“I Was Wrong” on Reopening Ottawa

Roughly two weeks ago I wrote a post Against Reopening Ottawa. Since then, my predictions have turned out to be mostly wrong.

First, here’s same the chart I published at the time showing a moving average of new cases in Ottawa:

And now here’s the same chart updated as of the data available today from Ottawa Public Health:

At first glance I sort of feel vindicated, since I predicted cases would keep going up, and they definitely did. But pretty much all of the particulars of my predictions were wrong. In relation to Ottawa cases, my three predictions were:

  • Based on the original data I inferred a steady doubling time of two weeks; actual new cases went up way faster than that, and then flattened completely, and now seem to be decreasing again.
  • I predicted a brief slowdown in new cases starting around July 21st, due to Ottawa’s mandatory mask bylaw going into effect on July 7th. If you squint a little it kinda looks like that part might have been right, but on closer inspection it isn’t. The timing is off (the graph flattens well before masks would have an impact, and starts decreasing well after) and the sharp drop in new cases is so recent that it’s liable to disappear in the next few days anyway (Ottawa reports cases by first symptom where possible, so each new day’s data tends to backfill more cases over the last week or so).
  • I predicted that starting August 3rd or shortly after, cases would spike again due to stage 3 of reopening. This one isn’t right or wrong yet as we haven’t gotten there, but I’d no longer make the same prediction today.

I also made some less rigorous province-level predictions that turned out to be (mostly) wrong:

  • Ontario did see a brief more general increase in cases after my post, but that has stopped, and if you zoom out cases still seem to be trending clearly downward everywhere but Ottawa. I’ll give myself a win for predicting that Peel and other previous hot spots were mostly under control at that point, but Ottawa is now one of the regions with the most reported daily cases, making it an outlier, not a pack leader.
  • British Columbia cases have continued to pick up again, but apparently in the same ways as Ontario; they have one region that’s become a hot spot and everywhere else is well under control. The main difference is that BC’s infections are so low to begin with that their hot spot is swamping their general numbers and making it look like the whole province is in trouble.
  • Alberta I admit I don’t really understand. Their data seems to have done the same thing as Ottawa: a sharp increase followed by an immediate flattening. But they only have two major population centres (Edmonton and Calgary) and neither city on its own shows this pattern. If anybody has a better understanding, please leave a comment.

Finally, there’s a few other miscellaneous points I wanted to make:

  • As I mentioned parenthetically above, Ottawa reports cases by date of first symptom where possible (or even date of infection, if the source is known), which means that new cases reported on a given day are almost always backdated by a few days and up to two weeks. Thus we should expect the last week or so of numbers to show up as lower than they will be in the final tally. This is why I’m not convinced by the magnitude of the recent “dip” in new cases in my chart above, and explains why what was a gradual uptick in my previous post turned into a sudden spike. It does also vindicate me “reading too much into the graph”, although I didn’t realize this at the time so I can’t take credit for it.
  • Ottawa Public Health addressed the recent spike shortly after I published my last post, claiming that it was unrelated to reopening and more related to private parties and lack of distancing in private spaces (in particular among younger age groups). This seems like a generally plausible explanation, and I don’t have any better guesses that explain the weird shape of the data. There’s been so much garbage floating around from the WHO and the CDC, it’s nice to get evidence that my local health authorities actually know more than I do about COVID.

Against Reopening Ottawa

I.

This is not the COVID post I thought I would be writing a few weeks ago. I honestly didn’t think I’d be writing a COVID post at all.

A few weeks ago, most of Canada seemed to be under control. There were a few hot spots left in e.g. Southern Ontario, but almost any other graph you looked at showed a nice, clean downward trend. How quickly things change. I live in Ottawa, so I’m going to focus there. This data comes directly from Ottawa Public Health (though the graph is mine):

You can see things getting consistently better in May, staying quite steady throughout June, and then starting to creep back up around the beginning of July.

This is a huge problem.

It may seem like I’m exaggerating slightly – after all, Ottawa (a city of 1 million people) went from roughly 4 new cases a day, to roughly 8 new cases a day, over the span of two weeks. That hardly seems comparable to the huge surges being seen in the southern United States or other problem areas. Ottawa’s health care system and hospitals have plenty of capacity. Our testing turn-around time remains under 48h, and our testing throughput remains high. We have a mandatory mask bylaw in place. There seem to be a lot of things going right.

In point of fact there are a lot of things going right – I’d still rather be in Ottawa than in any part of the US. But just because some things are going well, doesn’t mean we’re not still in trouble. Going from 4 to 8 new cases a day is an increase, and any increase is really bad news.

II.

