p value calculated by ggsurvplot using Gehan-Breslow test(log.rank.weights = "n") is wrong

[stroke1989]

when I use ggsurvplot to plot a K-M curve and make statistic test, I set log.rank.weights = "n" (Gehan-Breslow test, not default "log rank test"), however, the p value is same to that calculated by log rank test method, and the plot still showed words "log-rank". I use survdiff and surv_pvalue to recalculate p value (both use Gehan-Breslow test), the p-value calculated by survdiff is same to that calculate by surv_pvalue, and same to p value calculated by SPSS K-M(Breslow method. that is to say p value(survdiff- breslow method) = p value( surv_pvalue - breslow method) = p value(SPSS k-m breslow) ≠ p value (ggsurvplot, log.rank.weights ="n", also breslow method). I don't know why.

code in following

library(survival)
library(survminer)
View(lung)
attach(lung)
fit <- survfit(Surv(time, status) ~ sex, data= lung)
ggsurvplot(fit, data = lung, ggtheme= theme_bw(), risk.table = T, pval = TRUE, pval.method = TRUE, surv.median.line = "hv", tables.height = 0.2, xlim = c(0, 110),
log.rank.weights= "n" )
survdiff(Surv(time, status) ~ sex, rho = 1)
surv_pvalue(fit, method = "Gehan-Breslow")

[jgarces02]

Any new about this?? Thanks

[stroke1989]

Any new about this?? Thanks

I don't know what you mean?

[jgarces02]

No, it's for @kassambara... if he's already solved this problem you commented (or have you found another solution?). Sorry the confusion!

[stroke1989]

No, it's for @kassambara... if he's already solved this problem you commented (or have you found another solution?). Sorry the confusion!

did you also found this problem when you use ggsurvplot package? I assumed that just I meet this problem

[Estefy-987]

I had the same problem. It's like surv_pvalue is not updated, but if you write the full name "gehan-breslow" it works.

surv_pvalue(fit_SG_tto, method = "1")
variable pval method pval.txt
1 TB 0.002100802 Log-rank p = 0.0021
surv_pvalue(fit_SG_tto, method = "log-rank")
variable pval method pval.txt
1 TB 0.002100802 Log-rank p = 0.0021
surv_pvalue(fit_SG_tto, method = "n")
variable pval method pval.txt
1 TB 0.002100802 Log-rank p = 0.0021
surv_pvalue(fit_SG_tto, method = "gehan-breslow")
variable pval method pval.txt
1 TB 0.0018 Gehan-Breslow p = 0.0018

However in ggsurvplot if you write gehan-breslow" doesn't work

Error in (function (fit, data = NULL, fun = NULL, color = NULL, palette = NULL, :
log.rank.weights %in% c("survdiff", "1", "n", "sqrtN", "S1", .... is not TRUE

[kassambara]

fixed now, thanks