After the regression , I want to estimated death from the effect of
age and period, while setting the effects of gender to zero.
Compute the prediction using the sample mean of gender in the equation.
It would be easiest if you created e.g. male=(gender==1) and ran the
with male instead of i.gender. Then the effect of gender is the
magnitude of the male coefficient. If the sample were equal #s M,F,
the effect of age and period ignoring gender would be to set gender
to its sample mean of 0.5. But if the sample mean of male is other
than 0.5, set it to that and calculate the prediction appropriate for
every individual having "average" gender. Calculate the linear
prediction and then transform it into poisson's standard if that is
what you want.