Opinion: Proposed U.S. census changes could be catastrophic for public health

In 2030, U.S. mortality and morbidity rates will suddenly jump. This will happen not because there is a new pandemic or a sudden rise in cases of existing diseases or injuries, or a sudden decline in access to health care. These jumps in rates will occur throughout the U.S., more in some states than others. Within states, these rate changes will be especially noticeable among four groups: college-age people, retirement-age people, people of color, and immigrants. Even more remarkable, these shocking changes in U.S. health statistics could be averted without any new treatments or changes in access to or quality of health care. What is the secret to prevention? Ensuring that there is an accurate 2030 U.S. decennial census. The time is short to carry out this prevention campaign. The threat is imminent and profound. On Sept. 10, the U.S. Federal Register published an extraordinary notice blandly titled “Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions.” It announced plans for three major unprecedented and untested changes to the U.S. decennial census, which together will reduce, selectively, the number of people counted — and where. Since rates are computed by dividing the number of cases by the population at risk in a specified time period, then by definition, if the number of cases stays the same but the size of the denominator shrinks, the rates will be higher. The initial window for comments was a scant 33 days, with comments due by Oct. 13. Public pushback has resulted in the comment period being extended to Nov. 2. What are the three critical changes being proposed? 1) Instead of counting the “whole number of persons in each state,” as commanded by the 14th Amendment of the Constitution, the count will be restricted solely to “citizens or lawful permanent residents of the US.” This excludes all foreign citizen residents who do not meet these criteria, including international students, work visa holders, and recognized refugee and aslyees, and also erases undocumented immigrants. 2) Instead of counting people in relation to their usual residence on Census Day (April 1 of the census year), the changed “rules of residence” will require counting people in relation to where they resided the most between Jan. 3 to April 1, as tied also to the address used in tax records. This will likely mean that college students will be counted as living at their parents’ address, instead of at college, and “snowbirds” and seasonal workers will be counted in the warmer states where they winter, not their primary non-winter residence. 3) Prohibit questions about race and ethnicity (race has been asked about since the first U.S. census in 1790) and prohibit any questions about sexual orientation (first appearing in the 2020 census in a question about household relationships). Have any of these changes been carefully tested, following usual protocols for reviewing any changes to the census? No. Are they consistent with the Constitution and its amendments? No. The only reason to make these changes is partisan politics. The Constitution allocates political power — specifically representation in the House of Representatives — and draws political districts in relation to decennial census counts of population size. Change the rules on who is counted where and when, and states’ population counts, political power, and political districts will change. The bet is that the proposed changes will provide partisan advantage to those proposing the change in rules. Altering the count will affect more than just denominators and the distribution of political power. The predictable data distortions will rupture any understanding of the realities of — and trends in — U.S. morbidity and mortality rates. They will also undermine the data used in federal funding formulas for program eligibility and resource allocation, even as removing people from the census count by political sleight of hand does not disappear their needs. Furthermore, the distortions will impair the conduct and use of population survey data involving economics, health, education, transportation, etc., by distorting the population-based sampling frames and weights on which all such surveys rely. The far-ranging potential harms are starkly revealed by the U.S. Census Bureau’s list of 50 common uses of census data for allocation of resources and planning, spanning every sector of the U.S. Those especially germane to health and health services include: “Planning for hospitals, nursing homes, clinics, and the location of other health services”; “Estimating the number of people displaced by natural disasters”; “Creating maps to speed emergency services to households in need of assistance”; “Facilitating scientific research”; and “Assessing the potential for spread of communicable diseases.” There is no substitute for the granular U.S. decennial census data painstakingly collected by the U.S. Census Bureau, in accord with their longstanding and now threatened mission to “count everyone once, only once, and in the right place.” Over the past 40 years, virtually all health outcomes tracked in the benchmark U.S. annual federal documents Health, United States and in the Healthy People objectives have used census population data to report rates for both the total U.S. population and by race/ethnicity, as well as occasionally by socioeconomic position. For those who think economic data are sufficient to identify who is at highest risk of poor health outcomes in the U.S. absent racial/ethnic data, think again. A case in point is maternal mortality. The U.S. stands as an outlier with rates higher than those in all other high-income countries. Reflecting intertwined histories of racialized and economic injustice, not only are there stark disparities by race/ethnicity and by educational level, but there are also racialized inequities within each educational level, and educational inequities within each racialized group. Moreover, these inequities interact: Maternal mortality rates are higher among Black women with a completed college education compared to white women with less than a high school diploma. Maternal mortality is a critical and preventable outcome — but prevention requires accurate data. Prohibit collecting data on race/ethnicity in the census, and the net result will be to render invisible longstanding and well-known American racialized health inequities, overall and by socioeconomic position, as well as their changes over time. For example, between 1966 and 1980, U.S. socioeconomic and racialized inequities in rates of death before age 65 shrank, only to rise thereafter. These patterns likely reflect the beneficial impacts of the 1960s Great Society programs, followed by the post-1980 conservative and neoliberal backlash. Erasing data does not erase need or harm, but instead is a time-disgraced method of avoiding accountability via the shameful ruse of “no data, no problem.” For these and many more reasons, the proposed changes to the 2030 U.S. decennial census must be blocked before they can ever be implemented. Everyone who cares about the health and wellbeing of the U.S. population and the integrity of data can submit comments by Nov. 2. Nancy Krieger is professor of social epidemiology and American Cancer Society clinical research professor at the Harvard T.H. Chan School of Public Health. She represents her own views and does not speak on behalf of her institution.