[R] Survival dummy variables and some questions

[Frank E Harrell Jr]

Stephen wrote:
> Hi 1. To clarify: There is a posting saying that dummy regression using 
> the coxph function is not possible... That posting may be outdated.... 

That does not make sense.

> 2. Q. You say 'Make sure that eventbefore is a "pre time zero" 
> measurement' please explain: Do you mean that if someone who is not left 
> censored and has no eventbefore then their value is zero? So John is in 
> my study from 1978 to 1992 and has no events prior to 1980 then John's 
> eventbefore is 0 prior to 1980. 3. .... just kidding... Thanks S ----- 

I'm not sure you can handle left censoring by constructing something on 
the right hand side of the model.  But if you have a simple historical 
covariate such as "previous history of disease" before the start of 
follow-up time that's ok.

Frank

> Original Message ----- From: "Frank E Harrell Jr" To: "Stephen" Cc: 
> Sent: Monday, July 18, 2005 2:46 PM Subject: Re: [R] Survival dummy 
> variables and some questions > Stephen wrote: >> Hi All, >> >> I am 
> currently conducting some survival analyses. I would like to >> extract 
> coefficients at each level of the IVs. I read on a previous >> posting 
> that dummy regression using coxph was not >> possible. > > I'm not sure 
> what that means. > >> >> Therefore I though, hey why not categorize the 
> variables (I realize some >> folks object to categorization but the 
> paper I am >> replicating appears to have done so ...) > > The fact that 
> some people murder doesn't mean we should copy them. And > murdering 
> data, though not as serious, should also be avoided. > > Make sure that 
> eventbefore is a "pre time zero" measurement. > > Frank > >> >> and turn 
> the variables into factors and then try the analysis. >> >> E.g., 
> Dataset <- read.table("categ.dat", header=TRUE) >> >> 
> Dataset$eventbefore2c <- factor(Dataset$eventbefore) >> >> .. other IVs 
> here >> >> ... >> >> surv.mod1 <- coxph(Surv(start, stop, event) ~ sex2 
> + ageonset2c + >> eventbefore2c + daysbefore2c, data=Dataset) >> >> 
> Strangely enough, I receive a warning message when the variables are >> 
> treated in this way: X matrix deemed to be singular; variable 11 in: >> 
> coxph(Surv(start, stop, event) ~ sex2 + ageonset2c + eventbefore2c + I 
>  >> don't receive any warnings just treating the variables in their >> 
> initial continuous format. >> >> I am currently using version >> >> 
> platform i386-pc-mingw32 >> >> arch i386 os mingw32 system i386, mingw32 
>  >> status major 2 minor 1.1 >> year 2005 month 06 day 20 >> language R 
> Is this approach to dummy variable using coxph erroneous? >> >> Is there 
> another way to conduct dummy variable regression with coxph? >> >> Also, 
> if I include frailty (id) does anyone know of a useful way to >> 
> investigate frailty? >> >> If one were to plot recurrent events does 
> anyone know of a way of >> interpreting them? >> >> References & code 
> appreciated. >> >> BTW. not too familiar with R, less so with survival 
> analysis .... but >> well worth the effort. >> >> Many thanks in 
> advance... >> >> Regards >> >> Stephen >> >> >> ???? ?"? ???? ???? >> 
> http://mail.nana.co.il >> >> [[alternative HTML version deleted]] >> >> 
>  >> >> 
> ------------------------------------------------------------------------ 
>  >> >> ______________________________________________ >> 
> R-help at stat.math.ethz.ch mailing list >> 
> https://stat.ethz.ch/mailman/listinfo/r-help >> PLEASE do read the 
> posting guide! >> http://www.R-project.org/posting-guide.html > > > -- > 
> Frank E Harrell Jr Professor and Chair School of Medicine > Department 
> of Biostatistics Vanderbilt University >
> 
> × ×©×œ×— ×¢"×™ ×“×•××¨ × ×¢× ×¢
> http://mail.nana.co.il


-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University