It’s epidemiology week here at savemaple.org. I hope you’re as excited about that as I am.
Let’s kick it off with a practical piece outlining the goal of this. So, what’s the point of reviewing what is known about the epidemiology of COVID-19? It’s to help you decide which precautions make sense for you, if you are determined to return to something approaching your pre-COVID-19 lifestyle.
The second-biggest waste of time in the USA
Decades ago I attended a time-management seminar at a rural health conference. The speaker said two things that stuck with me, and that seem to encapsulate the core tension in re-opening the economy.
The first thing he said was: “The second-biggest waste of time in the USA consists of people doing something really well, that doesn’t need to be done at all”.
That intentionally begs a question, doesn’t it? And when an audience member asked the obvious question, it was his reply that made the whole thing stick. “The biggest waste of time in the USA is premature death”.
It was, after all, at a health care conference.
As we go through various phases of re-opening, I think the tension between those two viewpoints is going to occupy a lot of our thinking. You can’t take every possible precaution, all the time. Not if you want to return to something like a normal life. On the one hand, it’s inefficient to spend time and effort protecting yourself against non-existent risks, particularly if that imposes cost or hardship. On the other hand, it’s crazy not to protect yourself against real risk of disease or death, particularly if such protection is cheap and easy.
In short, you probably want to take every reasonable precaution when the situation calls for it. But to do that well, you need know the odds. You need some idea of whether the chances of infection are low or high in any given situation, and then make up your mind on what risk you’ll tolerate.
But the damnable thing is, nobody will tell you the odds. So it’s not even a question of your personal tolerance for risk. It’s that nobody can give guidance as to where the risks are. And that’s why you see a lot of guidance that just boils down to “take every precaution, all the time”. Not a heck of a lot more helpful than “be scared, be very scared” as a guiding principle.
Four factoids to get you started.
First, YMMV. Obviously, if I’m citing some benchmarks, they have to be for some “average” person, however defined. Someone who is older (younger), with more (fewer) chronic conditions, will have more (less) risk than average.
But I think it’s helpful to put down a few benchmarks, for just how frequently infection occurs. On average, given average people, and average behavior. Just to start to get a handle on risks.
Second, these are single-source estimates. Ideally, I’ll find other references to back these up. For now, I’m starting with what I can find. These first two are from a recent NEJM piece showing that hydroxychloroquine, by itself, does not prevent COVID-19 infection (Post #707, last item).
Odds that you’ll get infected, if someone in your household is infected: 10% to 15%. This is often referred to as the “secondary attack” rate, that is, the rate at which you get second COVID-19 infections within a household, after one individual in that household is infected. That’s cited in this New England Journal of Medicine article, and was the basis for their power calculation (estimate of required sample size.)
Odds that you’ll get infected from a 10-minute-or-more face-to-face chat with an infected person, at less than 6′, not wearing any protective gear: about 15%. That’s from the NEJM article cited above. Their sample population consisted of people meeting that criterion, and just under 15% of them eventually got COVID-19 or had symptoms consistent with COVID-19.
Odds that any one random infected person will ever infect anyone else: less than 30%. The entire spread of disease is due to less than one-third of the infected population. Roughly 70% of infections are “dead ends” from the standpoint of spreading the virus. That’s from a recent New York Times summary of epidemiological research on COVID-19.
That seems to dovetail with the 15% above, doesn’t it? If 70% are duds, from the standpoint of spreading infection, a 15% infection rate from a single long up-close conversation with a random infected individual seems kind of plausible.
And there is an inevitable flip side to the fact that 70% of infected individuals don’t spread the disease. That means that most of the disease spread has to occur through just a small number of people. Leading to this approximation.
About 80% of all infections are caused by just 10% of the infected population. That’s from the NY Times article cited above, and is my brief summary of the several research articles cited. The bulk of infections are spread by a handful of highly infectious individuals, with the worst case being “superspreader” events where such an individual attends a large public gathering such as a party, funeral, or church service.
A major caveat about that last one is that this appears to be at odds with the Commonwealth of Virginia’s own epidemiological analyses. But I believe that’s because the Commonwealth only tracks “outbreaks” in institutional settings. (“Outbreaks” are instances of two or more related cases in an institutional setting, including long-term care facilities, other congregate living settings, correctional facilities, health care settings, and educational settings).
