Avian 'flu - maybe not so mild after all
Argument, simply put, is that only acute cases were being diagnosed and those are necessarily the ones where the patient is in danger of dying, so the mortality rate is artificially inflated by selection bias. We only know of the sparse really bad cases.
If only, the argument goes, we did a random sample of a population where a case had been found, we would likely find many asymptomatic or mild cases that never came to medical attention.
Or, unfortunately, maybe not...
Not definitive but curious.
Good news is then, that the H5N1 strain really has a hard time spreading; the bad news is that it is very lethal. As is.