Medicare Data Inaccuracies Hinder Health Disparity Assessment

A new report published by the Department of Health and Human Services (HHS) Office of the Inspector General (OIG) has found that race and ethnicity data gathered as part of Medicare enrollment are inaccurate for some groups and these inaccuracies hinder the government’s efforts to assess and address health disparities.

Medicare is a key part of the American healthcare system, with approximately 66 million beneficiaries currently enrolled. The Centers for Medicare and Medicaid Services (CMS) has made advancing health equity a top priority for the agency, and its ability to assess health disparities is crucial to achieving that goal.

The HHS-OIG report notes that the ability to accurately assess health disparities relies on the quality of the race and ethnicity data used, so the agency set out to evaluate Medicare data.

OIG began by analyzing all race and ethnicity data in the Medicare enrollment database, which is the only source of this information for all enrolled beneficiaries. The accuracy of the data was then evaluated by comparing it to self-reported data, which is considered the most accurate type of data in the context of race and ethnicity reporting. The self-reported data was gathered from a subset of beneficiaries residing in nursing homes.

The report indicates that Medicare’s enrollment data was inaccurate in terms of race and ethnicity for some groups of beneficiaries, such as those identifying as Hispanic, Asian/Pacific Islander, or American Indian/Alaska Native, as compared with others.

The report also highlighted two types of errors that created the most inaccuracies in the data: enrollment data sometimes identifying beneficiaries as a race and ethnicity that they do not identify themselves with on the nursing home assessment and the data not capturing the race and ethnicity with which these beneficiaries do identify.

The report shows that 28 percent of the beneficiaries identified as Hispanic in the enrollment data do not identify themselves as Hispanic on their nursing home assessments, and the same error occurs for 46 percent of beneficiaries identified in the enrollment data as American Indian/Alaska Native and 17 percent identified as Asian/Pacific Islander. Only 4 percent of Black and 1 percent of White beneficiaries were found to be subject to this same error.

Similarly, 13 percent of beneficiaries who self-identified as Hispanic on their nursing home assessments are not identified that way in the enrollment data, and this error also occurs for 35 percent of beneficiaries who self-identified as American Indian/Alaska Native on their nursing home assessments and for 24 percent who self-identified as Asian/Pacific Islander. These figures are much smaller for Black and White beneficiaries, which represent 3 and 4 percent of this type of error, respectively.

These errors may be due in part to limitations in the source data, the report notes. The primary source of Medicare race and ethnicity enrollment data is the Social Security Administration (SSA), which stopped routinely collecting this data in 1989. Even while the data was routinely being collected, race and ethnicity categories used by SSA were limited. Before 1980, beneficiaries could only choose from Black, White, or other, and anyone who did not complete the question were labelled as “unknown.”

As a result, SSA data is lacking for over 3.3 million Medicare beneficiaries who are categorized as “other” or “unknown.” Because SSA stopped routinely collecting thie data, the completeness of the data will decrease over time, making it even less useful for Medicare enrollment in the future, according to the report.

To address these issues, the OIG recommends that CMS develop its own source of race and ethnicity data, use self-reported race and ethnicity information to improve data for current beneficiaries, develop a process to ensure that the data is as standardized as possible, and educate beneficiaries about CMS's efforts to improve the race and ethnicity information.