It strikes me that in Ontario we’ve reached a turning point in our fight against the novel coronavirus, SARS-CoV-2. As we enter Phase 1 of restriction easing, for better or worse, we are becoming lab rats in our own potentially dangerous human biology experiment. This phase should, in theory, provide information in terms of which activities do truly lead to spikes, and will (hopefully) showcase our government’s capacity for contact tracing, widespread testing and the surgical isolation of infected individuals.
In light of the delay between sweeping policy changes and learning their consequences, I thought it might be interesting to keep an eye on something that I haven’t been able to find anywhere else (although it must exist, any dummy can throw up an article these days, I hear). I was curious about the rate of positive cases v.s. the absolute number of positives shared more frequently in the media (if we only did 3 tests one day, only finding 3 positive cases in Ontario is NOT a success story, “The News”). More importantly, revisiting these graphs in roughly 3 weeks time will retroactively tell us if the re-opening was a good idea (insert 2020 hindsight joke here).
Since I could not find a downloadable repository of testing data anywhere, I figured now was the time to start collecting data. I will drop some links that I’ve been visiting for information below. The numbers I am using come from Global News, which says it grabs its data from the Ontario Ministry of Health. Missing data was manually found in backdated repositories, and numbers before March 31st were not included, due to inconsistent test-number reporting.
Notice that the bottom plot is the one I was looking for and failed to find elsewhere online. The top plot is the absolute number of positive tests found, (often reported in the media, along with the death toll). The middle plot is the number of tests done per day (also not readily available to my knowledge), which, as expected, climbs over time as testing capacity increases.
My first observation regarding the percentage plot is that the peak of positive tests doesn’t correspond with the absolute positive’s peak. This is likely a failing of the percentage plot, because only symptomatic patients were tested towards the beginning of April. As expected, the rate falls off when testing ramps up, i.e., more generalized population testing is introduced (healthy healthcare workers, etc.).
As more generalized testing is introduced, I think the percentage plot overtakes the accuracy of the absolute daily positives plot. When we fast-forward past the days where the percentage plot hits its peak at around a 17% infection rate (on a day that only around 3,000 tests were done), we hit more recent data, where around 15,000 tests are performed daily. Suddenly our percentage plot becomes less noisy and we can extract better observations. We see a positive rate fall from around 5% positives to a seeming plateau at around a 2% positive rate. I will caution that we continue to look at a skewed population of people seeking out tests, so the total population incidence rate is likely significantly lower.
Without a very large ramp-up in testing capabilities, however, a continuation of the 2% rate in future days should measure the same skewed population, and would likely indicate a continued plateau. I believe that this curve is the one to watch in coming weeks, as a benchmark for our performance. If the proportion of infected patients begins trends significantly up in 2–3 weeks time, it will probably mean that we were hasty in easing restrictions. If it stays constant or reduces (as testing either increases or stays the same), we can be much more confident that Phase 1 is humming along smoothly, and begin to spend our time more productively, by feeling apprehensive about Phase 2.
Final Thoughts (for now)
The goal of this analysis is to be vigilant in case the data we are receiving drops in quality in future weeks. From what I can see, present cases seem to have genuinely trended downwards and settling at a fixed rate. Ideally, we’d see less of a plateau and more of a continued dip, indicating a trend towards eradication, however, asking for more than fairly low in the face of the threat of an exponential explosion, without a full sacrifice of rights and freedoms might be just a bit greedy.
This is a more recent snapshot of the data:
Re-updated after May 22, when I said: It probably does not need to be said that a decrease in testing and increase in absolute positives is not a good thing. This is especially given the fact that this is still the impact of pre-relaxation data that we are looking at. Hoping we can test more asymptomatic people and that the numbers start looking better in the future, because the trend is looking like the second wave is already building with less of a lull than anticipated.
