Covid Geography and Wave Height
Why the fifth wave may have unexpectedly peaked without emergency interventions
Last week I wrote very pessimistically that I expected we’d have emergency interventions here in Georgia and that the fifth wave would therefore peak this weekend.
In fact, we had no extra interventions, but the fifth wave seems to have peaked anyway: the 7-day average has fallen for two days in a row. That’s great news, if the trend continues. It also requires some consideration: why did we peak, if not for interventions?
First, it’s possible that people decided to change their behavior in response to frightening numbers and dire warnings from people like me. Maybe the combined effect of people deciding to order in rather than dine out, or defer non-emergency medical care, or at least wear a mask when in public, added up to enough of an effect to get transmission rates down. The government may also have stepped up enforcement of existing regulations.
Second, people are self-isolating. Part of this is due to government-imposed isolation measures - at various points over the last few weeks, we’ve seen news that over half of Tbilisi’s kindergartens have closed, that over 100 schools nationwide have closed entirely, and that over 700 classes have gone remote due to covid cases. But also, businesses may be taking similar measures, and individuals may be self-isolating in response to cases in themselves or others close to them. Currently about one in seventy Georgians has a diagnosed active infection - just over 1.4% of the population. If each of those people self-isolate, and each has an average of two non-overlapping close contacts who also self-isolate, you’re looking at almost 5% of the population self-isolating. Add people who have symptoms but didn’t test but self-isolate anyway, and people who have flu or a chest cold or some other non-covid illness and self-isolate, and it’s plausible that upwards of 10% of the population could be self-isolating.
That sounds like a lot, but I can say anecdotally that about 10% of my students have gone into voluntary self-isolation due to covid or covid symptoms, and about 10% of the kids in my son’s class have done the same. This could be selection bias - I’m not sure that “private school kids” is the most representative sample for figuring out the behavior of the general public with respect to self-isolation - but I think it works as a very basic sanity check.
Third, we may be approaching some kind of local immunity/isolation threshold for this wave due to network effects. Studies have found that the people who have the highest number of social connections tend to get sick earlier in disease outbreaks (which intuitively makes sense) but of course once these people are “removed” (in self-isolation, or immune due to having recovered) from their social networks, the network as a whole is less connected, and therefore the disease spreads more slowly. Once you’ve removed enough connections in a particular community, the wave will have to slow down.
This third factor - local network effects - probably explains some of the geographical factors of waves. For example, I’ve noted previously that small island nations and microstates tend to have higher (per capita) waves. If a wave moves through a population slowly, based on geography - if it takes, for example, a month for the hotspot to move from Batumi to Kutaisi - then any given set of sparsely-connected regions will have lower per capita waves than any particular highly-connected region within that set. In other words, if places like Adjara, Imereti, Tbilisi, and Kakheti are densely connected, but relatively isolated from each other, then we’d expect that the highest per capita wave in Adjara would be much higher than the highest per capita average wave for all of Georgia. Just as in the US we’d expect the highest per capita wave in any given state to be higher than the highest per capita wave for the nation as a whole.
Indeed, Georgia’s highest wave peaked at 1665 cases per million in a single day (August 17th, 2021). The same day, Tbilisi hit 2112 cases per million. On August 13th, Adjara had had 2291 cases per million. On October 19th, Imereti had 2390 cases per million and Kakheti had 2946.
This is happening because higher waves in one region are “balanced out” by lower waves in other regions - which in turn happens because the average person in any given region only has close contacts within that region.
Breaking the numbers down by region also helps us to notice something: in absolute numbers, the 4th wave was about 10% higher than the 5th wave. However, per capita, per region, the 5th wave was much higher - 13% higher in Imereti than in Tbilisi, and 29% higher in Kakheti than in Adjara.
I don’t want to place too much credence in these numbers because of trends in population mobility - for example, in August people tend to travel to the coast, so the population of Adjara grows and the population of other regions (including Tbilisi) shrinks. That trend would mean that Adjara’s per capita number would need to be adjusted down by a lot, and Tbilisi’s up by a little. Also, I’m not sure how good the government’s regional population numbers are in general - how many people, for example, officially live in the regions but unofficially live in Tbilisi (e.g. through informal rentals which are never reported to the government to avoid paying rent tax).
But the broad trend we’re seeing is that the 4th and 5th wave seem to have occurred mainly in different regions, and that the 5th wave - which was centered in the less populated regions - was more severe per capita in those regions, but less severe overall for Georgia as a whole.
This is my answer to critics of government intervention who argue that interventions have no effect. Having the 5th wave peak with no apparent intervention presents evidence in favor of this view. However, I would argue that if intervention had come sooner, numbers in Imereti and Kakheti wouldn’t have hit 2390 and 2946 per million, respectively - just like they never got that high in Tbilisi in the 4th wave.
This geographical perspective has implications for the next wave. NCDC is still saying we’re going to have a tough November. If this wave truly is over, we should actually expect numbers to decrease throughout most or all of November (although death numbers will continue to rise, again, until the middle of the month). But the question is, what if covid just moves again? What if there are still major communities or regions of Georgia where there hasn’t been a significant Delta wave? What if covid rebounds in Tbilisi as soon as people stop taking social distancing and self-isolation measures?
As to the first question: Samegrelo, Kvemo Kartli, and Shida Kartli haven’t been hit nearly as hard as the four regions I discussed above. Together, they account for about a million people. We could see a wave hit there next. There isn’t any obvious reason why a wave would hit those three regions simultaneously, though. I would say that geographical conditions are now in our favor: if the Delta wave has in fact mostly run its course in most of Georgia’s most populous regions, then the next wave or two should be relatively mild unless a more contagious strain arrives.
As to the second: I do think that people will again relax their behaviors and take on more risk, but I don’t necessarily see this causing a major wave - instead, I think it will manifest as high floor after the fifth wave, or maybe a small sixth wave, similar to what we saw after the second wave last winter. Of course, how the government and church handle the holidays may play a role, and winter tourism season may play a role - I wouldn’t be surprised if there are outbreaks in and around the major ski resorts this winter. But overall, my feelings about the next few months are a bit more optimistic than they were last week.
I still think we’re going to continue seeing intermittent closures and all that, but now that we’ve seen some unambiguous empirical evidence that the waves are self-limiting at around 2400 - 3000 cases per million in any given region, I think we can revise our worst-case scenario down significantly compared to last week when we didn’t know how bad things could potentially get.
That said, political and epidemiological factors are still against us. Georgia still has a very low vaccination rate and a very disengaged government, and probably about a third of the population is still totally susceptible, probably concentrated in regions that haven’t been hit hard yet. I wouldn’t be complacent yet and I think Georgians should press the government on things like vaccinations for kids, boosters for adults, flu shots, and better enforcement of basic public health precautions.
I also think the government should be a bit more transparent and public health officials should communicate more effectively - as opposed to forcing people get their covid information from people like myself or Bidzina Kulumbegov. Not that there’s anything wrong with private citizens taking the initiative to provide good information to the public, but when the government doesn’t do this effectively it contributes to the sense that the country is rudderless and devoid of leadership, which I think is bad for the kind of social trust and cohesion that would actually help us fight the pandemic effectively.
But overall, in light of the trend from the past two days, the outlook from this week is much less pessimistic than the outlook from last week.
Just as an administrative note: I know I’ve done two covid posts in a row now, and I’m trying not to have this substack be wall-to-wall covid. I’ve got posts brewing on lots of other topics in culture and politics and I’ll try to make sure the next couple are non-covid related.