Trying to Make Sense of Saskatchewan’s COVID-19 Modeling, AKA Educated Guess

If you’re confused or overwhelmed by the influx of unsettling COVID-19 infection and death rate projections for Saskatchewan, you’re not alone. So let’s try to make some sense of it.

COVID-19 modeling provides us with infection and death rate projections which are based on assumptions, which are based on variables.

This process isn’t unique to viruses. For example, the annual provincial budget is a projection – a model of future planned government spending and revenue, created using assumptions about what might happen in the year ahead by analyzing variables such as historical government spending and revenue from the taxes we pay.

Similarly, COVID-19 models are created based on assumptions made by analyzing what variables are available, such as patterns of transmission in other countries, to plan what Saskatchewan’s healthcare system might need to manage the pandemic here.

So put simply, modeling – whether of provincial budget or COVID-19 projections – is a fancy way of saying “Educated Guess”.

The main problem with COVID-19 modeling is good data for assumptions is not readily available, given none of us, even the Dr Faucis and Dr Tams of this world, have experienced anything like this. Even as research trickles out of countries like China and Italy that have already (hopefully) been through the worst of it, there are still unknown variables, especially between regions, that impact its spread.

At this point it feels prudent to point out that obviously I’m not an epidemiologist. I am not a statistician or medical professional or fortune teller or mathematician. All I’ve done is (way too much) reading on COVID-19 modeling, which I will attempt to distill here.

So let’s start with looking at why Saskatchewan and the Saskatchewan Health Authority (SHA) bothered to produce COVID-19 models.

The short answer is because the most responsible thing they can do right now is make an Educated Guess on how many ventilators, staff and empty ICU beds they’re going to need to be ready for a potential wave of coughing, gasping COVID-19 patients, like those which have overwhelmed hospitals across the globe.

Yes, that wave may never hit Saskatchewan, but not being ready if it does is not an option.

I suppose you could argue the SHA’s responsibility ends there and the Saskatchewan public doesn’t need to know the unsettling modeling details, but this blog has rather strong feelings about government transparency and accountability. When the modeling, or Educated Guesses, are driving decisions as starkly life or death as the ones related to COVID-19 planning, we should know what data is being used to make them.

Adding to the confusion is the fact each province is using different variables for their assumptions and different language to describe their projections.

For example, Alberta’s model presents its projections in three categories, or scenarios, named “Probable”, “Elevated” and “Extreme”. BC’s model has three categories named, in ascending severity, “South-Korea-type”, “Hubei-type” and “Northern Italy-type”. Ontario went with “No Intervention”, “Current Intervention” and “Full Future Intervention” as well as “Best Case” and “Worst Case”. The federal government’s model labelled its scenarios “no control”, “weaker controls (delay and reduce the peak)”, and “stronger epidemic control”.

Saskatchewan kept it simple with “Scenario 1”, 2 and 3, with 3 (somewhat confusingly) being the least severe.

Clear as mud? Great. Now lets look at the assumptions used by each.

Alberta’s Probable Scenario uses “1-2 more people” to describe its assumptive reproduction number, or the amount of people that one person diagnosed with COVID-19 would infect in a group that is not immune.

You may have seen this assumption described as the mathematical term R0(zero) or “R-naught“. Every virus or infection has a different R0, depending in part on various characteristics such as how its spread (i.e. droplets vs airborne), its incubation period, and what kind of public health measures are in place, such as a vaccine, or in the case of COVID-19, physical distancing.

The extremely contagious measles is thought to have a R0 of 12 to 18 in a susceptible group (AKA unvaccinated). Researchers have assigned the 1918 Spanish flu a median R0 of 1.8, ranging from 1.47 – 2.27. 2009’s H1N1 had an R0 of 1.2 to 1.3.

You don’t need to be an epidemiologist to know that Alberta’s “Probable” R0 assumption of “1-2” people isn’t exactly precise. BC doesn’t state what number they’re using for their lowest range South Korea scenario, but reputable research puts it at 1.5.

