Why I Am Not Optimistic About School Reopening in D.C.: The North Carolina Modelling Edition

And elsewhere too, but regular readers will know that I like to focus on data I understand and follow, so D.C. (not Wor-Shing-Tun) it is.

A recent modelling paper by a consortium from North Carolina and Georgia recently modeled what percentage of kids would be infected after 60 days. I’ll walk through some of the assumptions and caveats, but, long story short, I think it’s reasonable to assume that after 60 days, ten percent of D.C.’s entire student population will be infected (of course, with the low school testing rate, we might not know about many of these–we’ll return to this in a bit).

The model assumes a school of 500 students with one initial student who is infected. Some of the assumptions working in D.C.’s favor:

  1. In some wards, vaccination of 12-17 year-olds likely means those students have a higher percentage of protected students than is modelled (highest protected percentage was fifty percent).
  2. In some wards, the prevalence of COVID-19 in the community is likely less than 0.5%, meaning that not every idealized 500 student school would start with one infected student–for a while anyway. In other words, some schools after sixty days could be thought of as having run the model for less time.
  3. To the extent D.C. is doing cohorting in schools, that will lessen spread. If the school of 500 students should really be thought of as five schools of 100 students, that will slow down (or prevent) transmission throughout the entire school.
  4. D.C. is doing some testing, so the no testing at all model is somewhat pessimistic.

The assumptions working against D.C.:

  1. The flip side of point #1 above is that Wards 7-8, even if you assume twenty percent of students have been previously infected and that acquired immunity is highly effective against Delta, the wards with the lowest student vaccination are also those wards with the highest prevalence. This also weakens the protective effects of point #2 above.
  2. Related to the previous point, D.C.’s kids are more likely to live in higher-risk areas than lower risk ones.
  3. D.C.’s testing is inadequate. The opt-in requirement means that students likely at higher risk for COVID-19 will be less likely to give permission to be tested. More importantly, D.C.’s plan to test only ten percent of students is utterly inadequate to the task at hand–the model assumes a fifty percent weekly testing rate. At ten percent testing, if five students out of fifty are infected, that ten percent testing rate means we would miss all of the five infected students seventy-two sixty percent of the time (assuming no false negatives). With fifty percent of students tested, the five infected students are missed only four percent of the time. So ten percent testing is better than zero percent testing, but it’s not like fifty percent at all. Essentially, it will tell you that you have had an outbreak (note the past tense).

It’s also unclear to me after reading the mayor’s and D.C. Health’s presentations if the model’s assumptions about isolation and quarantine would be met: D.C.’s close contacts definition (see p. 12) seems very lax. All those caveats aside, here’s what the model predicts:

Screen Shot 2021-08-24 at 5.13.06 PM

As I noted, some schools will be better than fifty percent of students protected, but many will be at thirty or forty percent–and some lower than that. Remember parts of D.C. are doing well, though not well enough (unfortunately, COVID-19 doesn’t grade on a curve), and other parts resemble Southwest Missouri.

So if we assume 50,000 kids in DCPS–and note this completely excludes the nearly equally sized public charter system–a ‘very good’ case scenario is looking at 5,000 student infections over sixty days (ten percent). These student infections will lead to ‘parental’ infections (who then, of course, can spread COVID-19 to other adults). The student infections alone would average 85 cases per day, though this would not be a constant average, meaning there will be surges greater than that. Keep in mind, ‘the good place’ for the entire city is around seven cases per day (one new case per 100,000 people per day). And those 85 cases don’t include subsequent household infections (“Guess what I brought home, Mom?”), or the ongoing infections before school has started.

(Aside: D.C. very likely will significantly undercount children’s infections because, if they’re not tested at school, there’s a reasonable chance won’t be tested at all–like so many other services, it will happen at schools or not all. If a child is sent home for quarantine and has minor symptoms develop at home, it is unlikely that those cases will be tested or reported).

Add to this several other cheery observations, such the majority of DCPS students live in high-prevalence, low-vaccination areas, those under forty (i.e., parents) are far less likely to be vaccinated than the city-wide average, and Black people under forty have even lower vaccination rates (and higher morbidity and mortality rates due to COVID-19), and this could get really, really bad.

As I always write in my doom-and-gloom posts, I hope I’m wrong. I desperately hope there’s something obvious (or not obvious) I’m missing, or that an assumption is wildly off-base. But if I’m not, September and October could be awful.

Have a nice day.

This entry was posted in COVID-19, DC. Bookmark the permalink.