWD16
Topic
Real World Data & Information Systems
Topic Subcategory
Health & Insurance Records Systems
Disease
SDC: Oncology