How to plot cumulative incidence with number at risk from survival analysis with competing risk usin

[lih]

How to plot cumulative incidence with number at risk from survival analysis with competing risk (see code below) 

ods graphics on;
proc lifetest data=dataname plots=cif(test ) notable timelist=0 6 12 18 24 36 48 60 OUTCIF=cifnrm_out;
time time*event(0)/eventcode=1;
strata group/order=internal;
run;
ods graphics off;

[OsoGris]

When using the EVENTCODE= option for a competing risks analysis PROC LIFETEST does not support the ATRISK plot option.

[Ksharp · Accepted solution]

You want this ?

proc lifetest data=sashelp.bmt plots=cif(test ) notable  OUTCIF=cifnrm_out ;
time t*status(0)/eventcode=1;
strata group/order=internal;
run;



data time;
do t=0 to 2500 by 500;
 output;
end;
run;
proc sql;
create table temp as
select * from (select distinct Group from cifnrm_out),time order by 1,2;
quit;
data want;
 set cifnrm_out temp(in=inb);
 by Group t;
 retain temp;
 if first.Group then call missing(temp);
 if not missing(AtRisk) then temp=AtRisk;
 if inb then time_AtRisk=temp;
 keep Group t AtRisk CIF time_AtRisk ;
run;
options missing=' ';
proc sgplot data=want;
step x=T y=CIF/group=Group;
xaxistable time_AtRisk/class=group colorgroup=Group location=inside;
xaxis label='Disease-Free Survival Time';
yaxis values=(0 to 1 by 0.2);
run;

[lih]

This is great!

Thank you

[J_Park]

This is great info to create the plot including number of risk.  I followed the steps but in the end, I only got this (attached).  No numbers shown.  I am not sure why.  Any insights?

Thank you!

SAS Code I used:

proc sgplot data=test;

step x=timeBDD_3 y=CIF/group=pred_opt_ter;

xaxistable time_AtRisk/class=pred_opt_ter colorgroup=pred_opt_ter location=inside;

     xaxis values=(0 to 3.25 by 0.5) labelattrs=(size=13) valueattrs=(size=13)  ;  

     yaxis  values=(0 to 0.2 by 0.05) label= "Probability" labelattrs=(size=13) valueattrs=(size=13);

xaxis label='Years';

keylegend  /  title='Predicted Probability' location=outside position=bottom across=3  valueattrs=(size=12) TITLEATTRS=(size=12); 

    title h=12pt 'Probability of MACE by Optimal CT Predicted Probability Tertiles';    

format pred_opt_ter gp.;

run;

[Ksharp]

Plz post your sample data,so I can test it .
And post your FULL code, you should notice that I have one more data step to get the number of risk before PROC SGPLOT ,but you didn't post it yet.

[J_Park]

Hello,
This is the full sas code I used.  I was able to add the number at risk but they are identical across the time.   I can't figure out why.   I would appreciate any insights.   The data attached is a mock up data as I can't send the actual data.
Thank you!JP

data model1_tert2(keep= T Group eventstatus );  set model1_tert ;     rename         timeBDD_3 = T        pred_opt_ter = Group          eventBDD_3 = eventstatus;  run;

proc lifetest data=model1_tert2  plots=cif(test ) notable  OUTCIF=cifnrm_out ;     time T*eventstatus(0)/eventcode=1;  strata Group/order=internal; run;

data time;do t=0 to 3.0 by 0.5; output;end;run;
proc sql;create table temp asselect * from (select distinct Group from cifnrm_out),time order by 1,2;quit;
data test; set cifnrm_out temp(in=inb); by Group ; retain temp; if first.Group then call missing(temp); if not missing(AtRisk) then temp=AtRisk; if inb then time_AtRisk=temp; keep Group T AtRisk CIF time_AtRisk ;run;

proc format ; value group0   = "<0.02"1   = "0.02-0.03"2   = ">0.03";quit;

options missing=' ';
proc sgplot data=test noborder;styleattrs datacontrastcolors=(green blue darkred);   step x=T y=cif /  group=group name='plaque'  lineattrs=(thickness=4 pattern = SOLID) ;   keylegend 'plaque' /  title='Predicted Probability' location=outside position=bottom across=3  valueattrs=(size=12) TITLEATTRS=(size=12);
  xaxistable time_AtRisk/class=group colorgroup=group location=inside;    * xaxis values=(0 to 3.25 by 0.5) labelattrs=(size=13) valueattrs=(size=13)  ;  xaxis min=0 max=3.25 values=(0 0.5 1.0 1.5 2.0 2.5 3.0 )           labelattrs=(size=13)           valueattrs=(size=13);      yaxis  values=(0 to 0.2 by 0.05) label= "Probability" labelattrs=(size=13) valueattrs=(size=13);    xaxis label='Years';    label T =  "Years"   cif= "MACE";      title h=12pt 'Probability of MACE by Optimal CT Predicted Probability Tertiles';     format group group.;run;

[Ksharp]

"The data attached is a mock up data as I can't send the actual data."
Where is your dataset? I don't see any attachment.

[Ksharp]

With using the builded-in sas dataset SASHELP.BMT as a sample dataset to test your code.

