“Pandemic of the unvaccinated”? At midlife, white people are less vaccinated but still at less risk of Covid-19 mortality in Minnesota

Introduction

The introduction of effective vaccines for Covid-19 in early 2021 resulted in a general consensus, for a time, that future pandemic surges would lead to a “pandemic of the unvaccinated.”¹ If this supposition held, Covid-associated mortality by group would become highly reflective of vaccine status. However, limited data are available to allow for the evaluation of this supposition for different racial/ethnic groups, despite substantial evidence that vaccination status and mortality both differ dramatically by race/ethnicity across the US.^2,3 In the present study, we evaluated whether COVID mortality is reflective of vaccination rates for different racial/ethnic groups in Minnesota using death certificate data on all Covid-19 deaths from March 2020 to December 2021.

Methods

This analysis uses death certificate data from Minnesota, March 2020 through December 2021. We examined Minnesota because of the unique availability of near-real-time data on both vaccination status and Covid-19 mortality that are simultaneously separated by race/ethnicity and age and cover individual time periods within the pandemic. In contrast, national vaccination data that are race/ethnicity-specific are not separated by age. Minnesota also stands out for its prolonged and deadly surge of the Delta variant, which did not end until it was supplanted by the Omicron surge at the end of 2021.⁴ We leverage the resulting high mortality in Fall 2021 to explore mortality in specific racial/ethnic groups at key midlife ages, when many Covid-19 deaths occurred in the U.S.⁵

Deaths were defined as Covid-19 deaths if there was any mention of U07.1 on the death certificate. We calculated Covid-19 death rates by race/ethnicity for three pandemic periods corresponding to pre-vaccination (March 2020-January 2021), mid-vaccination (February-June 2021), and late-vaccination, Delta-dominated (July-December 2021) time periods. The “Delta period” also includes the very early stages of the Omicron surge in Minnesota. We defined “vaccinated” as “fully vaccinated,” which means two doses of an mRNA-based vaccine or one dose of adenoviral vaccine. We used “fully vaccinated” rather than “single-shot vaccinated” as our vaccination metric because full vaccination provides greater protection and is likely to be reported with higher data quality.⁶ Booster shot uptake is not available by race/ethnicity in Minnesota. Additional methodological considerations are outlined in the Appendix.

Results

By the end of 2021 in Minnesota, vaccination among white Minnesotans was outpaced by vaccination among Black, Indigenous, and People of Color (BIPOC) Minnesotans at midlife ages (ages 45-64). Yet in all age groups and in each phase of the pandemic, white mortality was substantially lower than mortality among Minnesotans of color (Figures 1A-1C).

• Download figure
• Open in new tab

Figure 1. Period-specific vaccination rates and Covid-19 mortality rates for white and BIPOC Minnesotans by age.

The lines in panels A-C depict the cumulative vaccination progress of each group over time as measured by the percentage of that race-specific, age-specific group that have completed their vaccine series (left axis). The bars in panels A-C depict the race-specific, age-specific mortality rates (right axis) during the pre-vaccination period, mid-vaccination period, and Delta period of the pandemic. Mortality rates are annualized to facilitate comparison across periods. Vaccination rates and mortality rates are presented for Minnesotans ages 19-44 (Panel A), ages 45-64 (Panel B), and ages 65+. Panel D shows how Covid-19 mortality rates increase with increasing age for BIPOC women, white women, BIPOC men, and white men. The dashed lines show the relatively lower age at which BIPOC groups experience the same mortality rates as white groups at ages 65-74.

White undervaccination at midlife ages is pronounced: at the end of 2021, “fully vaccinated” rates were 82% for BIPOC Minnesotans compared to only 70% for white Minnesotans (Figure 1B). Yet, the gap in BIPOC-white mortality at those same ages was extreme; for example, during the Delta period, for both men and women, BIPOC mortality at ages 55-64 was higher than white mortality at ages 65-74 (Figure 1D). In other words, despite relatively high vaccination rates, BIPOC populations as a whole had, not only greater Covid-19 mortality than same-age white populations, but Covid-19 mortality equivalent to more than 10 years of additional age (at white mortality levels).

