key: cord-0756329-wvfw94n1 authors: Zelner, J.; Trangucci, R.; Naraharisetti, R.; Cao, A.; Malosh, R.; Broen, K.; Masters, N.; Delamater, P. title: Racial disparities in COVID-19 mortality are driven by unequal infection risks. date: 2020-09-11 journal: nan DOI: 10.1101/2020.09.10.20192369 sha: 5506a16ddf1d6de4673ada96071c2673e2e985ce doc_id: 756329 cord_uid: wvfw94n1 Background. As of August 5, 2020, there were more than 4.8M confirmed and probable cases and 159K deaths attributable to SARS-CoV-2 in the United States, with these numbers undoubtedly reflecting a significant underestimate of the true toll. Geographic, racial-ethnic, age and socioeconomic disparities in exposure and mortality are key features of the first and second wave of the U.S. COVID-19 epidemic. Methods. We used individual-level COVID-19 incidence and mortality data from the U.S. state of Michigan to estimate age-specific incidence and mortality rates by race/ethnic group. Data were analyzed using hierarchical Bayesian regression models, and model results were validated using posterior predictive checks. Findings. In crude and age-standardized analyses we found rates of incidence and mortality more than twice as high than Whites for all groups other than Native Americans. Of these, Blacks experienced the greatest burden of confirmed and probable COVID-19 infection (Age- standardized incidence = 1,644/100,000 population) and mortality (age-standardized mortality rate 251/100,000). These rates reflect large disparities, as Blacks experienced age-standardized incidence and mortality rates 5.6 (95% CI = 5.5, 5.7) and 6.9 (6.5, 7.3) times higher than Whites, respectively. We also found that the bulk of the disparity in mortality between Blacks and Whites is driven by dramatically higher rates of COVID-19 infection across all age groups, particularly among older adults, rather than age-specific variation in case-fatality rates. Interpretation. This work suggests that well-documented racial disparities in COVID-19 mortality in hard-hit settings, such as the U.S. state of Michigan, are driven primarily by variation in household, community and workplace exposure rather than case-fatality rates. Funding. This work was supported by a COVID-PODS grant from the Michigan Institute for Data Science (MIDAS) at the University of Michigan. The funding source had no role in the preparation of this manuscript. SARS-CoV-2 in the winter and spring of 2020, and where the epidemic has been marked by unmistakable racial and 23 socioeconomic iequality. Data 25 We used data from 73,441 people with confirmed and probable COVID-19 infections from the Michigan Disease 26 Surveillance System (MDSS) from March 8 th , 2020 through July 5 th , 2020. Each case in MDSS is tagged with a unique 27 identifier, sex at birth, age, race, the date their case was referred to MDHHS, whether the individual died, and the 28 date of death. In order to mitigate the impact of right censored deaths on our case-fatality rate estimates, 15 we 29 truncated the data at the 97.5% quantile of time to death from case referral date, or forty-six days. This results in 30 truncating all data with a case referral date later than May 20 th , 2020, after which our data comprise 58,428 individ-31 uals. 32 From this dataset, we excluded 25 cases that did not reside in Michigan or were missing a state of residence, 8,613 33 people for whom race or ethnicity was not recorded, and 27 people who did not have age recorded or had age 34 > 116 years old indicating entry errors. We combined 68 pairs of records that had duplicate patient identification 35 numbers, resulting in 34 fewer cases. Finally, we dropped 28 patients whose sex at birth was unknown, leading to a 36 final dataset of 49,701 people with a confirmed or probable COVID-19 infection, with known age, race or ethnicity, 37 state of residence, sex at birth, and state prisoner status. After filtering the individual-level MDSS case data, we binned age by 10 year intervals to age 80, while we combined 39 ages 80 and above in one bin. We also created the race/ethnicity categories of Black/African American, Latino, 40 Asian/Pacific Islander, Native American,Other, and White, where Other comprised the census category of 'other', 41 and mixed race people. In order to model per-capita rates of disease we used tract-level population data from the 42 2010 U.S. Census aggregated to the state level for Michigan to define the population in each age-sex-race stratum. COVID-19 cumulative indidence rates. To calculate age-specific rates of COVID-19 infection in each age ( ), sex 45 ( ), race ( ) bin, per 100,000 population, we fit a poisson regression model with a population offset term, where is the size of the population for the -th group from the 2010 U.S. Census. We included age x sex, age 47 x race, and sex x race interaction terms to capture the full spectrum of potential heterogeneity in our outcome data. 48 2 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . For all analyses of per-capita age-specific incidence rates, we 53 used a log-Gaussian prior distribution with a mean of 0 and standard deviation of 0.1. Age and sex standardization. To characterize racial disparities unrelated to differences in population age struc-55 ture or sex ratios, we employed a post-stratification approach to direct standardization. Specifically, we generated 56 weights for each age x sex bin in the data, using the marginal age and sex distribution for the state of Michigan. We 57 then