Cold Covid - part the eleventh

Helen Branswell of STAT interviews a real disease expert. A computational one, at that. And, naturally, his views coincide nicely with Constant Power^©; not that he outright uses such a term.

But I still could easily imagine 100,000 people [in the United States] dying every year from Covid in the endemic state just because you have a very big number of infections, even if the infection fatality rate is quite low.

One source, Novavax, has been touting its various technologies as being capable of the fastest turn-around among all drug companies. They may just get the chance to prove it. If their vaccine works, of course.

In order to have doses available in September you'll need to have made your strain selection decision for an mRNA vaccine, say, in June.

But, it may be that turn-around for mRNA vaccines is just as rapid. We'll see.

And, of course, a dig at CDC. Deserved in this case, although the 'science' reporters out there who read through the CDC PRs and have made the Big Mistake deserve some blame too.

What happens is there's an algorithm ticking along behind the scenes at the CDC website. It's a fine algorithm. Because the genomic data takes two weeks to come in, it tries to do a "nowcast" to predict what the variant distribution is today. And then that algorithm does its update and comes up at 73% Omicron with some wide confidence intervals [CIs], and now the media cycle is reporting that the CDC data finds that 73% of current cases are Omicron.

It's a model. It's not data. And they're wide CIs. We don't know what proportion of cases are Omicron in the U.S. right now, because genomic surveillance lags.

So, what are the numbers? 34.0-94.9%
Wide enough to drive a cruise ship through. He has reason to bitch.