[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure.  Examples of US Centers for Disease Control and Prevention (CDC) Geographical Population Weighting
Examples of US Centers for Disease Control and Prevention (CDC) Geographical Population Weighting

The figure shows examples of CDC geographical population weighting using a fictitious state with 3 counties (A) and actual CDC weighting of counties in New York State (B) by population, household crowding, and percentage of county population who are people of color (POC). The impact of the CDC’s method of geographical reweighting is demonstrated by juxtaposing the hypothetical example in panel A with actual county population data in panel B. By up-weighting counties such as county A (eg, Bronx), down-weighting counties such as county C (eg, Albany), and excluding counties such as county B (eg, Saratoga), the CDC inflates the proportion of residents of color in the weighted population, making their risk of death appear lower, while deflating the proportion of White residents, making their risk of death appear greater.

Table.  Percentage Distribution by Race/Ethnicity for COVID-19 Deaths, CDC-NCHS–Weighted Population, and US Census Population and Absolute and Relative Differences Using Data as of May 13, 2020
Percentage Distribution by Race/Ethnicity for COVID-19 Deaths, CDC-NCHS–Weighted Population, and US Census Population and Absolute and Relative Differences Using Data as of May 13, 2020
1.
Krieger  N, Gonsalves  G, Bassett  MT, Hanage  W, Krumholz  HM. The fierce urgency of now: closing glaring gaps in US surveillance data on COVID-19. Health Affairs Blog. Published April 14, 2020. Accessed July 7, 2020. doi:10.1377/hblog20200414.238084
2.
National Center for Health Statistics. Provisional death counts for coronavirus disease (COVID-19). Published May 5, 2020. Accessed May 8, 2020. https://www.cdc.gov/nchs/nvss/vsrr/covid19/tech_notes.htm
3.
Gee  GC, Ford  CL.  Structural racism and health inequities: old issues, new directions.   Du Bois Rev. 2011;8(1):115-132. doi:10.1017/S1742058X11000130PubMedGoogle ScholarCrossref
4.
Acevedo-Garcia  D.  Residential segregation and the epidemiology of infectious diseases.   Soc Sci Med. 2000;51(8):1143-1161. doi:10.1016/S0277-9536(00)00016-2PubMedGoogle ScholarCrossref
5.
Sewell  AA.  The racism-race reification process: a mesolevel political economic framework for understanding racial health disparities.   Sociol Race Ethnicity. 2016;2(4):402-432. doi:10.1177/2332649215626936Google ScholarCrossref
6.
Roy  A, Brooks  O, Boyd  R. Health care inequality: confronting racial and ethnic disparities in COVID-19 and the health care system (virtual hearing). House Energy and Commerce Committee of the 116th US Congress. Published June 17, 2020. Accessed July 7, 2020. https://www.congress.gov/event/116th-congress/house-event/110812?s=1&r=20
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    1 Comment for this article
    EXPAND ALL
    Weighted Versus Unweighted Assessments of COVID-19 Death Inequities
    Michael McAleer, PhD(Econometrics),Queen's | Asia University, Taiwan
    The critical and incisive analysis by public health experts distinguishes carefully between weighted (or "indirect standardization") and unweighted data in assessing inequities inCOVID-19 surveillance and mortality data according to race and ethnicity data provided by the US CDC.

    The weighted data are given relative to COVID-19 deaths, while the unweighted are given relative to the total population.

    If the shares of COVID-19 deaths for each country are proportional to the total population across geographic, racial and ethnic distributions, there would be no need for weighting.

    As the publicly available aggregated cross section data were analysed relative to two population distributions, unless
    the two populations are proportional to each other across the cross sections, it is a foregone conclusion that at least one, if not both, of the weighting structures, which were reported as unchanged between May 13 and July 7, 2020, will lead to biased estimates of the unweighted distribution.

    This will undeniably bias, upward or downward, according to race and ethnicity, the estimated inequities according to race and ethnicity across countries, which will mitigate against correcting the inequities through appropriate allocation of healthcare resources.

