Race and Ethnicity are Essential Health Data

Amelia Kermis, who holds a master’s degree in public health, is a former branch chief at the U.S. Census Bureau and a doctoral student at George Washington University’s Milken Institute School of Public Health. Meeta Anand is senior director of census and data equity at the Leadership Conference on Civil and Human Rights. 

In early 2020, Chicago released a breakdown of its COVID-19 deaths by race. Despite Black people accounting for 30% of the city’s population, they represented 72%  of Chicago’s COVID deaths. Then-Chicago Mayor Lori Lightfoot called the gap a “public health red alarm” and activated a Racial Equity Rapid Response team to combat the racial disparities in COVID-19 deaths by concentrating screening and outreach efforts to reach those most at risk. Without census data, Chicago would have had a count of deaths but no way to know who was dying at a higher rate. 

On September 10, 2026, the U.S. Department of Commerce proposed a rule that would take away Chicago’s ability to know who was dying at a higher rate and coordinate its response to those at highest risk, among other things. The rule – 91 FR 57524, docket USBC-2026-0628 – states that “no questions about race, ethnicity, or sexual orientation shall appear on the short form of the decennial census questionnaire or any questionnaire used for the enumeration of the population.” The only demographic questions left would be biological sex, date of birth, and household relationships. 

The proposal comes four months after the Census Bureau quietly reversed its plan to adopt updated federal race and ethnicity standards (Statistical Policy Directive No. 15) on the 2027 American Community Survey (ACS), despite its own research showing the new questions produce better race and ethnicity data. Together, the two decisions would leave the country with outdated race data from the ACS and none from the decennial census. 

Race is a social determinant of health 

The rule’s summary says the count should not be “distorted in any way by questions about immaterial personal characteristics, such as race.” Public health doesn’t treat race as biology. Instead, it treats race as a social determinant of health because it shapes where people live, what they’re exposed to, and the health care they receive. Social determinants of health are the conditions in which a people are born, grow, live, work, and age, along with their access to power, money, and resources, all of which can influence their health. 

For public health programs to succeed, they must be tailored to the people they serve. General data may show that a health problem is widespread but won’t indicate who is most at risk. To do that, public health experts need to compare rates by demographics such as race, which requires the total population as the denominator. 

Amelia Kermis: What hepatitis B screening in Philadelphia taught me

My first exposure to the importance of race at the total-population level came during my public health training in Philadelphia, Pennsylvania. I worked as part of a student-led initiative, Hepatitis B United Philadelphia, to lower rates of liver cancer due to hepatitis B by providing free screenings and vaccinations. We knew that rates of undiagnosed chronic hepatitis B were highest in Asian Americans and Pacific Islanders, so we focused on that community in Philadelphia. Our messaging wasn’t in accessible languages, and we focused our activities based on community-based organization interest rather than on the community’s needs. 

Race and ethnicity data changed our approach. Of those screened from 2007 to 2010, infection rates ranged from 4% of Koreans to 23.5% of Chinese immigrants, which led us to shift our focus to Chinese residents. Rather than developing education content in multiple languages, we developed tailored messages in various Chinese dialects. We partnered with community leaders to find times and locations to host our events. This meant, for example, that we scheduled screenings around Chinese restaurant workers’ shifts and after Sunday church services.

However, our work didn’t end there; to decide on further screenings, research vaccination rates, and report outcomes, we needed population data on specific AAPI groups within Philadelphia. Our only source of accurate, disaggregated data on these communities was the decennial census (rather than the annual ACS, which is a census estimate based on samples). The decennial census, with its race and ethnicity questions, provides the most detailed and complete picture of our communities. 

When programs aren’t tailored, health inequities stay

The National Institute of Child Health and Human Development launched the nationwide Back to Sleep campaign in 1994 with one message for all parents regardless of race. Ten years later, sudden infant death syndrome, or SIDS rates had “declined by about 50 percent,” but “the SIDS rate for African Americans is twice that of whites.” The campaign had helped everyone but left the racial gap in place. Reaching those families meant taking the message to specific communities, and knowing where Black families with infants lived required census data by race and age. Three decades later, the gap is wider, and only race data shows it. Black infants die of sudden unexpected causes at roughly three times the rate of white infants. A national count without race would show a modest rise and hide which families are losing babies. 

HIV prevention shows the same pattern. Among adults who should be taking pre-exposure prophylaxis, or PrEP, 5.9% of Black adults and 10.9% of Hispanic adults are using it, compared with 42.1% of white adults. Without the ability to track race data, a prevention tool like PrEP can look like it’s working while those at highest risk go without it. 

Why the ACS can’t fill the complete-count gap

The annual ACS can provide an idea of a community’s residents. It cannot replace the complete census count because it’s a sample of residents that is used to estimate the larger population, and it’s calibrated to the census. The National Center for Health Statistics uses those same estimates as denominators for birth and death rates by race, and a Centers for Disease Control and Prevention surveillance system weights its data by race and ethnicity to county population estimates that are based on decennial counts. 

Take race out of the 2030 Census, and these important public health data drift further from an actual count each year, with small groups drifting fastest. The Census Bureau itself cites this risk in the Preliminary Regulatory Impact Analysis (PRIA) attached to the proposed rule: “To produce and release these data [fertility/mortality rates, cancer rates, poverty rates, unemployment rates, labor force participation], the decennial census counts of the population by race and ethnicity are relied upon, as there are no other comparably accurate, comprehensive, and reliable sources of this information. Omitting race and ethnicity from the 2030 Census questionnaire would require the development of estimates from a base population derived from outdated and/or less accurate data and would deteriorate the accuracy of the estimation method itself.” 

The proposed rule points to the ACS as a fallback, noting that the bureau “is not proposing any restrictions on its ability to ask demographic questions” there. But this spring the bureau declined to update those questions, and without decennial race counts, the ACS loses the population totals it’s calibrated to. The rule also says, “data from administrative records can fill information gaps,” but administrative records are weak for public health use cases. Census Bureau researchers found that “minorities, especially Hispanics, are more likely to have non-matching Hispanic origin and race responses” in administrative and third-party records, and that “minority groups and individuals ages 0-17 are more likely to have missing race or Hispanic origin data.” When someone other than the person in question reports their race, inaccuracies occur; a Census Bureau study found that observers identified single-race American Indians and Pacific Islanders correctly only 62% of the time, and multiracial people 8% to 36% of the time.

Removing race from the 2030 Census removes the denominators health departments need to publish rate data. 

Two things public health practitioners can do before November 2

1. Comment and name your use case. A comment describing a concrete impact is harder to dismiss than a form letter. Say which rates you publish by race and ethnicity; which grant, needs assessment, or surveillance system depends on them; and what happens to that work without the denominator. 

Agencies are required to consider significant comments, and those comments become part of the record if the rule is challenged in court. In 2019, the Supreme Court blocked the citizenship question from the 2020 Census because the Census Bureau’s stated reasons didn’t hold up against the record. A comment with a concrete use case is evidence in the record that can change a rule. A form letter shows how many people care. 

2. Bring your institution. Ask your health department, your hospital’s community health needs assessment team, your school of public health, or any other group you are affiliated with to file its own comment. 

To get started, see the comment guides at Roadmap to 2030 and Hands Off Our Census. Submissions are due to Regulations.gov by November 2, 2026.