A version of this article first appeared in Emily Oster’s newsletter, ParentData.
The biggest safety questions of late, for all of us, are about the coming holidays. How can we see family safely? Should we? What are the actual risks? I’m going to dive in. (I should say that on this topic, as often, I found some inspiration and good ideas from Kelly Fradin, whom you should follow and read.)
In May, I wrote about my five-step decision-making strategy: question framing, risk mitigation and evaluation, benefit evaluation, deciding. In the case of the holidays, like most decisions, that’s not a bad place to start. However, many are struggling with the middle steps on risk.
We kind of know the question in most cases: See family in whatever way is the “tradition,” or skip it. And we know the benefits. And we know we do need to decide—there is a clear deadline.
But people seem to be craving both an understanding of how to mitigate risk and a better sense of the size of the risks.
The more I thought about this, the more I realized the right way to think about risk here may not be far off from the “Safety Lasagna” idea I wrote about a few weeks ago in the case of school. (Later, some people told me this was called the “Swiss Cheese Model,” but I do not care for Swiss cheese so I’m going to ignore that.) There are layers of protection we want to think about around the holidays, too. Both mitigating and evaluating risks rely on thinking through the layers.
It’s the holidays, though, so let’s skip the lasagna and go with a turducken. Very broadly, here are the layers:
Layer 1: Limiting chance that the virus enters your holiday gathering.
Layer 2 (relevant if Layer 1 fails): Limiting spread when you are together.
Layer 3 (relevant if Layers 1 and 2 fail): Preventing spread when you return.
Basically, the key to a safe holiday for you—and for society—is to limit chance you bring the virus, limit chance you spread if you do bring, and limit chance you spread it out if someone does get it. This, Layer 3, is key. I talk a lot about individual risk, and that’s important here, but there are also big issues of social risk. We have to try to figure out how to keep the holidays from generating COVID-19 spikes. This is our responsibility.
So let’s dig into this: what goes into the layers, how can they be as safe as possible, and can we say anything about how big the risks really are.
The chicken layer focuses on the risks of the virus getting to your gathering. At the core, the risk is:
Chance someone is infected x Chance the infection gets through whatever barriers you put in place
There is some base rate chance each person is infected, dependent largely on the rate in the geographic area. And then there are things you can do to lower the risk in the lead up to the holiday.
I know what you are saying: What are those numbers?
What’s the actual chance you’re infected with COVID-19 at a given moment? Let’s assume you are an average person in your area, neither especially high or low risk for exposure. There are two ways you could evaluate this. First, you could look at case rates in your area. Here’s a nice map. Where I am, here in Rhode Island, we are averaging 12.6 cases per 100,000 people per day over the last week. What does that mean? If we assume that people are infectious for 10 days (this is based on CDC isolation criteria), then that means 126 out of 100,000 people are infectious at any given time, or about 0.12 percent. So there about a 1.2 in 1,000 chance of you having COVID-19, by this metric.
Of course, not everyone is tested, so case rates are probably low. You could also look at the positivity rate—the share of people tested in your area who test positive. In Rhode Island, that’s currently about 1.5 percent. That’s likely to be much higher than the actual prevalence in the population, especially in areas with limited testing. (If we have very few tests, we only test sick people.)
Bottom line: The number is probably something higher than the case rate but lower than the positivity rate. So something between 1.2 in 1,000 and 15 in 1,000.
OK, so that’s your base rate (or a range, anyway). If you’re an average person and you do nothing in this chicken layer, that’s the risk of bringing COVID-19 to your family gathering.
What can you put in your chicken to lower the risk?
1. Self-Isolation. Go out less, do not send kids to school (if they ever go), limit playdates, make it more like lockdown. This will lower the risk. How much? If you fully locked down for 14 days, let’s say it basically lowers to zero. If you did a week, you’d get about half the impact—half of infections show up within five days, so if you are quarantined for a week, you rule out about 50 percent of infections. If you do just a few days, you get a bit less than that. So let’s say you “lock down” for a week, and you started with a risk of 5 in 1,000; now you’re at around 2.5 in 1,000.
2. Testing. If you are going to self-isolate for 14 days, there isn’t a lot of reason to test. If you are not, testing may pick up some infections. Let’s say you are going to self-isolate for seven days and then test before going. Half of infections would have shown up already.
Let’s say the test picks up 80 percent of the remainder. A little calculation, and your 2.5 in 1,000 risk is down to 0.5 in 1,000.
3. Safe Travel. This is a tricky bit. Flying, driving a long way with stops—these all add something to your risk. I’ve written a bit about this before and my sense is that these risks are really quite small if you do them safely. With the caveat that kids under 2 cannot wear masks (and have trouble refraining from touching everything), so it’s likely best if they are not the ones traveling.
The way I’ve written it, using Rhode Island as the example, makes the numbers seem small. But if you’re in a place with high and rising rates, these numbers could be an order of magnitude higher—we’d be talking about 5 or 10 in 1,000 even with quarantine and testing, which is a much larger risk. (A useful comparison may be flu in terms of prevalence: About 10 percent of the U.S. population gets the flu in a given year. If we think of the flu season as about 35 weeks, this means a risk of 2.8 per 1,000 in each week.)