At the risk of rehashing a topic everybody is sick of at this point: virus spread is modelled exponentially (this exponent is the “R0” everybody keeps talking about). R0 isn’t a magical fixed value; when something changes (e.g. people start wearing masks) then R0 for the virus changes too. When R0 is less than one, the virus gradually fades out of the population, as was happening in Ottawa in May. When R0 is exactly one, then the number of new cases stays flat, as was happening in June. When R0 is greater than one, the virus starts picking up steam and spreading again.

R0 in Ottawa is, clearly, greater than one at this point, and has been since roughly the beginning of July (or maybe a little later, depending on how much you’re willing to smooth the graph). This isn’t a big deal for today – we can still handle 8 new cases a day, or 10, or 12. But the thing about positive exponential growth is that it keeps going up, faster, and faster, and faster. As a very rough approximation, let’s assume that the last two weeks are representative of R0 in Ottawa right now, and that our new cases continue to double every two weeks. That would mean that August 1st we’d be dealing with 16 new cases per day. Not too bad. By the end of August, 64 new cases a day – bad (until very recently we had fewer than 60 active cases total), but still manageable. By the end of September though… 250 cases a day, which is probably more than we can handle. And so on. If the trend continued through to Christmas (which is, granted, very unlikely) then we’d be looking at roughly 16000 new cases a day.

Now there’s a lot of reasons we’re very unlikely to hit 250 cases a day, let alone 16000. If new cases increased that much I assume the city would re-institute some kind of lockdown, and at that kind of load other factors like herd immunity would start kicking in as well. But when cases are already increasing it seems like a really bad time to start reopening even more. And yet…

III.

The general consensus seems to be that there’s a minimum of two weeks of lag between the development of actual cases, and reporting. This time accounts for how long it takes somebody who’s been exposed to incubate the virus, develop symptoms and go get tested. Of course in some places where testing is overloaded, the delay can be much more than that, but Ottawa is not overloaded, so let’s assume two weeks.

Ottawa officially entered “phase 2” of our reopening on June 12th, though in practice most businesses were not ready on the day of, and reopened piecemeal over the following week; let’s take an average reopening date of June 15th. Two weeks after that brings us to June 29th, which is, (surprise!), right at the beginning of our uptick in cases. This is a bit of evidence that our “phase 2” reopening was in fact too much; R0 is now back above one, and the virus is spreading.

In worse news, despite phase 2 already being too much, Ottawa officially entered “phase 3” of our reopening yesterday (July 17th). Not only does this seem like a bad idea in general given that R0 is already back above one, but phase 3 includes a huge swath of very risky activities whose impact on R0 will almost certainly be far greater than the impact of phase 2: indoor service at restaurants and bars, movie theatres, museums, etc. As with phase 2, a lot of businesses weren’t ready day of; if we take an actual phase 3 reopening date of July 20th, and add two weeks, it’s easy to see an even bigger spike of cases coming down the pipe, starting around August 3rd. (August 3rd is a statutory holiday here, so in practice I expect the data might not show up until a few days later.)

The one bright spot in all this is that Ottawa made masks mandatory while indoors, starting July 7th. That will presumably have a big impact on transmission rates, and was less than two weeks ago, so we won’t see it in the data yet. Hopefully new cases start to drop again around July 21st (two weeks after mandatory masks), and if we’re very lucky then that decrease will entirely counter-act the increase from both stage 2 and stage 3 reopenings. But that seems like a lot to ask, especially since many people were already voluntarily wearing masks even before the mandate. Only time will tell.

Ultimately, I predict continued increases in Ottawa until around July 21st, at which point the trend will reverse due to mandatory masks, and we see decreases again until August 3rd or shortly afterwards. Then we’ll see the impact of phase 3 reopening, but I can only imagine that it’s going to be bad. I suspect by late August it will be clear that phase 3 is unsustainable and will have to be rolled back. I only hope we don’t learn that lesson too much the hard way.

IV.

[This section is more of an appendix of other little things I didn’t fit in the main post.]

Somebody I discussed this with argued that I’m reading too much into the graph I presented. The data from end of June up to July 9th actually looks well in line with the rest of June, and the growth after that point could very well just be random variance as was likely the case with the brief spike from June 7th to 13th. I suppose this is possible, though it feels unlikely to me. Time will tell, if cases continue to rise or not.

Alberta and BC (two other provinces) are also seeing recent spikes in cases after reopening, though oddly Ontario (the province that Ottawa is actually in) has not. I haven’t dug into the regional data to back this up, but I imagine it’s because the Peel region and the other “hot spots” I referenced earlier are finally under control and decreasing rapidly, which is balancing out the gradual increase in Ottawa and other places. Again, time will tell. I expect we’re already at the bottom of that particular wave, so Ontario-wide cases should start ticking up again (slowly) this week.