Confused about the number of 'people tested' vs. 'tests performed' for COVID-19? You're not alone

In late November, when Alberta was reporting a COVID-19 positivity rate around 8.5 per cent, the Public Health Agency of Canada was reporting a figure for the province that was roughly double: 17.9 per cent.

This major discrepancy was a source of confusion, not just for everyday Canadians, but for health officials, themselves.

In early December, PHAC removed Alberta's data from its calculations of the national positivity rate, citing "differences in methods of computation."

Then the federal agency backtracked, saying it had made an error.

"Following dialogue with the province of Alberta, it was determined that a field of submitted data had been interpreted incorrectly by PHAC," Canada's chief public health officer, Theresa Tam, explained in a written statement.

So what's going on here?

It has to do with how COVID-19 tests are counted.

There are two different ways of doing this and like many jurisdictions Alberta does both.

Maintaining these two separate counts allows for different types of data analysis, but ithas also been a source of confusion for laypeople, data scientists and public-health experts alike.

So, to hopefully clear things up, here's a quick explanation of how Alberta calculates its own positivity rate and why other calculations can yield such wildly different results.

'Daily test positivity rate'

Alberta Health says it calculates its "daily test positivity rate," quite simply, as the number of positive tests divided by the number of total tests on a given day.

That gives you a decimal number, which is then multiplied by 100 to give you a percentage.

So if there were 10,000 tests performed yesterday and 800 of them came back positive, that would be a test positivity of 8.0 per cent:

• 800 / 10,000 = 0.08
• 0.08 * 100 = 8.0%

Pretty simple, right?

But there are a few complicating factors that can make things a bit more difficult, including the factthat COVID-19 tests are counted in two different ways.

'Tests performed' vs. 'people tested'

The first method is counting the number "tests performed."

This is exactly what it sounds like:if there were 10,000 tests conducted yesterday, the number of "tests performed" is 10,000.

Theothermethod is counting the number of "people tested." And what this means is not as obvious.

It actually refers to the number ofuniquepeople tested. In other words, the number of people testedfor the first time ever.

So imagine you got a test back in April because you had some concerning symptoms, and it came back negative.

Now imagine you went for another test in the summer, just for the heck of it, because Alberta was offering asymptomatic testing to everyone at that time.

Both of your tests would count in the daily talliesof "tests performed."

But only your first testwould count in the tally of "people tested." Any subsequent tests you get would not count in this category.

Early in the pandemic, the vast majority of people being tested were being tested for the first time. But today, the number of people who have already been tested at least once is a large and ever-growing group. And the gap between these two measures continues to grow ever wider.

According to Alberta Health, this is why the province's positivity rate looked so much higher in the federal reports.

In doing its calculations, the Public Health Agency of Canada was dividing by the wrong number, says Alberta Health spokesperson Tom McMillan.

"The issue that PHAC experienced was the result of their staff using the incorrect denominator," he said in an email.

Instead of dividing by the number of "tests performed," they were dividing by the number of "people tested."

"This is a smaller number than the daily number of tests,which will result in a higher per cent positivity," McMillan said.

This type of confusion isn't unique to Canada.

Why comparing jurisdictions is so hard

Around the world, testing protocols have varied widely and so, too, havemethods for counting tests and calculating positivity rates.

Similar discrepancies between federal and local jurisdictions have cropped up in the United States, according to theCenters for Disease Control and Prevention (CDC).

The CDC, itself, uses Alberta's method for calculating positivity but it notes some state governments have taken different approaches.

Instead of dividing the number of positive tests by the number of total tests, some states divide the number of new people infected by the number of total tests, while others divide the number of new people infected by the total number of unique people tested.

"Any of the three methods for calculation (tests/tests, people/testsor people/people), when applied consistently and with clear communication, will allow public health officials to follow magnitude and trends effectively, and the trends will be useful for local public health decision-making," the CDC says on its website.

The challenge comes when trying to compare positivity rates from one place to another.

Whenever doing so, it's important to understand the methods of calculation.

But eventhat's not enough.

McMillan said the method of calculation is just one of "many other complex factors that will impact positivity rates from jurisdiction to jurisdiction."

It's important to understand the local testing protocols, as well.

When Alberta was doing widespread asymptomatic testing during the late spring and summer, for instance, the large number of tests drove the positivity ratedown.

Roughly 659,000 tests were done on people with no symptoms during that time, and just 0.11 per centof them came back positive.

Comparing Alberta's rate fromthat time to another province where only symptomatic testing was being done would lead to a distorted picture.

Even comparing Alberta's current rate (during a time when asymptomatic testing has been scaled back) to Alberta's rate over the summer requires a similarasterisk. Because of the differences in who can be tested, it's not an apples-to-apples comparison.

All this is probably enough to make your head swim. But at least you can take some comfort in the fact that you're not alone.