R help - Oct 2009 - cox regression extract strata as numeric

Hi there,

I perform a stratified cox and then I need the strata as a  numeric array
"straft.ln"
ft.ln <-
coxph(Surv(times,deaths)~ages+chemos+chemos:f1+chemos:f2+horms+horms:f1+horms:f2+grades+grades:f1+grades:f2+positives+positives:f1+positives:f2+sizes+sizes:f1+sizes:f2+strata(stra),data=ddd)

basehazzft.ln=basehaz(ft.ln,centered=FALSE)
H0ft.ln=c(basehazzft.ln[,1])
timeft.ln=c(basehazzft.ln[,2])
straft.ln=c(basehazzft.ln[,3])

I notice that basehazzft.ln[,3] is not the same with straft.ln with respect to
the values i.e.:

(basehazzft.ln$stra[1])
[1] stra=1
134 Levels: stra=1 stra=10 stra=100 stra=101 stra=102 ...
stra=99
> c(basehazzft.ln$stra[1])
[1] 1


so far so good but:

(basehazzft.ln$stra[285])
[1] stra=2
134 Levels: stra=1 stra=10 stra=100 stra=101 stra=102 ...
stra=99
> c(basehazzft.ln$stra[285])
[1] 47

while the desired value is 2, I get a 47. What am I doing wrong? I tried the
as.numeric function but I have the same problem..

I just want instead of the [ "stra=1" "stra=1"
"stra=1" etc]  the [1,1,1 etc]

Thanx in advance for any answers!




      
___________________________________________________________ 
×ñçóéìïðïéåßôå Yahoo!; 
ÂáñåèÞêáôå ôá åíï÷ëçôéêÜ ìçíýìáôá (spam); Ôï Yahoo! Mail 
äéáèÝôåé ôçí êáëýôåñç äõíáôÞ ðñïóôáóßá êáôÜ ôùí åíï÷ëçôéêþí 
ìçíõìÜôùí http://login.yahoo.com/config/mail?.intl=gr

	[[alternative HTML version deleted]]