OK, so you’re all together. How can you mitigate risk of spread? Again, we’re back to
Baseline risk of spread x How much lowered by mitigation
Here, you’ve got control over both. If you see your family outside for a hike, that’s low risk, and it’s not hard to wear masks to lower it to basically zero. If you’re having dinner together, though, or staying in a house together, these risks are higher.
How high? This study in India puts transmission risk for household contacts around 10 or 11 percent. My sense is this is consistent with other studies. Which is to say: If someone shows up with COVID-19 at your holiday, and you all stay together in a house and have close contact, there is a 10 percent chance they will pass it to you. Part of what is difficult is that there is huge variation across individuals in how much they pass it—some people are just really good spreaders—which is to say that it’s likely that either a lot of people in the house get it or none do.
How can you mitigate this? Honestly, if you are staying together, it will be hard. Which isn’t to say impossible. You can try to limit physical interaction and avoid obvious things like sharing glasses. If it’s not too cold out, keep windows open for fresh air circulation. You can wear masks, but this may be impractical over several days. You can wash your hands a lot and not touch your face. All good ideas and likely to lower the risk.
But the biggest risk lowering activities are going to involve setting up the holiday differently—meeting outside, having a meal but not staying together, etc. If you can be in a hotel or an Airbnb and see extended family in a more distanced way, that is lower risk. Whether these options are feasible depends on your family.
In my view, this intermediate layer is the thinnest. Unlike schools, where there is a lot you can to do mitigate risk while people are there, in-household family gathering mitigation is tough. Especially when people are drinking which, let’s face it, they are.
Imagine the virus gets through Layer 1, and then Layer 2. You or someone in your family returns from the holiday with the virus. This is scary, and obviously your first concern here will be to get better. But this layer is really about a second concern—you return with a mild or asymptomatic case, or one that doesn’t show up for a few days. If you jump back into “normal” life, whatever that is, then the virus can get beyond your family and into the broader community.
If this happens—and if it happens a lot—we’ll see spikes associated with the holiday, like we did after the Fourth of July: younger people, mostly, getting together, spreading the virus, not knowing they are sick and spreading it. Investing in this layer is crucial for protecting our communities.
The turkey layer is basically the same as the chicken layer (I know, blasphemy!): quarantine, possibly testing. The longer you do it, the more benefit. The value depends also on the baseline risk. If you spend the holiday in a high prevalence area, especially coming from a low prevalence one, or meet up with a lot of people from areas with outbreaks, your chance of infection is higher.
What does this mean? It depends on your situation. If you’re home and your children are doing remote school, this means virtually nothing—you’re already pretty isolated. But if your kids are in school or day care, or you are back at work in person, this consideration may suggest more caution in the week after Thanksgiving. And for schools that are in person, this may suggest reason to encourage families who travel for the holidays to spend a few days virtually. I’ve argued elsewhere that schools are relatively low-risk environments, and I think they are, but it doesn’t mean people can’t come in from outside with the virus, and this will drive the need to quarantine, etc.
The layers are all about safety and risk, and my suggestion here is you think carefully about what is feasible in terms of mitigation at each step, and then evaluate the overall risk of infection spread. You’re not done, of course, since there are the benefits to consider, as well as the risk of serious illness if infected. If everyone attending is under 60 and healthy, a risk of 2.5 in 1,000 looks very different than it does if you have an immune-compromised older parent.
I wanted to address just a few specific questions that people have raised.
My husband’s family refuses masks and will not take any precautions, and they make fun of us if we wear masks. They want to all get together for the holiday. Should we go? (It’s me and my husband and our toddler, all healthy.)
Yikes. But everyone’s family has their stuff, I guess. I’m going to assume for the purposes of this that you basically cannot argue with them productively. Given that, you’ve kind of destroyed Layer 1 (at least on their end) and made Layer 2 very limited. Yes, you can wear a mask and be made fun of, which is fine, but as I note above this may not be practical anyway, and Layer 2 is already pretty thin if you’re planning to stay with them.
To me, the two key questions here are your own disease risk (which seems like it is low), and the broader spread risk. If you spend the holiday with your husband’s family, you should do it knowing that there is some chance you’ll get sick, although given your age and so on, it is unlikely you’ll be very sick. But I’d also urge you to invest strongly in Layer 3. You can choose to take on personal risk here, but you should still do what you can to mitigate the risks to the broader community.
We’d like to be tested, but where could we do that?
Some states have asymptomatic testing available for travel, so you could seek that out. There are also private market solutions to this, although they are expensive (see here, for example). Cue my usual complaints about the U.S. and testing.
Is 14 days really necessary?
I wish we knew more about this. The best data we have suggests more than half of cases show up by five days, nearly all (like 98 percent) within 11 days, and all by 14 days. But this is based on a single, fairly small sample. I suspect we’ll learn more about this from universities that are doing more testing and monitoring, but for now this is the best we have.
Some of you will think through the turducken and you’ll decide, basically, you cannot do Thanksgiving this year. It isn’t feasible, the risk isn’t worth it, or there’s something else.
This will be sad. Others of you will decide to spend the holiday with family; unless you’re an unusual person, or you’ve been able to really mitigate the risks, this will make you anxious.
It’s important to acknowledge this. Neither of these choices is likely to make you feel the kind of unbridled happiness that the holidays (usually) bring. So take a deep breath, make your turducken, make a choice, and try to move forward.