I'm hoping that someone reading this list might have a solution to
something that I'm having trouble finding an answer to.
What I have is a dataset of asbestos mine workers. I'm keen to explore the
impact of time since first exposure (date they started at the mine) on
various cancer incidences. This is my time varying exposure as it
increases every year that passes. I also have date of birth and date of
entry to study. What I have been attempting to do is stsplit to create
bands of current age (5 year agebands), then stsplit on the date they
started at the mine to get lots of records for time since first exposed (ie
a new record for every year that has passed). Then I thought I could
create a nested case control study (matching on ageband) and time in
study(automatically) and then include the time since first exposed in a
conditional logistic regression model. However, the stsplits seem to be
working ok, but when it comes to the nested case control - I don't seem to
be picking up many/any controls. One suggestion was that all my cases
could be falling towards the end of the follow up period, but I've explored
this and this isn't the case. My thought is that my time periods are too
narrow or too specific (ie 16.23954 years etc ), but I'm not sure how to
get around this.
Below are the commands I've been using. I'd be really grateful for any
suggestions/assistance or alternatives for how I might deal with this problem.
Alison Reid
Occupational & Environmental Epidemiology Group
School of Population Health
University of Western Australia
35 Stirling Highway, Crawley WA 6009
