Strange result of Proc lifetest in one strata

[YYK273]

Hi everyone,

I am currently facing an issue while working with a longitudinal health dataset to calculate the Cumulative Incidence Function (CIF) of chronic disease among four cohorts with competing risks (outcomes coded as 0. censor, 1. event, 2. death). I am trying to add personal weight, but ERROR The WEIGHT statement is not available for competing-risks data. Is it OK if I don't apply weight?

The specific concern lies with cohort=6, mid babyboomers, which is displaying an unusually high CIF of 79% compared to the more typical values ranging from 8% to 10% for other cohorts.

Summary of Failure OutcomesStratum cohort FailedEvents CompetingEvents Censored Total1 3.Hrs2 4.WarBabies3 5.Early BabyBoomers4 6.Mid BabyBoomersTotal  

Summary of Failure Outcomes
Stratum	RACOHBYR	Failed	Competing	Censored	Total
Events	Events
1	3.Hrs	70	274	1316	1660
2	4.WarBabies	108	543	1996	2647
3	5.Early BabyBoomers	128	510	2167	2805
4	6.Mid BabyBoomers	120	531	2882	3533
Total		426	1858	8361	10645

Upon reviewing the Summary of Failure Outcomes, there seems to be no significant difference between cohort=6 and the other cohorts.  Given this, 120 deaths out of a total of 3533 observations, the CIF for cohort=6 should not be 79%.

I am reaching out to the community for insights into why the CIF for cohort=6 is producing such a high value. I have included a snapshot of my dataset and the relevant portion of the output for your reference. I am wondering if there might be an issue with my code or if there are specific considerations for cohort=6 that I might be overlooking.

Here is the relevant portion of the OUTCIF output: [Include the portion of the output related to CIF for cohort=6]

If anyone has encountered a similar issue or has any suggestions on troubleshooting, I would greatly appreciate your input. Thank you for taking the time to read my post. 

This is the code I'm using for CIF with competing risk:

proc lifetest data=Phreg12121   
plots=cif(test cl)  outcif=survival_data;
time T_FI1YR*outcome(0)/failcode; 			
strata cohort ;
*WEIGHT weight0_SD;
run;

[OsoGris]

Shouldn't you be specifying a FAILCODE= value in your code?

proc lifetest data=Phreg12121   
plots=cif(test cl)  outcif=survival_data;
time T_FI1YR*outcome(0)/failcode; 			
strata cohort ;
*WEIGHT weight0_SD;
run;

[YYK273]

I appreciate your reply. Since the User’s Guide said If you specify the FAILCODE option without the equal sign, PROC LIFETEST produces a separate analysis for each distinct event value. 

I also added failcode=2;

proc lifetest data=phreg12121   
              plots=cif(test cl)  outcif=survival_data ;
   time T_FI1YR*outcome(0)/failcode=2; *format outcome eventf.;
			strata RACOHBYR ;
run;

However, it still showed a high CIF for cohor=6, while the eventcode=1 showed normal CIFs.

[PaigeMiller]

I don't know the answer ... but in situations like this, it helps to plot the raw data for Cohort 6 and then separately plot the raw data for the other cohorts, using the same axes for all plots. Do you see any major differences in the raw data?