State-level data obscures important variations in how cities and counties experience COVID-19

A special team within The COVID Tracking Project has been focused on gathering, cleaning, analyzing, and reporting on COVID-19 data at the city, county, and ZIP code level since May 28, 2020. Our data collection by metro area and by race and ethnicity shows virus transmission patterns in finer detail. It also creates a public archive of city-based COVID-19 data for the many cities that provide only current information, with no historical time-series to show changes through time; this allows us to identify days when cases or deaths begin increasing at a higher rate. It also lets us compare cities and counties across the nation and pinpoint how certain neighborhoods and communities of color have been affected.

When we began collecting data for a small subset of metropolitan areas in late May, widespread protests were happening across the US, raising concerns that the public gatherings could become COVID-19 “spreader events.” We collected data on total COVID-19 cases and deaths in metropolitan areas with large and frequent protests to see if the data suggested that group gatherings were driving virus transmission. We’ll be publishing an in-depth look at what we found soon, but as a preview, we’re publishing the case and death data we collected for Minneapolis, where protests originated after the killing of George Floyd and went on daily for two weeks. When protests broke out in late May, Minneapolis represented 13 percent of the state’s COVID-19 cases. One month later, that percentage had not increased and, in general, the data did not reflect increased growth in either cases or deaths.

Beyond tracking cities with protests, we also began tracking new geographies in order to measure the effects of local government’s reopening plans. We currently have historical data for 64 cities and counties in 27 states, and are actively collecting from 53 at this time. Before we begin releasing more city-based data, we want to highlight a few things that have made this dataset so challenging to assemble and work with.

Why the data is so difficult to collect

Like US states and territories, cities and counties are allowed to individually decide how to report their COVID-19 data. Each locality determines which and how many metrics to present, what demographic nuances to report, and how to display them. Some cities, like Chicago, report this data through their own city health department. Some cities report in collaboration with the county health department, like San Antonio in Bexar County, Texas. Some states, like Virginia, report city-level data on the state’s COVID-19 dashboards. Our data-gathering processes take this variation into account as we standardize the data for analysis. 

When determining the sources to use for data collection, we start at the city level. If that data is unavailable, we gather information at the county level. While most cities are contained within one county, a city like Atlanta spans two counties and is reported via two different county-level sources and one city-level source, making it difficult to collect complete data.

Our city data research also uncovered some peculiarities in city and county reporting that has obscured critically important data. In the COVID Racial Data Tracker, examining the racial and ethnic makeup of US counties with the highest COVID-19 infection and death rates helped us understand more about how the spread of COVID-19 highlights racial disparities. We expected that our city- and county-level data would also provide useful information on where COVID-19 was disproportionately affecting communities of color. What we didn’t expect was that city and county data for a single area could tell dramatically different stories about the pandemic’s effects. One such case, in Wayne County, Michigan, underlines the need for detailed and complete COVID-19 data by metro area as well as county. 

Detroit and Wayne County, Michigan: how the exclusion of a city affects a county’s COVID-19 data

Detroit, where Black people constitute 79 percent of the city's population, became an early example of the disproportionate impact of COVID-19 on Black communities. The city of Detroit sits within Wayne County, representing 38 percent of the county’s total population. But Wayne County does not include Detroit when reporting cases and deaths on their COVID dashboard. Text in the lower left corner of the dashboard reads: “This dashboard represents cases for every Wayne County community aside from the City of Detroit.” This raises the questions: How does excluding Detroit affect the data? Does the geographically segregated reporting make the effects of COVID-19 on certain populations appear less severe?

On August 18, 2020, The COVID Tracking Project analyzed data presented by both Wayne County and Detroit. Detroit alone reported 1,962 cases per 100,000 people—about 600 more cases per 100,000 people than all of Wayne County except for Detroit. Yet, when stratified by race, Wayne County alone indicates a much higher case rate for Black people than Detroit. 

The data for Wayne County without Detroit suggests that Black people are about three times as likely to contract COVID-19 than all other residents of the county. Despite the high risk ratio¹ for cases, the numbers indicate that Black people who contract COVID-19 have an 18 percent lower risk of dying from it than white people.

