How Serious is COVID-19? It's Time to Randomize Tests!

Let me start with a provacative question: How serious is COVID-19?  Clearly it is deadly, and you need look no further than the constant headlines of the NYT, follow just about anyone on Twitter, own a Facebook account, or just basically not live under a rock (Maybe your safest place currently).  But, we have a really really hard time right now understanding the magnitude of this.  It really shouldn't be quite this hard though- we need to make some short-term tradeoffs for a clearer understanding of long-term tradeoffs.

I propose two ideas, both involving random samples of the population.  First, we randomly sample ~1K to ~10K individuals to receive tests.  We may need to provide incentives to these individuals.  We need as much participation as is possible.  Because, those who are less likely to do it may have behaviors that are different than those who do and related to being positive.  We need to track everyone we asked to participate.  This will give us a better sense of the true percentage of the population that are currently affected.   Maybe better, Abbott Labs has apparently just produced an antibody test that would allow us to know who is immune as a result of already having it.  Of course, there is some percentage of the population that had it and perished because we didn't test early enough; however, if we did a comparable random sample we could track the sub-sample that did not have it over time and see what happens as we re-open the economy.  We would also have a sense of how many in society currently have it- by the way we may be able to use their blood to help others recover.  Given that 25%-60% of individuals may be "silent carriers" an Economist article suggests the infection rate may be 200X what we think it is. 

In the US, this could mean more than a 1/3rd of the population has already had it.  That would actually be wonderfully welcoming news (Like very seriously some of the best news you have heard since this started) because it means that most of us are immune and a really large percentage of us are asympotomatic and it won't be fatal.  This doesn't mean to go out now and celebrate- we can still overwhelm our health care system and many will die who would otherwise survive.  Our elderly will still be at great risk, and we can still protect them through smart policies.      

We don't know if the economist article is right, but I believe we would have a much better understanding of whether this is the case if we randomly test about 10K individuals across the US with these Abbott tests.  Even now, that would be taking one days worth of tests or less and using it to this end.