It was late Friday afternoon when Quebec’s public health research institute quietly released its latest round of projections for what the COVID-19 pandemic holds in store for the province.
Though it took a few hours before reporters caught on and had a close look at them, those projections have been much discussed ever since — especially what’s on the horizon for greater Montreal, should they hold true.
In one scenario, the researchers estimate the city could see as many as 100 deaths per day by midsummer.
It gets worse: that number would skyrocket if the Quebec government went ahead with its plans to lift confinement measures, according to another scenario outlined in the report.
The projections were released at the end of a week in which Legault twice announced delays to his plan to loosen restrictions in Montreal. As it stands, elementary schools and businesses in the region are still set to reopen on May 25.
Critics of Legault’s approach have been holding up the projections as evidence that his plan should be abandoned outright, at least until the outlook improves for Montreal.
With those projections at the centre of this debate, we decided to take a closer look at how they were made and what they say.
Building the model
The model is the work of researchers at Quebec’s public health research institute, the INSPQ, and a team from Universite Laval led by Marc Brisson, a professor of health economics.
The model is, basically, a complicated math equation. The answers to the equation are predictions about how many people will catch the virus, how many will be hospitalized and how many will die.
Like many researchers modelling the spread of COVID-19, Brisson is using a variation of what’s known as the Susceptible-Infected-Recovered, or SIR, model.
Models like these allow you to calculate how many people within a given population, and at what rate, will go through each of these categories (known as compartments).
This has been a standard way of modelling epidemics since the 1920s, though in the case of COVID-19, the compartments are usually susceptible, exposed, infected and recovered, and dead.
The problem, of course, is that so much remains unknown about the novel coronavirus, such as how many people have been exposed at any one time.
However, we do have some fairly concrete data, such as the number of confirmed cases, hospitalizations and dead.
So what researchers do is make the assumption that some combination of parameters — values for things like transmission rates and social contacts — will equal the known data.
With the help of supercomputers, Brisson’s team tried around 50 million different combinations of these parameters, once for the Montreal area and again, for the rest of Quebec.
Though each combination is a prediction, they also covered periods of time that have already passed. That means the researchers can check them against real-world data.
They then selected the 500 predictions that most closely aligned with the data for hospitalizations and deaths.
What they published Friday was what these predictions said will happen next.
What if we leave the confinement measures intact?
Brisson’s team used the models to answer two questions: what will happen if we leave the current confinement measures intact? And what will happen if we remove some of these measures?
The vast majority of the 500 predictions about the rest of Quebec estimated that cases, hospitalizations and deaths will decrease if confinement measures were left in place. No surprise there.
When it comes to Montreal though, only 40 per cent of the 500 predictions foresaw a decrease in the effects of COVID-19 if the confinement measures remained.
The rest of the predictions estimated things would actually get worse, with around 100 people dying daily by midsummer. And that doesn’t include potential deaths in long-term care institutions.
A 60-40 split, Brisson said, is too close to draw a definitive conclusion about where Montreal is headed. The city, according to the model, is at a crossroads.
“It’s still unclear whether we’re on an optimistic or pessimistic slope. But what is clear is that we’re in a fragile situation,” Brisson said.
The doomsday scenario
The second question — what happens if we lift confinement measures — is the one Legault is currently grappling with.
In order to model that, Brisson’s team examined the plan to ease restrictions that Legault announced last month, which called for elementary schools and daycares to reopen first in the regions outside of Montreal, then inside the city.
It also called for the gradual resumption of activity in the construction, manufacturing and retail sectors.
The researchers estimated that Legault’s plan translated into a 15 to 30 per cent increase from our current level of social contacts.
That estimate is based on a number of different sources, Brisson said, including mobility data from Google.
“We have an average number of contacts for each type of employment, and we have a proportion of people who work in these different types of jobs,” he said.
When those values are plugged into the equation for the rest of Quebec, the results showed only a slight increase in hospitalization and deaths.
It’s a different story when it comes to Montreal.
One important variable in the model is what’s known as the R0, the average number of people who are infected by one person with the disease. When it’s over 1, the epidemic continues; it fades away if it’s less than 1.
While Brisson’s team is confident that outside of the Montreal area R0 <1, they believe it is hovering around 1 in the Montreal area.
In that context, if you increase the number of interactions people have with each other, then there is a greater chance “susceptibles” come into contact with “exposed” or “infecteds.” The number of cases, hospitalizations and deaths will, therefore, jump rapidly.
That’s what produces the doomsday scenario for the Montreal area in the model.
It predicts that if social contacts were to increase between now and May 25 — as Legault had originally planned — more than 150 people would be dying daily by July and hospitals would be overwhelmed by August.
From there, the curves are literally off the chart.
Brisson’s model is dynamic, meaning it is constantly being refined and will hopefully become more precise over time.
Simply measuring the impact of easing restrictions by increased social contacts is rather blunt.
It doesn’t, for instance, take into account testing and contact tracing. An effective testing regime would involve isolating exposed and infected people from those susceptible to the illness, which would contribute to lowering the R0.
The province has gradually been increasing the number of tests conducted daily, particularly in Montreal. That’s something the next version of the model will try to account for, Brisson said.
It will also have a better idea to what extent Quebecers are respecting physical-distancing guidelines.
Testing, physical distancing, masks, washing hands — all are measures that reduce the rate of infection. If followed widely, they hold out the possibility that easing confinement measures and increasing social contacts won’t necessarily lead to huge spikes in the disease predicted in the model.
As is, though, the projections were sobering even for seasoned experts.
“It was definitely enough for me to really take a deep breath,” said Erin Strumpf, a professor in McGill’s department of epidemiology, biostatistics and occupational health.
“I’m not sure that it’s cause for panic but it is cause for a pretty serious rethink about whether Montreal’s ready for the policy changes that a few weeks ago were imminent.”