The Commonwealth shows that just 17% of cases in Virginia can be traced to “outbreaks” as defined above. As of today (6/78/2020), outbreaks account for 8,846 cases (out of a total of 51,521 (both pieces of information available on this Virginia web page). (Of those, fully 80% are in either long-term care facilities or other congregate living facilities).
In other words, something like a super-spreader event at a funeral or party would not be tracked by the Commonwealth as an outbreak. My guess is that the spread of cases in Virginia is much like the spread documented elsewhere, it’s just that the Commonwealth’s “outbreak” definition excludes most of the instances where a single individual infected multiple other persons.
The upshot of all of that is that transmission of COVID-19 in the community is a lot more “lumpy” than you might naively assume. The great majority of infected individuals never go on to infect anyone else. Infections within (e.g.) nursing homes account for only a small fraction of cases. And so, pretty much the entire chain of infections is carried by a small number of individuals who manage to infect many others.
My takeaway from all of that is that one-on-one interaction with (e.g.) a cashier at the grocery store probably isn’t very dangerous. Particularly not if both parties wear masks.
You can start with my prior estimate of the fraction of all Fairfax County residents who are plausibly walking around in an infectious state. I estimated that at 0.8% (Post #680), and the numbers today would still be in that ballpark. Given that (in my experience) grocery store cashiers are younger than average, the rate there may well be lower. Now multiply by the 15% chance (per above) that you’d pick up an infection, if you stood face-to-face and chatted for 10 minutes with no protective gear. That works out to maybe one chance of infection in 1,000 encounters, in a total worst-case scenario where you chit-chat with the cashier for ten minutes, at less than 6′ distance, without masks.
I’m not even going to try to do the arithmetic with proper social distancing and masks. Not to mention sneeze guards. Needless to say, if I worked out the plausible odds for of a once-weekly grocery shopping, I’m pretty sure the odds are I’d be dead long before COVID-19 could kill me. Which is good news, in a way, given the question at hand.
But I would still be wary of any large indoor gatherings. It’s hard to say how much mask use mitigates risk in crowd situations. But research shows shows that the vast bulk of cases come from a relative handful of people. And that seems to be a pretty strong argument for staying out of crowds.
It’s not just that there are more people in a crowd. That’s part of it, but only part of it. So, you might naively think that, given the cashier example above, hanging out in a crowd of 100 cashiers would generate 100 times the risk of infection.
But that’s incorrect. It’s incorrect because that naive calculation assumes that each infected individual can infect just one other — as was the case of the face-to-face interaction at the cash register. That’s how you get from risk of infection from one encounter, to risk of infection with 100 encounters.
To the contrary, the epidemiology seems to show that transmission is quite “lumpy”, in that most transmission is due to a handful of individuals who each infect many people. Such superspreaders of disease may infect many individuals in a crowded situation. That’s the lesson of the Mount Vernon, Washington choir practice event, and many similar reported superspreader events.
For the one-on-one interaction, that doesn’t matter. You’re only facing one cashier, you can only get infected by that one individual. But in a crowd, each superemitter can infect multiple individuals. In effect, we’d have to factor superspreaders into the prior calculation as if they were several different single-infection-passing individuals.
The upshot is that the presence of superemitters (superspreaders) in the population makes indoor crowd scenes somewhat more-than-proportionately risky, compared to one-on-one interactions, from the perspective of an individual (you) getting infected. But even then, not hugely so. If (roughly) 15% of the population are aerosol superemitters, and each of them can infect 10 in a crowd, the result is just over double the risk of 100 single-person encounters.
That all assumes that the behavior at the party is equivalent to the behavior at the grocery store. So I’m not sure that the arithmetic here is telling the true story on the odds. The bottom line is that indoor crowds are more-than-proportionately riskier than the equivalent number of one-on-one interactions. Just how much riskier is difficult to quantify, at best.
My bottom line? I’m continuing to shop for groceries in-person. I’m staying out of anything resembling an indoor crowd. I remain undecided on anything else.
More fun with epidemiology tomorrow.