COVID Transmissions for 1-8-2021
Modeling spread
Good morning and welcome to COVID Transmissions.
It has been 418 days since the first documented human case of COVID-19. We’re done with the first week of 2021!
In the US we are still dealing with a tense political situation. I hope it gets better. But please do try to have a relaxing weekend—no matter how things turn out, the weekend is an opportunity to prepare yourself through rest.
A bit of a dive into a modeling paper today.
As usual, bolded terms are linked to the running newsletter glossary.
Keep the newsletter growing by sharing it! I love talking about science and explaining important concepts in human health, but I rely on all of you to grow the audience for this:
Now, let’s talk COVID.
Asymptomatic transmission of COVID-19
By this point in the pandemic most of us should be aware that people who have no symptoms can still transmit the virus. In the event that this is the first you’re hearing of it, this is quite important. You can be a risk to others even if you do not feel sick at all. Even if you tested negative yesterday, you can be dangerous to other people.
A recent modeling paper from the Journal of the American Medical Association tries to understand just how possibly dangerous you might be: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2774707
Modeling is always unreliable. You have to accept that all models are wrong, but some are useful. In my opinion, this model is useful because it states upfront its assumptions and comes to conclusions that resemble what we have seen in reality. And importantly, it looks at a question we cannot ethically ask in a real experiment. When studying transmission, we cannot just give a large realistic population of people COVID-19 and see what happens. That would be very, very wrong. Modeling—either through the use of mathematical models or animal models—is the best we can do. This model takes the mathematical approach, and I think it does it rigorously.
Specifically, the model was set up in the following way (emphasis mine):
This decision analytical model assessed the relative amount of transmission from presymptomatic, never symptomatic, and symptomatic individuals across a range of scenarios in which the proportion of transmission from people who never develop symptoms (ie, remain asymptomatic) and the infectious period were varied according to published best estimates. For all estimates, data from a meta-analysis was used to set the incubation period at a median of 5 days. The infectious period duration was maintained at 10 days, and peak infectiousness was varied between 3 and 7 days (−2 and +2 days relative to the median incubation period). The overall proportion of SARS-CoV-2 was varied between 0% and 70% to assess a wide range of possible proportions.
I bolded the fact that they considered three conditions—symptomatic, presymptomatic, and never symptomatic. This is important, because people without symptoms are not necessarily people who will never develop symptoms. They can transmit before they are sick, but still get sick later. Or they might never get sick. It’s good to keep those separate.
They also importantly used published research to think about the rates at which asymptomatic people actually transmit, so that they were basing this in reality. They used a few parameters for that, because there is probably not one single correct value for that. At least not that we know of; there is always variation in measurement. They also did the same for the period during which people are infectious, varying not its duration but instead the period when infectiousness was at its highest within that window.
I bolded the last sentence because for the life of me I can’t figure out what they’re talking about. Reading further in the methods, it appears that what they mean is that they varied the proportion of transmission from never symptomatic people to see how this impacted the model. There must be a typo in that part of the abstract. That sentence doesn’t make sense. This happens sometimes in scientific publications, and I’m sure it makes them harder to approach. One time, I read a paper where an entire figure was accidentally duplicated, but with the legend for a different figure appearing the second time. That was confusing.
Anyway, having set up a well rigorous model, they looked at transmission under various scenarios for their assumptions.
This all culminates in the following figure:
This is an amazing image that is very hard to understand at a glance. It’s important to note that as things get more yellow, it’s a sign of fewer asymptomatic transmissions. As they get more red, it’s a sign of more. The white color would be zero asymptomatic transmissions. Please note that no part of the graph is white.
Along the X-axis, they are tracking their different assumptions for when someone is at peak infectiousness relative to when they typically begin to have symptoms. Positive means a condition where most people—except those who are “never symptomatic”—are at their most infectious after symptom onset. That doesn’t mean they’re uninfectious before that. But, the peak matters. The negative sign means peak infectiousness is before symptom onset—in those conditions people are most able to transmit before they get sick. This axis only indirectly affects the color of the graph because it increases the number of presymptomatic people with a chance of transmitting.
Along the Y-axis, we look at varying the proportion of transmissions from people who are never symptomatic. This is going to directly affect the color of the graph because it adds strictly to the number of transmission events that are happening from asymptomatic people.
Now, let’s think about the extremes of the graph. If we set peak infectiousness to 2 days before symptom onset, and “never symptomatic” transmission to 70%, well, it looks like 100% of cases are from asymptomatic people. That’s kind of wild! We know that 30% must be coming from people who are presymptomatic in that condition, since 70% was our maximum for never-symptomatic people.
Let’s walk down the proportion from never-symptomatic a bit. That white line only crosses the Y-axis at a very low point. If peak infectiousness is 2 days before symptom onset, then we still see more than 75% of cases coming from asymptomatic infections even if only 20% of never-symptomatic people spread the virus. So far this exercise is suggesting that asymptomatic spread can be very influential.
This is also clear from the fact that most of the graph is between the two lines. It looks like no matter how you vary these conditions, asymptomatic spread tends to make up about 50% to 75% of transmitted cases (at least in this model). In fact, asymptomatic transmission only falls below 50% in conditions where peak infectiousness is after median symptom onset. If more than 20% of cases come from never-symptomatic cases, also, it’s impossible for less than 50% of cases to come from asymptomatic patients. At least, in this model.
The first possible interpretation of this is that the model is wrong. Now, all models are wrong. But I don’t think this one is terribly wrong. This is based on some good published information on the peak date of infectiousness, and on other parameters about the virus. So I wouldn’t expect reality to be too far different from what we see here. But, it will be different.
