COVID Testing in the US: the Danger Ahead
Weekly Highlights of the Geography of COVID by the US COVID Atlas Team
While the world awaits an effective tool (be it a vaccine or a drug) to address the public health challenge that arises with the pandemic, testing remains an integral part of identifying and controlling the spread of COVID-19.
The Importance of Testing
The World Health Organization (WHO) has encouraged countries to do as much testing as possible. The more tests that are conducted, the easier it is to track the spread and reduce transmission. Testing paired with contact tracing, and early detection of hotspots is essential to curbing the spread of the pandemic. While the US might be towards the top in terms of the ratio of the population tested for COVID-19 (4th in the world, 138K tests per million population, according to worldometer), it also has the largest amount of people infected in the world.
A third of people surveyed by a Business Insider poll conducted in late April said they thought they’d had the disease; just 5 percent of those people were able to get tested. Currently, with the sunbelt states (states across the southwest and southeast) experiencing a surge in cases, states are also struggling to keep up with testing. News reports pointed out that those states have cars lining up for tests, and the wait is so long that tests are “yielding results so far after the fact that they’re useless.”
Testing has also been attributed to as a key reason why other countries are doing a better job at curbing the spread of COVID-19. South Korea managed to have new cases of single digits per day even without an extensive lockdown, and their extensive testing has been appraised, with over 100 tests per confirmed case as of July 15th. The US is only doing 12 tests per confirmed case to date. This trend is also evident in other Asian and other Oceanic countries, including Australia and New Zealand, who are conducting well over 100 tests per confirmed case. Given these statistics and the current surge in cases, there is a concern that at least parts of the US are falling behind on testing.
Looking at Available Testing Data
Currently, the most commonly-cited data source for US testing data is the COVID Tracking Project, which reports testing data at the state level. This data source aggregates testing information from the state health department websites. Third-party data sites like the COVID Tracking Project are unique to the data reporting of the US as there is no official data source at the federal level. CDC uses state health departments’ data, coupled with third-party information sources to maintain their database.
Each state has its own way of reporting testing data. Some states report the total number of tests performed (for example, New Mexico is reporting a “total tests” number), and others will distinguish between serology/antibody and PCR tests (states such as Michigan and Georgia are reporting both PCR and antibody testing numbers; New Jersey and other states are not reporting antibody testing data, but are noting testing numbers as PCR tests only.) Some states also separate tests performed in government labs and those performed in commercial labs. These could all lead to confusion in determining what “tests performed” actually means.
There is currently no comprehensive source for testing data at the county level. County level data is important because it gives more detailed local information, helping officials pinpoint places where resources are most needed. The Corona Data Scraper and Worldometer provide county-level testing statistics for some areas, but those sources are not complete. Many states’ health department websites do provide testing information for counties, although the data would require manual-scraping and cleaning. The existence of such data, however, creates the possibility of county level analysis.
How Much Testing is Good Enough?
Different parts of the country have varying infection rates and population density. A place where more cases are reported would inevitably need more testing to make sure that they are tracking as many people as possible. Rather than looking at the proportion of the population that has been tested, a more common metric for the scope of testing is “testing positivity”. This is a concept that the WHO has been using to guide testing, and is implemented in COVID Act Now, a COVID risk warning system designed by data scientists and medical experts from Georgetown University, Stanford University, Grand Rounds and beyond. This “testing positivity” indicator is essentially a positive test rate, with a lower rate meaning better testing is being implemented.
In COVID Act Now, this metric is calculated by:
Some US states report data on a bi-daily or even weekly basis. When tracking COVID cases, most states also show a weekly oscillating cycle. Using a 7-day trend measure is an intuitive choice.
The WHO mandates a 10% maximum positive test rate for effective testing, and many countries who were successful at containing the outbreak were achieving rates much lower (e.g. South Korea, with a testing positive rate of 1.4%). The COVID Act Now project established the following thresholds for positivity rates which can be useful in decision-making:
Below 3%: Low and Safe
Between 3–10%: Medium risk
Between 10–20%: High risk
Above 20%: Critical
Using COVID Tracking Project’s state level data, incorporating this positivity rate method, the status quo of testing positivity (as of July 17th) in the US shows that the majority of the country is at a level of “medium” risk and above (see Figure 1).
Much of New England, Illinois, Michigan, and Hawaii have the lowest positivity rates at 3% and below. Much of the South — from Texas through Louisiana, Arkansas, Alabama, Mississippi, Georgia, South Carolina, and Florida — have high risk positivity rates above the accepted WHO threshold of 10%. Notably, these areas are also in persistently high and/or emerging statistical hotspot regions of COVID previously reported on by the Atlas team.
County level data is limited, but for the approximately 650 counties that have data readily available (either from the Corona Data Scraper, or Worldometer), their cumulative testing positivity statistics are depicted in Figure 2, categorized by the thresholds detailed before. (The 7-day statistic is not used here because of different datasets not all providing historical data.)
This county level map further confirms the state testing trends, with some more detailed information:
- Critical positive test rates counties confirm known emerging hotspots of Florida, Arkansas, Texas, and Southern California;
- Yakima county and Franklin county in Washington are showing critical positive test rates, confirming our discovery about rural Washington in last week’s post;
- Pennsylvania could be on a “high risk” threshold, with overall medium positive test rates, and high risks on the eastern edge along the state border;
- Regionally, Illinois, Wisconsin, New York and Northern California have lower positive test rates, but these states also have medium or high risk counties. This once again illustrates the need for county-level data to identify and isolate higher risk counties before they progress any further.
Testing is essential to mitigating harms of the pandemic, along with effective contact tracing programs and early hotspot detection. Using a positivity rate metric to capture testing capability and preparedness can be helpful to estimate the spread of COVID-19. Getting access to county-level testing data, despite its challenges, remains crucial. To better enable county level discovery, the COVID Atlas project team is working on a more comprehensive county level testing dataset, combining existing sources with our manually scraped state health department website data, expected to be released later this summer.
- Ryan Wang is a Spatial Data Science Summer Fellow at University of Chicago.
- Geographical analysis uses COVID Tracking Project, Corona Data Scraper, and Worldometer data.
- All analyses and data are as of July 17th, 2020.
