t-threads
private beta
search about

Infectious disease epidemiologist and microbiologist, aspirational barista. mlipsitc@hsph.harvard.edu Director @CCDD_HSPH

Jun 7, 2020 4:11 PM
copied to clipboard

This thread (and the update that follows) makes a good attempt at putting – as @trvrb admits – very rough numbers to what could happen as a result of the protest. A few comments.

Trevor Bedford (@trvrb):
The appalling killing of George Floyd and video after video after video of police brutality cements the elemental rightness of the Black Lives Matter cause. 1/21
https://twitter.com/trvrb/status/1269395225228664832


First on the math: this is critically dependent on the value chosen for “R~1” because in a branching process with R<1, the ultimate number of cases eventually caused by one case is N=1/(1-R) – the sum of a geometric series with a=1 and r=R. Thus for R=.95, N=20.


But R near 1 was arbitrarily chosen to be 0.95, and this has a huge influence on results. with R=0.9, N=10, so half as many cases / deaths; with R=0.98, N=50, so 5 times as many. This number is probably the most sensitive – and least supported – input into the calculation.


The other part I’m unsure about is that each day of protest adds another transmission per infected person. @trvrb lays out the reasons it could be small (masks, distance, outdoors) and large (tear gas, shouting, arrests), but


to me for a mildly transmissible virus (R0~2-5 uncontrolled) this seems like a big jump. It’s a matter of intuition in the absence of data.


But now to the politics of blaming the protestors, as @trvb certainly doesn’t but people will use any numbers to do. Let’s consider the impact of other activities. @trvrb notes that gambling in Las Vegas is back. Las Vegas has 100K hotel rooms https://www.pbs.org/wgbh/pages/frontline/shows/gamble/etc/facts.html.


Casinos are indoors, crowded, and I bet most gamblers aren’t wearing masks. What numbers of transmissions are happening there? At Lake of the Ozarks (again, shouting, crowding, no masks)? etc.?


Second: suppose 0.95 is the right estimate for the current R. Clearly there is a lot of variation across the country, but let’s stick with the example. Suppose that the states that all the states that have failed to control the epidemic so far (21 or so https://www.pbs.org/wgbh/pages/frontline/shows/gamble/etc/facts.html)


managed to do so. This would massively overshadow the effects of the protests. Lots more assumptions required to quantify that, but let’s think on the average level over the country. Suppose that the mean R went down from assumed 0.95 to 0.94. Then (again very roughly)


the nearly 1000 daily deaths now in the US would go down in a geometric series that had 16.7 instead of 20 times as many before it ended. This means sustained a ~1% reduction in transmission would save 3.3x1000=3300 deaths, or 3-15x the daily toll @trvrb gets for protests