    The use of such arbitrary weighting densities begs the question as to why such weights were ever considered, let alone used, regardless of the questionable explanatory statement given by the US CDC.
    CONFLICT OF INTEREST: None Reported
    READ MORE
    Views 8,841
    Citations 0
    Research Letter
    Public Health
    July 28, 2020

    Comparison of Weighted and Unweighted Population Data to Assess Inequities in Coronavirus Disease 2019 Deaths by Race/Ethnicity Reported by the US Centers for Disease Control and Prevention

    Author Affiliations
    • 1FXB Center for Health and Human Rights, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
    • 2Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
    • 3Population Health Sciences Department, Harvard Graduate School of Arts and Sciences, Cambridge, Massachusetts
    • 4Department of Social and Behavioral Sciences, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
    JAMA Netw Open. 2020;3(7):e2016933. doi:10.1001/jamanetworkopen.2020.16933
    Introduction

    Surveillance and mortality data show large inequities in the impact of coronavirus disease 2019 (COVID-19) by race/ethnicity.1 Currently, the US Centers for Disease Control and Prevention (CDC) does not report mortality rates by race/ethnicity. Instead, the percentage distribution of COVID-19 deaths by race/ethnicity is presented alongside a weighted distribution of the population from the CDC’s National Center for Health Statistics,2 which weights each county’s population by its share of COVID-19 deaths, not population (Figure). We investigated whether the resulting magnitude of inequities using the weighted population underestimates those observed using the total population (unweighted).

    Methods

    This cross-sectional study used publicly available, aggregated data downloaded May 13, 2020.2 Because the data were deidentified, institutional review board approval and informed consent were not required, in accordance with 45 CFR §46. This study follows the relevant portions of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.

    We compared the distribution of COVID-19 deaths by race/ethnicity with 2 separate population distributions provided by the CDC: National Center for Health Statistics weighted population and US Census unweighted population. Data analysis was performed from May to June 2020 using R statistical software version 3.6.3 (R Project for Statistical Computing).

    Results

    In total, 54 861 COVID-19 deaths were reported as of May 13, 2020. Applying the US Census population distribution, Black individuals were the most overrepresented among COVID-19 deaths, accounting for 9.9% greater than their share of the US Census population, whereas White individuals were underrepresented (−8.1%). In contrast, comparisons with the weighted data suggest that White individuals are most overrepresented among COVID-19 deaths (10.9%) (Table). Discrepancies were also noted when comparing deaths with the unweighted vs weighted populations among Latinx (−1.7% vs −10.2%) and Asian (0.1% vs −5.7%) individuals (Table).

    The CDC’s weighting approach inflates the proportion of residents of color in the weighted population, as shown in our hypothetical example in panel A of the Figure, where the state’s true population is 30% people of color, but the CDC’s weighted population is 46.7% people of color. For example, in New York, large urban counties with higher percentages of crowded households and residents of color are weighted more heavily compared with their share of the population than smaller, suburban, and rural counties, where residents are predominantly White, as shown in panel B of the Figure.

    Discussion

    Use of the CDC’s weighted population distributions to evaluate racial/ethnic inequities in COVID-19 mortality underestimates the excess burden of COVID-19 among Black and Latinx individuals compared with analyses conducted using the total population (unweighted) in the US Census data. According to the CDC, weighting was conducted because “COVID-19 deaths are concentrated in certain geographic locations where the racial and ethnic population distribution differs from that of the United States overall.”2

    The indirect standardization procedure implemented by the CDC is misleading and obviates a key mechanism by which structural racism operates to produce health inequities: social segregation.3 The CDC approach heavily weights large, urban counties because of their high proportion of COVID-19 deaths (eg, New York City) and excludes counties without any COVID-19 deaths (Figure). In effect, the CDC treats the geographical clustering of COVID-19 deaths as a nuisance parameter that must be controlled for to accurately compare the distribution of deaths across racial groups in the same geographical areas. However, the same mechanisms that pattern the geographical distribution of COVID-19 mortality also operate to produce racial/ethnic inequities in mortality.

