key: cord-0863092-wza7brt0
authors: Jia, K. M.; Hanage, B.; Lipsitch, M.; Swerdlow, D. L.
title: Excess COVID-19 associated deaths among the unvaccinated population >= 18 years old in the US, May 30 through December 4, 2021
date: 2022-02-13
journal: nan
DOI: 10.1101/2022.02.10.22270823
sha: c54a35442e4bc29f5dca3b9857bf3a0e75de5953
doc_id: 863092
cord_uid: wza7brt0

Vaccines against SARS-CoV-2 were authorized at the end of 2020 and are effective in preventing deaths; however, many persons remain unvaccinated. We used weekly publicly available CDC data from 26 U.S. jurisdictions, which include COVID-19 death rates by age and vaccination status, to estimate the number of excess deaths that might have been averted by vaccination from May 30 to December 4, 2021 in unvaccinated persons >= 18 years old. We subtracted the death rate in the vaccinated from rates in the unvaccinated to estimate the death rate each week that could be attributable to non-vaccination and multiplied this rate difference by the number of people in the unvaccinated group for each age group and each week, to estimate the excess mortality among the unvaccinated. Then, we extrapolated the number of deaths due to non-vaccination in the 26 jurisdictions to the whole US population using 2020 census estimates. In the 26 participating jurisdictions, there were an estimated 83,400 excess deaths among the unvaccinated populations, from May 30 to December 4, 2021. The largest number of excess deaths occurred in those 65-79 years old (n=28,900; 34.7% of total), followed by those 50-64 years old (n=25,900; 31.1%). Extrapolated to the US population we estimated approximately 135,000 excess deaths during the study period in persons >= 18 years old. Our estimates are an underestimate of all excess deaths that have occurred since vaccine became available because our analysis period was limited to May 30 to December 4, 2021 and many excess deaths occurred before and after this period. In summary, we used retrospective data to provide an estimate of the substantial number of COVID-19 deaths among the unvaccinated illustrating the importance of vaccination to prevent further unnecessary mortality during this pandemic.

due to non-vaccination in the 26 jurisdictions to the whole US population using 2020 census estimates, after excluding those partially vaccinated [2] (Appendix III).

The 26 US jurisdictions had similar age structure to the overall US population and a similar cumulative incidence of COVID-19 per 100 persons. However, the 26 jurisdictions had a higher proportion unvaccinated compared to that in the US overall, when considering only the fully vaccinated and unvaccinated. ( Table 1 ).

In the 26 participating jurisdictions, there were an estimated 83,400 excess deaths among the unvaccinated populations, from May 30 to December 4, 2021. The largest number of excess deaths occurred in those 65-79 years old (n=28,900; 34.7% of total), followed by those 50-64 years old (n=25,900; 31.1%) (Appendix IV). We extrapolated the estimate to the US population after excluding those partially vaccinated based on the national vaccination rate data [3] , and estimated approximately 135,000 excess deaths during the study period in persons >18 years old. In our sensitivity analyses, the excess deaths varied by the assumed size of the unvaccinated population in the non-26 jurisdictions, which dropped to 119,000 were the population unvaccinated 30% lower and increased to 150,000 were it 30% higher for the rest of the country (Appendix V), and also varied by proportion partially vaccinated (Appendix VI).

Handling of partial vaccination is a challenge because the proportion partially vaccinated is not reported for the 26 jurisdictions and because of possible reporting lags and misclassification in the national dataset [3] . This was further evaluated in Appendix VI. We used data that represented over half of the total US population, and the 26 US jurisdictions have age structure and incidence similar to the national overall. In addition, rates of underlying conditions and prevention behaviors may vary between vaccinated and unvaccinated individuals and confer differing risks for acquisition of SARS-CoV-2 infection and mortality. Although we cannot be certain that all the excess deaths in unvaccinated could have been prevented with vaccination, subtracting rates of death among vaccinated from deaths in unvaccinated should partially account for deaths that would have occurred even if persons were vaccinated. Our estimates are an underestimate of all excess deaths that have occurred since vaccine became available because our analysis period was limited to May 30 -December 4, 2021 and many excess deaths occurred before and after this period. In addition, we only estimated direct effects of vaccination, however the number of excess deaths could be augmented by indirect effects since vaccines likely reduce transmission to others. In summary, we used retrospective data to provide an estimate of the substantial number of COVID-19 deaths among the unvaccinated illustrating the importance of vaccination to prevent further unnecessary mortality during this pandemic.

We acknowledge the CDC for curating and sharing the data for this analysis, with special thanks to Akilah Ali, Avnika Amin, Amelia Johnson, and Heather Scobie, for coordinating with individual states, establishing standardized methods, and collating the data. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Footnotes: a The partially vaccinated individuals were excluded from the data set on the 26 jurisdictions, so only those fully vaccinated and unvaccinated individuals counted towards the total. b From the US 2020 population estimates by year-age c Total number of unvaccinated divided by total number of fully vaccinated and unvaccinated, after excluding those partially vaccinated among those 18 years old or above, averaged across the study period d Cumulative incidence was calculated as the total number of new reported cases from May 30 to December 4 divided by the mean number of the population in the period (12 years old or above) in 28 jurisdictions (the 26 jurisdictions plus New York and North Carolina) e This estimate was obtained by dividing the total number of new reported cases during the study period by the US population estimates in August 2021 (including the <12 years old) f Included deaths occurring among those 18 or above with a positive specimen collection date from May 30 to December 4, 2021 g Included all-age COVID-associated deaths in the US with a date of reporting from May 30 to December 4, 2021. Note that this is not directly comparable to the figure from 26 jurisdictions (due to different age range and use of report date rather than test date). h Extrapolated based on the US 2020 population estimates for 18 years old or above, under the assumptions that the rest of the country had the same proportion unvaccinated as the 26 participating jurisdictions after excluding those partially vaccinated (Appendix II).

