The engines of SARS-CoV-2 spread
science.sciencemag.org
science.sciencemag.org
"For SARS-CoV-2, studies suggest that ∼10% of cases cause 80% of infections (1). Overdispersion is characterized by a large number of people who infect no one, and most people who do transmit infect a low-to-moderate number of individuals."
"overdispersion means that most infected individuals who enter a community will not transmit, so many introductions may occur before an epidemic takes hold; likewise, overdispersion also increases the probability of disease extinction as the epidemic recedes and fewer people are infected."
"If transmission is highly overdispersed, broad and untargeted interventions may be less effective than expected, whereas interventions targeted at settings conducive to superspreading (such as mass gatherings and hospitals) may have an outsized effect."
They then make the argument that this pattern of transmission makes it unlikely for things like travel bans to have an effect:
"Only a small number of such long-distance connections are needed to create a “small world” network in which only a few infection events can transmit the virus between any two individuals worldwide. This is one reason why early travel bans could not stop the global spread of SARS-CoV-2, although they may have slowed the pandemic."
Then they move on to the importance of preventing spread in the household, via relatively inexpensive interventions:
"Moving from international to household scales, the burdens of interventions are shared by more people; there are few international travelers, but nearly everyone lives in households and communities. Measures to reduce household spread may appear particularly challenging, but because they directly affect so many, they need not be perfect. Household mask use and partitioning of home spaces, isolation or quarantine outside the home, and, in the future, household provision of preventive drugs could have large effects even if they offer only modest protection."
So overall, everything in this article draws a (IMHO convincing) conclusion that our interventions should be focused on household transmission. Then they go on to say that community transmission control measures must be draconian in order to be effective:
"Conversely, control measures at larger spatial scales (for example, interregional) must be widely implemented and highly effective to contain the virus. Indeed, few nations have managed to curb infection without stay-at-home orders and business closures, particularly after community transmission is prevalent."
Unfortunately, they've worded this in a weird way, which makes it sounds like they're advocating for such measures, when they're actually doing the opposite; the article is fundamentally about how even if you miss one transmission event, it's enough to keep the graph connected. So overall, this article is saying that we're doing it wrong: we should be focused on the middle of the bell curve, where modest efforts can yield greater results, instead of spending huge amounts of money to unsuccessfully filter out infrequent, improbable events.
Said in a way that might be more intuitive for programmers, it's similar to the expenses required to ensure system uptime. While it's possible to obtain (e.g.) 95% uptime with very little effort, getting to 99% requires significantly greater costs (redundant servers, etc). Getting to 99.9% requires significantly more expense (e.g. datacenter redundancy, hot deploys, etc.), so on and so forth. Every additional fraction of a percent of bad stuff that you try to stop costs orders of magnitude more effort. The same thing applies here: it's relatively inexpensive to ask people to do things at home, but shutting down all businesses and international travel is an incredibly expensive intervention, aimed at the tails of the distribution.
Also, with the caveat that Sweden is seeing a surge in case (albeit less than Europe as a whole), something that makes its more open policy viable is that a majority of households are single-person.
We could have easily done the same thing in the U.S., on a voluntary basis, using otherwise empty hotel rooms, and it was discussed in March/April, but essentially never implemented.
This strategy is being employed at the University of Illinois, where undergraduate students have to be tested three times per week, test results are available within about 8 hours, and the university has a goal of quarantining any positive cases in a segregated dorm within 30 minutes of a positive test result.
The University of Illinois has had the most successful college opening in terms of controlling COVID despite it being known as somewhat of a party school.
It's also in the middle of fscking nowhere.
It's a lot easier to contain a pandemic when you're not buried in a large city as well.
Here's what I figure:
- If I caught it a week ago, I would be sick by now
- So, I didn't have it a week ago
- I haven't had contact with anyone outside my house for a week
And it's the same for my spouse. We can't catch it from each other if we don't have it.
So, the payoff for distancing and masks inside the house seems very low for a very high cost. I can only stand 6 months and counting of lockdown because my best friend (spouse) is locked down with me.
We did have a scare once, and did a 2-week quarantine within the house. I just don't see any point doing that all the time when neither of us have it right now.
More changes costs some incremental fuel (as it siphons pressurized bleed air), but the effect is quite small.
In order to see such a story, someone would have had to do contact-tracing of a large number of people with covid back to a single flight they all were on. Given the haphazard nature of contact tracing (in the U.S. at least) it is quite plausible that large-scale transmission events on airplanes are happening but simply aren't being noticed because we're not looking for them.
That said, the ventilation in planes is pretty good, and if everyone is wearing a mask it's equally plausible that air travel is relatively safe.
As with anything else, including sitting in a restaurant or bar, any single flight is unlikely to infect you and statistically "safe" from the perspective of individual risk. But air travel is absolutely spreading this disease.
The whole notion of treating a pandemic via the lens of individual risk ("will I get sick if I do this one thing") is broken analysis, of course. But at this point I despair of our ever being able to free ourselves of that nonsense. It's just too ingrained.
How would you determine this without solid contact-tracing? Nearly all of the cases where I'm at are associated with "international travel", so at least some of that has to come from planes.