We’re Switching to Prevalence. Now We Just Need to Get It Right.

Many moons ago, which is to say in early June, some asshole with a blog noted that, instead of using fourteen-day declines–a policy pushed by the conservative failtank AEI and adopted by the CDC–we needed to use prevalence measures. That is, scaled to population, how many people are infected?

This really mattered. In the D.C. area, states and local governments juked the numbers to claim they had fourteen days of decline (or close enough), and then reopened too early. Now, cases are climbing again, and all of the hard work people did was pissed away.

For prevalence, the metric is the number of cases per 100,000 people during a one- or two-week period (ECDC reports two-week prevalences). For context, Germany institutes restrictions once the one-week prevalence is over 50/100,000 people (as a daily rate, that would be ~7/100,000 per day), a threshold most U.S. states currently breach.

Prevalence is a scary statistic for politicians because it’s really hard to wiggle out of. If you require adequate testing (less than five percent positive), and a certain prevalence threshold, there’s really no way to juke the numbers (though many governors will still try!). You’re either there, or you’re not, and it’s very simple for voters to understand. It’s also scary because, in the U.S., if prevalence were routinely discussed, it would clearly illustrate how much we suck at this. But it is what we should be using.

The good news is that we’re finally starting to talk about prevalence. Between the White House Task Force using prevalence statistics, D.C. using prevalence to put states on a high-risk list, and other people talking about prevalence, we’re finally starting to use the right statistic. Now we need to use the right thresholds. Which brings us to this recent op-ed by Ezekiel Emanuel, Saskia Popescu, and James Phillips.

There’s a lot that’s good in the piece–they lay out what activities are high and low-risk, and offer some concrete suggestions. The problem is this paragraph (boldface mine):

First, schools cannot reopen safely when community transmission is high and climbing. In our view, schools should open only in places that have fewer than 75 confirmed cases per 100,000 people cumulatively over the previous seven days, and that have a test positivity rate below 5 percent. By our count, 12 states and the District of Columbia meet both metrics. In many larger states, some counties or cities meet those criteria. Even with those numbers, about one in 1,300 people might return to school with a case of the coronavirus, meaning a school of 350 students, faculty and staff will have roughly a one-in-four chance of someone coming in with Covid-19. (Many countries, such as Japan, Austria and Italy, have suppressed the virus to the extent that they have fewer than one in 10,000 people with confirmed cases.)

First, most schools are much larger than 350 total people. To put this another way, the D.C. school system with 4,000 teachers (never mind students and staff) is pretty much guaranteed to have multiple teachers who are sick on the first day. It doesn’t have to be like that. It shouldn’t be like that. That last sentence shows it’s possible to get to 10 per 100,000, and if we were willing to do the hard work, we could do that.

If we crushed the curve further, we could reopen far more safely. And by requiring a lower threshold, it would be safer for everyone else too. This seems like a political compromise, but viruses don’t do political compromises. Crush the curve to at least 10 per 100,000 in a week (I would prefer lower), and then we all can truly get back to, if not normal, then normal-ish.

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