Although many studies have investigated the accuracy of Medicare race variables, there are few methodological studies or reports that confirm the consistency of race within CMS enrollment data over time. After comparing Medicare beneficiaries that were enrolled in both 2010 and 2019, this methodological report concluded that over 99.9% of beneficiaries are consistent over time. Further stratification by entitlement and sex did not justify the changes for the <0.1%.
Can a beneficiary's race change in the MBSF: Base?
The Master Beneficiary Summary File: Base file (MBSF: Base), a research identifiable file offered by CMS, contains two annual race variables: the Beneficiary Race Code and the Research Triangle Institute (RTI) Race Code. There are studies that look at the accuracy of these race variables(1,2), but there is little information about whether the information changes within the data itself.
Both race codes are sourced from Social Security Administration (SSA) data. However, the RTI Race Code has additional algorithms to further identify Hispanic and Asian populations that would otherwise be mislabeled(3,4). Historically, the SSA collected race data on a voluntary basis when individuals applied for a social security number (SSN)(5). The options for race have changed since the inception of SSA in 1936, meaning that race data utility varies by the date of SSN application(6–8). At the time of writing, the author has not found a pathway for individuals to request a race change to the SSA.
Based on the above information, we may assume that race should never change in the data. This brief seeks evidence that supports or contradicts this assumption by comparing race variables for Medicare beneficiaries enrolled in 2010 and 2019.
Methodology
The study population includes Medicare beneficiaries who were enrolled in both 2010 and 2019. The data source is the 2010 and 2019 MBSF: Base. The MBSF: Base is a research identifiable file offered by CMS that contains demographic information for all Medicare enrollees by calendar year. The Beneficiary Race Code and RTI Race Code are reported once per year per beneficiary.
To quantify stability over time, we compared the race reporting for beneficiaries who were enrolled in both 2010 and 2019. The 2010 MBSF: Base contained approximately 50 million beneficiaries. The 2019 MBSF: Base contained approximately 64 million beneficiaries. The total number of beneficiaries that could be identified in 2010 and 2019 was approximately 32 million beneficiaries. For everyone in our cohort, the values of 2010 Beneficiary Race Codes were compared to the 2019 Beneficiary Race Code, and the values of the 2010 RTI Race Code were compared to the 2019 RTI Race Code.
We further stratified beneficiaries with non-matching RTI Race codes by 2010 Current Reason for Entitlement and Sex. This was an attempt to identify potential reasons for race changes. The enrollment process for Old Age and Survivor’s Insurance (OASI) is less intensive than the enrollment process for beneficiaries who qualify via Disability (DIB), End-Stage Renal Disease (ESRD), or both (DIB & ESRD). Potentially, more intensive application requirements could give beneficiaries a greater opportunity to report their race. We also stratified by Sex in response to RTI’s methodology of surname matching, specifically to see if beneficiaries’ race changed based on surname changes.
Results
Out of the 32 million beneficiaries that were enrolled in Medicare in 2010 and 2019, we found that fewer than 1% have different race values between 2010 and 2019 (Table 1). We further stratified the beneficiaries with different RTI race values by 2010 Current Reason for Entitlement (Table 2) and Sex (Table 3). Both stratifications showed very little change compared to the changes seen in Table 1.
Table 1: Beneficiaries with Different Race Values between 2010 and 2019
| Variable | N | Total | (%) |
|---|---|---|---|
| Beneficiary Race Code | 3,080 | 31,937,417 | 0.0% |
| RTI Race Code | 28,716 | 31,937,417 | 0.1% |
Table 1 caption: This table describes the total number of beneficiaries with different beneficiary race and RTI race codes between 2010 and 2019. The total column describes the total number of beneficiaries that were enrolled in Medicare in 2010 and 2019.
Table 2. Stratifying Beneficiaries with Different RTI Race Code Values by 2010 Current Reason for Entitlement
| 2010 Reason for Entitlement | N | Total | (%) |
|---|---|---|---|
| OASI | 22,036 | 27,996,754 | 0.1% |
| DIB | 6,606 | 3,904,983 | 0.2% |
| ESRD | 56 | 20,594 | 0.3% |
| DIB & ESRD | 18 | 15,086 | 0.1% |
Table 2 caption: This table describes the total number of beneficiaries with different RTI race code values by 2010 reason for entitlement. The total column describes the total number of beneficiaries enrolled in both 2010 and 2019 with a given Reason for Entitlement value in 2010.