    From macrogeographical regions to microneighborhoods within cities, structural racism has determined the distribution of Black, Latinx, and Native American communities and is a key mechanism that produces and maintains inequities in infectious disease outcomes.3-5 Specifically, historical and contemporary policies and processes, including land theft, racial terrorism, redlining, and gentrification, determine the location, quality, and density of residence for people of color.3,5 Consequently, Black and Latinx individuals are clustered in the same high-density, urban locations hardest hit in the first months of the pandemic, with these areas weighted most heavily by the CDC’s procedure (Figure). By adjusting for the geographical distribution of racial groups, the CDC effectively compares inequities that would remain had all racial and ethnic groups lived in the same geographical areas. Controlling for this major pathway understates COVID-19 mortality among Black, Latinx, and Asian individuals and overstates the burden among White individuals.

    This study is limited by the fact that conclusions comparing inequities in weighted and unweighted populations may change as the epidemic evolves. However, as of July 7, 2020, the CDC’s weighting method remains unchanged.

    In summary, the CDC’s presentation of data on race/ethnicity and COVID-19 deaths is misleading, with consequences for resource allocation for mitigating health inequities.6 We urge the CDC to drop the misleading weighted counts and publish mortality rates per race/ethnicity group stratified by age, gender, education, and ZIP code characteristics1 to adequately equip epidemiologists and policy makers with the data to mitigate inequities.

    Back to top
    Article Information

    Accepted for Publication: July 3, 2020.

    Published: July 28, 2020. doi:10.1001/jamanetworkopen.2020.16933

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Cowger TL et al. JAMA Network Open.

    Corresponding Author: Tori L. Cowger, MPH, FXB Center for Health and Human Rights, Harvard T.H. Chan School of Public Health, 677 Huntington Ave, Boston, MA 02115 (vcowger@g.harvard.edu).

    Author Contributions: Ms Cowger and Mr Makofane had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Ms Cowger, Ms Davis, Ms Etkins, Mr Makofane, and Ms Lawrence contributed equally to this work.

    Concept and design: Cowger, Davis, Etkins, Makofane, Bassett, Krieger.

    Acquisition, analysis, or interpretation of data: Cowger, Davis, Etkins, Lawrence.

    Drafting of the manuscript: Cowger, Davis, Etkins, Makofane, Lawrence, Krieger.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Cowger, Davis.

    Administrative, technical, or material support: Davis, Etkins, Lawrence.

    Supervision: Bassett, Krieger.

    Conflict of Interest Disclosures: None reported.

    Additional Contributions: Jarvis Chen, ScD (Department of Social and Behavioral Sciences, Harvard T.H. Chan School of Public Health), provided thoughtful comments on this article and trained Ms Cowger, Ms Davis, Ms Etkins, Mr Makofane, and Ms Lawrence in social epidemiology.

    References
    1.
    Krieger  N, Gonsalves  G, Bassett  MT, Hanage  W, Krumholz  HM. The fierce urgency of now: closing glaring gaps in US surveillance data on COVID-19. Health Affairs Blog. Published April 14, 2020. Accessed July 7, 2020. doi:10.1377/hblog20200414.238084
    2.
    National Center for Health Statistics. Provisional death counts for coronavirus disease (COVID-19). Published May 5, 2020. Accessed May 8, 2020. https://www.cdc.gov/nchs/nvss/vsrr/covid19/tech_notes.htm
    3.
    Gee  GC, Ford  CL.  Structural racism and health inequities: old issues, new directions.   Du Bois Rev. 2011;8(1):115-132. doi:10.1017/S1742058X11000130PubMedGoogle ScholarCrossref
    4.
    Acevedo-Garcia  D.  Residential segregation and the epidemiology of infectious diseases.   Soc Sci Med. 2000;51(8):1143-1161. doi:10.1016/S0277-9536(00)00016-2PubMedGoogle ScholarCrossref
    5.
    Sewell  AA.  The racism-race reification process: a mesolevel political economic framework for understanding racial health disparities.   Sociol Race Ethnicity. 2016;2(4):402-432. doi:10.1177/2332649215626936Google ScholarCrossref
    6.
    Roy  A, Brooks  O, Boyd  R. Health care inequality: confronting racial and ethnic disparities in COVID-19 and the health care system (virtual hearing). House Energy and Commerce Committee of the 116th US Congress. Published June 17, 2020. Accessed July 7, 2020. https://www.congress.gov/event/116th-congress/house-event/110812?s=1&r=20
    ×