Over the past two years, the official count of coronavirus deaths in the United States have increased and are now approaching 1 million lives. A large majority of Americans, they say personally know someone who has been hospitalized or died of coronavirusand it has affected – to varying degrees – almost every aspect of life.
A new Pew Research Center analysis of official reports of COVID-19-related deaths across the country, based on mortality data collected by The New York Times, shows how the dynamics of the pandemic have changed over the past two years.
A timeline for the changing geography of the pandemic
The Pew Research Center performed this analysis to understand how the geography of the coronavirus outbreak has changed over the course of its course. For this analysis, we relied on official death reports attributed to the new coronavirus collected and maintained by New York Times.
The estimates in this report are subject to several sources of error. There can be significant differences between the true number of deaths due to COVID-19 and the officially reported counts of these deaths. There may also be variation across states in the quality and types of reported data. For example, most states report deaths based on the deceased’s residence rather than the place where they died. The New York Times collects data from many different local health agencies, and this is likely to lead to some additional measurement errors.
This analysis is based on county-level data. Counties in the United States vary greatly in their population sizes, so in many places in the essay we divide counties into approximately equal groups (in terms of their population) to be able to compare or report population-corrected death rates rather than total death rates: Deaths.
The pandemic has rolled across the United States unevenly and in waves. Today, the death toll from the pandemic looks very different from what it looked like in the country early 2020. The first wave (approximately the first 125,000 deaths from March 2020 to June 2020) was largely geographically concentrated in the Northeast and especially the New York City region. During the summer of 2020, the largest proportion of the approximately 80,000 deaths that occurred during the second wave of the pandemic were in the southern parts of the country.
The autumn and winter months of 2020 and early 2021 were the deadliest of the pandemic to date. More than 370,000 Americans died of COVID-19 between October 2020 and April 2021; the geographical boundaries that characterized the earlier waves became much less pronounced.
By the spring and summer of 2021, the nationwide death rate had dropped significantly, and vaccines were widely available to all adults who wanted them. But by the end of the summer, the fourth and fifth waves (marked by new variants of the virus, the delta and then the omicron) came in rapid succession, claiming more than 300,000 lives.
In many cases, characteristics of societies that were associated with higher death rates at the beginning of the pandemic are now associated with lower death rates (and vice versa). Early in the pandemic, urban areas were disproportionately affected. During the first wave, the coronavirus death rate in the 10% of the country living in the most densely populated counties was more than nine times higher than the death rate among the 10% of the population living in the least densely populated counties. In each subsequent wave, however, the country’s least densely populated counties have recorded higher death rates than the most densely populated places.
Despite the staggering death toll in densely populated urban areas in the first months of the pandemic (average 36 monthly deaths per 100,000 inhabitants), the total death rate during the pandemic is slightly higher in the least populated parts of the country (an average of 15 monthly deaths per 100,000 among the 10% living in the least densely populated counties against 13 per 100,000 among the 10% in the most densely populated counties).
As the relationship between population density and coronavirus death rates has changed during the pandemic, the relationship between counties’ voting patterns and their COVID-19 death rates has also changed.
In the spring of 2020, the areas that recorded the largest number of deaths were much more likely to vote Democratic than Republican. But by the third wave of the pandemic, which began in the fall of 2020, the pattern was reversed: Counties that voted for Donald Trump rather than Joe Biden suffered significantly more deaths from the coronavirus pandemic than those who voted for Biden rather than Trump. This reversal is likely to be the result of several factors, including differences in remedial efforts and vaccine intake, demographic differences, and other differences correlated with county-level bias.
During this third wave – which continued into early 2021 – the coronavirus death rate among the 20% of Americans living in counties that supported Trump with the highest margins in 2020 was about 170% of the death rate among one in five Americans living. in counties that supported Biden by the largest margins.
As vaccines became more widely available, this discrepancy between “blue” and “red” counties became even greater as the virulent delta strain of the pandemic spread across the country during the summer and fall of 2021, even when Total the number of deaths fell somewhat from its third wave peak.
