hang on Re: childhood CMV encephalitis --> 16x psychosis risk

[Guest guest]

> I guess, to find a risk ratio of 16, at least one

non-encephalitis subject and 16 CMV subjects had to come down with

psychosis (because you cant have a fraction of a person) - so that

sounds as if it would be very highly significant.

Wow, I really botched that one - what I said was waaay illogical. It

would have be correct only if half the population had had a history of

CMV encephalitis, which is not true at all.

In reality - there were just two cases of psychosis among 63 people

with prior CMV encephalitis history. That gives the rate 2/63, which

is 16.6 times higher than the rate of psychosis among the normals (ie

those without history of CMV encephalitis). The raw p is still under

0.05, but I also forgot the Bonferroni issues, which are significant

here. In total, the CMV finding is far less impressive than I thought,

on its own. If it were confirmed in a separate cohort, it would be a

finding with greater confidence. Still, childhood CMV encephalitis -

at least, childhood CMV encephalitis requiring hospitalization, which

is the criterion in this study - doesn't appear to be one of the main

causes of psychosis, because it is after all so very rare: only 63

people out of 1.2 million had childhood CMV encephalitis, whereas

2,270 of those 1.2 million had psychosis.