What I take away from this is that it’s very easy for a situation to arise where half or more of cases are coming from asymptomatic people. In fact, that’s the conclusion that the authors make, too:
In this decision analytical model of multiple scenarios of proportions of asymptomatic individuals with COVID-19 and infectious periods, transmission from asymptomatic individuals was estimated to account for more than half of all transmissions.
Please note that the conclusion doesn’t focus on specific numbers. It talks about general trends—because the model is designed not to be right, but to be useful. It is a guide. It tells us the general trends.
So, let’s accept the conclusion. It’s easy for asymptomatic cases to be the source of at least half of infections. What does this mean for us?
Well, it means you could be dangerous to other people, and other people who seem perfectly well could be quite dangerous to you. It tells me there is no way of knowing if another person can give you COVID-19. Which, leads us to the next part of the authors’ conclusions:
These findings suggest that measures such as wearing masks, hand hygiene, social distancing, and strategic testing of people who are not ill will be foundational to slowing the spread of COVID-19 until safe and effective vaccines are available and widely used.
Yes. They’re correct. Because there is no obvious pattern by which you can recognize people who can transmit COVID-19, we have to be vigilant with protective measures that can stop the spread. While the precise details of the model may not be hyperaccurate, it demonstrates with effective estimation a fact which we have suspected to be true from observation. Here, the model is useful because it helps us to confirm the hypothesis made from our observations in a context where it would be impossible to conduct an experiment to test that hypothesis. It tells us something that we suspected but could not ethically confirm.
And even though it may feel like old news, with deaths in the US reaching record highs, it’s old news we need to continue to hear.
What am I doing to cope with the pandemic? This:
Mastering the stir fry
Well, trying to.
I like to take on cooking challenges. For much of the pandemic, I was working on perfecting the crispy fish skin. Before that, I spent a lot of time working on different ways of cooking rice. When I take on a challenge, it’s about really exploring and understanding a technique or ingredient; finding its limits and pushing them one way or another to the extent that my small apartment kitchen allows.
Recently, I reached my one-year anniversary at my job. This entitles me to an award! Specifically, I can get some kind of product from a site that my company maintains, as well as a lapel pin. The only thing that looks to be worth getting is a cast-iron wok, and it’s about time that I took on the challenge of learning to properly stir-fry.
“Stir-frying is simple!” you may be thinking. To an extent, this is true—if you are stir-frying the same way that you might also sauteé, since the techniques have a lot of similarities. But stir-frying is really not that simple. Done properly, it is a subtle technique that uses extreme heat and heat-resilient oils, varying pan architecture, and careful attention to ingredient size and cooking order to bring on chemical reactions that create deliciousness you can’t get any other way.
One problem with mastering the stir-fry is that really, there are two “styles,” one which is a little more welcoming to steam and liquid than the other. Steam is generally the enemy of the stir-fry because you want to sear the food quickly in very hot oil, and use variation in temperature to manage your ingredients. But sauces are hard to introduce without generating some steam; hence a more steam-allowing technique vs another technique that is less tolerant of moisture.
I’m still learning more about this, but I’ll share my progress with you as I go. Especially after the wok arrives.
Reader Carl Fink had a comment today, regarding single-dose vaccine efficacy (as usual—thank you Carl for always making sure I have something I can put here!):
Hey, John, would you like to correct Nature? In the Nature Briefing, they wrote, "The COVID vaccine developed by Moderna, which was authorized by US regulators last month, can provide protection against COVID-19 within two weeks of the first dose, according to the results of a large clinical trial. (Reference: New England Journal of Medicine paper)"
However, the actual paper (https://www.nejm.org/doi/10.1056/NEJMoa2035389) says, "The finding of fewer occurrences of symptomatic SARS-CoV-2 infection after a single dose of mRNA-1273 is encouraging; however, the trial was not designed to evaluate the efficacy of a single dose, and additional evaluation is warranted."
I think Nature's description is at best misleading. What about you?
It’s tempting to correct Nature. But, unfortunately, I can’t. I want to note that scientists and scientific publications often choose their words very carefully. The Nature briefing is no exception to this, and the words here are definitely careful. Here’s my take:
I see how it could appear misleading, but the way they have phrased it is accurate. 50% vaccine efficacy is still efficacy. People working on an HIV vaccine would be amazed at 50% efficacy. The problem is that for COVID-19, 50% protection is probably not adequate given how contagious the disease is. I think perhaps they should have considered noting that this did not exceed the expected minimum bar for adequate protection.
Still, strictly speaking, what they said here is true.
50% is what I consider the bare minimum protection offered by a single dose. Recall, please, that I don’t think the studies had enough followup to show us the true protective value of a single dose. However, I think 50% is still evidence of protection. The question really is, then, how much protection is offered. That, we don’t know. But here, Nature hasn’t made a claim about the answer to that question. Instead, they just say that protection is offered. It is.
You might have some questions or comments! Send them in. As several folks have figured out, you can also email me if you have a comment that you don’t want to share with the whole group.
Join the conversation, and what you say will impact what I talk about in the next issue.
Also, let me know any other thoughts you might have about the newsletter. I’d like to make sure you’re getting what you want out of this.
Part of science is identifying and correcting errors. If you find a mistake, please tell me about it.
Though I can’t correct the emailed version after it has been sent, I do update the online post of the newsletter every time a mistake is brought to my attention.
No corrections since last issue.
See you all next time. Have a great weekend!
Always,
JS
Lee, can you speak to this Twitter thread from the other day? I found it...alarming. https://twitter.com/drericding/status/1346899021621813249?s=21
Note that I also wrote carefully! I wrote, "misleading." I did not write, "Incorrect." If you can read it as meaning something other than what (as you say) should be taken from it, it's misleading.