Excess Relative Risk (ERR) per unit of dose is commonly used in epidemiology (an example of the cohort study on Cancer risk due to exposure to radiation https://www.ncbi.nlm.nih.gov/pmc/articles/PMC558612/ ). Although specific statistical instruments are often used to obtain ERR's, is it probably can be performed using STATA as well?
The model to be fitted is linear product additive excess risk model
Poisson regression for IRR assessment using additive risk model (λ = exp(λ0 + λ1 +... + D)) with exposure to time at risk can be relevant tool, but the question in this case is how to separate the background (baseline?) risk from IRR estimate. In other words, can we simply treat the (IRR-1) as an Excess Relative Risk (or, more accurate, Excess Relative Hazard) per unit of dose?
reri command used with Poisson regression ( https://www.stata.com/new-in-stata/r...k-interaction/ ) can show the ERR's. However, it actually deals with a pair of interacting factor variables, and I have no idea how to use it in case of several dose categories.
The model to be fitted is linear product additive excess risk model
Code:
λ = λ0*(1+ERR*D) where λ0 - background risk function for those not exposed (multiple terms, expressing the risk factors, such as sex and age at exposure, including their interactions) ERR - excess relative risk per unit of radiation dose D - radiation dose (continious or stratified variable)
reri command used with Poisson regression ( https://www.stata.com/new-in-stata/r...k-interaction/ ) can show the ERR's. However, it actually deals with a pair of interacting factor variables, and I have no idea how to use it in case of several dose categories.
