National Post (National Edition)
Virus numbers tell cautionary story
Coronavirus modelling from the Public Health Agency of Canada has a strong track record of predicting where the virus is going, a troubling trend given that the agency now predicts a potentially steep rise in cases.
By Sunday, Canada is predicted to have seen between 841,650 and 878,850 cases of coronavirus and recorded between 21,510 and 22,420 deaths, according to the latest modelling.
On Sunday that modelling will almost certainly be right, because as of Friday, there were 858,217, cases and 21,865 deaths.
A review of the modelling numbers the Public Health Agency releases approximately every month shows its predictions have been within range of the actual results most of the time.
The agency's estimate for the number of cases on Jan. 24 missed the mark by about 5,000 cases. Canada recorded 747,383 cases, lower than the agency predicted, but a day or so later the pandemic reached that target.
Last September, the agency slightly underestimated how quickly the virus would grow — it predicted a high of 155,795, while Canada actually saw 160,535. The model has also twice slightly undercounted the number of deaths the country would see.
Outside of those estimates, the agency has accurately predicted where Canada would be each month since last June.
In addition to the estimates, the agency has also issued worst case scenarios with each new modelling update. Those scenarios haven't come to pass, but experts say that's exactly what is supposed to happen.
November's update warned, for example, that in a worst case the country could see 60,000 cases a day in early January, if the government eased restrictions and Canadians gathered in big numbers over the holidays. It predicted with no change, cases could climb to 25,000 cases a day.
Neither of those numbers came true, but early January was the highest peak of new cases for the virus, with just over 14,000 cases reported on Jan. 4, 2021. Provincial government across the country put in restrictions and many Canadians avoided holiday parties That 14,000 caseload was between the middle of the road scenario of 25,000 cases and the best case the agency predicted, which was about 5,000 cases per day.
Dr. Howard Njoo, the country's deputy chief public health officer, said the worst case scenarios in the modelling are not meant to panic Canadians, but to show them they need to take action.
“I am not sure it's so much a warning as it is to just basically further reinforce the point to Canadians that our actions matter,” he said. “What you do today certainly will have consequences down the road.”
Caroline Colijn, a professor at Simon Fraser University who holds a Canada 150 Research Chair in Mathematics for Infection and Public Health, said it's important to imagine models as headlights on a dark road.
“You have your headlights on and you see that there's a boulder in your path, so you do something,” she said.
She said knowing what the models predicted as a worst case gave politicians the opportunity to adapt their approach.
“The idea is to pass that information to decision makers, and they get to decide whether we keep doing what we're doing or we do something else,” she said. “We can't predict what an individual policy-maker will do, so we don't do that, and that's why the models are not crystal balls.”
Daniel Coombs, head of the mathematics department at the University of British Columbia, said no politicians would allow cases to rise to 60,000 cases a day. But the models let them know when they're headed in that direction.
“I think everybody knows, before they got that high, we would see a strict lockdown or something else would happen that would prevent that,” he said.
Canada has been recording a rise in COVID-19 variants, including a type that originally spread in the U.K. Coombs said that will make modelling more difficult, because it can be hard to know when they will fully emerge as dominant strains.
“The problem is knowing when that wave is going to rise. So in Ontario, there have been more cases for longer and I kind of suspect that that rise is probably happening right now.”
He said while the U.K. strain spread rapidly there last year, it might not do the same here in Canada.
“The question is how is that strain going to interact with the local environment, the conditions, physical distance, what's open, and what's closed, which are going to be different than they were in the U.K. in the fall.”
Canada's new daily caseload has been falling since that January peak, but with new variants and a lagging vaccine rollout the agency fears the downward slope may not stay that way.
The most recent model attempted to factor in new variants of the virus and forecast that in either the worst case or middle of the road scenarios, Canada could see 20,000 cases per day.
Njoo said the variants are an added wrinkle into attempting to chart the course of the virus.
“They're more challenging to predict, but at the end of the day, it all depends also on sort of human behaviour, the number of interactions, and whether there's a sustained community transmission of the variants.”
Colijn said the variants will have an impact in Canada, but the variants are an X-factor, because they may spread differently in Canada than elsewhere, but it doesn't change the need to track and model where cases are headed.
“We don't really know you know in Canada how much more transmissible these variants will be,” she said. “It's important to use modelling to explore that even if there are sources of uncertainty.”