kottke.org posts about David Dowdy

Why Are Covid-19 Cases Declining So Quickly in the US?

posted by Jason Kottke   Feb 17, 2021

Over the past week or two, I’ve read a number of articles and threads about why Covid-19 cases are falling so rapidly in the US. The explanations have all been somewhat unsatisfying to me. Cases have indeed dropped off quite quickly and it happened pretty uniformly all over the country. Look at the mini state graphs on the NY Times Covid page — they all look about the same. Hospitalizations and positivity rates have dropped too, so while the number of daily tests has fallen too, this appears to be a real drop and not just an artifact of a lack of testing. Which is great news! Imagine a February and March that looked like December — a disaster compounded.

So what’s going on here? For The Atlantic, Derek Thompson lists four reasons for the decline in cases and hospitalizations that mirror the arguments I’ve seen elsewhere: “social distancing, seasonality, seroprevalence, and shots”.

The vaccine explanation is the weakest one for me: not enough people outside of healthcare workers had gotten them early enough to start bending that curve sharply downward in early January. But as Thompson notes, it could be having more of an effect on hospitalizations because the folks getting shots (and therefore immunity against severe infection) are those most likely to end up in hospitals due to infection. And obviously, vaccines are going to become the dominant factor in falling case numbers as more and more people get jabbed.

I’m also skeptical of the seasonality argument, but (again, as Thompson notes) there’s a lot we don’t know about how temperature, sunlight, humidity, and this specific coronavirus interact. Obviously Covid-19 is a seasonal thing and that’s definitely a contributing factor here, but that sharp of a drop in early January? I don’t know if it’s the primary driver here. Also, the seasonal flu typically peaks in February in the US.

The seroprevalence argument is an interesting one. Here’s Johns Hopkins infectious disease epidemiologist David Dowdy in a great Twitter thread about the US case decline:

I think the most logical explanation is one proposed initially by @mgmgomes1 and others — namely that we are seeing the effects of population immunity with heterogeneous mixing + strong behavioral effects. Take a(n overly) simple example. Assume 60% of a population has zero respiratory contacts, while the other 40% lives life as normal. If 75% of that high-mixing group has immunity (e.g., 30% population seroprevalence), you could easily see herd effects.

Basically, a large percentage of the folks at the greatest risk of getting Covid-19 in the US (i.e. folks who aren’t able or willing to keep from seeing other people and/or take proper precautions) have gotten it, resulting in a sort of localized “herd immunity” among those folks. After the massive holiday surge in cases (more on that in a sec), this hypothesis suggests, the virus started running out of people to infect and rates dropped quickly. This is the first explanation I read that really made sense to me.

Thompson leads off his piece with the behavioral explanation: “Maybe Americans finally got the hang of this mask and social-distancing thing.” I do not buy that people who previously weren’t doing so before suddenly started wearing masks (or better masks), keeping distant, spending less time indoors with others, and staying home from work started doing so in numbers large enough to cause such a sharp downturn. But you can’t consider the decline without also looking at how cases got so high in the first place. Here’s Steven Johnson on Twitter, zooming out a few months:

[It’s] not so much that people got the hang of social distancing, but rather that the holiday season compelled people to relax social distancing for in-person family gatherings. So the current decline is mostly reversion to where we were in Oct-Nov.

Yes, this. Without these holidays, we may have seen much more of a winter plateau than a spike. So here’s what seems plausible to me. As the cold weather made the coronavirus more effective at infection, people gathered for Halloween, Thanksgiving, Hanukkah, Christmas, and New Year’s — each subsequent holiday building on the previous one — and it pushed cases much higher than they would have been without those major gatherings. After two months of massive infection rates, the virus burned itself out among the high-mixing group and everyone else retreated back into their homes and pods to hunker down, resulting in the steep decline we’re seeing.

Obviously, careful scientific study will be necessary to tease out how significant each of these (and other!) causes were to the holiday spike and subsequent decline. But for now, the way forward is continuing to social distance, wear (better) masks, limit close contacts, and get people vaccinated — before B.1.1.7 and the other variants hit.