Pandemic curve to flatten by mid-June in central Canada, model predicts...other provinces way ahead

AboutAnything  | Greg McComb


   As Canada hits the mid-point of the second month of the coronavirus pandemic, the question on the lips of the quarantine-weary is: when will this thing end!? Several provincial premiers have provided tentative plans for easing of lockdowns as conditions improve, however, we have not heard any hard-and-fast dates....and there is a reason.

Photo by Jusdevoyage on Unsplash

  Although medical experts have a general idea of how the coronavirus virus spreads - the famous COVID-19 curve - much of this movement is random, so it's difficult to know how the disease will unfold over several months. 
 
    Many forecasts done early on by both federal and provincial governments have been well off the mark, most large overestimates of deaths and cases. As we are half way through the pandemic, more data is available to improve forecast accuracy.....so, as promised in my last blog, I took a crack at one. During a deep data dive, I came across an important aspect of this forecast that needs to be highlighted. I'll talk about this first.

Canada forecast is really a forecast of Quebec, Ontario...many provinces have already bottomed out

   As I combed through the coronavirus data, one thing that stuck out was how the vast majority of pandemic deaths were in either Quebec or Ontario; with a small percentage in other provinces. 

Data sources: Public Health Agency of Canada, Ontario and Quebec government reports and databases * Four Toronto-area Public Health Units: Toronto, York, Peel and Durham

The bar graph (May 3) above shows 93% of COVID-19 deaths (3,505 out of 3,760) were in central Canada, including the 'hot spots' of Montreal (1,410) and Toronto (762). Five provinces and three territories had less than 10 deaths. While B.C. and Alberta had small outbreaks: combined, those two provinces had 1/6 the deaths of Montreal alone.

  As discussed in the last blog, the virus spread quickly in Italy and the United Kingdom because they have large, densely populated cities with a lot of viral-laden international travelers traipsing through. The reverse is true for the Prairies, Maritimes and northern territories. Separated by thousands of miles from major cities,  their isolation saved them from any major outbreaks.

  This is important because it means the 'Canada' forecast presented here essentially becomes a forecast for central Canada, not other provinces. The case-based pandemic curves for four provinces illustrate this

Graphs generated by: ctvnews.ca/health/coronavirus

point, (left, April 28). While much is being discussed in the media about curve 'flattening,' these provinces (SK, MB, NB and PEI) have already peaked and come down to less than 10 cases-a-day, the watermark discussed by healthcare professionals to start loosening quarantines. P.E.I. and New Brunswick haven't had a COVID-19 case in several days.  This is likely one of the reasons Saskatchewan is discussing opening up its economy in early May, while Ontario's education minister said schools will stay closed until the end of May.
      
    Pandemic curve to flatten by mid-June in central Canada
                        
 So, back to the forecast model (updated May 4):  the results are in the bar graph below and I will walk you through those before explaining the methodology. The graph below is Canada's COVID-19 curve with the blue bars on the left real data, and the red bars a forecast, based on a 'quick-and-dirty' simulation I developed.

 Forecast model predicts flattening of pandemic curve in mid-June, with two-thirds the deaths of optimistic scenario in federal model.

Data sources: Public Health Agency of Canada (blue), date forecast by author (red)

     Like other COVID curves, this forecast curve follows a pattern: a rapid increase in deaths/day starting at three deaths on March 16th, peaking at 207 deaths on May 1st. Next, is the oft-repeated 

Photo by visuals on Unsplash

'flattening' (horizontal black line) in the middle - which should last about three weeks. We're are near the end of that now. Next, a gradual come-down should last about six or seven weeks, with deaths/day dropping from just over 150 deaths/day to less than 20/day by mid-June. Notice two things: the model predicts regular weekly 'waves' or cycles in deaths -- and the curve is not symmetrical: the left-side has a sharp incline while the right is less sloped, and so has a slower decline in deaths.

So, there are two important findings from this model: 

• The coronavirus outbreak is forecast to wind-down in mid-June, with deaths/day leveling off below 3/day by June 18. 

•  Total deaths from this 'first wave' are estimated at 7,107 or just over 7,000 deaths. That's about two-thirds the deaths predicted by a Public Health Agency of Canada (PHAC) model published in early April. The PHAC model predicts 11,000 deaths for an optimist  scenario that assumes 2.5% of the population is infected over the course of the pandemic. For 5% infection, PHAC's model predicts 22,000 deaths.

                        The methodology: how'd I do that?
  
