Read on Twitter
@bhrenton argues that tracing an infection back to a restaurant is harder than tracing it back to a family member.
This is true.
But how do we actually find and investigate restaurant clusters?
Shall we do a "choose your own adventure"?
THREAD 1/ https://twitter.com/bhrenton/status/1337882897039159301?s=20
This is true.
But how do we actually find and investigate restaurant clusters?
Shall we do a "choose your own adventure"?
THREAD 1/ https://twitter.com/bhrenton/status/1337882897039159301?s=20
On Wed, #1 comes into your system. You give him a call. He says he started developing symptoms on Monday.
Do you:
(a) ask him about his close contacts (6 ft, 15 min) starting 48 hours before symptom onset
(b) ask him where he thinks he got infected
2/
Do you:
(a) ask him about his close contacts (6 ft, 15 min) starting 48 hours before symptom onset
(b) ask him where he thinks he got infected
2/
He's a college student who lives alone. On Sat he went out with 5 friends. 1 is already in the system as positive; 2 tell you they recently tested positive.
Do you:
(a) end the interview and close his record?
(b) ask about school?
(c) ask where he went on Sat?
3/
Do you:
(a) end the interview and close his record?
(b) ask about school?
(c) ask where he went on Sat?
3/
Answer (c): Sat is more likely since most of that group is infected.
Your supervisor is pinging you complaining about the large number of cases that need to be picked up today.
Where do you focus?
(a) Bowling alley with 10 lanes
(b) Restaurant/sports bar with 12 tables
4/
Your supervisor is pinging you complaining about the large number of cases that need to be picked up today.
Where do you focus?
(a) Bowling alley with 10 lanes
(b) Restaurant/sports bar with 12 tables
4/
Answer (b): crowded, drinking, shouting
You search for the name of the sports bar in your system.
Bingo! Your colleagues have already found #5, #6 who visited the week before and a server (#7).
Do you call the sports bar? Or go back to the queues?
5/
https://twitter.com/miekocakes/status/1327727953820618753?s=20
You search for the name of the sports bar in your system.
Bingo! Your colleagues have already found #5, #6 who visited the week before and a server (#7).
Do you call the sports bar? Or go back to the queues?
5/
https://twitter.com/miekocakes/status/1327727953820618753?s=20
The owner tells you that she's been worried because there was another server (#8) that also got sick about the same time as #7 (already in your system, but didn't mention working at the bar).
You: "Do you keep track of customers?" (recommended in state safety guidelines)
6/
You: "Do you keep track of customers?" (recommended in state safety guidelines)
6/
Owner: "Yes we do. We ask every party to write down the name and phone number of one person. We keep a notebook at the entrance."
You: "Well, could I have those names/numbers?"
Owner: "Of course, that's why we do it. Which days?"
(a) Sat
(b) Sat and last Sat
(c) Sat to Sat
7/
You: "Well, could I have those names/numbers?"
Owner: "Of course, that's why we do it. Which days?"
(a) Sat
(b) Sat and last Sat
(c) Sat to Sat
7/
You: (c)
Owner: No problem, I'll send you >200 names/numbers
You:
(a) Go back to the queues (your supervisor is really annoyed with you)
(b) Call >200 people
The end of the adventure depends on this last choice.
8/
Owner: No problem, I'll send you >200 names/numbers
You:
(a) Go back to the queues (your supervisor is really annoyed with you)
(b) Call >200 people
The end of the adventure depends on this last choice.
8/
(b) is a huge amount of work. Each name is the head of a party, so that could represent >1000 people. Some of those numbers will be wrong. Some of them will not answer the phone.
Some might even have been tested already and in your system but never mentioned the sports bar. 9/
Some might even have been tested already and in your system but never mentioned the sports bar. 9/
But these are mostly young people who could have no or mild sxs. Many might not get tested.
@BillHanage would argue you should call them all because you already know this is a cluster. This is higher yield than the other cases waiting in the queue. 10/ https://twitter.com/kj_seung/status/1323693468070236160?s=20
@BillHanage would argue you should call them all because you already know this is a cluster. This is higher yield than the other cases waiting in the queue. 10/ https://twitter.com/kj_seung/status/1323693468070236160?s=20
Let's not be hysterical. The risk for any of these people is low.
It's not hundreds of people. It's probably less than 40 cases (you won't even know until after a few days of calling people).
But small clusters can have large downstream effects. 11/ https://twitter.com/EllenBarryNYT/status/1337229682396311552?s=20
It's not hundreds of people. It's probably less than 40 cases (you won't even know until after a few days of calling people).
But small clusters can have large downstream effects. 11/ https://twitter.com/EllenBarryNYT/status/1337229682396311552?s=20
Finding
4 person cluster: not easy, but we are doing it.
8 person cluster: hard but can be done retrospectively.
40 person cluster linked to restaurant: really hard.
Is it worth the effort? With our current case loads, we'll never be able to find all of these clusters. 12/
4 person cluster: not easy, but we are doing it.
8 person cluster: hard but can be done retrospectively.
40 person cluster linked to restaurant: really hard.
Is it worth the effort? With our current case loads, we'll never be able to find all of these clusters. 12/
But as @zeynep says, "Knowledge is power, and sometimes small data is more power."
Or as my team says, "Never underestimate the power of a good cluster map." https://zeynep.substack.com/p/small-data-big-implications?utm_campaign=post&utm_medium=web&utm_source=copy
Or as my team says, "Never underestimate the power of a good cluster map." https://zeynep.substack.com/p/small-data-big-implications?utm_campaign=post&utm_medium=web&utm_source=copy