During the fourth wave of the pandemic, the death rates in the most pro-Trump counties were about four times what they were in the most pro-Biden counties. When the highly transmissible omicron variant began to spread in the United States in late 2021, these differences narrowed significantly. However, the death rates in the most pro-Trump counties were still around 180% of what they were in the most pro-Biden counties in late 2021 and early 2022.
The cumulative effect of these divergent death rates is a major difference in the total number of deaths from COVID-19 between the most pro-Trump and most pro-Biden parts of the country. Since the pandemic began, counties representing the 20% of the population where Trump ran up to its highest margins in 2020 have experienced nearly 70,000 more deaths from COVID-19 than counties representing the 20% of the population where Biden did best. Overall, the death rate for COVID-19 in all ccounties Trump won in 2020 are significantly higher than those in counties Biden won (at the end of February 2022, 326 per 100,000 in Trump counties and 258 per 100,000 in Biden counties).
The partisanship in COVID-19 deaths widened as more vaccines became available
Partisan differences in COVID-19 death rates increased dramatically after the availability of vaccines increased. Unvaccinated people are at far higher risk of death and hospitalization from COVID-19, according to the Centers for Disease Control and Prevention, and vaccination decisions are strongly associated with bias. Among the vast majority of counties for which reliable vaccination data are available, counties that supported Trump with higher margins have significantly lower vaccination rates than those that supported Biden with higher margins.
Counties with lower vaccination rates recorded significantly higher death rates during each wave, with vaccines widely available.
During the fall of 2021 (roughly equivalent to the delta wave), about 10% of Americans lived in counties with adult vaccination rates lower than 40% in July 2021. The death rates in these low-vaccinated counties were about six times as high as death rates in counties where 70% or more of the adult population were vaccinated.
More Americans were vaccinated by the winters of 2021 and 2022 (roughly equivalent to the omicron wave), but nearly 10% of the country lived in areas where less than half of the adult population was vaccinated by November 2021. Death rates in these low-vaccinated counties were about twice what they were in counties that had 80% or more of their population vaccinated. (Note: The statistics here reflect the death rates in the county as a whole, not the rates of vaccinated and unvaccinated persons, although data at the individual level show that the death rates among unvaccinated persons is much higher than among the vaccinated.)
This analysis is based on official reports of deaths attributed to COVID-19 in the United States collected and reported by New York Times.
COVID-19 deaths in Puerto Rico and other U.S. territories are not included in this analysis. In addition, deaths without a specific geographical location have been ruled out.
Data were retrieved from the GitHub repository maintained by The New York Times on March 1, 2022 and reflect reported coronavirus deaths through February 28th.
There are several irregularities in the death data. Many localities submit their reporting on weekends and holidays. In addition to the rhythm of the reporting cycle, there are many instances where a site will revise the number of deaths downward (usually only by a small amount) or release a large batch of previously unreported deaths in a single day. The downgrades were identified and applied retrospectively in earlier days.
Large batches of cases were identified by finding days that increased by more than 10 deaths and were 10 standard deviations above the norm for a county within a 30-day window. Deaths reported in these abnormal batches were then evenly distributed over the days up to when they were released.
Population data for U.S. counties come from 2015-2019 American Community Survey estimates published by the Census Bureau (available via tidycensus package in R on Feb. 21). The vote share in 2020 for each county was purchased from Dave Leips Valgatlas (downloaded November 21, 2021).
The analysis looks at deaths among counties based on their poll in 2020. Counties were grouped into five groups with approximately equal populations. For analyzes that include voting in 2020, Alaska’s counties are excluded because Alaska does not report its election results at the county level. The table below provides more details.
This essay benefited greatly from thoughtful comments and consultations with many people around the Pew Research Center. Jocelyn Kiley, Carroll Doherty, and Jeb Bell provided invaluable editorial guidance. Peter Bell and Alissa Scheller contributed their visuals expertise, Ben Wormald built the card animation, and Reem Nadeem provided the digital production. Andrew Daniller drew careful attention to the quality check process, and David Kent’s attentive copy-editing eye brought clarity to some difficult concepts.
Main photo: Kent Nishimura / Los Angeles Times via Getty Images