@danluu I suspect it's less that we have better understanding of the risks in a purely quantitative sense than we understand "no idea what the error bars are *really* like on this, but it's a cheap precaution and the downside of not doing this could be really bad. This is a no-brainer, right?"
(And as an ex-quant-dev, I have strong opinions about the quants who understood risk vs. those who believed their models. Approximately right vs. precisely wrong.)
Re subtle long-term effects: people were finding correlations between number of (major) infections in one's past life and IQ. Obvious alternate hypothesis that's hard to eliminate: higher IQ causes people to be better at avoiding major infections.