Holiday Reporting Lags Interrupt Positive Trends: This Week in COVID-19 Data, Sep 17

This week in US COVID-19 data

Six months after the COVID-19 pandemic was officially declared, the United States has seen 188,802 deaths and 6.2 million cases. This week, cases and deaths were both up, at 4.7 and 17.9 percent, respectively. The sharp increase in reported deaths was partially the result of last week’s Labor Day holiday, which delayed reporting even more than usual, as well as a release of probable deaths data from Arkansas. 

Hospitalizations, however, continued their downward trajectory; 30,684 people on average were hospitalized with COVID-19 this week, about half the number seen in the second half of July.

Precise testing figures remain out of reach

While the US has yet to reach the testing levels recommended by the Harvard Global Health Institute, the number of tests administered in the past week did increase 2.6 percent. According to data collected by the COVID Tracking Project, the United States administered 5 million tests this week, one for every 66 people in the nation.

As Alexis Madrigal and Robinson Meyer wrote in The Atlantic this week, the decline in testing nationwide since the summer may be a question of test visibility rather than capacity. “So far, the U.S. has reported only about 200,000 antigen-test results,” Madrigal and Meyer wrote. “But some evidence suggests that these tests are being used on a much wider scale than is understood: Thousands, if not tens of thousands, of antigen tests may already be happening every day without their results appearing in any public data.” As a reminder, antigen tests are diagnostic tools that can help detect positive cases. While PCR tests remain the gold standard for diagnostic tools, antigen tests can help identify people when they are most infectious to others. This ability, combined with the comparatively lower cost and the quick turnaround of antigen tests, explains why these tools are becoming used more widely. However, antigen tests are less sensitive, and thus less reliable, than PCR tests. As a result, some states count positive antigen results only as probable cases—if they are counting them at all. 

Some data sources, like Carnegie Mellon’s COVIDcast tracker, are beginning to report large swaths of antigen testing across the US. As seen on the map below, antigen test positivity rates are generally under 10 percent, with some notable exceptions in Oklahoma, Tennessee, and Alabama.

Problem spots remain in the Midwest

Across many states in the Midwest, cases are again on the rise. For the majority of these states, tests are also rising—but a few, such as South Dakota and Wisconsin, are seeing case growth rapidly outpace new testing.

After a drop earlier this month, South Dakota’s cases are rising again, and some of the numbers are alarming: According to CMS.gov data from Aug 27–Sep 2, 23 counties in the Mount Rushmore state had test positivity rates above 20 percent. Test positivity rates that high strongly suggest that more people are carrying the virus than are being tested. Meanwhile, after a few days of comparatively low test positivity rates over the Labor Day weekend, Wisconsin’s cases grew this week by 65 percent.

Hospitalizations continue to decline

There is some good news: The number of currently hospitalized patients with COVID-19 fell in 39 states this week. Improvements were particularly rapid on the West Coast, as hospitalizations and new cases continued their decline in California, Nevada, Arizona, and Washington. In the coming weeks, we’ll have more insight into what effect, if any, the catastrophic wildfires in Oregon, Washington, and California have had on COVID-19 transmission.

Earlier this week, the COVID Tracking Project’s City Data team released the data we’ve collected for Minneapolis and Detroit and described some of the benefits—and pitfalls—of collecting data at the city and county level. As the team writes, city-level data can help us understand the effects of protests and other gatherings on COVID-19 case rates in ways that state-level data sometimes obscures. It can also help us find gaps and peculiarities in datasets that we might not otherwise understand, like what happens when Wayne County, Michigan, decides to omit Detroit’s COVID-19 data from its accounting of cases, deaths, and demographics. We’ll be releasing more city data, along with case studies, in the weeks to come.

To keep up to date on our work, follow us on Twitter and join our low-frequency email list.