Thread by @statesdj

Let’s talk about Fall Semester. Colleges and universities have to make some big decisions soon
1/

And it won’t affect just students. For example, the University of Michigan #GoBlue has 46,000 students but it also has 25,000 faculty and staff. While students may be mostly young, many faculty and staff are in vulnerable age groups
2/
cupahr.org/wp-content/upl…

There’s also real meaning to “university community”, schools are densely linked social networks with multiple paths linking students, e.g. classes. Eliminating large classes can reduce but not eliminate the potential for epidemic spread
3/
osf.io/6kuet/

And then there are social activities. No, I don’t think masks and social distancing are going to be the norm for campus parties, and one super spreader partying in Korea infected 119 others
4/
nzherald.co.nz/world/news/art…

Lots of people are saying we need to reopen campuses. But what happens if we just use “students are young and resilient, and the fatality rate in this group is low, they’ll do OK” strategy?
washingtonpost.com/opinions/why-w…
5/

The US has about a million active cases of COVID-19 in a population of 330 million, about 0.3%. Let’s assume students and staff reflect this national average come Fall term. That means a campus like Michigan should expect about 200 infected people on campus
6/

The number of cases in the US doubled every 3 to 5 days in February and March. If we allowed that to happen on campus, 0.3% would grow to the 80% required for herd immunity in 8 doublings. It would take just 24 to 40 days
7/

Of course that means 80% of the 25,000 faculty and staff on campus will also be infected. Several thousand will become seriously ill and hundreds will die. Bad news.
8/

Many people are saying “testing, testing, testing” will be the answer, with some good behavior and tracking thrown in. But how much testing and how often?
9/

About the widely used qRTPCR test. We talk about it as if it’s a perfect binary positive or negative, but nothing works that well in the real world. At best it’s about 80% sensitive if you're tested around the day you develop symptoms
10/
acpjournals.org/doi/10.7326/M2…

But an RTPCR COVID-19 test has a 100% false negative in the first few days after exposure and detection wanes later in disease. Let's estimate that a round of testing will on average pickup maybe 60% of cases
11/

If we test 71,000 students and staff of whom 200 are truly infected with a test that has a sensitivity of 60%, we will find 120 of the 200 positives but miss 80.
12/

If we wait a week before testing again, the 80 we missed will have infected 2.6 to 5 more people and we will now have 211 to 500 new cases in addition to the 200 already had. Oops.
13/

And let’s not forget the false positives. Technically RTPCR is highly specific, but in the real world false positives happen. If we go with a conservative estimate of 0.8%, when we test the 71,000 students and staff, we’ll expect 568 false positives.
14/
medrxiv.org/content/10.110…

That many false positives is a big problem. It’s a lot of tracking & quarantining. More importantly, most people labeled infected weren’t actually infected. After their 14 day quarantine, they’ll have a false impression that they are now immune
15/

If we test everyone twice a week, we can get on top of case growth, but there will be twice as many false positives. Good lab practices to drive the FP rate down will be critical
16/

A campus like Michigan may need 142,000 tests/week, ~20,000/day. That’s a lot of testing, but with clever molecular biology, we can do it. There are many strategies in play, the challenge will be getting a solution in place within a few months
17/fin
pubs.acs.org/doi/10.1021/ac…

Addendum: it will be important to have contingencies in place. What’s the plan if you get 200 positives on week one, 300 on week two, and 450 on week three?