At key midlife ages (45-64), Minnesota’s white population is its second-least vaccinated racial/ethnic group (following Native Americans) (Figure 2A). However, despite low vaccination rates, Minnesota’s white population aged 45-64 has lower mortality than all other racial/ethnic groups, which ranged, during the period dominated by the Delta variant, from 133% (Asian) to 497% (Native) of white mortality (Figure 2B). Due to the relatively old age of the white population even within this age group, in age- and sex-standardized rates, the disparities in this period are even starker, with other racial groups ranging from 166% (Asian) to 519% (Native) of white mortality (Figure 2C). Figure 2D underscores this discrepancy: there is a generally close relationship between racial/ethnic groups’ mortality during the Delta period and their vaccination rates at that period’s midpoint—with the striking exception of white Minnesotans, who are advantaged relative to even the highest-vaccinated group, Asian-Americans.

• Download figure
• Open in new tab

Figure 2. Period-specific vaccination rates and Covid-19 mortality rates at midlife (ages 45-64) by race/ethnicity in Minnesota.

Panel A shows the cumulative vaccination progress of each racial group over time at midlife as measured by the percentage of that group who have completed their vaccine series. Panels B-D show midlife Covid mortality; mortality rates are annualized to facilitate comparison across periods. Panel B depicts the race-specific mortality rates during the pre-vaccination period, mid-vaccination period, and Delta period of the pandemic. Panel C compares crude and age- and gender-standardized mortality rates during the Delta period. Panel D compares the ranking of racial groups, from worst to best performing, by percent unvaccinated and mortality rates during the Delta period. The vaccination data in Panel D is from the midpoint of this period (October 2, 2021) and the mortality rates cover the entire Delta period (July 2021 to December 2021).

Discussion

The present study found that despite lower vaccination rates (compared to all but Native Minnesotans), during the Delta surge and early Omicron surge in late 2021, white Minnesotans had lower Covid-19 mortality at midlife than did Black, Hispanic, Asian, and Native Minnesotans. We note two broad possible explanations for these puzzling results. One possibility is that racial inequity in Covid-19 mortality risk—due to differential transmission, comorbidities, or unequal medical access⁷—among the unvaccinated, the vaccinated, or both, may be so great that it simply overwhelms the differences in vaccination status. A second possibility is that findings may reflect vaccine differences within the “fully vaccinated” population, with people of color potentially less likely to have received booster shots and less likely to have received mRNA vaccines in their primary series.⁸ These possibilities reflect the continuing inequities of the pandemic even in the period of widespread vaccination.

Regardless of the precise mechanism, the findings suggest that the “pandemic of the unvaccinated” formulation is incomplete and that Covid-19 also remains a “pandemic of the disadvantaged.” If population mortality primarily reflected population vaccination rates, then we would have expected to find that white communities in midlife would have an equal or greater burden of Covid-19 mortality than communities of color. The fact that we did not observe such a finding indicates that structural racism, as manifested through systems and policies that affect health care access, occupational risk, and housing conditions, continue to fundamentally shape risk of Covid-19 mortality even in the Delta/Omicron period.^9–12

While a “pandemic of the unvaccinated” framing may be used as a rationale for current policy actions at the federal and state levels, which include phasing out vaccine and mask mandates and accelerating a “return to normal,” a “pandemic of the disadvantaged” framing emphasizes the need for sustained population-based Covid-19 prevention and mitigation strategies that center health equity. Such measures could aim to further increase vaccination with intensive communication campaigns organized with community partners, widespread access to walk-in vaccine clinics (including for those working atypical hours), paid sick leave facilitating vaccination in the context of feared side effects, and vaccine delivery programs targeting low-income neighborhoods, schools, and workplaces. Such measures might also aim to mitigate Covid-19 spread—through approaches including improved ventilation in workplaces and public buildings, paid sick leave to prevent work while ill, Medicaid expansion and universal health care, economic payments to marginalized and medically high-risk populations, protective equipment and increased pay (limiting the need to work at multiple jobs) for long-term care workers, eviction moratoriums and other housing support, continued mask mandates, and public funding for community testing programs and scientific research on Covid—all acknowledging that, even in contexts in which vaccine uptake among people of color is relatively high, the mortality burden of the pandemic remains unequally borne. The “pandemic of the disadvantaged” framing suggests that a sole emphasis on individual behavior will not be adequate for reducing health inequities at midlife.

