Je veux savoir à propos de…

Federal Register / Vol. 86, No. 147 / Wednesday, August 4, 2021 / Rules and Regulations
lotter on DSK11XQN23PROD with RULES5

help to overcome the current limitations of demographic information and enable timelier reporting of equity results until longer term collaborations to improve demographic data quality across the health care sector materialize. The use of indirectly estimated race and ethnicity for conducting stratified reporting does not place any additional collection or reporting burdens on facilities as these data are derived using existing administrative and censuslinked data.
Indirect estimation relies on a statistical imputation method for inferring a missing variable or improving an imperfect administrative variable using a related set of information that is more readily available.54 Indirectly estimated data are most commonly used at the population level such as the facility or health planlevel, where aggregated results form a more accurate description of the population than existing, imperfect data sets. These methods often estimate race and ethnicity using a combination of other data sources which are predictive of self-identified race and ethnicity, such as language preference, information about race and ethnicity in our administrative records, first and last names matched to validated lists of names correlated to specific national origin groups, and the racial and ethnic composition of the surrounding neighborhood. Indirect estimation has been used in other settings to support population-based equity measurement when self-identified data are not available.55
As described in section IV.D.2, we have previously supported the development of two such methods of indirect estimation of race and ethnicity of Medicare beneficiaries. One indirect estimation approach, developed by our contractor, uses Medicare administrative data, first name and surname matching, derived from the U.S. Census and other sources, with beneficiary language preference, state of residence, and the source of the race and ethnicity code in Medicare administrative data to reclassify some beneficiaries as Hispanic or Asian/
Pacific Islander API.56 In recent years, 54 IOM. 2009. Race, Ethnicity, and Language Data:
Standardization for Health Care Quality Improvement. Washington, DC: The National Academies Press.
55 IOM. 2009. Race, Ethnicity, and Language Data:
Standardization for Health Care Quality Improvement. Washington, DC: The National Academies Press.
56 Bonito AJ, Bann C, Eicheldinger C, Carpenter L. Creation of New Race-Ethnicity Codes and Socioeconomic Status SES Indicators for Medicare Beneficiaries. Final Report, Sub-Task 2. Prepared by RTI International for the Centers for Medicare
VerDate Sep<11>2014

21:11 Aug 03, 2021

Jkt 253001

we have also worked with another contractor to develop a new approach, the Medicare Bayesian Improved Surname Geocoding MBISG, which combines Medicare administrative data, first and surname matching, geocoded residential address linked to the 2010
U.S. Census, and uses both Bayesian updating and multinomial logistic regression to estimate the probability of belonging to each of six racial/ethnic groups.57
The MBISG model is currently used to conduct the national, contract-level, stratified reporting of Medicare Part C &
D performance data for Medicare Advantage Plans by race and ethnicity.58 Validation testing reveals concordances with self-reported race and ethnicity of 0.96 through 0.99 for API, Black, Hispanic, and White beneficiaries for MBISG version 2.1.59
The algorithms under consideration are considerably less accurate for individuals who self-identify as American Indian/Alaskan Native or multiracial.60 Indirect estimation can be a statistically reliable approach for calculating population-level equity results for groups of individuals such as the facility-level and is not intended, nor being considered, as an approach for inferring the race and ethnicity of an individual.
However, despite the high degree of statistical accuracy of the indirect estimation algorithms under and Medicaid Services through an interagency agreement with the Agency for Healthcare Research and Policy, under Contract No. 500000024, Task No. 21 AHRQ Publication No. 080029EF.
Rockville, MD, Agency for Healthcare Research and Quality. January 2008.
57 Haas, A., Elliott, M. et al 2018. Imputation of race/ethnicity to enable measurement of HEDIS
performance by race/ethnicity. Health Services Research, 54:1323.
58 The Office of Minority Health 2020. Racial, Ethnic, and Gender Disparities in Health Care in Medicare Advantage, The Centers for Medicare and Medicaid Services, pg vii. https www.cms.gov/
About-CMS/Agency-Information/OMH/researchand-data/statistics-and-data/stratified-reporting.
59 MBISG 2.1 validation results performed under contract GS10F0012Y/HHSM5002016
00097G. Pending public release of the 2021 Part C
and D Performance Data Stratified by Race, Ethnicity, and Gender Report, available at: https
www.cms.gov/About-CMS/Agency-Information/
OMH/research-and-data/statistics-and-data/
stratified-reporting.
60 Haas, A., Elliott, M. et al 2018. Imputation of race/ethnicity to enable measurement of HEDIS
performance by race/ethnicity. Health Services Research, 54:1323 and Bonito AJ, Bann C, Eicheldinger C, Carpenter L. Creation of New RaceEthnicity Codes and Socioeconomic Status SES
Indicators for Medicare Beneficiaries. Final Report, Sub-Task 2. Prepared by RTI International for the Centers for Medicare and Medicaid Services through an interagency agreement with the Agency for Healthcare Research and Policy, under Contract No. 500000024, Task No. 21 AHRQ Publication No. 080029EF. Rockville, MD, Agency for Healthcare Research and Quality. January 2008.

PO 00000

Frm 00023

Fmt 4701

Sfmt 4700

42629

consideration there remains the small risk of unintentionally introducing bias.
For example, if the indirect estimation is not as accurate in correctly estimating race and ethnicity in certain geographies or populations it could lead to some bias in the method results. Such bias might result in slight overestimation or underestimation of the quality of care received by a given group. We feel this amount of bias is considerably less than would be expected if stratified reporting was conducted using the race and ethnicity currently contained in our administrative data. Indirect estimation of race and ethnicity is envisioned as an intermediate step, filling the pressing need for more accurate demographic information for the purposes of exploring inequities in service delivery, while allowing newer approaches, as described in the next section, for improving demographic data collection to progress. We expressed interest in learning more about, and solicited comments about, the potential benefits and challenges associated with measuring facility equity using an imputation algorithm to enhance existing administrative data quality for race and ethnicity until self-reported information is sufficiently available.
c. Improving Demographic Data Collection Stratified facility-level reporting using dual eligibility and indirectly estimated race and ethnicity would represent an important advance in our ability to provide equity reports to facilities.
However, self-reported race and ethnicity data remain the gold standard for classifying an individual according to race or ethnicity. The CMS Quality Strategy outlines our commitment to strengthening infrastructure and data systems by ensuring that standardized demographic information is collected to identify disparities in health care delivery outcomes.61 Collection and sharing of a standardized set of social, psychological, and behavioral data by facilities, including race and ethnicity, using electronic data definitions which permit nationwide, interoperable health information exchange, can significantly enhance the accuracy and robustness of our equity reporting.62 This could potentially include expansion to 61 The Centers for Medicare & Medicaid Services.
CMS Quality Strategy. 2016. https www.cms.gov/
Medicare/Quality-Initiatives-Patient-AssessmentInstruments/QualityInitiativesGenInfo/Downloads/
CMS-Quality-Strategy.pdf.
62 The Office of the National Coordinator for Health Information Technology. United State Core Data for Interoperability Draft Version 2. 2021.
https www.healthit.gov/isa/sites/isa/files/2021-01/
Draft-USCDI-Version-2-January-2021-Final.pdf.

E:FRFM04AUR5.SGM

04AUR5