Re: st: Analysis of longitudinal data with timevarying covariates and late entry

[Steve Samuels <sjsamuels@gmail.com>]

"search lexis, all" and "search apc, all" might lead you to some solutions.

Steve

On Nov 15, 2013, at 8:47 AM, roland andersson wrote:

I am analysing the effect of migration on the incidence of
appendicitis in a large national cohort. The inclusion into the cohort
started at 1990 from a national census, and to this population are all
births and immigrations added up till 2010. The total number of
individuals are>11.000.000
As a consequence there is no fixed point in time when the study
starts. The incidence of appendicitis is decreasing over the
studytime, and is also dependent on age.

My intention is to stset the data with age as studytime and late entry
for the immigrants, stsplit in age-bands and do the analysis using
strate and stmh. I assume that this will control for the age-effect.

I have not been able to figure out how to control for the time-period
effect. Ideally I would like to split the data also according to
calender year, but I can not see how this can be done as I use the age
as analysis time. I can stsplit on age but I cannot figure out how to
stsplit on year of birth as the studytime is age.

I have also considered to make a matched analysis where the follow up
starts at the time of immigration for both the natives (controls) and
immigrants (cases) but have not found any easy way of matching the
data n:1 on year of birth and sex.

Appreciate all help

Roland Andersson
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/