During the course of the pandemic, new or unfamiliar terms have swirled around all of us: contact tracing, case incidence, confirmed vs. probable cases, public health orders. Perhaps the most widely mentioned—and misunderstood—term is percent positivity. Percent positivity helps us assess disease spread in our community, but it is influenced by factors like who is able to get tested and lab timeliness. This can make it difficult to interpret percent positivity without more context.

To further complicate things, there are multiple ways to calculate it. Let’s take a deeper look into how percent positivity can be calculated and what it can tell us.

Calculating Percent Positivity

There are three ways to calculate percent positivity, and CDC does not recommend any particular calculation over another. Each method creates a fraction of “positives” (either people or tests) over a total (either total people tested or total number of tests).

Test Over Test

CDC uses this method. To calculate it, take the number of all positive tests and divide by the number of total tests (both positive and negative), then multiply by 100 to make a percentage.

This method counts duplicates—people who are tested multiple times. For example, if one person is tested three times, with two tests being positive and one test being negative, those two positive tests are both counted. This isn’t a big deal if most people are only getting tested once. But as testing availability has increased and people are getting tested multiple times, this method makes less sense to use at this phase in the pandemic. This might be the only option for an entity, like CDC, that doesn’t have person-level data that can be deduplicated.

People Over Test

Public Health Madison & Dane County uses this method. With this method, the number of new people with positive tests is divided by the total number of tests (both positive and negative), then multiplied by 100 to make a percentage.

The advantage of this method is that it accounts for all retests taken in the denominator but only counts a positive test once in the numerator. In other words, a positive person is only counted once.

As of September 30, in Dane County, 202,984 people have been tested for COVID-19, and there have been 366,896 tests. This means many people have been tested multiple times, which is good: we want people, especially those in high-risk groups and the people who work with them, to be tested more than once. We include all those tests in our calculations to gauge the spread of the virus and to know whether there is enough testing happening.

This method of calculation will yield the lowest percent positivity of the three methods. 

People Over People

Prior to September 30, this is the sole method the Wisconsin Department of Health Services used. As of September 30, DHS displays 7-day percent positive by both People Over People and People Over Test. The visualization DHS provides on their website illustrates how these values can diverge over time as more people get tested more than once.

To calculate the People Over People method, the number of new people with positive tests is divided by the total number of people tested (both positive and negative), then multiplied by 100 to make a percentage.

This method does not count duplicates, but it also does not account for retesting. For example, if someone tests negative, they are counted as a unique person. Nothing would be added to the numerator, but a count of one would be added to the denominator. If they come back and test negative two more times, nothing would happen to the percent positivity; they’ve already been counted as a unique person.

But say that same person who tested negative later comes back and tests positive twice. A count of one would be added to the numerator as a new positive person but the denominator wouldn’t change since they were already counted as unique person being tested. If we have enough people who test positive after having a negative test, this can increase the percent positivity because the numerator is increasing but the denominator is staying the same.    

Example 1

This simple example outlines how 10 total people (2 positive and 8 negative) with 11 total tests (2 positive from one person, one positive from another person) would be calculated with each method:

10 Total People 2 people positive, 8 people negative. 11 Total Tests 3 tests positive, 8 tests negative. Test Over Test (number positive tests/number total tests)*100=27%. People Over Test (number unique positive people/number total tests)*100=18%. People Over People (number unique positive people/number total people tested)*100=20%

Example 2

This example is a little more complicated, with more people being tested multiple times, and one of them having tested negative once then positive later. Notice how People Over Test can start to look quite different from People Over People method.
10 Total People 3 people positive, 7 people negative. 13 Total Tests 4 tests positive, 9 tests negative. Test Over Test (number positive tests/number total tests)*100=31%. People Over Test (number unique positive people/number total tests)*100=23%. People Over People (number unique positive people/number total people tested)*100=30%

What Can Impact Percent Positivity?

No matter the method used, the reason we calculate percent positivity is to give us some sense of disease spread in our community.

What makes percent positivity go up?

Say percent positivity in Badger County is 20%. That’s high! This could mean there are widespread infections in the community.

But then you might wonder, well who is able to get tested? If only people who are hospitalized with symptoms of COVID-19 are able to get tested, it’s likely a good chunk of the people we test will test positive (this is why early in the pandemic, when testing was hard to come by, our percent positivity was high). This doesn’t necessarily mean there are widespread infections in the community; it could just mean we don’t have enough testing to really get a good picture of COVID-19 in our community.

Reporting processes and delays can also impact percent positivity, which is why it’s important to look at trends in percent positivity, such as over a 7-day or 14-day average, instead of day by day.  

What makes percent positivity go down?

If the number of infections in a community goes down or testing is expanded to more people who are not infected, percent positivity will decline. We would expect percent positivity to go down as more people are screened in non-outbreak settings (such as routine screening in schools, long-term care facilities, and workplaces) and the results are reported on time. Keep in mind this isn’t foolproof: if a community has widespread transmission and testing becomes more accessible, testing might find more people who are infected and percent positivity will go up.

Does Percent Positivity Give Us a Complete Picture of COVID-19 in Dane County?

Percent positivity tells us some information about spread, but as noted above, it can also depend on factors like how it’s calculated, testing accessibility, and lab timeliness. No one metric can give us a complete picture of COVID-19 spread in our community. That’s why we look at percent positivity along with eight other metrics each week. When comparing percent positivity across different communities, we recommend paying attention to the trends, rather than only focusing on the numbers. Ask, “What patterns am I seeing over time? What could be driving these patterns?” A great way to stay up-to-date—and find answers to these types of questions!—is to subscribe to our blog posts. Each Thursday we release Data Notes for the week. To read more about percent positivity, visit the CDC’s website.

This content is free for use with credit to the City of Madison - Public Health Madison & Dane County and a link back to the original post.