The extent to which findings in Minnesota may resemble those of other state contexts is unclear. If vaccination rates are generally higher in metropolitan areas compared with rural areas, other states with very urban populations of color and large rural white populations may show similar vaccination disparities. At the national level, aggregated over all ages, the white population does not have a lower vaccination rate than all BIPOC communities, although in most states, white vaccination is lower than the high average age of white populations would predict.¹³ However, the lack of publicly-available data on the age composition of vaccine status by race/ethnicity for the U.S. as a whole limits the ability to know how widespread the midlife patterns identified here may be.

Conclusions

Recent research emphasizes the exceptionally young age distribution of Covid-19 deaths in the United States relative to other countries.^14,15 The results in the present study highlight how this distinctive risk at midlife may be intertwined with the deep inequality in U.S. Covid-19 mortality. Populations of color may be at notably high risk at midlife—even when they have greater vaccination rates than white people at the same ages.

Data Availability

Data are available at https://osf.io/bxjkh/?view_only=80e912ce2b624ab7a78e51754c3952e3

Methods Appendix

The death certificates analyzed were those included in a data release from MDH on May 1, 2022. Among 2021 Covid-19 deaths included in the death certificate database, 91% were added to these regularly-updated data within three weeks following the date of death. Thus, these data are likely close to complete for 2021 deaths.

To create death rates from the death certificates, we used 2020 population estimates from the National Center for Health Statistics (NCHS), as the most recently available estimates. We chose not to project denominators for 2021, instead using official 2020 estimates, since 2020 population dynamics (shaping 2021 populations) are so distinctive relative to the dynamics of prior years. In order to report death rates across periods of unequal length (e.g., an 11-month pre-vaccination period vs. a 6-month Delta period), we annualized all rates. To construct age- and sex-standardized death rates at midlife for Figure 2C, we performed a “direct” standardization that reweights each race/ethnic population’s mortality using a shared set of weights.¹⁶ As our “standard” population weights, we used the 5-year-age and sex distribution of the full 2020 Minnesota population at ages 45-64.

Vaccine data were reported at the adult age categories 19-44, 45-64, and 65+; our main mortality analyses accordingly used these groups. Some racial group populations are too small to examine separately for ages 65+ in Minnesota. Thus, we show individual racial groups at ages 45-64 (Figure 2) and dichotomous white vs. BIPOC when examining all age groups separately (Figure 1). Mortality analyses support the use of the BIPOC category because, in Minnesota, the primary mortality divide at all high-mortality ages, all time periods, and for both women and men is between white Minnesotans vs. all others.¹⁷ We report mortality rates using race-bridging for multiracial populations; vaccination rates are reported by the state with single-race populations. We use denominators constructed with the appropriate racial categories for these respective numerators.

Minnesota vaccination data are reported as population percentages using 2019 population denominators estimated by the Minnesota Department of Health (MDH). These MDH denominators arederived from the 2019 American Community Survey (ACS) 5-year estimates and Census Population Estimates Program (PEP) estimates. These denominators can result in unrealistically high vaccination estimates for growing subpopulations; for example, state data report that 100% of Black and Asian Minnesotans aged 65+ are fully vaccinated. We adjusted these vaccine rates using the ratio of 2020 population sizes (as estimated by the NCHS) to the MDH estimates. We note that, because state vaccination data for Black and Asian populations aged 65+ are top-coded at 100%, our adjusted estimates likely under-estimate vaccination for these populations; moreover, under-estimating these two groups will produce likely under-estimation in the BIPOC category as a whole since Black and Asian populations together constitute 69% of the 65+ BIPOC population. This under-estimation is conservative for our argument because it implies that the seeming similarity in vaccination among white and BIPOC populations aged 65+ (Figure 1C) may or may not disguise greater true vaccination among BIPOC populations.