Thread by @BillHanage

Lots of chatter about how early data show #Omicron is "less dangerous". But it is important to correct for the lag between infection and hospitalization when comparing the severity of variants with very different properties when it comes to transmission, here’s why 1/n

Imagine two viruses exactly the same in terms of severity, but one of them infects on average 3 people, and the other 1.1 (I know you can’t have 0.1 of a person, it’s an average). These are the effective reproductive numbers or Re and they determine how quickly they spread 2/n

If we start with 1000 infections with the less transmissible variant and just one of the more, we would expect the latter to overtake the former by the 7th generation (assuming nothing changes). Exponentials are wild 3/n

(yes those are both exponentials, it's just the exponent in one is smaller than the other and that makes a big difference) 4/n

Now assume the sickest people only get hospitalized 4 generations after they are infected. Almost all the hospitalizations we we see are with the less transmissible variant. The hospitalizations from the more transmissible have not yet started to increase rapidly 5/n

This is not only an issue at the start of the exponential but for anything growing at a faster rate. because new cases enter the denominator faster than the ones that subsequently turn bad are counted in the numerator 6/n

To account for this when comparing variants you have to account for different growth rates *and* the lag time to severe illness (thx @baym for emphasizing this). Of course the lag time might change adding another complication 7/n

And many other complicated things can go on. For now, remember that we are seeing the hospitalizations that resulted from infections that started about 2 weeks ago. And that was only a few days after Omicron was described and before it was reported in many places 8/n

The rate of increase may slow a bit because people, especially vulnerable people, alter their behavior. I also want to emphasize we don’t have a good handle on severity in highly vaccinated populations. But a lot of people in many parts of the US remain unvaccinated 9/end