Multiple KM curves for overall & disease related mortality?

[Csaba Csobay-Novak]

Dear Users,
Seems to be basic... I'm struggling to create a KM survival graph that has multiple curves on it. I've found an option to have multiple curves according to a grouping variable, but how can I create two curves that are associated to the very same group of patients, but based on different variables: overall mortality and disease-related mortality? Thx for your help.

[Girish Venkataraman]

Csaba,

If you are using Stata (which I presume you are), you might be better off using -twoway- line plots generating your survival instead of sts graph. I sense you will have to do two separate stset statements. This below was me adding a flexible parametric curve to the the K-M plot generated by sts graph looking at OS at 5 years after diagnosis for both. You are welcome to use bits and pieces of the code depending on your specific use case. If you share your de-identified data via -dataex- , you might get more help in the forum for your specific use case.

local x rt1chembtk
keep if f17==1

// Run flexible parametric analysis and KM and save meansurvival
*--------------------------------------------------------------------------------------
cap drop temptime 
stpm2 i.`x', df(4) eform scale(hazard)
cap drop s0* 
cap drop s1* 
range temptime 0 5 1000 
/*predict the survival for both therapies*/
forvalues i = 0/1 {
    predict s`i', meansurv at(`x' `i') timevar(temptime) ci
}
list s0 s1 in 1000
sts graph, by(rt1chembtk) ///
                    risktable(0(1)5 , size(small) order(1 "Frontline NA" 2 "Frontline chemo only")) ///
                    plot1opts(lcol(navy%22)) plot2opts(lcol(cranberry%21))                             ///
                    addplot(line s0 s1 temptime, sort color(navy%88 cranberry%85) legend(1 "FPM" 2 "FPM2"))                     ///
                    ylab(0 "0" .25 "25" .5 "50" .75 "75" 1 "100")                                     ///
                    ytitle("Survival probability (KM)/ Predicted OS (FPM)", size(small))             ///
                    xtitle("Time since transformation (years)", size(small))                         ///
                    title("NA in del(17p) positive subset") ///
                    legend(off) ///
                    subtitle("Frontline RT therapy (P = 0.085 by FPM and Log-rank test)") ///
                    text (.6 3 "BTKi/BCL2i (OS60: {bf: 47%})") 

graph export ".\Manuscripts CORTI\Fig_kmfpm_OS60_`x'_sub_f17_1.png", width (2000) height(1600) replace as(png)

drop s0* s1* temptime _rc* _d_*   

This gives: