Wednesday, November 1, 2023

What happened with Covid NPIs? By Matthew Yglesias

— Read time: 11 minutes

MATTHEW YGLESIAS

NOV 1, 2023


What happened with Covid NPIs?

Learning the right lessons from 2020


I know there’s a lot going on, but I thought today would be a good time to debate the hottest policy topic of three years ago: how to fight the SARS-Cov-2 pandemic.


As Nate Silver wrote a month ago, after-action reports on the pandemic show a very clear pattern of blue states having lower mortality than red states. But examining this in more detail shows that the liberal Covid advantage accrues essentially entirely after the rollout of vaccines. While vaccination was highly effective at saving lives and conservative anti-vax sentiment seems to have cost a lot of lives, pre-vaccination variation in Covid policy appears to explain very little about Covid mortality.


In other words, the non-pharmaceutical interventions didn’t work.


Silver pitched this as a contrarian claim, but it strikes me as conventional wisdom. Either way, though, the inefficacy of NPIs in this case is important to note. We’re certain to see more pandemics in the future, and it’s essential to try to learn useful lessons from the one we just experienced. And my take (which, unlike Silver’s, may actually somewhat contrarian) is that we need to be more careful going forward about distinguishing between NPIs and NPI rules.


What was “social distancing”?

Armed with three years of retrospective knowledge, it’s fascinating to re-read this New York Times article by Eric Lipton and Jennifer Steinhauer titled “The Untold Story of the Birth of Social Distancing.”


The point of the story is that social distancing as a public health strategy is a newer concept than you might have thought, with very specific origins in recent American political history. It turns out that George W. Bush read John Barry’s book, “The Great Influenza” in 2005 and was very impressed. His interest in the book, along with the general post-9/11 climate of concern about bioterrorism, led him to ask the government to develop a plan for a future pandemic. They came up with the strategy of social distancing based on a mix of mathematical modeling and study of the 1918 flu pandemic. And these recommendations were initially very controversial among well-qualified people:


The concept of social distancing is now intimately familiar to almost everyone. But as it first made its way through the federal bureaucracy in 2006 and 2007, it was viewed as impractical, unnecessary and politically infeasible.


“There were two words between ‘shut’ and ‘up’” initially, said Dr. Howard Markel, who directs the University of Michigan’s Center for the History of Medicine and who played a role in shaping the policy as a member of the Pentagon research team. “It was really ugly.”


But in February 2007, the CDC adopted “Interim pre-pandemic planning guidance: community strategy for pandemic influenza mitigation in the United States: early, targeted, layered use of nonpharmaceutical interventions” as official policy and updated it in 2017.


These recommendations were largely based on a pair of papers that both studied city-to-city variation in NPI adoption during the Great Flu and concluded that the NPIs were effective.


It’s interesting, though, to look at the particular NPIs in question.


One of the studies, by Howard Markel et. al., finds that “School closure and public gathering bans activated concurrently represented the most common combination implemented in 34 cities (79%); this combination had a median duration of 4 weeks (range, 1-10 weeks) and was significantly associated with reductions in weekly [excess death rate].”


The other study (this one from Richard Hatchett, Carter Melcher, and Marc Lipstitch) found that “cities in which multiple interventions were implemented at an early phase of the epidemic had peak death rates 50% lower than those that did not and had less-steep epidemic curves.” But they also find that “few cities maintained NPIs longer than 6 weeks in 1918.” Across 17 cities, the most common interventions were bans on public gatherings (15 cities) and shutdowns of places like theaters (15), churches (15), and schools (14). Eight cities implemented “staggered business hours to reduce congestion in stores and on transit systems.” And just one city used “community-wide business closures.”


Note that in 2020, early reopening states like Georgia kept things shut down for about three weeks. And the Atlantic’s “Human Sacrifice” article depicted a masked man getting a haircut from a masked barber.



Which is just to say that even a lax state like Georgia was actually pretty strict relative to the sample that formed the basis of the recommendations. To the extent that people genuinely wanted to replicate the success stories of 1920, an early and intense adoption of bans on big crowds in public spaces that lasted for 3-6 weeks is basically what was on the menu.


Curve-flattening mostly worked

One slight quibble I had with Silver’s analysis that the blue states did no better, on average, than the red ones during the pre-vaccine phase is that it’s clear the New York City metro area did dramatically worse than average. Three of the 10 deadliest states — and two of the top two — during this early period are in the tri-state area.


That area was hit first and hardest in the United States and, critically, it was hit before NPIs were put into place. If you wave that away, the blue states in the aggregate did do slightly better.



This isn’t really to quibble with the red versus blue bits.


It’s just to note that the early wave that hit Greater New York was a scary exemplar of how bad things might have gotten. That was a situation in which the vulnerable were taking few countermeasures and the virus was truly spreading uncontrolled. Most of all, Greater New York had the pandemic nightmare scenario where there were so many patients in the hospitals that it wasn’t possible to treat them all. When that happens, medical professionals have to triage away the least-promising cases and the death rate soars.


The point of the pre-polarization rallying cry to “flatten the curve” was that we wanted to adopt enough NPIs to prevent what had happened in New York from recurring in Los Angeles, Chicago, Dallas, and Houston. And while that didn’t work perfectly, it did work. Looking at the early devastation in New York, it was easy to think things would only get worse as multiple cities were hit simultaneously. But it didn’t get worse.


If you take the Great Flu model seriously, essentially every state did adopt meaningful NPIs. And if you judge their success by the curve-flattening criterion, it worked. So it’s arguably less that NPIs failed everywhere than that NPIs actually worked everywhere. At least relative to reasonable expectations.


Variation wasn’t that large

A related issue is that while Covid became a highly polarized issue, I don’t think the actual situation across jurisdictions was quite as different as the discourse would lead you to believe.


For starters, even in states that were “open,” lots of stuff that would normally happen wasn’t happening. Taylor Swift could have done a Lover tour exclusively in red states but she didn’t. Basically no major musician toured. The NBA played its games in “the bubble,” they didn’t try to play to packed arenas in Texas and Florida. Hollywood was basically not putting out movies, so even if your state’s policies weren’t formally forcing movie theaters to close, there just weren’t a ton of people out at the movies.


Many churches stayed voluntarily remote, even in places where they were allowed to open.


And plenty of blue states, meanwhile, did start opening stuff up well before vaccines were available.


I did not personally dine indoors until I was vaccinated, but indoor dining opened up in DC in June of 2020, a few weeks after it had already opened in Virginia and Maryland. My understanding is the west coast states kept that shut down way longer, but they generally have milder weather so people weren’t being driven indoors by the cold.


The really stark divide was over schools. But precisely what outraged some of us about prolonged school closures is that by September of 2020, in places like DC, they weren’t closed as part to a general lockdown. By that time, places that had lobbied to be allowed to reopen (because they needed the revenue) had largely been allowed to reopen. What was closed was schools — teachers and administrators were getting paid anyway and the interest group pressure from the Washington Teachers Union was to stay closed. I don’t know that it would have been better, all things considered, to have a European-style strong lockdown. But a school closure in that context would at least have been part of a coherent Covid policy. What we actually got was relatively soft policy paired with a socially costly disruption to education.


Weird rules, weak enforcement

A few months into the pandemic, a lot of the rules that were on the books were honestly kind of weird.


You had to wear a mask on an airplane, for example, but they were still serving drinks and snacks on the flights, so plenty of people were taking their masks off. Same deal for airports — mask mandate in place, but the bars and restaurants were open. I stayed in a hotel in Austin pre-vax to go on the Joe Rogan Experience, and the hotel had lots of signs up in the lobby about how you had to wear a mask. But the hotel bar was open.


