In one further thump on a drum the New York Times has been beating for a while,
a Friday editorial calls for routine HIV testing for all patients entering clinics or hospitals. Maybe this is ultimately a good idea—the CDC seems to think so. But one thing I feel pretty sure of is that the Times‘ argument here, “look how well this worked in Botswana,” is not terribly compelling. One rather obvious distinction: something like a quarter of the adult population in Botswana has HIV. By the upper end estimates, we’re talking maybe half a percent of the U.S. adult population. And that means that (applying our friend Bayes’ rule) even for extremely accurate tests, you’re going to get substantially more false-positives than accurate hits in the U.S. And each of those has got to have costs—at the very least in terms of peace of mind, possibly also in terms of unnecessary, expensive treatment.

Still, if the CDC thinks routine screening is better than any more narrowly targeted form of testing, maybe it is, since they’ve surely considered precisely this factor. But Botswana analogies don’t seem likely to help settle the question.

Addendum: I suppose I should recall this caveat to some extent. It looks as though the accuracy rate of contemporary HIV tests is so high (they’re claiming 99.8 percent) that even for a relatively low frequency illness, the false positive problem will be limited.