If we add Detroit to Wayne County so that the data represents the county’s total population, the infection risk for Black residents drops, but the death rate for the county becomes much higher—and more disproportionately concentrated within the county’s Black population. This full county-level reporting reveals that Black people in Wayne County are twice as likely to contract COVID-19 and have a 10 percent increased risk of dying of COVID-19 compared to white residents of the county.

The drastic impact of Detroit on Michigan and Wayne County’s COVID-19 data illustrates the importance of granular and localized data reporting. By collecting this data, we can see where the virus is disproportionately impacting certain populations. On August 5, 2020, 60 percent of COVID-19 deaths among Black people in Michigan were from Wayne County if we include Detroit. If we exclude Detroit, as the county’s official reporting does, it appears that only 13 percent of deaths among Black people statewide statewide were from the county. The reasoning behind Wayne County’s decision to exclude Detroit, a city with nearly half of the county’s population, is unclear. This difference shows the county’s metro areas outside the city limits, and prevents the public from understanding COVID-19’s impact on the entire county, including Detroit.

Using our data: possible insights and potential pitfalls

Below you can find the data we compiled for Minneapolis, Wayne County, and Detroit. Note that our list of cities and counties has evolved since we began tracking, so some datasets begin later than others. Also, due to inconsistencies in city and county reporting, some days may be missing counts. 

We suggest that you pair our results with 2018 ACS five-year-estimates from the US Census Bureau to have a sense of the overall population of each county and city.

Notes:

Minneapolis data

We began tracking Minneapolis on May 29, 2020. Data from July 29 to August 17, 2020, is incomplete due to reporting inconsistencies. Minneapolis reports a combined racial category of “Asian and Pacific Islander,” which is different from the two separate categories reported by the US Census Bureau (“Asian” and “Native Hawaiian and Other Pacific Islander”). While we have combined these categories to try to get an accurate population denominator for Minneapolis’s “Asian and Pacific Islander” category, it is possible that there is some misclassification due to the differences in reported racial categories. 

Detroit and Wayne County data 

Data is from June 3 to August 18, 2020. Wayne County case and death data, broken down by race, age, and municipality, was collected daily from the Wayne County COVID-19 dashboard. Detroit case and death data, broken down by race and age, was collected daily, Monday through Saturday, from the Detroit COVID-19 dashboard.

The current Wayne County estimates (as they are reported by the county) do not include the city of Detroit. The ACS reports Wayne County values with Detroit included, while COVID-19 metrics are reported separately for Wayne County (without Detroit) and Detroit. Thus, Detroit’s population values must be subtracted out of Wayne County’s ACS estimates to have an estimate of the underlying Wayne County population without Detroit. 

Spreadsheet

CSV

Public Notes CSV

Authors: Nicki Camberg, Artis Curiskis, Catherine Pollack, Kara Oehler, and Aarushi Sahejpal

Data analysis: Catherine Pollack

Graphics: Júlia Ledur, Matt Worthington, and Clarissa Wong

Top illustration: Nicolas Wright

Wayne County and Detroit data entry analyst: Hannah Cummins

Minneapolis data entry analyst: Aarushi Sahejpal

Additional contributions: Eliot Brody, David Eik, and Sharon Wang

Editors: Hannah Birch, Alice Goldfarb, Erin Kissane, Joanna Pearlstein, Jessica Malaty Rivera, and Aviva Geiger Schwarz 

Acknowledgments

Many people at The COVID Tracking Project contribute to the research and data compilation efforts that make the City Data special project possible. Special thanks to the City Data team that compiles and cleans this data every day. The team and spreadsheets are managed by Nicki Camberg. Shift leads are Nicki Camberg, Aarushi Sahejpal, and Sharon Wang. Team leads are Artis Curiskis and Kara Oehler. Contributors include Sonya Bahar, Eliot Brody, Nicki Camberg, Hannah Cummins, David Eik, Rebecca Glassman, Hannah Hoffman, Pat Kelly, Jeffrey Ndubisi, Judith Oppenheim, Rick Palmer, Noah Parker, Catherine Pollack, Aarushi Sahejpal, Isabel Sepúlveda, Eva Sher, Ganda Suthivarakom, Katharine Teigen, Sharon Wang, Allysa Warling, and Jessie Zhang.