weighted posterior draws from the incidence and case-fatality models described above to generate adjusted Table 1 shows per-capita case and mortality rates by race/ethnic group, as well as corresponding case-fatality rates. Notably, the raw incidence rate among all non-White groups is substantially higher than among Whites for all groups 71 identified in the data except for Native Americans. than Native Americans, the risk of COVID-19 infection was uniformly higher than for Whites. However, because 80 these groups have differing population age and sex distributions, standardization is necessary to ensure that these 81 reflect differences in risk rather than being a function of the distribution of population. Standardized incidence and mortality rates. population. These results suggest that although there are meaningful differences in case-fatality by race and age, 116 that the large raw and standardized disparities in COVID-19 mortality cannot be explained by case fatality rates 117 alone. One way to understand the relative importance of exposure vs. case-fatality on the disparate burden of mortality 119 by race is to examine the counterfactual scenario in which each race/ethnic group has the age-and sex-specific 120 COVID-19 incidence rate of the corresponding age/sex group among Whites but their original age-and sex-specific 121 case-fatality rates. When we do this, we find that this would result in a decrease of 83%, 95% CI = (82%,84%) of 122 deaths among Blacks, 63%, 95% CI = (53%,72%) among Latinos, and 59%, 95% CI = (46%,69%) among Asian/Pacific 123 Islanders. These results suggest that while differential case-fatality rates can account for some of the disparity in Black vs. White 125 mortality rates, the large majority of COVID-19 deaths among African-Americans in Michigan can be attributed to 126 the large differences in age-specific incidence illustrated in Fig 1. Similarly Table 2 : Age and sex-standardized incidence and mortality rates and corresponding rate ratios. The table shows incidence rates and mortality rates and 95 percent posterior credible inervals, as well as corresponding standardized incidence rate ratios (IRRs) and mortality rate ratios (MRRs). For all comparisons, the incidence and mortality rate among Whites is used as the reference group. 4 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . Figure 1 : Incidence of COVID-19 infection per 100K population by age and race. Dashed lines indicate the crude overall rate for each group, dotted lines indicate group-specific age and sex-adjusted rates. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . Figure 2 : COVID-19 case-fatality rate by age and race. Dashed lines indicate the crude overall rate for each group, dotted lines indicate group-specific age and sex-adjusted rates. 6 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . Figure 3 : Disparities in COVID-19 incidence and case-fatality rate by age and race. Dashed lines indicate the ratio in the crude overall rate for each group, dotted lines indicate group-specific age and sex-adjusted rate ratios. The solid gray line is a guide for assessing the strength of association. 7 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . https://doi.org/10.1101/2020.09.10.20192369 doi: medRxiv preprint risks by sex that vary over race/ethnic categories. However, the age-standardized results in Table 2 suggest this is 131 unlikely to be the case. We also explicitly examined sex-specific differences in incidence, overall mortality and case-132 fatality (See Figure S1 in the supplementary materials), and found no meaningful differences by sex or race, although 133 the risk of death from COVID-19 is significantly higher for men than for women across all groups, echoing other 134 findings. Our results highlight yawning gaps in COVID-19 incidence and mortality in Michigan that cannot be explained 137 away by differences in population age and sex composition. Our results also suggest that the stark differences in seeking of all risk averse individuals, it is quite likely that these patterns of excess death reflect underlying disparities Blacks/African Americans are 5 Times More Likely to Develop COVID-208 19: Spatial Modeling of New York City ZIP Code-level Testing Results Racial demographics and COVID-19 confirmed cases and deaths: A 210 correlational analysis of 2886 US counties Socioeconomic Deprivation, and 212 Hospitalization for COVID-19 in English participants of a National Biobank American Indian Reservations and COVID-19: 214 Correlates of Early Infection Rates in the Pandemic Racial Disparity in COVID-19 Deaths: Seeking Economic Roots with Census data Inequality in acute respiratory infection outcomes 218 in the United States: A review of the literature and its implications for public health policy and practice COVID-19 Cases and Deaths in Federal and State 221 County Jail Incarceration Rates and County Mortality Rates in the United States Residential segregation and the epidemiology of infectious diseases Infectious Fear: Politics, Disease, and the Health Effects of Segregation COVID-19 and African Americans A systematic review and meta-analysis of published research data on COVID-230 19 infection-fatality rates Besides population age structure, health and other demographic factors can contribute 232 to understanding the COVID-19 burden Assessing differential impacts of COVID-19 on black communities Potential Biases in Estimating Absolute and Relative Case-Fatality Risks during Outbreaks