Similarly the federal government’s R0 assumption doesn’t share a precise number, but this diagram from their model seems to suggest that the “stronger control” scenario represents a R0 of 1.75ish.

Refreshingly, Saskatchewan used specific numbers:

Saskatchewan’s lowest range R0 of 2.4 was based on that of Wuhan, China.

That is, in part at least, why Saskatchewan’s projections, even our lowest or “best-case” scenario, are higher than in other regions – we used a higher R0.

So with Alberta’s population roughly four times higher than Saskatchewan’s, their lowest projected number of 800,000 total cases isn’t that different than Saskatchewan’s 153,000. Still doesn’t make a ton of sense though, given we used the higher R0 of 2.4 and they supposedly used from “1 – 2”, but this is the best I can do to explain it.

The other main difference between our modeling and theirs is that Alberta added the assumption of time, namely a mid-May Probable peak of approximately 21,000 total cases (see the shortest, black-line curve on the above slide).

Saskatchewan’s modeling was presented without a timeline, meaning those 153,000 total cases could take years – or weeks – to accumulate, or maybe we’ll never get there because a vaccine will emerge.

When it comes to other assumptions for modeling, unfortunately there just isn’t enough reputable information out there yet. I think for many of us, this is a discombobulating notion because we’re used to the opposite: endless, too-much information on pretty much everything.

What we do know (kind of) is that COVID-19 is spread by getting droplets* from an infected person’s cough or sneeze in your own nose or mouth, or picking those droplets up off a surface and wiping them in your eyes, nose or mouth.

Gross.

*It’s important to note that droplet contact is different than airborne transmission (think about what comes out of your mouth when you sneeze vs. what comes out of an aerosol spray), the latter of which hasn’t been definitively linked to this virus. Not yet.

The median incubation rate (the time it takes from a person being exposed to the virus to showing symptoms) is thought to be around 5 days, but can stretch up to 13.

What’s most important, in my opinion, is that as the virus travels the world, replicating itself again and again, researchers do not believe that COVID-19’s genetic makeup is changing for the better or worse.

This means that given the opportunity to jump from one person to the next, the COVID-19 virus in Saskatchewan is behaving exactly the same as it did when it ravaged Wuhan, Italy, Spain, Washington State and New York City, to name a few.

We are not special or different – COVID-19 is not weaker in Saskatchewan, nor are we somehow more resilient.

Right now the primary difference in how COVID-19 spreads here versus the rest of the world are the public health measures in place to promote physical distancing from each other.

The fact we have remarkably low population density doesn’t hurt either.

We’ve also had the blessing of being one of the last places in the Western world, or at least the European and North American continents, to “catch” COVID-19. We have surfed on the wake of other countries’ devastating hindsight and have been able to apply lessons here learned the hard way elsewhere.

Not that we didn’t learn some lessons of our own. The Saskatchewan government was subject to pressure that got us moving earlier on physical distancing measures than other provinces. The public, the Saskatchewan medical community and the opposition had to beg for the snap election to be cancelled, the schools to close etc.

But I digress.

Back to the models – the projected deaths also don’t make a ton of sense to me.

For example, in their Extreme scenario Alberta projects 1.6-million cases with a death rate of 1 – 2%, or 16,000 to 32,000 deaths. Their Probable scenario of 800,000 cases projects 400 – 3100 deaths, or a 0.05 – 0.4% death rate.

As of writing this, Alberta has had 1651 confirmed cases and 44 deaths, which is a 2.7% death rate.

Globally, as of today, the rate sits at 6%, undoubtedly disproportionally influenced by explosive rates in places like Italy, where 13% of all confirmed cases ended in death.

Saskatchewan’s projected death rate is 2% of all confirmed cases in all three scenarios.

Ambitious? Probably, though its the same rate Ontario is using and far more realistic than Alberta’s 0.05%. The word “death” doesn’t even appear in BC’s model.

