Dear all,
I have just joined the group and would like to ask for advice on 2 areas:
1) Adjusted prevalence using logistic regression
I came across adjustment of prevalence rates of a disease using logistic
regression. How does one get the prevalence using logistic regression and
compute the associated 95% confidence interval from this?
What puzzle me is that if I have continuous and categorical predictors in
the logistic regression and among these are age and gender which I want to
adjust the prevalence for, how do I do it? Are there any statistical
software specifically for this?
In addition, adjusted means for continuous dependent variables are based on
ANCOVA, but the problem is that the mean of categorical covariate (e.g.
gender: 1=male, 2=female) are substituted for the computation, which doesn't
sound logical. Any advice on this too?
2) Time-dependent covariate in Proportional Hazard (PH) model
For time-dependent covariates in Cox (PH) regression model with dependent
variable time1, if I have only the baseline value for blood pressure
measurement bp1 (which is time-dependent), can I just include as one of the
term on the model as bp1*time1? If not, can anyone please advise?
Thanks a lot for your attention & best regards.
_________________________________________________________________
STOP MORE SPAM with the new MSN 8 and get 2 months FREE*
http://join.msn.com/?page=features/junkmail
|