Testing Data is Released for the US COVID Atlas
The latest state-level testing data and features now on the Atlas
A few months ago, the US COVID Atlas team initiated an effort to collect, analyze, and incorporate testing data into the Atlas’s map functionality. We previously posted on the usefulness of testing data, including using testing positivity to gauge whether an area is safe to reopen, and the inconsistencies within US testing data. We had also noted that we would incorporate the testing data into the US COVID Atlas map later in the year. At last, this feature is now live on our website at theuscovidatlas.org!
This post will detail some highlights of the testing data, explain the variables included, and some usage tips for this feature. Our team believes that testing data could become very useful for public health professionals and administrators in getting a better understanding of the pandemic and make more informed decisions.
Testing data features in the US COVID Atlas
Currently, the COVID Atlas provides state-level testing data in our map interface, based on data from the COVID Tracking Project. To access the data in the Atlas, select DATA SOURCE — “By State (UsaFacts.com)” and MAP VARIABLES — testing related variables, like 7 Day Testing Positivity Rate %, to reveal the testing data features (see the figure below.) The Atlas uses states’ testing numbers and positive test reports every day, averaged over a 7-day period, as our raw data inputs. Three variables are then presented to the user: 7 Day Testing Positivity Rate, 7 Day Testing Capacity, and 7 Day Confirmed Cases per Testing. Clicking on a state/region/territory, or hovering over the area, will reveal the tool-tip for that state, which is also updated with the exact numbers for the three testing variables.
Our website currently does not provide county-level testing data because of data integrity issues at the county level: many states are not reporting county-level testing data, and for the states that do, not all the counties have data available. Still, our team is scraping and storing available county-level testing data on a daily basis from a variety of data sources for future use.
New color schemes are adopted for more accurate visual representation when mapping out the testing variables (see figure below). Specifically, we use the darker blue to represent higher testing capacity, signaling better testing resources and a lower pressure for testing. For positivity rate variables, we use a blue to yellow color scheme, making the differences in positivity rates easier to identify on the map. Fixed bins are applied to testing variables. For positivity rate and positive case per testing, the bins are 3%, 5%, 10%, 15%, 20%, >25%. Note that 3%, 5%, and 10% are different standards adopted by various official agencies as the “safe” threshold of testing positivity rates. Like other variables, one can check testing variables in the right-side detailed data panel when clicking on a state, as well as looking into temporal changes by moving the time slider (the testing data are provided all the way back to March 2020).
How we calculate key metrics and scopes of measurement
The US COVID Atlas now shows three testing-related metrics by state: (1) positivity rates (seven-day average of the number of positive tests per number of tests done); (2) testing capacity, (seven-day average number of tests per 100,000 population); and (3) new confirmed cases per number of tests done (seven-day average new confirmed cases per number of tests).
There is currently no federal level report of testing data, and no federal standards have been established to report testing data. Hence, using data and methods from validated sources, our Atlas reports three testing related variables to reflect multiple aspects of testing. The state-level testing raw data is sourced and validated by the COVID Tracking Project, providing total testing numbers and positive test numbers. For consistency and a better interpretation, we report molecular (PCR) test numbers when available.
We also distinguish between criteria of testing reported, as states report total testing numbers and positive test numbers using different criteria: either by people (number of people who got tested), encounters (people tested with molecular tests, with multiple tests on the same person removed), or specimen (total number of tests conducted). Some states do not have this information available. This criterion is shown in the tool-tip for individual states. When calculating the “7 Day Testing Positivity Rate” variable, we prioritize the people-based criterion (in the order of “people”, “encounter”, and “specimen”) for states reporting numbers using multiple criteria. For example, if a state reports testing numbers in both people tested and specimens collected, we would choose people tested when calculating the 7-day moving average of testing positivity rates. Preferring people-based criterion leads to less-optimistic positivity rates, and in most cases, more accurate results. This means that officials will be more cautious in making re-opening decisions. (See this article for more detailed reasoning, and an example, behind why we prefer the people-based criterion.)
When calculating “positivity”, the CDC and Johns Hopkins University teams have both acknowledged inconsistencies in the way that states report data, and consequently, different ways to calculate positivity. Based on these analyses, we have also included a “7 Day Confirmed Cases per Testing” variable. Technically speaking, this is not a “positivity rate” as a positive case does not necessarily equal a positive test. However, providing this alternate variable allows us to cross-check data validity, as an erratic difference between our positivity variable and positive case per test variable would signal poor reporting quality by states regarding testing data. Currently, the difference between the two variables is generally very small, with the positivity rate slightly lower than the positive case per test in a few states.
The significance of testing data
Testing positivity is viewed as an important metric to gauge whether an area is safe to reopen or resume activities. Five percent is commonly used as a positivity rate threshold. According to the World Health Organization, before reopening, rates of positivity should remain at this 5 percent threshold or lower for at least 14 days. A high positivity rate or a low testing capacity suggests more testing should be done, more restrictions should be implemented, and more interventions should be designed to slow the spread of the disease.
The latest data incorporated in the US COVID Atlas show that many states are above the 5 percent positivity threshold. As of the end of November, as many as 41 states are currently above the 5 percent threshold, with 27 states over 10 percent positivity.
Testing positivity is associated with other metrics such as cases and deaths. States that have high positivity rates also have low testing capacity, high population adjusted case and death numbers, as shown in the figures above. Therefore, testing positivity and testing capacity could not only be used as a measure to see the stress on the healthcare system and other related infrastructures but also as valid variables to measure general situations of the pandemic.
As the nation approaches colder weather, people visiting families for the holidays, and more indoor gatherings, accuracy in testing reporting is imperative in painting a more precise picture of the pandemic. Hence, the COVID Atlas feels the need to incorporate testing data into our map tools, and our team will continue to update testing data following the changing data availability to reflect accurate, useful, and up-to-date information regarding COVID testing data in the US.
Related Articles:
- Detailing testing positivity and its impacts
- What is “testing criterion” and why we include this measure in our data
- How testing data could be used and what is the official process after a positive case has been identified
Explore the US Covid Atlas on your own:
https://theuscovidatlas.org/map.html
This project was brought to you by our Data Analyst Fellow Ryan Wang and Postdoctoral Fellow Qinyun Lin, with write-up by Ryan Wang and Qinyun Lin, edits by Marynia Kolak.
