Numbers
2437 cases reported in Tbilisi this week. My projection was 2379 (with margin of error: 1856-2903). The positive test rate for the week went up from 2.01 to 2.5.
The weekly infection rate (the growth from one week to the next) averaged 1.22 last week and 1.27 this week. We can expect it to continue increasing as the UK strain becomes more prevalent and the impact of relaxed restrictions continues to set in.
Due to the holiday this week (April 9th) the reported case numbers may be slightly lower than normal, meaning we might be slightly worse off than we seem.
Projection and Risk Assessment
My exponential growth model predicts 3315 cases reported in Tbilisi next week (with margin of error: 2586 – 4045). Including the increase in infection rate from UK cases, I’m thinking we’ll see at least 3500. Depending on how much impact other news events (like opening movie theatres) have, we might see even more. Also there are no holidays coming up, that I know of. So my guess is the numbers next week will be on the high end of the prediction range.
Risk assessment based on worst case scenario is about 1 in 250 people in Tbilisi will be infected and transmitting.
At current infection rates, we’ll see roughly 5900 cases in Tbilisi the week ending on Easter (Sunday, May 2nd). That’s the equivalent of around November 6th – 7th in the second wave. Hospitals filled up when we got to around 10,000 cases per week in Tbilisi, which, at current rates, we’ll hit by the second week of May. Hopefully cases won’t go much higher than that, but it means we’re probably looking at a very hard lockdown by the end of May and continuing through June, or else some kind of unprecedented collapse of the medical system. This might happen a bit earlier depending on the state of the UK strain. Speaking of which:
News
Gamkrelidze says the UK strain makes up 20-30% of new cases. Unfortunately there doesn’t seem to be any context for this number – is it an educated guess? Is it based on random sampling of all new tests? I have no idea how much to rely on it. If it’s correct, however, then the UK strain should overtake the original strain by around Easter (that is, more than 50% of cases will be UK), and that process would cause about a 10% rise in the overall infection rate. That sounds small, but in practice what it looks like is an additional 900 cases that week in Tbilisi. It also gets us to 10,000 cases per week in the first week of May rather than the second. This might mean we get into – and out of – lockdown more quickly. Before you celebrate, note that an earlier peak means that people who otherwise would have lived to get vaccinated won’t. The goal of public policy should have been to delay the third wave as much as possible to get as many people vaccinated as possible. Instead, the timing couldn’t be worse – we’re heading into a peak at the exact same time as the year’s biggest holiday and we’ll be forced into lockdown at exactly the time when we should have been opening up for tourism and mass vaccination is scheduled to start as soon as the third wave has passed.
Again, I have no idea how accurate Gamkrelidze’s number is, but it’s definitely plausible given the progression of the UK strain elsewhere. Note that based on the last numbers released before this statement (about 72 confirmed UK cases) I had projected that the UK strain wouldn’t be dominant yet by Easter, so my overall prognosis is now a little bit worse than before in terms of how quickly the numbers will rise.
And look – as I wrote this, apparently Gamkrelidze has increased his estimate to 50-70%! It’s breaking news, people! At this rate, by the time I finish this post it will be 80 – 110%! (That was a joke, obviously – it can’t be 110%). This is just unbelievably sloppy, and Gamkrelidze needs to be more transparent as to where he’s getting this information from. To give estimates ranging from 20% – 70% in a period of four days with no explanation is just absolutely maddening. Where’s the data? Where’s the transparency? Apparently not even Gamkrelidze has any idea how accurate Gamkrelidze’s number is. The UK strain didn’t jump from 20% to 70% in four days; this is clearly a case of Gamkrelidze not owning up to his mistakes. If any of my students are reading this: this is why you need to cite your sources.
Speaking of a maddening lack of transparency, the Georgian government has decided to embiggen Easter this year in order to fight the pandemic. Apparently May 3rd and 12th were already public holidays, so now they’ve added the 4th through 11th, meaning the entire Easter holiday will last from Friday, April 30th (Good Friday) to Wednesday, May 12th (St. Andrew’s Day). Things will then go back to normal on Thursday, May 13th – unless, for some reason, you need to access emergency medical care, in which case you will probably be SOL.
What to make of this extended “holiday”? Is this an Orwellian spin on the lockdown I’ve claimed we’ll need by the end of May? Is this a cockamamie scheme that’s just crazy enough to work? Is this a morale booster for beleaguered workers desperate for some more time off? Or a poke in the eye to restauranteurs who have demanded early openings? The fact that this announcement has apparently been met with such universal confusion that a government spokesman had to clarify what they meant by “holiday” tells me that nobody really knows. What will be open? What will be closed? It’s anybody’s guess. Again: no data, no transparency, no real clarity on what any of this means. The government’s communication here is just egregiously bad and getting worse.
A Slight Doubling
The Ministry of Education and Science (MES) reports that there has been a slight increase in the number of coronavirus cases in schools. The number of cases among teachers and administrators has increased from 0.1% to 0.2%. “Twice as many cases” seems like a bit more than a “slight” increase, but who knows – maybe it went from 0.14% to 0.16% and they’re just rounding the numbers to 0.1% and 0.2%. Apparently the MES is releasing these numbers via facebook post anyway.
