Skip to main content

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

Our take —

This study sought to estimate the degree of bias in the CDC-weighted estimates of COVID-19 mortality by race/ethnicity. The CDC approach, motivated to adjust for geographical clustering of COVID-19 deaths in racial/ethnic comparisons, inappropriately removed the impact of a major driver of racial/ethnic inequities in COVID-19 mortality, that of residential segregation, which is driven by historical and contemporary forms of structural racism. This study used the CDC-weighting methodology versus unweighted U.S. Census-based estimates, and found that the CDC-weighting likely overweights the number of non-white residents in a county, which then underestimates their mortality rates. This was illustrated by white people representing -8.1% of the proportion of COVID-19 deaths compared to their US Census proportion in the overall population, but being 10.9% overrepresented in the CDC-weighted estimates. This is an important “proof of concept” paper showing that using the appropriate denominators is extremely important to obtaining accurate COVID-19 mortality estimates by race/ethnicity.

Study design

Cross-Sectional, Ecological

Study population and setting

In lieu of reporting COVID-19 mortality rates by race/ethnicity, the U.S. Centers for Disease Control (CDC) reports rates alongside a weighted population distribution, using weights based on the distribution of COVID-19 deaths. The purpose of this study was to understand whether using this weighted distribution, instead of that obtained from the U.S. population, obscures the magnitude of racial/ethnic inequities. The study team obtained aggregate data from the beginning of the pandemic through May 13, 2020, which included 54,861 COVID-19 deaths, along with the CDC-derived weights, from the CDC’s National Center for Health Statistics. The national distribution of COVID-19 deaths was compared to the national weighted distribution of racial/ethnic populations (based on share of deaths) versus the actual distribution from the U.S. Census.

Summary of Main Findings

The study found that the CDC weighting by percentage of total COVID-19 deaths inflates the proportion of non-white residents. Using the census population distribution, Black people account for >9.9% of COVID-19 deaths compared to their proportion of the population, while conversely, white people were underrepresented by about -8.1%. The weighted data suggested the opposite, with white people accounting for 10.9% more and Black residents overrepresented in the CDC-weighted estimate, but only by 4.2%. Latinx people were underrepresented in the unweighted (-1.7%) and the CDC-weighted (-10.2%), and for Asian individuals an inflation of 0.1% vs. a deflation of -5.7%, respectively. The study concluded that overall, because the CDC weighting system often inflates the proportion of non-white residents in the population, it then deflates their mortality rate.

Study Strengths

This study uses publicly available data to estimate the mortality rate for race/ethnicity categories in the US, across all counties. Using both the US Census and the CDC-weighting, it calculates both absolute differences and ratio differences. By using the whole population for each county, it is representative of the whole US.

Limitations

The study does not present confidence intervals or error estimates, which means it is unclear how much uncertainty there may be in these estimates.

Value added

This is an important proof of concept that the CDC weighted standardized mortality estimates that use this weighting method are likely biased, and underreport the mortality risk in Black, Native American/Indigenous, Latinx, and populations of color.

This review was posted on: 17 August 2020