Table 3. Stratifying Beneficiaries with Different RTI Race Code Values by Sex
| Sex | N | Total | (%) |
|---|---|---|---|
| Female | 18,593 | 17,881,969 | 0.1% |
| Male | 10,123 | 14,055,437 | 0.1% |
| Unknown | 0 | <11 | 0.0% |
Table 3 caption: This table describes the total number of beneficiaries with different RTI race code values by sex. The total column describes the total number of beneficiaries enrolled in both 2010 and 2019 with a given sex value.
Discussion
Race does not change frequently in the MBSF: Base. Intuitively, this makes sense because there seems to be no method for updating one’s race and there is no incentive to change one’s race with the SSA. When it comes to data stability, race is a very stable source.
However, for the small proportion of race changes over time, it is still unclear how or why these occurred. Stratifying further by 2010 Reason for Entitlement and Sex did not yield any clear insight.
Limitations
This article looks at two specific points in time, so there are limitations that restrict the confidence of our conclusion. For example, this methodology cannot track changes that occurred from year to year and is missing information from more recent years of data. More detailed information could provide greater details into beneficiary race changes over time. Additional exploration into the characteristics of the beneficiaries who did display race changes could also provide more information about how race changes occur.
This article is also limited to beneficiaries who were enrolled in 2010 and 2019. Beneficiaries who lose enrollment between 2010 and 2019 are not counted. This could result in under-counting the total number of beneficiaries with changes in race.
Conclusion
This methodological report looks at the consistency of the two race variables available in the MBSF: Base over time. After comparing 2010 and 2019 race variable values, we confirmed that over 99% of all beneficiaries retain the same race over time. This supports that race is generally consistent within the MBSF. An exploratory stratification by entitlement and sex did not yield any insightful reasons for why a beneficiary’s race could change over time. Overall, race appears to be consistent in the MBSF: Base over time, but the rare instances where it is inconsistent require additional review.
- Jarrín OF, Nyandege AN, Grafova IB, Dong X, Lin H. Validity of Race and Ethnicity Codes in Medicare Administrative Data Compared With Gold-standard Self-reported Race Collected During Routine Home Health Care Visits. Med Care. 2020 Jan;58(1):e1–8.
- Waldo DR. Accuracy and Bias of Race/Ethnicity Codes in the Medicare Enrollment Database. Health Care Financ Rev. 2004;26(2):61–72.
- Medicare Beneficiary Summary File (MBSF) Base with Medicare Part A, B, C, and D Codebook [Internet]. Chronic Conditions Warehouse Virtual Research Data Center; 2025 [cited 2025 Dec 12]. Available from: https://www2.ccwdata.org/documents/10280/19022436/codebook-mbsf-abcd-v2.pdf
- Eicheldinger C, Bonito A. More accurate racial and ethnic codes for Medicare administrative data. Health Care Financ Rev. 2008;29(3):27–42.
- Puckett C. The Story of the Social Security Number. Social Security. 2009 July;69(2):55.
- Maxwell A. A Resource Guide for Using Medicare’s Enrollment Race and Ethnicity [Internet]. U.S. Department of Health and Human Services Office of Inspector General; 2023 June [cited 2025 Dec 15]. Report No.: OEI-02-21-00101. Available from: https://oig.hhs.gov/documents/evaluation/2715/OEI-02-21-00101-Resource%20Guide.pdf
- Data Brief: Inaccuracies in Medicare’s Race and Ethnicity Data Hinder the Ability To Assess Health Disparities [Internet]. U.S. Department of Health and Human Services Office of Inspector General; 2022 June [cited 2025 Dec 15]. Report No.: OEI-02-21-00100. Available from: https://oig.hhs.gov/documents/evaluation/10115/OEI-02-21-00100-Complete%20Report.pdf
- Social Security [Internet]. [cited 2025 Dec 15]. Employee SSN Application Form. Available from: https://www.ssa.gov/history/ssn/ss5.html