Seasonality of human coronaviruses (other than COVID-19) in Stockholm, Sweden. Source:
Place your bets
College is starting early in Virginia this year. The Ancient and Honorable College of William and Mary will start its fall semester this year on August 19, and will finish up just prior to Thanksgiving. That matches what U VA announced just a few days ago.
My guess is that both colleges are doing this to take advantage of the expected seasonality of COVID-19 infection rates. If you look at the graph above, and you are charged with keeping live-in college students as safe as you can, you’d gladly trade the infection rate in August (month 8) for the infection rate in December (month 12).
And that’s starting to look pretty smart, to me. I think we’re now seeing a seasonally-related decline in COVID-19 infection rates in Virginia. In effect, in Northern Virginia, I think we’re now following something like the red line in the right-hand graph above. That’s my explanation for the downward trend in new cases overall in recent weeks, despite our having done virtually nothing more to reduce disease spread. And a few things that should have increased it. There’s no way to be sure of that, but that’s my best guess. After a month of 300-cases-per-day in Fairfax, we now seem to be solidly in 100-cases-per-day territory. FWIW, I can’t think of any other explanation other than “natural course of the epidemic”, which is just the scientific version of “stuff happens”.
And I think that seasonally-related decline is probably occurring in most parts of the country. You just can’t see it, against all the other things going on. Unlike the other, well-established coronaviruses pictured above, for COVID-19, that seasonality is just one of many factors that may strongly affect incidence of new infections. That seasonal “signal” can easily be masked by the “noise” of any number of things. Initially, it’s masked by the initial spread of this novel disease — all the trend lines are up, until such time as the infection rate stabilizes. And it’s masked by the “noise” of changes in behavior. A state full of people who refuse to wear masks, that rapidly re-opens its economy, may well see a resurgence of infections, no matter what the slow, underlying seasonal trend is.
And so you can only see that seasonality by eye when it occurs against a backdrop of otherwise stable factors. Stable ongoing infection rate, steady and prudent behavior in the population, and so on. You can only see it, by eye, in a handful of places, because only a handful have nothing else going on.
If all you have to rely on is your eye — a plot of new cases, over time — at this point, seasonality of COVID-19 is pretty much a matter of faith. But after some time spent this week playing with the data and pondering the evidence, that’s my conclusion as to what’s driving the numbers, right now, in Northern Virginia. For what that’s worth. Backed up by what I see as as two major Virginia colleges planning their re-opening in way that only makes sense in light of significant expected seasonality of infections.
I think you can see the seasonality playing out here in Northern Virginia — but maybe not elsewhere — for at least four reasons. First, the spread of infection had stabilized. We’d reached the point where we were consistently getting 1000 new cases in Virginia, more-or-less, week-after-week. We we no longer in the “ramping up” phase of infection spread. Second, we now seem to have stable, smart behavior going on with respect to minimizing spread of infection. When I venture out to the store, I’d say 98% of all persons involved (staff and customers) are masked. Third, our behavior is changing slowly. Re-opening isn’t really doing much, at present, and so should have only a modest impact on the count of new infections. And finally, we don’t have point sources for large, essentially random outbreaks, such as meat processing plants or crazy mega-churches.
The complexity of what we’re actually observing across the states requires its own, separate posting. That’s coming up next, in epidemiology week. So the fact that we’re not seeing a nice, smooth decline across the country doesn’t really dissuade me from making this assertion. I think it’s just an indication that seasonality is one of several forces that have a strong influence on the rate of new infections. I think you need some fairly special conditions to be able to see it in the data. And NoVA just happens to satisfy those conditions, right now.
As with hydroxychloroquine, this is one where you simply have to ignore the fact that the President brought this up. Per the research pictured above, there was plenty of reason to expect to see seasonality. (Though no basis from which to expect to see this entirely novel infection disappear entirely.) Doesn’t really matter that the President’s people latched onto that as a way to minimize the potential dangers of the pandemic. Either it’s seasonal or it’s not, regardless of who is talking about it.
Finally, another way to look at this is that Mother Nature may give us a few months to get our act together, with respect to COVID-19. You can be assured that if this disappears entirely in late summer, all the usual idiots will declare that it’s all over, and we’ve won. I think that would be imprudent. As with our local universities, we ought to think about how best to take advantage of this haitus, if it occurs. Because if the current decline is due to seasonality, then as the winter follows the summer, we ought to start preparing now for the likely rise of infections rates this winter.
There are many open questions. Is COVID-19 really as seasonal as other coronaviruses? Does that seaonality vary based on the local climate? Will it follow the same pattern as other coronaviruses? How large will the seasonal decline be, for COVID-19? And will that seasonal decline be swamped by other factors, such as changes in behavior?
More posts on this topic will follow. For now, I think it’s enough to put my marker down and say, to my eye, what we’re seeing now in NoVA seems to fit the concept of seasonal decline in coronavirus infection rates. And, as importantly, doesn’t seem to have any other likely source. Only time will tell if that interpretation is correct or not.