That facebook post initially reported that 0.4% of students had tested positive. This would be an astronomical increase (it was 0.02% two weeks ago, and 0.03% last week, so we could expect something like o.04% this week) and when I saw it reported on 1tv.ge I assumed it was a transcription error. Then I checked and found the same number elsewhere (on.ge, newsreport.ge, imedinews.ge) and assumed maybe the MES had made a mistake. I looked at their contact page and noticed that they had a facebook page, and so I went there and left a comment asking if the number was correct. Their social media person responded quickly – within about 30 minutes – confirming that it was a typo, and the real number was 0.04%. Note that most of these sources have now corrected the error, without printing an acknowledgement of it.
What’s the takeaway? It’s nice having a public agency be responsive and quickly correct small mistakes. I’m tempted to be critical of a Ministry using facebook as its primary PR tool but on the other hand it does allow for this kind of quick back-and-forth. A hypothetical journalist who covers coronavirus in Georgia would have been able to do what I did – notice the number didn’t comport with previous weeks’ statistics, asked what was up, and received a response before publication time – rather than reprint the mistake.
On the other hand, that didn’t happen, so…
This story also shows us that it’s important to follow news critically and conscientiously. I usually prefer analysis to raw news because the analysts are going to double-check the press releases for typos and other types of mistakes. The analysts are going to know enough about a topic to notice when a number just sounds wrong. However, in Georgia all I can get is the raw news, so I have to do the analysis myself. On the other hand, reading analysis means you get the news filtered through more than just fact-checking. Every analyst, myself included, has biases and blind spots. The same factors that cause me to pay extra attention to numbers in schools (the fact that I’m both a teacher, and a parent of a child who is learning remotely) might cause me to pay less attention to other things.
That’s why I try to present my reasoning in addition to my conclusions – and why it’s frustrating when an actual expert like Gamkrelidze presents his conclusions (like the “20-30% of cases are UK cases”) without his reasoning or evidence or data. If I don’t know how Gamkrelidze got that number, it’s impossible for me to independently evaluate his analysis in light of his biases and blind spots – or to predict that he’ll change his mind four days later. If I don’t know why Gamkrelidze keeps refusing to say that we’re in the third wave, then I don’t know if I’m mistaken about something, or if he is, or if he has access to more/different information, or if he’s working on a different definition of “wave”, or if his statements are motivated by a communication strategy to manipulate the public rather than to tell the truth, or something else entirely. Maybe I should just try writing to Gamkrelidze on facebook too.
If I were doing this full time (and spoke better Georgian) I’d be calling the Ministry of Health and the NCDC and various other agencies daily and following up with all the open questions I’m writing about here. I don’t know why there apparently isn’t a journalist in the whole country who gets paid to cover covid.
Anyway, let’s run the same analysis on the MES numbers as I did two weeks ago. If 0.2% of adults in schools had undiagnosed covid cases (which were caught by screening tests) then about 0.2% of the population at large will have undiagnosed covid cases. 0.2% of about 4 million is about 8000. That suggests there are probably about 8000 active cases of covid nationwide that have not been diagnosed. The number of known active cases is 7190. So, the MES numbers suggest, again, that we’re missing about half of cases. I know it looks like we’re missing slightly more than half, but honestly 0.2% isn’t enough precision to make this calculation at that level of significance. If they’re rounding up from 0.18% then that suggests 7200, and I don’t have any evidence that they aren’t rounding up from 0.18%. They could also be rounding down from 0.24%. Also, teachers aren’t a random sample of the population so despite studies that teachers are generally infected at about the same rate as non-teachers there might be some confounding factor in Georgia in particular. The MES number should give us a small amount of additional confidence that our other estimates – based on national reported testing and confirmed cases – are approximately in the right ballpark; it shouldn’t be used independently as a precise estimate of incidence in the general population.
Vaccines
There has been an increase in evidence that the AstraZeneca vaccines may cause blood clots. Earlier I had said this was almost certainly not true, so obviously I’m really glad I put that “almost” in there. Still, I think I’ll take the hit on this one and say I may have been too quick to dismiss the early evidence as a statistical artifact. I don’t pretend to have any special knowledge about vaccines and my position on any vaccine is “the experts are probably right about whatever they say about it”, and so when the relevant experts were saying the number of clotting cases didn’t reach statistical significance I basically adopted that position, and now that they’re saying there might be a link after all I’m adopting this position. Should I be less certain about the claims of vaccine experts in the future? Probably not. I think this type of mistake is probably rare enough that I’ll still be right more often if I believe whatever vaccine experts say about vaccines than if I believe what strangers are telling me on facebook, even if in this one particular crazy case the strangers on facebook turn out to have been right.
Georgia is still working on getting a bunch of vaccines. I’m not going to go into specifics about most of them because, again, it’s not going to matter much for the third wave, which will peak by late May, how many vaccines we end up with in July. The exception here is the 100,000 Sinopharm vaccines which the country has now and plans to start giving by April 20th. That’s obviously good news for the 50,000 people who will get vaccinated, but it’s not going to move the needle much in terms of overall infection rate or the course of the third wave.
Well, I’m going to end on that. It’s been a hectic news day, especially for a Sunday, and this post has run very long. Good luck, everyone!
Daily numbers from stopcov.ge and 1tv.ge.
Georgia On My Mind 