Three weeks or so after Canada largely shut down to try to deal with the novel coronavirus, there seem to be many more questions than answers.
Among them: Is the huge social dislocation we’ve undergone helping? Are we starting to get the disease under control? The best place to look for answers is the number of deaths, University of Toronto epidemiologist Ashleigh Tuite explains.
“It’s alarming,” she says of the graph below.
“It’s what you expect to see when you have an outbreak that is growing. It suggests, because the cases are lagged, that we are in the middle of the growth phase of an epidemic.”
Of the three ways of measuring the epidemic, ‘deaths’ is probably the best
There are three ways of measuring the outbreak, Tuite explains: positive tests, hospital or ICU admissions, and deaths.
Of these, measuring deaths is the most reliable method.
“There are a number of different data sources that we can look to,” she says. “The reason that people are focusing on deaths is because they tend to be identified the best.”
For a number of reasons, positive tests aren’t a great way of measuring the spread of coronavirus, she says.
“In Ontario, for instance, testing right now is fairly restrictive. You have to meet certain criteria to get a test. We know that we’re likely missing a lot of the milder infections, so looking at reported cases is problematic, because it’s only the tip of the iceberg in terms of reported infections. It’s going to reflect changes in our testing policy.”
Hospitalizations and ICU admissions are better, she says. That data is useful because it reflects infections that are more recent than those that led to deaths.
“We can look at …. hospitalizations for COVID-19 and people who are in the ICU for COVID-19. Those outcomes, because they happen a bit more upstream, the time from infection to hospitalization will, in general, be earlier than mortality.”
But because of the formality around accounting for death, epidemiologists think it gives them the highest-quality data.
“Looking at something like deaths is useful because we think that most if not all of the people who die of COVID-19 are going to be identified,” she says.
Something else to bear in mind, however, is that deaths happening now reflect infections in mid-March, when the shutdown hadn’t fully taken effect.
“The numbers are showing us that three weeks ago there was maybe a bit of a sense of ‘Why are we doing this now?’”
“That sort of illustrates why we did what we did when we did it.”
How would we know if things were getting better?
“Ideally we would like to have no deaths from COVID-19, and hopefully we can reach that point, but for now I think success would be a flattening those numbers and eventually a downward trend,” Tuite says.
A rise in cases may not be a bad thing (It may just mean that we can see better)
A few weeks ago, Quebec showed a dramatic jump in cases that turned out to be related to a change in testing procedures. As Ontario expands testing, which at the moment is being done at the lowest rate in the country, it may well show something similar.
Tuite says we shouldn’t necessarily interpret rising numbers of known cases as a change for the worse.
“There are definitely challenges with interpreting the reported case data, because they are so tied to testing and testing capacity and the rules around who gets tested. As we expand testing in Ontario, we do anticipate that we’ll see an increase in cases.
“That is not necessarily a bad thing.”
Take very recent data with a grain of salt
Data about the coronavirus reflecting the last day or two can seem lower than it is because the data is incomplete. When Global News was assembling data on deaths recently, for example, the number for Tuesday was in the low 20s. By the following morning, it had risen to 57.
If you’re not aware of this quirk of the numbers, coronavirus data can give a misleading impression that it’s gotten better in the last few days when that isn’t the case.
“If you look at that epidemic curve it may look like cases are going down,” she says. “That’s not because cases are actually going down, it’s because there are lags in the reporting. The data takes a little bit of time until it is complete, so it gives you a false sense that it has peaked, and it’s going to go down.”