Even with the strongest control measures in place, COVID-19 could kill between 11,000 and 22,000 Canadians over the course of the pandemic, the agency said in its forecast "COVID-19 in Canada: Using Data and Modelling to Inform Public Health Action."
The modelling lays out three scenarios: one with strong control measures such as high rates of social distancing and testing, 1-10 percent of the population will be infected; one with weaker measures, 25-50 percent infected; and one with no measures taken, 70-80 percent infected.
They are estimates based on the best available data so far.
"Data and models can help Canadians see how our collective efforts ... can determine the trajectory of Canada's COVID-19 pandemic," said Theresa Tam, chief public health officer of Canada, at a press conference Thursday morning.
Tam said the objective for Canada is to be in the strong control measure camp, which the modelling suggests would result in roughly 11,000 deaths if 2.5 percent of the population is infected and 22,000 deaths if that infection rate rises to 5 percent.
If the infection rate hits 25 percent under the weaker control measures scenario, those deaths could spike to more than 100,000 or 250,000 if 50 percent of the population is infected.
"These stark numbers tell us we must do everything in our power to stay in the control model," Tam said.
Even if the country is successful in getting the epidemic under control, measures like physical distancing, restrictions on international and domestic travel, and contact tracing will need to remain, Tam said.
Ontario's modelling released last week predicted the province could see 1,600 deaths by the end of April even with the current social distancing and shutdown measures in place while Quebec's modelling forecast it could see between 1,263 and 8,860 deaths by the end of April.
Tam said the federal modelling is "imperfect" but it can help understand the state of the pandemic and where it might go, along with the effect of public health measures on the transmission of the coronavirus.