Skip to content

Calculate local RR from GAM

2 messages · Przemysław Jura, BXC (Bendix Carstensen)

#
Dear Group,
I am just a beginner in Epidemiology, epi-statistics and this group hence, in advance
I apologize if I ask trivial question.

I would like to ask how is it possible to calculate from GAM (general
additive model) Relative Risk
the model has following formula:
log(E[O_i]) ~ log(N_i) + someCovariates_i, where Oi is number of cases
of some disease in i-th area, N is population at risk in i-th area,
other covariates are i-th region specific.
I one paper authors calculated RR for each i-th region out of this model, but I have no
idea how can I do it myself.



Kindest regards,

pj
#
In the paper:

@article{Carstensen:Plummer:2010:JSSOBK:v38i06,
  author =	"Bendix Carstensen and Martyn Plummer",
  title =	"Using Lexis Objects for Multi-State Models in R",
  journal =	"Journal of Statistical Software",
  volume =	"38",
  number =	"6",
  pages =	"1--18",
  day =  	"4",
  month =	"1",
  year = 	"2011",
  CODEN =	"JSSOBK",
  ISSN = 	"1548-7660",
  bibdate =	"2010-09-16",
  URL =  	"http://www.jstatsoft.org/v38/i06",
  accepted =	"2010-09-16",
  acknowledgement = "",
  keywords =	"",
  submitted =	"2010-02-09",
}

there is a section which explains how to do this for parametric spline models which are not dramatically dissimilar from GAM models.

Best regards,
Bendix Carstensen