But even right now, 2% isn’t isn’t far off – as of publishing this piece, with 298 confirmed cases and 4 deaths, Saskatchewan’s death rate is currently 1.3%.

Finally, let’s look at hospitalization rates, which is such a big part of what this is all about: protecting the healthcare system from total collapse.

It’s also wherein lies the most frightening data, because not unlike other provinces, Saskatchewan’s hospitalization and ICU projections are a hot mess.

The above chart puts the daily COVID19 inpatient hospital population (census) in Saskatchewan at 390 and the ICU at 120, at the peak of Scenario 3.

But because Saskatchewan’s modeling does not provide a timeline, I can’t explain where or how they came up with those numbers, or when they’re projecting that peak might happen.

I can tell you it’s weird, because using Hubei’s numbers (the province in China that is home to Wuhan, so the same R0 Saskatchewan uses in Scenario 3), BC (with 5x the population of Saskatchewan) puts its ICU peak census at 166.

And Alberta (4x our population):

To add just a touch more confusion to the pile, Saskatchewan’s model goes on to provide us with this:

Is that 890 a daily or cumulative count? And what happened to the 120 peak ICU number we just discussed?

If the answer on this discrepancy lies within the modeling document, I can’t find it. I can only assume it’s a daily count (in which case I’m sorry, but we’d be f**ked).

Those pink boxes should send chills down your spine.

INH stands for “Integrated Northern Health”, which is basically the entire top half of the province, as seen on this map:

IRH stands for “Integrated Rural Health” and basically consists of everything else that’s not the Integrated North, or Saskatoon or Regina.

Unlike the INH, however, the IRH will at least get some dedicated COVID-19 facilities.

According to those pink boxes, if we ever reach that “surge” peak, both regions – aka everywhere in the province except Saskatoon and Regina – simply will not have enough ICU beds.

The plan?

I mean, I get that right now most ICU patients from outside Saskatoon and Regina are transported to a hospital in the city if need be – but I have many, many questions about how the SHA plans to move over 200 patients in at once.

So yeah, that’s a problem, IMO.

As for acute (non-ICU) hospital beds, we know that basically all non-emergent patients and elective (aka scheduled like knee replacements, not necessarily cosmetic or whatever) surgeries in Saskatchewan have been cancelled, which isn’t sustainable in the long-term but for now has freed up about 1000 beds currently awaiting COVID-19 patients.

Above and beyond that, the plan for increasing capacity in Saskatoon and Regina, presumably in part to accommodate the influx of patients from outside of those cities, is two new “field hospitals”: a 400-bed facility at EVRAZ Place in Regina and 250 beds at Merlis Belcher Place on the U of S campus.

Additionally, Regina is planning on using, if necessary, 80 beds at Wascana Rehab Centre for non-COVID related medical treatment, and Saskatoon has some plans to free up beds at the Parkridge Center.

I’m not getting into ventilators. We don’t have enough, neither does anywhere else in the world. But it has already taken way too long write this and I still don’t think I was able to really provide much clarity for you, so I’m going to leave it here.

All I know for sure is this: right now, what we’re doing in Saskatchewan is working, which feels like a horrible thing to say when “working” still equals four people dead. A phased-in approach to returning to “normalcy” without a vaccine means gambling peoples’ lives.

Because if we do decide to remove restrictions without a vaccine, what is the benchmark for shutting it all down again? One death? Five?

I keep hearing references to the “cure being worse than the disease” in relation to the economic pain being felt in Saskatchewan and across the world.

Then I read the stories of the four (well three, we don’t know much about the fourth) beautiful, vibrant people who have already been killed by this virus in Saskatchewan and can barely contemplate how we could deliberately sacrifice one more friend, neighbor, grandparent or fellow member of the Rider Nation.

I’m Tammy Robert. I’m a writer, but pay the bills consulting in political strategy, media and public relations. Feel free to email me anytime at tammyrobert@live.ca.

To Share this post on social media, please scroll down to beneath the pesky ads that someday I will get around to removing.