How many people do I kill with COVID-19 when I go to get groceries?
Quick post today: A back-of-the-envelope calculation of how worried I should be about going to the grocery store that I’m giving somebody COVID-19.
(These numbers are a little bit out-dated now — I originally did this calculation a week ago, when I was in Chicago. I am now in Des Moines IA, and am not going to the grocery store any more.)
Chances I kill somebody through COVID-19 when I go to the grocery store = (chance I have COVID-19) x (chance I transmit the virus if I have it) x (chance the person I give the virus to dies of the virus)
There is a 0.4% chance I am carrying COVID-19
1.7% of people in Cook County are carrying the disease
5.2M people live in Cook County
93,000 likely carriers in Cook County
186 deaths in Cook County (and 8,034 confirmed cases)
.2% death rate (Source). 186 / .2% = 93,000 carriers
93,000 / 5.2M = 1.7%
~1/4 people who have the disease are asymptomatic, and I am asymptomatic. So there is a 1.7%/4 = 0.4% chance I am carrying COVID
4.5% chance that I give someone COVID when I go to the store if I have COVID-19
2.5% chance I transmit by coughing on someone
This number is totally made up. It is likely lower, because I have not been coughing at all when going out, have been wearing a cotton mask when near people, and have been staying more than 6 feet away from people
2% chance I transmit by touching someone with my hands. These numbers are all totally made up as well
15% chance virus goes onto something I touch with my bare hands
15% chance someone else picks up virus after touching that thing
2.5% + 2% = 4.5%
Sense check: R0 is ~1, you have the disease for 14 days, so the average odds you give someone else the disease each day is ~7% if you have COVID. I would estimate that I’ve had far less social contact with people since the outbreak started than most of the people who are driving the spread of the disease, so my personal R0 would probably be much lower than 1
(I’m not sure if “my personal R0” is a meaningful thing to say — maybe R0 is only defined for a population? But I can’t fit the research that would be needed to find that out on the back of this envelope)
.6% chance that person dies from COVID-19
.2% chance someone dies if they have COVID-19
Of course, if they have the disease, then there is some chance they give it to someone else, and on it goes. So let’s multiply this number by 3 to get .6%. Perhaps it should be much higher — I think this is the biggest weakness in my methodology
This means my chances of killing someone when I to the grocery store is 1.1 x 10^-6
For context, this is approximately equal to your chances of killing someone while driving ~100 miles if you get in the average number of fatal car accidents / mile.
For a different kind of context, if a human life is valued at $10M (I believe this is what the department of energy uses when doing nuclear power cost/benefit calculations), then the equivalent cost of my trip to the grocery store is $11.
Now I’m not saying that the value of human lives can be directly translated to dollars, but I think that these kinds of utilitarian calculations are useful to get a sense of the magnitude of what we’re talking about. I don’t really have a sense for what 1.1 x 10^-6 means — my imagination isn’t that good. But I know what $11 means. It means that if I would be willing to pay an $11 “grocery shopping fee” then it’s probably alright to go get groceries. I should be moderately worried about going outside, but if I get groceries or go for a run, I don’t need to feel waves of guilt, as long as I am taking precautions.
You can plug in your own updated numbers or use different assumptions to get a different result for your likelihood of killing people when you do various activities. It’s fun! And allows you to compare the relative riskiness of different activities, and to prioritize your worrying towards the things that are truly the most worrisome.