Thread by @EricTopol

That people can have #COVID19 without symptoms is one of the most challenging aspects of preventing spread.
A recent preprint meta-analysis says the rate is 16% medrxiv.org/content/10.110…
But we (@danieloran and I) believe it is >30%.
scripps.edu/science-and-me…
Which is right?
1/x

It's unclear. But first let's review the meta-analysis.
They found 460 studies and winnowed it down to 5 (1%), for various reasons 2/x

The only 5 included are here. These are rated for different biases, but not included was whether they're of a representative population (1/5?)
A goal was to have some follow-up of the people to be sure that they weren't actually pre-symptomatic, which is the big confounder
3/x

We took a different approach; it included those 5, plus 9 additional cohorts which we believe add important context to the conundrum.
all links/ details:scripps.edu/science-and-me…
The heterogeneity is remarkable, from 6-96%.
But only 1 is <30%. And many are more representative
4/x

Our >30% estimate is actually conservative, acknowledging that 12/14 cohorts with >40% must be admixed with pre-symptomatics.
The real % is important beyond spread concerns. We know from the Diamond Princess that people w/o symptoms had significant lung injury by CT scans 5/x

It's amazing that it's May and we don't really know the answer.
An @NIH study in 10,000 people has launched
nih.gov/news-events/ne… but it uses seroprevalence as a proxy for missed infections.
We need a large, prospective, random sample of those w/ active infection to know. 6/x