. CC-BY-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 13, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 13, 2022. ; https://doi.org/10.1101/2022.02.10.22270823 doi: medRxiv preprint

Data on COVID-19 associated mortality by vaccination status. Data on vaccination status and COVID-19 associated mortality are from the U.S. CDC, aggregated across participating US jurisdictions (CDC, 2022a). Included jurisdictions had to regularly link their case surveillance and immunization data with CDC. Twenty-six of the jurisdictions reported deaths among vaccinated and unvaccinated persons and are included in the analysis.

Vaccination status. Individuals were considered as fully vaccinated if they completed a primary series of FDA-authorized COVID-19 vaccines at least 14 days before the date of record (Scobie et al 2021) . The number of unvaccinated is estimated by subtracting the numbers of fully and partially vaccinated (which were determined by vaccine administration data) from the US 2019 population estimate . However, the partially vaccinated individuals were excluded from the data set (CDC, 2022a), and thus we excluded the partially vaccinated during extrapolation to the US total.

A COVID-19 associated death was defined by any person who had a positive documented SARS-CoV-2 diagnosis and died. The data set recorded deaths among vaccinated and unvaccinated individuals based on the week of a positive specimen collection, and for most cases this was a few weeks ahead of the date of death (CDC, 2022a). Methods of outcome ascertainment varied across jurisdictions: it was common to use vital records, while some jurisdictions combined it with provider reporting and/or case investigations (Scobie et al, 2021) .

For each week in the study period, we estimated the mortality rate difference, by subtracting the rate of deaths among unvaccinated individuals with that among vaccinated individuals, for each age group in each week. This measure will account for both the changing attack rate over the course and the proportion vaccinated that increased over time.

Mortality rate difference !"#,%##& = # deaths among the unvaccinated !"#,%##& # unvaccinated !"#,%##& − # deaths among the vaccinated !"#,%##& # vaccinated !"#,%##&

We then multiplied the mortality rate difference to the total number of unvaccinated individuals in the 26 jurisdictions to estimate the number of excess deaths.

Excess deaths !"#,%##& = Mortality rate difference !"#,%##& * # unvaccinated !"#,%##& Thirdly, we extrapolated the excess deaths to the whole US population, after excluding those partially vaccinated. The percentage partially vaccinated was obtained by subtracting the percentage fully vaccinated from the percentage vaccinated with at least 1 dose(s) by week, for each of the age groups: the 18-24, 25-39, 40-49, 50-64, 65-74 , and the 75+ years old (CDC, 2022b). We multiplied the total US resident population of each year-age with the factor (1 -%partially vaccinated), for each age group (U.S. Census Bureau, 2022). The excess mortality in the 26 jurisdictions was then extrapolated as follows is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 13, 2022. ; https://doi.org/10. 1101 Lastly, we summed the weekly number of excess deaths by age and by week to obtain the total number of excess deaths among the unvaccinated population from June to December 2021, which captured the delta wave. Footnote: a Extrapolated by age based on the US 2020 population estimate, assuming that the proportion unvaccinated in the remaining of the country was the same as the 26 participating jurisdictions, considering within the population of either fully vaccinated or unvaccinated only.

The excess mortality for the whole US population is given by Excess deaths in the US !""#,%&" = Mortality rate difference !""#,%&" * 7# unvaccinated in the 26 jurisdictions !""#,%&" + # unvaccinated in the non-26 jurisdictions !""#,%&" @ By allowing the number of unvaccinated in the non-26 jurisdictions to change by a percentage m%, we have

Excess deaths in the US !""#,%&" = Mortality rate difference !""#,%&" * 7# unvaccinated in the 26 jurisdictions !""#,%&" + (1 + A) % * # unvaccinated in the non-26 jurisdictions !""#,%&" E The resulting estimates are: is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 13, 2022. ; https://doi.org/10.1101/2022.02.10.22270823 doi: medRxiv preprint

In this sensitivity analysis, we extrapolated the excess deaths among the unvaccinated by considering different percentage changes in the proportion partially vaccinated in the whole US population. We allow it to change by p% for each scenario, as in:

Excess deaths among the unvaccinated in the US '(),*))+ = Excess deaths among the unvaccinated in the 26 jurisdictions '(),*))+ * (1 − % partially vaccinated * (1 + p)% ) Total US population '() Fully vaccinated + unvaccinated in the 26 jurisdictions '(),*))+

As seen from the plot below, the estimated excess deaths decreased if there were a higher percentage of partially vaccinated in the population, vice versa. Figure S4 . Estimated excess deaths among the unvaccinated population by varying the proportion partially vaccinated in the US is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 13, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 

Rates of COVID-19 Cases or Deaths by

Age Group and Vaccination Status Public Use Data (version date

COVID-19 Vaccination Demographics in the United States

Rates of COVID-19 Cases and Deaths by Vaccination Status

COVID-19 Incidence and Death Rates Among Unvaccinated and Fully Vaccinated Adults with and Without Booster Doses During Periods of Delta and Omicron Variant Emergence -25

Monitoring Incidence of COVID-19 Cases, Hospitalizations, and Deaths, by Vaccination Status

National Population by Characteristics