Dear friends.
I have a paper (details below) examining the risk of renal failure after an
operation. A logistic regression was done, and the coefficients to two
regressors (age and creatinine) plus intercept with standard errors are
given. These coefficients must be dependent in estimation, and when no
details are given, I thought how I could most informatively get an
impression as to how these standard errors materialized in uncertainty in
predicting the risk of renal failure for a fellow age such and such etc.
The approach below gives very broad ranges but I can't see it is obviously
wrong ?
#simulations for Schepens: Risk assessment of acute renal failure after
#thoracoabdominal aortic aneurysm surgery, Annals of Surgery 1994;219:400-407
#Final logistic regression has u = -14.42+0.1723*age+0.005481*crea
#and prob = exp(u)/(1+exp(u))
#coefficients are given with se: 4.71, 0.0649 and 0.0029 respectively;n=84
sd <- c(4.71,0.0649,0.0029)*sqrt(84)
mean <- c(-14.42,0.1723,0.0055)
prob <- function(age,crea) {
u <- rnorm(1000,mean[1],sd[1]) +age* rnorm(1000,mean[2],sd[2])+crea*
rnorm(1000,mean[3],sd[3])
summary( exp(u)/(1+exp(u)))}
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._