Now its May 31st, the press is referring to the brief increase in positives as the “mother’s day” increase, as it seemed to coincide with the timeline roughly 1–2 weeks out from mother’s day, where people decided to relax their restrictions and visit home. Oddly, since the opening was May 19th, we are still currently looking a bit back in time, seeing if we retroactively got the numbers low enough to justify the relaxed restrictions. So far, we can cautiously say we do not currently seem to be exploding. I should also highlight that in order to increase capacity/make up for the lull in testing, we did recently relax the people getting tested, so it makes sense/is comforting to see our numbers dilute a bit and go down. That hopefully means we don’t have as many asymptomatic/sleeper cases as we could have! Going to end there on an optimistic note for now.
June 7: Its difficult to distill everything that’s happened so far. The government of Ontario has reported that a lot of the positives are actually backdated from the last couple days, but no word on from when. So the positives might look artificially high now, but might also have been artificially low at an earlier date. Difficult to draw conclusions re: Phase 1 based on that. Things didn’t seem to explode based on Phase 1 however, in general, which would be promising, if the recent protests weren’t likely to cause a much larger effect soon. Hopefully not, but we will just have to wait and see. The good is: testing rates appear to have improved, despite the hiccups in reporting, and no obvious case explosion thus far.
Sept 28: now that we are in phase 3 and schools have opened, the only good news seems to be our testing capacity is slightly ramping up. While our positive percentages remain below a 2%, there are way way more cases each day and the curve is starting to look exponential again. Not to mention, I’m graphing again, which is never a good sign. One more grey-area positive is that Ontario’s testing dashboard has gotten much cleaner and nicer looking: https://covid-19.ontario.ca/data. Their data doesn’t consistently go back quite as far as my aggregation above, but its now downloadable in a number of formats, making collection much easier! Lets hope we stamp out this spike and don’t need it any time soon…
Oct 21: things could look more positive. We’re back to phase 2 in many places in Ontario. Arguably, we should be in more. We’re over a 1% positivity rate (2.42%), which is a danger because it leads to high prevalence of multiple types of transmition. Deaths are up and a couple of clinical vaccine and treatment trials were recently halted (with the vaccine still halted in the States). There is a rumour that there’s been spine swelling (MS or MS-like symptoms) in two of the volunteers, but information is not being made publicly available, sowing distrust among the public and vaccine experts, alike. On top of that, a 28-year-old, male volunteer in the Brazil clinical study of the vaccine died yesterday, and they have not yet released information as to whether he received the treatment or the control (article here). The only silver lining is I think testing might be down in Ontario, because people don’t realize they can go get tested again since the backlog is now cleared. Hopefully those numbers, as well as positivity rates will start looking better soon.
Oct 29: 934 “new” cases, after the latest spike of 1042. Tests have dipped to 35,621 and 2.62% of individuals tested had positive results. Things are continuing to look worse. Seems like we’re trending to 1k/day. Hopefully the latest measures start to take hold before that happens. Testing is down, and far below expectation, and I’m not fully sure why.
Update: to be honest, this has been too depressing to post about for a few days. It seems like nobody is trying. Deaths have been in the double digits since last Tuesday, and the virus seems to be gaining momentum. There was a brief glimmer of hope earlier, with a positive vaccine announcement, but to be honest, it all seems really far off.
Websites I Like to Look at Obsessively (both courtesy of my little brother):
https://globalnews.ca/news/6804864/coronavirus-ontario-16k-tests-per-day/?fbclid=IwAR2v2IbHM10FDuRh2MQ_R4yQkdzm5lGKN9V_NLEFKsbN6EwHH7lbGJREwIw (scroll to the bottom and hover to see proportions)
A Note on My Tiny Data Collection
If you’d like the data I’ve collected from the earlier days (potential trend-altering typos and all) feel free to grab it out of the first lines of the code that’s now hosted on my GitHub (https://github.com/DaniRip/Covid-ON-Analysis). If that’s not convenient/well-formatted enough for your purposes, you are also welcome to just ask me and I can output it nicely for you!