samclifford
commented
2 years ago My understanding of how impact is calculated is that impact for age group j is
Impact_j = sum_{i=j}^{100} averted cases in age group i due to vaccination at age jwhere i goes up in five year increments.
And so relative impact should have a baseline of
cases_j = sum_{i=j}^{100} cases in age group i that could have been averted due to vaccination at age jso that if we are looking at the proportion of cases averted we have
relative_impact_j = Impact_j/cases_jor if we want to look at cases averted per vaccinee we have relative_impact_j = Impact_j/pop_j where
pop_j = sum_{i=j}^{100} population in in age group i that could have been vaccinated Or am I wrong and we actually want denominators that are the cases or population seen in the entire over-55s?
https://github.com/deusthindwa/optimal.age.targeting.pneumo.vaccines/blob/b3df9278609cad2b988cb62f986088a9b35ee205/script/6_impact_scenario.R#L54
The summing here is done with a denominator of the PCV20 Impact + PPV23 Impact at 65 years for vaccination age. Not sure what this is meant to indicate.
https://github.com/deusthindwa/optimal.age.targeting.pneumo.vaccines/blob/b3df9278609cad2b988cb62f986088a9b35ee205/script/6_impact_scenario.R#L86
Similarly, the above line calculates impact in each waning scenario at vaccination at 65 years old.
I'm writing some code that makes a function for the variables to group over when doing the relative impact calculation. Is the point of these to do an odds ratio or relative risk to show impact in one scenario vs another?
Additionally, there's zero uncertainty around these calculations due to the way that impact is calculated.