Results of a stratified, apparently random sample indicate that around 1.5 million Italians, or 2.5 percent of the population, have been infected by the coronavirus. That’s a lot more than the official case count of 250K, but a lot less than might have been expected based on their 35.2K death total.

How do the serology findings fit with my algorithm? I’ve been estimating a 0.6% age-adjusted mortality rate in the US, with a plus-or-minus 10 percent national adjustment to that rate for each year above or below the US median age of 38 years. The median age in Italy is 47 years, so the age-adjusted mortality rate for Italy would be 0.6% x 1.10^{9} = 0.6 x 2.36 = 1.4%. Divide the death total by the mortality rate = 35.2K/.014 = 2.5 million infections.

My algo estimates a much higher number of infections than what the Italian serology survey came up with. Either the serology survey methodology undercounted, or people previously infected have lost immunity, or the Italian covid mortality rate is much higher than my estimate — somewhere around 2.3%. Which would imply that the US mortality rate too may need to be revised upward…

### Like this:

Like Loading...

Backing into an age-adjusted fatality rate for the US based on the Italian findings, the formula would be 2.3% divided by 1.1 to the ninth, or around 1.1 percent. That’s nearly double my 0.6 percent estimate…

LikeLike