[R] validate (package Design): error message "subscript out of bounds"

[Wentzel-Larsen, Tore]

Dear R users 

I use Windows XP, R2.5.1 (I have read the posting guide, I have 
contacted the package maintainer first, it is not homework).

In a research project on renal cell carcinoma we want to compute 
Harrell's c index, with optimism correction, for a multivariate 
Cox regression and also for some univariate Cox models.
For some of these univariate models I have encountered an error
message (and no result produced) from the function validate i 
Frank Harrell's Design package:

Error in Xb(x[, xcol, drop = FALSE], coef, non.slopes, non.slopes.in.x,  : 
        subscript out of bounds

The following is an artificial example wherein I have been able to 
reproduce this error message (actual data has been changed to preserve
confidentiality):

library(Design)

# an example data frame:
frame.bc <- data.frame(time1 = c(9,24,28,43,58,62,66,107,116,118,123,
	127,129,131,137,138,139,140,148,169,176,179,188,196,210,218,
	1,1,1,2,2,3,4,8,23,32,33,34,43,44,48,51,52,54,59,59,60,60,62,
	65,65,68,70,72,73,74,81,84,88,98,99,106,107,115,115,117,119,
	120,122,122,122,122,126,128,130,135,136,136,138,149,151,154,
	157,159,161,164,164,164,166,172,172,176,179,180,183,183,184,
	187,190,197,201,201,203,203,203,209,210,214,219,227,233,4,18,
	49,113,147,1,1,2,2,2,2,2,3,4,6,6,6,6,6,6,6,6,9,9,9,9,9,10,10,
	10,11,12,12,12,13,14,14,17,18,18,19,19,20,20,21,21,21,21,22,23,
	23,24,28,28,29,29,32,34,35,38,38,48,48,52,52,54,54,56,64,67,67,
	69,70,70,72,84,88,90,114,115,140,142,154,171,195),
	status1 = c(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
	0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
	0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
	0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
	1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
	1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
	1,1,1,1,1),
	bc1 = factor(c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
	2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
	2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
	2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2,
	2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
	2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2),
	labels=c('bc.1','bc.2')),
	age = c(58,68,23,20,50,43,41,69,20,48,19,27,39,20,65,49,70,59,31,43,25,
	61,60,45,34,59,32,58,30,62,26,44,52,29,40,57,33,18,50,50,55,51,38,34,
	69,56,67,38,66,21,48,39,62,62,29,68,66,19,60,39,55,42,24,29,56,61,40,
	52,19,40,33,67,66,51,48,63,60,58,68,60,53,20,45,62,37,38,61,63,43,67,
	49,39,43,67,49,69,32,37,32,63,33,47,66,39,23,57,26,61,20,49,69,30,40,
	29,38,66,60,69,69,44,65,25,41,53,18,55,45,59,49,27,51,29,67,26,24,26,
	47,23,50,27,35,45,32,26,45,45,63,39,39,22,38,27,31,27,49,65,66,49,39,
	21,51,49,55,63,19,26,50,21,24,34,65,33,55,33,36,53,48,25,54,58,60,34,
	47,23,34,60,39,34,22,30,41,55,64,48,34,54))
frame.bc

# preparing for a simple univariate Cox regression:
dd.bc <- datadist(frame.bc[, c('bc1','age')], adjto.cat='first')
options(datadist = 'dd.bc')

# a univariate Cox regression:
cph.bc <- cph(formula = Surv(time1,status1)~bc1,
	data = frame.bc, x=TRUE, y=TRUE, surv=TRUE)
anova(cph.bc)
cph.bc
summary(cph.bc)

# the validate command for the Cox model:
val.cph.bc <- validate(cph.bc, B=200, dxy=TRUE , pr=TRUE)

----------------------
Output from the validate command:

       training       test
Dxy   -0.124360 -0.1423409
R2     1.000000  1.0000000
Slope  1.000000  0.7919584
D      0.016791  0.0147536
U     -0.002395  0.0006448
Q      0.019186  0.0141088
       training       test
Dxy   -0.191875 -0.1423409
R2     1.000000  1.0000000
Slope  1.000000  0.8936724
D      0.022397  0.0147536
U     -0.002339  0.0001367
Q      0.024736  0.0146169
       training       test
Dxy   -0.199514 -0.1423409
R2     1.000000  1.0000000
Slope  1.000000  0.8075246
D      0.025717  0.0147536
U     -0.002447  0.0005348
Q      0.028163  0.0142188
Error in Xb(x[, xcol, drop = FALSE], coef, non.slopes, non.slopes.in.x,  : 
        subscript out of bounds


Any help/suggestions will be highly appreciated.


Sincerely,
Tore Wentzel-Larsen
statistician
Centre for Clinical research
Armauer Hansen house 
Haukeland University Hospital
N-5021 Bergen
tlf   +47 55 97 55 39 (a)
faks  +47 55 97 60 88 (a)
email tore.wentzel-larsen at helse-bergen.no