The flurry of daily pandemic data can be overwhelming. Here's how to make sense of it
Statistics and charts can be enlightening — provided you know the key context of what they show
New cases, cumulative cases, death counts, testing rates.
Every day, Canadians are bombarded with fresh numbers on the COVID-19 pandemic from around the world. Maps and charts show a variety of statistics, but some of them lack context.
Here is a primer to help you make sense of the numbers.
Daily case numbers are a window to the past
On a daily basis, countries and regions are reporting new cases of COVID-19 that were confirmed by tests. News organizations report these in real time, adding them to a running tally.
What's important to understand is that these daily numbers are a window to the past.
COVID-19 symptoms can take anywhere from a few days to two weeks to manifest, according to Greta Bauer, a professor of epidemiology and biostatistics at Western University in London, Ont.
Then it can take a few more days for the symptoms to get serious enough for a person to seek medical help and get tested.
And then there's the wait for the lab results to come back.
"I'm on Day 11 of waiting for my own test results," Bauer said on Friday.
This means we'll have a better idea in a matter of weeks how many people are infected today.
"We call this the iceberg of diseases in epidemiology," she said. "We're looking in a rearview mirror."
The number of cases is higher
Because the number of daily confirmed cases is a window to the past, the number of actual cases today is likely much higher.
Compared to countries such as Singapore and South Korea, Canada is not testing as many of its people. To date, tests in this country have typically been reserved for patients who have recently travelled abroad, or patients showing severe symptoms.
Perhaps not surprisingly, until very recently, most confirmed cases in Canada were linked to travel.
"That's not a finding; that's a self-fulfilling prophecy," Bauer said.
The video below from the popular non-profit education site Khan Academy explains how to predict the actual number of cases based on the reported confirmed cases.
There are also infected people who don't show symptoms but can still shed the virus, known as asymptomatic spreaders. In Canada, they are not reflected in the daily counts.
There's no universal definition of a confirmed case
So, what makes a COVID-19 case official and part of the daily count? It depends on the country or region.
Some places define a confirmed case as one where a person was tested and the results were validated in a lab. Other jurisdictions consider a case confirmed if the patient exhibits all the major symptoms.
Sometimes, the definition of a confirmed case can change. Last week, Quebec changed how it counts new cases. Previously, a positive test at a hospital lab had to be validated at the Laboratoire de santé publique du Québec. The second step has been eliminated, clearing a backlog of pending results. This resulted in a big jump in new confirmed cases.
Alberta recently expanded its own testing criteria to include people with mild symptoms, said Stephanie Smith, director of infection control at the University of Alberta Hospital in Edmonton.
"It might change the numbers but won't impact what we do in terms of flattening the curve," she said.
Exponential growth is tricky to understand
In Canada, confirmed cases of COVID-19 have been doubling roughly every three days.
Last Friday, there were 1,087 confirmed cases in Canada. On Monday, it was 2,091. Seen on a chart, this looks like a curve that is rising toward the sky.
But when compared to some other countries, Canada's overall numbers may appear low, which may give a false sense of security. The rate of increase is roughly the same across most countries that are still seeing a steady rise in new cases.
"If a country has twice the cases as we do, it just means we're three days behind them," Bauer said.
Some news outlets try to solve this chart-making issue by putting the case numbers on a logarithmic scale instead of a linear one. This means the numbers increase not by regular intervals, such as 100, 200, 300, but by multiples of 10, such as 10, 100, 1,000, and so on.
This makes countries more comparable, because curves on a chart become slopes. And the slope will tell us whether we're getting the infections under control.
"We're still on the same exponential growth as the U.S. or the U.K.," Bauer said. "There's no evidence that we've slowed it down or that we identified most cases."
The slope helps us predict problems ahead, she said.
"If the slope stays steep, it will mean more hospitalizations," she said. "And that's when we'll see deaths go up, when hospitals are overwhelmed."
The chart below shows the number of cumulative deaths from COVID-19 by country on a logarithmic scale. The lines that are curving are countries that have gotten the disease under control.
Issues with dashboards
There has never been, in the history of humanity, so much data shared globally as a public health crisis unfolds. This is truly the first data-driven pandemic. And with a wealth of data comes a wealth of data visualizations.
Dashboards, or websites with a flurry of maps and charts, are especially popular. But just because numbers are converted to images doesn't mean they're accurately portrayed.
Take the dashboard by Johns Hopkins University, one of the first and most authoritative COVID-19 trackers. Its centrepiece is a world map showing cases by region. It has been shared widely to illustrate how bad the situation is in the U.S.
What it doesn't say is that in the U.S., cases are reported by county or city, while most other countries report a single number, or, in some cases, provide a breakdown by state or province, as Canada does.
This makes it look like COVID-19 is far more widespread in the U.S. than in other countries.
Alberto Cairo, director of the University of Miami's Center for Visualization, Data Communication and Information Design and author of How Charts Lie, said it would be more accurate to show the U.S. as a big bubble on a global scale "that splits apart by city or county when you zoom into the U.S."
Even the size of bubbles depicting the severity of the spread can be misleading, Cairo said. For example, New York City and Italy have circles of similar sizes, even though Italy has nearly four times the number of confirmed cases as New York.
Mild cases aren't so mild
It's widely reported that in 80 per cent of cases, people with COVID-19 experience "mild" symptoms. According to Bauer, this masks the reality of the illness.
"Mild cases can include pneumonia, but when people think of mild, they think it's like having a cold," she said. "But it just means they don't have to be hospitalized. It doesn't mean they can breathe normally."
Recovery numbers are problematic
Some news outlets and online trackers report the number of cases where patients recovered. But recoveries are even trickier to define than confirmed COVID-19 cases.
Does it mean the patient no longer has symptoms? Or that their latest lab test came back negative? Or that they were discharged from the hospital? There's no universal definition.
This is how some provinces define a recovery, according to public health bodies surveyed by CBC News:
- Ontario: two consecutive negative tests performed at the Public Health Ontario Laboratory at least 24 hours apart.
- Quebec: absence of acute symptoms for 24 hours, fever for 48 hours and at least 14 days since the onset of the acute illness.
- Alberta: two negative tests at least 24 hours apart, a discharge from hospital or 14 days since last experience of symptoms.
And this doesn't include the people with symptoms who self-isolate for two weeks at home. In all likelihood, they are not being tracked at all.
Don't focus on the case fatality rate
In various corners of the internet, people are discussing how many people have died of the disease out of all those who have been infected. This is what's known as the case fatality rate. There's a notion that if and when more thorough testing is implemented, the number of infected people will rise and the fatality rate will go down.
Bauer discourages this kind of thinking.
"Some people think it might be as low as one per cent here, not the 3.4 per cent in Hubei, [China], or the nearly 10 per cent in Italy," Bauer said.
However, she said, that won't change the most important fact: "The number of people who died will be the same."
As Smith pointed out, about 0.1 per cent of people who get the flu die.
"Even if the COVID-19 fatality rate is one per cent, it's still 10 times deadlier than flu."
- A previous version of this story referred to the number of people who died out of all those infected with the virus as the mortality rate. In fact, that is the case fatality rate.Mar 31, 2020 10:03 AM ET