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of ?e???da? ?pa?t??
> Sent: Tuesday, October 27, 2009 3:51 PM
> To: r-help at r-project.org
> Subject: [R] cox regression extract strata as numeric
> 
> 
> Hi there,
> 
> I perform a stratified cox and then I need the strata as a? numeric
> array "straft.ln"
> ft.ln <-
> coxph(Surv(times,deaths)~ages+chemos+chemos:f1+chemos:f2+horms+horms:f1
> +horms:f2+grades+grades:f1+grades:f2+positives+positives:f1+positives:f
> 2+sizes+sizes:f1+sizes:f2+strata(stra),data=ddd)
> 
> basehazzft.ln=basehaz(ft.ln,centered=FALSE)
> H0ft.ln=c(basehazzft.ln[,1])
> timeft.ln=c(basehazzft.ln[,2])
> straft.ln=c(basehazzft.ln[,3])
> 
> I notice that basehazzft.ln[,3] is not the same with straft.ln with
> respect to the values i.e.:
> 
> (basehazzft.ln$stra[1])
> [1] stra=1
> 134 Levels: stra=1 stra=10 stra=100 stra=101 stra=102 ... stra=99
> > c(basehazzft.ln$stra[1])
> [1] 1
> 
> 
> so far so good but:
> 
> (basehazzft.ln$stra[285])
> [1] stra=2
> 134 Levels: stra=1 stra=10 stra=100 stra=101 stra=102 ... stra=99
> > c(basehazzft.ln$stra[285])
> [1] 47
> 
> while the desired value is 2, I get a 47. What am I doing wrong? I
> tried the as.numeric function but I have the same problem..
> 
> I just want instead of the [ "stra=1" "stra=1"
"stra=1" etc]? the
> [1,1,1 etc]
> 
> Thanx in advance for any answers!
Here's one way to do it (using the example on coxph help page):
> require("survival")
Loading required package: survival
Loading required package: splines
> bladder1 <- bladder[bladder$enum < 5, ] 
> m1 <- coxph(Surv(stop, event) ~ (rx + size + number) * strata(enum) +
cluster(id), bladder1)
> basehazzft.ln=basehaz(m1,centered=FALSE)
> (basehazzft.ln$stra[1])
[1] enum=1
Levels: enum=1 enum=2 enum=3 enum=4
> basehazzft.ln$stra
 [1] enum=1 enum=1 enum=1 enum=1 enum=1 enum=1 enum=1 enum=1 enum=1 enum=1
enum=1 enum=1 enum=1
[14] enum=1 enum=1 enum=1 enum=1 enum=1 enum=1 enum=1 enum=1 enum=2 enum=2
enum=2 enum=2 enum=2
[27] enum=2 enum=2 enum=2 enum=2 enum=2 enum=2 enum=2 enum=2 enum=2 enum=2
enum=2 enum=2 enum=2
[40] enum=2 enum=3 enum=3 enum=3 enum=3 enum=3 enum=3 enum=3 enum=3 enum=3
enum=3 enum=3 enum=3
[53] enum=3 enum=3 enum=3 enum=3 enum=3 enum=4 enum=4 enum=4 enum=4 enum=4
enum=4 enum=4 enum=4
[66] enum=4 enum=4 enum=4
Levels: enum=1 enum=2 enum=3 enum=4
> as.character(basehazzft.ln$stra)
 [1] "enum=1" "enum=1" "enum=1" "enum=1"
"enum=1" "enum=1" "enum=1" "enum=1"
"enum=1" "enum=1"
[11] "enum=1" "enum=1" "enum=1" "enum=1"
"enum=1" "enum=1" "enum=1" "enum=1"
"enum=1" "enum=1"
[21] "enum=1" "enum=2" "enum=2" "enum=2"
"enum=2" "enum=2" "enum=2" "enum=2"
"enum=2" "enum=2"
[31] "enum=2" "enum=2" "enum=2" "enum=2"
"enum=2" "enum=2" "enum=2" "enum=2"
"enum=2" "enum=2"
[41] "enum=3" "enum=3" "enum=3" "enum=3"
"enum=3" "enum=3" "enum=3" "enum=3"
"enum=3" "enum=3"
[51] "enum=3" "enum=3" "enum=3" "enum=3"
"enum=3" "enum=3" "enum=3" "enum=4"
"enum=4" "enum=4"
[61] "enum=4" "enum=4" "enum=4" "enum=4"
"enum=4" "enum=4" "enum=4"
"enum=4"
> substr(as.character(basehazzft.ln$stra), 6, 999)
 [1] "1" "1" "1" "1" "1"
"1" "1" "1" "1" "1"
"1" "1" "1" "1" "1"
"1" "1" "1" "1" "1"
"1" "2" "2" "2"
[25] "2" "2" "2" "2" "2"
"2" "2" "2" "2" "2"
"2" "2" "2" "2" "2"
"2" "3" "3" "3" "3"
"3" "3" "3" "3"
[49] "3" "3" "3" "3" "3"
"3" "3" "3" "3" "4"
"4" "4" "4" "4" "4"
"4" "4" "4" "4"
"4"
> as.numeric(substr(as.character(basehazzft.ln$stra), 6, 999))
 [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 3 3 3 3 3 3 3 3
[49] 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4
>
HTH

Steven McKinney
Statistician
Molecular Oncology and Breast Cancer Program
British Columbia Cancer Research Centre


> 
> 
> 
> 
> 
> ___________________________________________________________
> ?????????????? Yahoo!;
> ?????????? ?? ?????????? ???????? (spam); ?? Yahoo! Mail
> ???????? ??? ???????? ?????? ????????? ???? ??? ???????????
> ????????? http://login.yahoo.com/config/mail?.intl=gr
> 
> 	[[alternative HTML version deleted]]

-- begin included message ---
> (basehazzft.ln$stra[285])
[1] stra=2
134 Levels: stra=1 stra=10 stra=100 stra=101 stra=102 ...
stra=99
> c(basehazzft.ln$stra[285])
[1] 47

while the desired value is 2, I get a 47. What am I doing wrong? I tried
the as.numeric function but I have the same problem..

--- end included ---

The variable in question is a factor.  The internal codes used for the
levels of a factor variable 'x', as reported by as.numeric(x) or c(x),
are not the same as the names of the levels, which appear from printing
x itself.

The confusing nature of factor variables makes a regular appearance on R
help.  I would suggest reading the section on factor variables in one of
the many introductory R texts. Your question is not per se an issue with
the Cox model routines.  

But to answer your underlying query, try substring(x, 6,10).  

PS If you use survfit() instead of basehaz() you will get back a
survival curve object, which is often much more useful.  I never use
basehaz myself, but there was a cry for a function with the same name
and same (limite) output as the SAS phreg option.  (Basehaz calls
survfit and then throws away 1/2 the information.)
 
Terry T.