0
$\begingroup$

I am wondering how to conduct longitudinal analysis of incidence rates.

A population was followed 5 years (2001-2005), and in each year "person-years" and "event" were calculated.

Now I am interested in seeing how the incidence rates changed accross the 5 years, and how the incidence rates are related to a categorical variable, "region" (with 4 levels).

Does Poisson regression analysis accouting for longitudinal data work? And how to control person-years in this particular analsysis? Any R packages for this?

Here is a set of hypothetical data:

Input = ("
 personyears  event  time    Region
   1263.2      26    2001     East
   2132.5      28    2002     East
   471.3       31    2003     East
   262.9       18    2004     East
   982.3       21    2005     East
   1245.1      22    2001     South
   339.9       30    2002     South
   791.2       18    2003     South
   1026.1      16    2004     South
   2324.6      28    2005     South
   1232.7      15    2001     West
   987.3       19    2002     West
   699.3       27    2003     West
   2000.5      40    2004     West
   729.9       13    2005     West
   914.9       26    2001     North
   725.6       31    2002     North
   765.9       32    2003     North
   982.6       21    2004     North
   814.7       26    2005     North
 ")

Data = read.table(textConnection(Input),header=TRUE)

Thanks.

$\endgroup$
  • $\begingroup$ There are a lot of related Q&A's on this site that should help you, e.g. stats.stackexchange.com/q/198527/241093. This person-years variable (exposure variable) needs to be entered in the model as an offset (calculated as a log of exposure). See, e.g. this Q&A: stats.stackexchange.com/q/311556/241093 for how it is denoted in R regression formulas. $\endgroup$ – AlexK May 24 at 22:53
  • $\begingroup$ @AlexK Thank you very much. $\endgroup$ – user248991 May 26 at 20:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.