Problems with comp(ten(x)) function

[Dadadu84]

I have recently updated R to 4.0.0 and then to 4.0.2 version.
In both cases, I was no longer able to use the function comp(ten(x)). Indeed, I retrieve only single digit integer number in the column of p value. See below the results with the example used in the supporting material of the package. I found also that the row with "Signif.codes" disappeared.

I tried to load only the survMiner package to avoid conflicts and I checked that all dependencies were updated. Results did not change.

data("leukemia", package="survival")
f1 <- survfit(Surv(time, status) ~ x, data=leukemia)
comp(ten(f1))
Q Var Z pNorm
1 3.68934 4.04897 1 .8335 1
n 50.00000 925.00000 1.6440 3
sqrtN 12.83792 55.81480 1.7184 2
S1 2.10453 1.64882 1.6390 4
S2 1.95449 1.45827 1.6185 5
FH_p=1_q=1 0.45686 0.14587 1.1962 6

                   maxAbsZ                Var                Q           pSupBr

1 3.68934 4.04897 1.8335 1
n 50.00000 925.00000 1.6440 3
sqrtN 12.83792 55.81480 1.7184 2
S1 2.10453 1.64882 1.6390 4
S2 1.95449 1.45827 1.6185 5
FH_p=1_q=1 0.45686 0.14587 1.1962 6

The problem is related to the newer versions of R because when I switched from 4.0.2 to 3.6.3 in RStudio (Tools > Global Options) the function comp(ten(x)) worked again.

data("leukemia", package="survival")

f1 <- survfit(Surv(time, status) ~ x, data=leukemia)
comp(ten(f1))
Q Var Z pNorm
1 3.68934 4.04897 1 .8335 0.066731 .
n 50.00000 925.00000 1.6440 0.100178
sqrtN 12.83792 55.81480 1.7184 0.085727 .
S1 2.10453 1.64882 1.6390 0.101222
S2 1.95449 1.45827 1.6185 0.105553
FH_p=1_q=1 0.45686 0.14587 1.1962 0.231616

---

Signif. codes: 0 ‘’ 0.001 ‘’ 0.01 ‘’ 0.05 ‘.’ 0.1 ‘ ’ 1

                    maxAbsZ                Var                Q           pSupBr

1 3.68934 4.04897 1.8335 0.13346
n 50.00000 925.00000 1.6440 0.20035
sqrtN 12.83792 55.81480 1.7184 0.17145
S1 2.10453 1.64882 1.6390 0.20244
S2 1.95449 1.45827 1.6185 0.21110
FH_p=1_q=1 0.45686 0.14587 1.1962 0.46257

[daverogers68]

I am also encountering this issue using R version 4.0.2 and survMisc 0.5.5.

[graueb]

I have had the same issue for several months, currently running R 4.0.3 and survMisc 0.5.5.
Seen for example from the last piece of example code:

[SGHege]

You can retrieve the actual calculated p-values after the comp function by using the attr function on the ten object used. It's not very clean but here is an example using the aml data. I'm not sure why the changes but most look like label changes and I hope this helps.

library(survMisc)
fit = survfit(Surv(time, status) ~ x, data = aml)

tenfit <- ten(fit)
comp(tenfit)

lrtfit <- attr(tenfit,"lrt")
supfit <- attr(tenfit,"sup")

[graueb]

Thanks a lot for the answer, @SGHege
But when I run your example code I still just get out the numbers 1-5 in the pNorm column, rather than actual p-values. Does it not look like this when you run the code?

[AlfredoBB]

for some reason, it is not shown
just do kable(attr(tenfit, "lrt"))