I think I found where the problem is:

You missed T variable in your code,check my original code.

data model1_tert2(keep= T Group eventstatus ); 
set sashelp.bmt;
rename status=eventstatus;
run;

proc lifetest data=model1_tert2  plots=cif(test ) notable  OUTCIF=cifnrm_out ;    
time T*eventstatus(0)/eventcode=1;  
strata Group/order=internal; 
run;

data time;
/*do t=0 to 3.0 by 0.5; */
do t=0 to 2500 by 500;
output;
end;
run;

proc sql;
create table temp as
select * from (select distinct Group from cifnrm_out),time
order by 1,2;
quit;

data test; 
set cifnrm_out temp(in=inb); 
by Group T;   /* <------- You missed the T variable*/
retain temp; 
if first.Group then call missing(temp); 
if not missing(AtRisk) then temp=AtRisk; 
if inb then time_AtRisk=temp; 
keep Group T AtRisk CIF time_AtRisk ;
run;

Open dataset TEST,you could see the different value from different T.

[J_Park]

OMG, I feel so stupid.  how can I miss the "T"?  Thank you soooooo much for your help!!!  Now it works great except one thing.  I wanted the xaxis values=(0 to 3.25 by 0.5) but it only shows 0, 1, 2, and 3.  I am not sure why.  Any insights?   

[Ksharp]

Then plz post the sample dataset , so I think test it .

Still using sas built-in dataset sashelp.bmt to test,and still not found any problem .

Referring to the following code,maybe you could found where the problem is by yourself.

proc lifetest data=sashelp.bmt plots=cif(test ) notable  OUTCIF=cifnrm_out ;
time t*status(0)/eventcode=1;
strata group/order=internal;
run;



data time;
do t=0 to 2500 by 100;
 output;
end;
run;
proc sql;
create table temp as
select * from (select distinct Group from cifnrm_out),time order by 1,2;
quit;
data want;
 set cifnrm_out temp(in=inb);
 by Group t;
 retain temp;
 if first.Group then call missing(temp);
 if not missing(AtRisk) then temp=AtRisk;
 if inb then time_AtRisk=temp;
 keep Group t AtRisk CIF time_AtRisk ;
run;
options missing=' ';
proc sgplot data=want;
step x=T y=CIF/group=Group;
xaxistable time_AtRisk/class=group colorgroup=Group location=inside;
xaxis label='Disease-Free Survival Time' values=(0 to 2500 by 100) fitpolicy=stagger; /*fitpolicy=none*/
yaxis values=(0 to 1 by 0.2);
run;

[J_Park]

Hello,
Thank you again for your response!    I am attaching my code and the sample file.    Thank you so much for your help!

proc lifetest data=sample  plots=cif(test ) notable  OUTCIF=cifnrm_out ;     time T*eventstatus(0)/eventcode=1;strata Group/order=internal; run;

proc freq data = cifnrm_out;tables  group*event group*censored;run;

data time;do t=0 to 3.0 by 0.5; output;end;run;
proc sql;create table temp asselect * from (select distinct Group from cifnrm_out),time order by 1,2;quit;
data test; set cifnrm_out temp(in=inb); by Group t; retain temp; if first.Group then call missing(temp); if not missing(AtRisk) then temp=AtRisk; if inb then time_AtRisk=temp; keep Group t AtRisk CIF time_AtRisk ;run;
proc freq data = test;tables  group*time_atrisk;run;


proc format ; value gp0   = "<0.02"1   = "0.02-0.03"2   = ">0.03";quit;
options missing=' ';
proc sort data=test; by descending group; run;
proc sgplot data=test noborder;styleattrs datacontrastcolors=(darkred blue green);  step x=T y=cif /  group=group name='plaque'  lineattrs=(thickness=4 pattern = SOLID) ;   keylegend 'plaque' /  title='Predicted Probability' location=outside position=bottom across=3  valueattrs=(size=12) TITLEATTRS=(size=12);
  xaxistable time_AtRisk/class=group colorgroup=group location=inside;    xaxis values=(0 to 3 by 0.5) labelattrs=(size=13) valueattrs=(size=13)  ;        yaxis  values=(0 to 0.2 by 0.05) label= "Probability" labelattrs=(size=13) valueattrs=(size=13);    xaxis label='Years';    label T =  "Years"   cif= "MACE";     title h=12pt 'Probability of MACE by Optimal CT Predicted Probability Tertiles';    format group gp.;run;

[Ksharp]

Where is your attachment for SAMPLE dataset?

And I noticed that you have TWO xaxis statements,why ? I think you should remove the second xaxis statement.

proc sgplot data=test noborder;
styleattrs datacontrastcolors=(darkred blue green);  
step x=T y=cif /  group=group name='plaque'  lineattrs=(thickness=4 pattern = SOLID) ;   
keylegend 'plaque' /  title='Predicted Probability' location=outside
                     position=bottom across=3  valueattrs=(size=12) TITLEATTRS=(size=12);

xaxistable time_AtRisk/class=group colorgroup=group location=inside;   
xaxis values=(0 to 3 by 0.5) labelattrs=(size=13) valueattrs=(size=13)  fitpolicy=none  label='Years'   ;     
yaxis  values=(0 to 0.2 by 0.05) label= "Probability" labelattrs=(size=13) valueattrs=(size=13);   
/* xaxis label='Years';    <--- Should get rid of it*/
label T =  "Years"   cif= "MACE";     
title h=12pt 'Probability of MACE by Optimal CT Predicted Probability Tertiles';   
format group gp.;
run;

[J_Park]

Hope you can see the sample.csv file.  Thank you!

[J_Park]

I just ran your code and it worked!   Thank you so much!!!