08 July 2021

Lagging indicators

 Vaccine effectiveness is measured; you give the vaccine to a number of people, and you count how many of them get sick from the thing you vaccinated them against, and you compare that to the expected number of sick people without vaccination.

The ethics panel won't let you deliberately infect your test group to see who gets sick from exactly how much measured exposure.  You're stuck using random exposure, and that is affected by behaviour.

We've known almost from the beginning that COVID-19 is a dose-dependent disease; how sick you get is influenced by how much virus you were exposed to when you were infected.

People who are not sure if they got the vaccine or the placebo, and who are further not sure that the vaccine works, go right on being cautious.  Caution reduces both their chance of being exposed and their dose if they are exposed.  (They are, after all, the sort of pro-science public-spirited people who sign up for vaccine trials; they don't want to get sick and they're at least aware of the "how not to get sick" guidelines and are probably following a bunch of them.)

People who think they're now immune are less cautious.  They get exposed more often and to statistically larger doses.  They get sick more than the test group does; the measured vaccine effectiveness is observed to go down.

Sure as death, the vaccine effectiveness has been reduced; that's exactly what you expect selection to do when you've got a functional vaccine, a big population of variously infected and vaccinated people, and long illnesses which give the virus a sustained period of reproduction in which to mutate.  But we can't tell how much is selection, and how much is a change in behaviour, without doing experiments to which that ethics board would rightly object.

In terms of things you can change, vaccine effectiveness is a lagging indicator of behaviour.  The vaccine you received is more effective if you distance, mask, and avoid going inside anywhere but your home like the plague.  If you act like COVID-19 doesn't exist, the vaccine is less effective.

(The medical statisticians trying to pull apart how much is behaviour and how much is virus variation have an unenviable job.  We'll never know without significant error bars.  But for planning purposes, here at individual scale, vaccine effectiveness is a lagging indicator of behaviour.)