Sunday, April 12, 2020

When is CVI worthwhile?

My previous post outlined some reasons to think that Robin Hanson's idea of low-dose controlled voluntary infection (CVI) should be explored further, despite widespread knee-jerk moral discomfort with the idea.  That's not to say that anyone should be attempting to implement this themselves right now.  Rather, careful experimentation (with consenting volunteers who understand the risk) is urgently needed to gather more data and allow us to better judge whether or not CVI would be a worthwhile policy to roll out more broadly.

What would make CVI worthwhile?  One central consideration, to begin with, would be a comparison of how much safer low-dose CVI turns out to be compared to uncontrolled infection, to be considered in conjunction with background estimates of how likely one is to suffer uncontrolled infection in the absence of CVI.  As a rough first pass, if we discover an optimal low-dose that results in an infection n times less dangerous than uncontrolled infections (on average), then it looks like anyone with a greater than 1/n chance of eventual accidental infection (within the time-frame in which CVI would grant immunity) would benefit (in the sense of reducing their ex ante health risk) from CVI.

Hanson estimated, based on other viruses, that covid's n is likely somewhere between 3 - 30.  A better-informed estimate of this value would make a big difference to when, and for whom, it makes sense to offer CVI.  But even on the lowest end, and even if most people can reasonably expect to avoid infection entirely due to ongoing suppression measures, targeted CVI could still prove beneficial for highly-exposed essential (e.g. health) workers.  Even more clearly, CVI would straightforwardly reduce risk for anyone who would otherwise recklessly seek out uncontrolled infection in order to subsequently qualify for an "Immunity Passport" or the like.

(Aside: an epidemiologist in the NY Times sensibly warns people against deliberately infecting themselves in an uncontrolled fashion. Their reasons mostly involve highlighting what a serious health risk Covid-19 poses, even for the young -- though appealing to hospitalization rates for "confirmed cases" is misleading here due to obvious selection effects in who gets tested.  At any rate, the seriousness of uncontrolled coronavirus infection is precisely why low-dose CVI could be so important for anyone who might otherwise suffer a high-risk uncontrolled infection.  I agree with the author that individuals should not be seeking out the latter!)

Beyond the "rough first pass", there I can see two main further factors to consider: (1) Timing the burden on the medical system, and (2) Indirect benefits.

re: (1), as discussed previously, familiar "flatten the curve" considerations would count in favour of redistributing Covid-19 infections away from times of peak burden on the medical system.  The value of data is so high that I don't think this is any reason to delay initial experimentation, but wider roll-outs should presumably be timed to match 'lulls' in the local state of the epidemic.  Otherwise, you'd generally only want to offer it during a 'peak' period to individuals for whom there's a greater than 1/n chance that they would otherwise still get infected during the peak period, in order for CVI to actually reduce hospitalizations during this time of limited medical capacity.

re: (2), the indirect benefits of returning sooner to "life as usual" are difficult to assess but shouldn't be underestimated (especially for the less privileged, who may be in economically precarious positions, live in cramped conditions, and otherwise lack access to greenspace and other vital goods). 

Looking at the "big picture" question of how society at large is going to deal with covid going forward: Policymakers have generally rejected the simple 'herd immunity' strategy as having too great a health cost. But if CVI can significantly reduce those health costs, then it might be worth revisiting how this then compares with the suppression strategies currently on offer (which all have their own significance downsides).

There are a lot of unknowns at present.  While far from certain, there's at least a non-trivial chance that widespread CVI could turn out to be the best strategy (when all costs are properly taken into account).  And there's a much greater chance that some more targeted uses of CVI (for high-exposure individuals) could be worthwhile as part of a broader suppression strategy.  Finally, of course it's always possible that we won't be able to identify a safer low-dose that helps at all.  But some initial experimentation is, I think, very clearly called for in order to provide policy-makers with the essential information they need in order to assess when, if at all, CVI should be used to help mitigate the risks of this pandemic.


Post a Comment

Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)