Re: st: Cox proportional hazard model

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Thu, 12/8/10, Nyasha Tirivayi wrote:
> I would like to look at the outcome for patients and also
> the non-patient family members to check if there are any
> spillover effects from the medication the patient is 
> receiving- theory being AIDS medication restores health to
> patient, helping them return to work, and in turn could
> also help other family members to shift labour from
> caring for the patient, to gainful employment. So I would
> like to have separate cox regressions for the patients and
> non-patient family members?

In that case you want to include all family members, patients 
and non-patient in your model. Use the -shared()- option to 
control for the fact that members of the same family are more 
similar to each other than randomly drawn people from the 
population. Add time varying dummies for patient status and 
medicine use, and where appropriate interactions. Time starts 
_not_ when people start using the medicine, but when they enter 
the labor market or become unemployed. You want to have 
families where the medicine is not used and families where the
medicine is used, otherwise how can you find the effect of 
medicine use? The most informative families, are those that 
change during the duration of your study.

The key thing is to keep the explanatory and explained variables 
separate, duration of unemployment is your explained/dependent/y 
variable and medicine use (whether or not, or duration) is your
explanatory/independent/x variable.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

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