We know that just over 60,000 Canadians have tested positive for the novel coronavirus.
It would be useful to know how many of us actually have the disease, but for a range of reasons — shortages of tests, the large number of people who don’t show symptoms — that’s a much harder number to gather.
But given that we know more or less what the coronavirus death rate is, it’s possible to use that to roughly estimate the total number of cases, epidemiologists say.
“I think it’s fair to take the deaths that you’re counting and infer back to get a general sense of how many cases there are likely to be,” says the University of Toronto’s Ashleigh Tuite.
“It’s a better estimate than the case fatality ratio, which is highly dependent on how much testing is happening.”
A German study released this week offers some idea of what the death rate might be.
Researchers from the University of Bonn tested 919 residents in Heinsberg, in western Germany near the Dutch border, for coronavirus. The area has become one of Germany’s worst-affected since a local carnival spread the virus widely.
About 15 per cent of the residents tested were positive. Cross-referenced with known coronavirus deaths in the community, researchers estimated a death rate of .37 per cent.
Applied to Canada, that would mean that about 1.2 million Canadians are, or have been, positive for coronavirus, and that testing is catching about 5 per cent of the cases that are actually in the community.
But Tuite cautions against treating those numbers too literally.
“It’s not going to be a precise value,” she says.
“There is going to be variability depending on the population you’re looking at, the health-care system in that country.”
“I think it’s appropriate to have fuzziness around that number, because it’s not a fixed thing.”
Estimates of a coronavirus death rate in different places have tended to be under 1 per cent, she says.
One issue with the German study is that Heinsberg has had fairly few coronavirus deaths, which means there is potentially a large error range in the death rate.
“The fact that it was a large outbreak is not an issue per se,” she says. “It’s the small number of deaths, because that’s really where the imprecision comes in. We can still learn from that, you just have to interpret it with caution, knowing that you’re dealing with small numbers in terms of the numerator.”
Another issue is that deaths aren’t counted as coronavirus deaths in a consistent way, she says.
Tuite and University of Toronto epidemiologist David Fisman cautioned that while many more Canadians likely have coronavirus than test results show, even the much higher numbers are still far short of what would be needed for herd immunity.
“It may mean we are further into this pandemic than we think, but we are still far from having an immune herd,” Fisman wrote.
Tuite agrees.
“Whether it’s 200,000 infections or a million infections (in Canada), it’s still very far away from what we would want and need to have in the population to shut down the spread.”
How could so many cases be missed? The German study, which found that 22 per cent of those who tested positive were asymptomatic, offers part of the answer. For Fisman, that provides an argument for wearing masks.
“If there’s a lot of asymptomatic infection, that means the masks aren’t just to protect you,” he says. “They’re to protect others from you because you may not know you’re infected.”
Systematic testing of groups has often found that large numbers of people were positive but asymptomatic. In testing at a Louisiana prison, nearly all the inmates tested positive for coronavirus, but two-thirds showed no symptoms. Testing of sailors on a U.S. aircraft carrier had similar results.
As society reopens, finding ways of estimating the true number of cases will become more important, Tuite says. Estimates that start with deaths and reason from there are one tool to do that.
“At this point, it’s the best we can do, until we start doing serosurveys (blood-based tests of a cross-section of the population) and looking within our own population,” she says. “We have to use numbers such as this to get a sense of where we may be.”
“If we’re missing 90 or 95 per cent of whatever number of cases, that’s a problem, because if we’re not finding those cases we’re not able to control it using the strategy that we’re planning to use. It basically indicates that we’re going to have to be doing a lot more testing to be able to find those missed cases.”