     In my last blog, I explained the nuts-and-bolts of my coronavirus

methodology: I use deaths/day, rather than cases because of

Andre Ouellet on Unsplash

reliability issues with cases, and I assumed a pandemic starts with at least three deaths. The exceptions are the case-based pandemic curves of four provinces, SK, MB, NB, PEI. There weren't enough deaths in those provinces to make a curve. With one exception, the graphs are generated by the author using data sources sponsored by governments or university researchers. Those sources are cited under each graph. As for the forecast model, the logic behind it follows: 
    
    There are many pandemic curves, especially for the Spanish flu of 1918. However, the COVID-19 curve has a unique signature based on how infectious the virus is, and mortality rates. Therefore, what is required is an existing curve or curves to base my forecast on.  Both Italy and Spain started their pandemics earlier than Canada so in recent weeks we got a glimpse into what the 'back

Italy's COVID-19 curve  (source: worldometers.info)

  end' of a COVID-19 curve looks like, (see graphs to the left). Using a mixture of 'black art' and statistics I merged those curves to forecast the last 1 1/2 months of Canada's pandemic curve. For example, I took an average of the slopes of Italy and Spain's curves -- and estimated the length of

Spain's COVID-19 curve (source: wordometers.info)

Canada's flattening taking into account what happened in those countries.  As well, I borrowed the wave-like tendencies of both curves as best I could. As I said, this methodology involved a bit of 'black-art' so there was no one formula I used for the simulation. It was a series of formulas cobbled together in a spreadsheet.

      One last thing, a disclaimer: I reviewed pandemic curves from many other countries and although they all roughly take the shape a COVID outbreak, each is unique.  For example, curves in Sweden and Germany have wave-like properties but the waves in both are much steeper. 

Photo by Macau Photo Agency on Unsplash

A similar shift could occur in Canada, resulting in a much different forecast. As well, if Canadian provinces move to loosen quarantines or open up businesses too quickly, this could result in higher death rates. This model assumes "business-as-usual" for the duration of the forecast. Finally, few countries have 'finished' their pandemics so there is little data on how stealthy or resilient the coronavirus is, once quarantines end. Will cases and deaths spike again? Essentially, the world has become a large experiment, and we will learn about those viral properties in coming months.

    Seniors not youth are affected by the coronavirus....over 95% of  deaths are people 60 and over

  One other thing that stuck out during my data dive was how older people have been disproportionately affected by the pandemic, (see graphs below from April 20). 
      Not just a little, a lot.

Data source: Situation of the coronavirus (COVID-19) in Quebec, Quebec Health

     In Quebec,  diagnosed cases of COVID-19 (blue bars) are spread evenly among age groups: 60% are under sixty and the rest are senior citizens, (40%).  The big difference is the survival rate of patients: younger people are healthier and have stronger immune systems so they get sick, but survive. Under one percent (0.2%) of  the 10,977

Photo by Michelen Studios on Unsplash

younger patients died, (see brown bars). That's in line with the regular flu.  Not so for seniors. Look at the blue/brown bar coupling on the far right, COVID patients over eighty years old. It shows this age group makes up one-fifth of diagnosed cases, yet they form 71.1% of deaths. In raw numbers, 623 of the 877 people who died from the coronavirus were the eldest age group, over 80. With weakened immune systems and comorbities like advanced cases of heart disease or diabetes, older seniors don't last long once infected. 

    The same thing is happening in Ontario, their graph is like Quebec (see graph below). In that province, 20.9% of diagnosed COVID

Data source: COVID-19 in Ontario, Public Health Ontario

cases were for people over 80, while 67% of total deaths are for this elderly group. By the numbers, 416 of the 622 total deaths were seniors over 80. 

 Over the past month, we have heard over-and-over about how this crisis has unfolded in senior homes, with dozens of deaths occurring in a single residence once an infection gets hold. This has been the most tragic part of this pandemic; especially the fact these parents and grandparents had to die in isolation, without the support of family members. The statistics don't tell this story...a complete overhaul of seniors' housing policy is-in-order.

                Pandemic policy over the next few months

   So, what does all of this mean for pandemic policy?  The federal government wants a phased and coordinated approach with the

Photo by Nathan Dumlao on Unsplash

provinces as Canada opens up over the next few months.  This makes sense.  As discussed above, provinces with pandemic curves that have 'bottomed-out' should end their lock downs first and act as test-cases for the rest of Canada, while 'hot spots' like Montreal and Toronto go last. Close monitoring of the early movers will let other provinces know just how fast-or-slow they can go, as they lift quarantines and get their economies moving again.

Second, the age analysis showed a stark divide between how old-and-young people react to a coronavirus infection.  Older people are extremely vulnerable while the young get sick, but survive.  As quarantines open, this would suggest that senior citizens (especially those eighty or over) should be the last to roam around freely in public: they should keep a social distance, and wear non-medical masks.  In contrast, society should be less worried about younger people, especially those under forty. With precautions, re-opening elementary and high schools could be done in an early phase.

  Finally, I have written little about the massive costs to the economy of closing everything down during a pandemic. Pandemics involve a harsh trade-off  between saving peoples lives during lock-downs, on the one hand -- and crashing the economy with millions of jobs lost, on the other.  Using statistics and graphs, I will try to try to make sense out of this in my next blog...

Email: gregmcc07@gmail.com