Model Description
We model the probability of death from covid as the
probability of covid infection times the probability of
death given infection.
For the probability of infection,
we use the same model as
microcovid.org, but with
three modifications. First,
we treat the microcovid.org activity risk as an
odds ratio, and transform it by
x -> x / (1 + x)
to get the
conditional probability of infection given that a person with
covid is at the event. This will make little difference
for small values of x (which microcovid.org emphasizes is
the range of validity for their model), and for large values,
it prevents the conditional probability from being greater than
1.
Second, we outsource the estimation of the true infection
rate in your area. And finally, we modify that infection rate
into one customized for your event, if you specify that
the event attendees are all vaccinated. (In that case,
we assume that vaccinated people in your area have an
infection rate equal to (1 - VEI) times to the infection
rate of unvaccinated people, where VEI is the vaccine
effectiveness against infection parameter.)
For the probability of death given infection, we start
with the age-specific infection fatality rate estimate
from
"Assessing the age specificity of infection fatality rates for COVID-19"
and further personalize it by multiplicative factors based
on sex and vaccination status.
Finally, to estimate the number of life hours lost, we
multiply the probability of death by an actuarial estimate
of expected lifespan based on age and sex. Specifically,
we use a Gompertz-Makeham equation
fit to
US lifespan data.