This is the second edit of my original post. The original post said, more or less, hey, there was an unusual jump in COVID-19 cases at William and Mary. Why was that? The first edit said, never mind, the W&M administration explained that cases rose as a result of testing everyone on campus. And this edit explains that not only did cases jump, they are going to continue to jump over the next few days.
Why? Looks like the same problem that has hit so many college campuses: Rapid spread of COVID-19 due to off-campus parties. In this case, St. Patrick’s Day parties.
You have to give W&M credit. Reading between the lines, they spotted this outbreak as students became symptomatic following St. Patrick’s Day. And they’re now staging the full fire drill, starting with testing every student on campus, followed by quarantine and contact tracing. This is just about as good as it gets, in terms of pandemic response.
Edit 2: That said, as of COB 3/25/2021, the count of cases on the William and Mary COVID-19 dashboard stands at 197 cases total, with 57 active cases.
And that won’t be the end of it. Meaning, don’t panic (yet) as the numbers continue to climb over the next couple of days.
The email that was circulated by the W&M administration (see below) states that all students are being tested. From that, it’s easy enough to calculate where this is heading.
So far, they’ve gotten back about half the tests (just over 3000). They’ve added 43 cases, based on those first 3000 tests. The natural assumption is that they’re going to find just about that many again when the rest of the tests come back.
You can reasonably expect another 40-ish cases to show up on that dashboard over the next few days as the rest of test results coming back. By the time all the tests come back, the total should be up to 250, possibly with around 100 active cases.
And then, even though they are testing all students, there will still be some infectious students circulating in the student body. The reason for this is that COVID-19 PCR testing has a fairly high false negative rate. (See Post #859, or the graph just below). So, even after this round of testing, there will be a non-negligible number of undiscovered infectious individuals on campus. If I had to guess, if they find 100, using PCR testing, they likely missed at least 25, for an overall false-negative rate of 20% (25/125).
So in addition to the lump of cases that’s showing up right now (and for the next few days), there should be a fairly long “tail” of new cases, from this outbreak, that will trickle out slowly, from the cases that PCR testing inevitably will miss.
To give this a name, let me dub it the “St. Patrick’s Day outbreak”. The email from W&M implicates off-campus St. Patrick’s Day parties, and subsequent socializing, for this uptick in cases. The timing seems about right for that, given that there’s a median five days between infection and onset of symptoms, for symptomatic cases.
William and Mary has now experienced the same problem as so many other colleges: Rapid spread of COVID-19 via off-campus parties. They had some of that at the start of the Fall semester. I think this is the first time this has happened mid-semester.
Addendum: COVID-19 PCR test false negative rate. A false negative occurs when a person has a disease, but the test does not detect it. Here’s the graph showing the false-negative rate of COVID-19 PCR (DNA) testing, by days following infection. This exact rate shown depends on the population used to create this graph, but the general gist of it is the same for any population. For the first couple of days after infection, you won’t test positive for COVID-19, even though you have it. By the time you’re four or five days into your infection, the odds of a positive tests are better, but still not 100%. And then, as you proceed further past the date of infection, the odds of false negative begin rising again.
Source: “Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction–Based SARS-CoV-2 Tests by Time Since Exposure:, Kucirka, Lauren M, Lauer, Stephen A, Laeyendecker, Oliver, Boon, Denali, Lessler, Justin doi: 10.7326/M20-1495 Annals of Internal Medicine, May 13 2020, https://doi.org/10.7326/M20-1495
Edit 1: Ignore the rest of this post. The College of William and Mary sent out an explanation by email this afternoon. The uptick is plausibly attributable to re-testing everyone on the campus. Here’s an excerpt from the email:
Dear W&M Community, William & Mary is undergoing a round of census COVID-19 testing of all students living on campus and in the vicinity. With roughly a third of results received, the university has seen a significant uptick in positives within the student population. We continue to update the COVID-19 dashboard to reflect current results every weekday. We anticipate this number to grow as additional results are returned, which is why we are contacting you today.
That’s the first such email I’ve received, I think, but the body of the email suggests that they’ve been sending these out all along.
Original post follows:
I’ve been tracking the COVID-19 case count at the College of William and Mary, comparing the actual case count (blue bars above) to an “expected” case count, based on the COVID new case rate observed for Virginia residents age 20-29.
A week ago, actual cases were about 15% below the expected level. But as you can see above, there was a spate of new cases at William and Mary in the past week or so. As a result, the actual case count (blue bars) is now back up to the “expected” level.
That’s not good. And that comes as something of a surprise.
Plausibly, this might just be some one-off event, such as an un-publicized outbreak on some sports team. But I’m not seeing any news reporting on this, nor did I find any explanation on the William and Mary website. In terms of public information, I’m not finding anything to explain it..
I tried changing up the methods that I use, to see if any plausible set of assumptions about timing of cases could make this, in effect, go away. The answer is no. If I vary (e.g.) the lag between the Virginia data and the William and Mary data, I still come back to the same conclusion: This has been a very bad week for COVID-19 cases on the William and Mary campus.
Coincidence or not, there was a sharp change in the Virginia data at about the same time. Our overall rate of new cases has been fairly steady. But for the past week or so, that’s been due to falling new-case rates for the elderly, and rising rates for everybody else. Like so:
So there was a bump-up in the new case rate among young adults generally, at this time. But even if I shift the data around to make sure that’s worked into the comparison-group number, this short-term change isn’t nearly enough to make a material change in the cumulative count of cases since the start of the semester.
Finally, to some degree, the graph of cumulative cases at the top of this posting disguises how different the past week was from the weeks prior. If I divide the data into seven-day periods, you can see that the count of new cases this week (33) is about three times the number you would expect, based on what had been the rate among 20-to-29-year-olds in Virginia.
I’m not worried yet, but I will be checking the William and Mary COVID-19 dashboard daily for a while.