One can understand how this happened.


It’s a compromise between the desire to have some public health rules in place and the demands of commerce. But while sometimes in life it’s good to have a simple horse trade, these rules didn’t have any underlying logic to them. And precisely because the rules had already been compromised to allow for commerce, they proved weirdly tolerable to keep in place. Air travel mask rules, first imposed by Donald Trump contrary to his stated views, lingered deep into the vaccine era only to be ended by a court decision that the administration didn’t bother to appeal.


That kind of public health theater ended up alienating a lot of people. But it’s of course not a huge surprise that poorly designed NPIs didn’t have benefits.


More broadly, even places that mandated masks didn’t mandate the use of good masks.


Realistically, though, they barely enforced the rules at all. Every time I took the metro or rode the bus, I’d see people who were maskless or chin-strapping. The view that it didn’t make sense to have cops arresting people for poor mask compliance made a decent amount of sense to me. But as Immannuel Kant wrote, “whoever wills the end also wills… the indispensably necessary means to it.”


If the goal of a mask mandate was to get people who were not Covid cautious personally to nonetheless act like they were Covid cautious, then you really would need strict enforcement. If you didn’t want strict enforcement, then what were you actually doing? Unlike in a real lockdown regime, nobody was stopping people from having house parties. To the extent that the point of NPIs is to get people to change actual behavior, it’s not clear to me how much of a difference the variation in policy meant.


The flip side of maskless metro riders in DC is that my mother-in-law in Texas was wearing high-quality masks without being required to do so. The various liberals I talked to in Austin weren’t actually behaving particularly differently from the liberals I know in DC, even though the rules were different, because the people I knew were personally cautious in their behavior regardless of the rules. Conversely, non-cautious people found ways to skirt the rules wherever they lived.


Lessons for next time

Rather than deciding that the NPI-based strategy totally flopped, I think it’s important to remember that the initial idea was that we were trying to replicate some relatively short-term NPIs adopted during the Great Flu.


On that relatively short 3-6 week time horizon, we really did replicate the strategy and it looks to me like it probably worked pretty well.


It would be bad to conclude from the lack of state-to-state variation that the whole thing was a mistake, because the most crucial part really didn’t feature that much state-to-state variation. Looking back on it, though, the great Known Unknown about the evidence is how long we could really sustain those NPIs. Here I think the important takeaway is that even the states that did nominally try to sustain them for months and months in practice mostly gave up.


Covid Zero did work in a number of places, but it required strong external controls on travel (which we didn’t really implement at any point anywhere in America) and really draconian enforcement.


At the same time, basically everyplace, including the dovish states, did dramatically curtail big events, whether or not they explicitly “wanted” to. It would be nice to have clearer evidence on how much that mattered since those large scale events (rather than restaurant meals) were really a huge point of emphasis of the historical NPIs.


I think the public health community adopted a basically unsound philosophy of “ask for as many restrictions as possible, and accept what we can get after political pushback.” That led to bizarre situations like bars being open (since there was pushback) while schools were closed (since the push was in both directions). You need to try to come up with a principled hierarchy of which things it makes sense to restrict, and then urge politicians to select a point on the spectrum that they think is workable, all things considered. You could be laxer or stricter, in other words, but you should try to pick something that is broadly coherent.


But most of all, we should keep in mind questions about enforcement and the reality that the ultimate target is people’s behavior. In practice, the biggest question in most cases is going to be what people feel like voluntarily complying with. And basic information about best practices for harm reduction and questions like “which air filters actually work?” or “how can you get a mask that isn’t counterfeit?” are going to have a lot more value that saying everyone has to wear a mask, then letting some people get away with wearing crappy masks and drinking at the hotel bar.


Update your profile

Only paid subscribers can comment on this post

Check your email

For your security, we need to re-authenticate you.


Click the link we sent to tka.lee@gmail.com, or click here to sign in.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.