I am comparing the long term survival of 2 groups by Kaplan Meier curves.
The difference is statistically significant.
However, the group with lower survival is also the group with older age at presentation (mean 75.4 SD7.8 vs 78.6 SD7.2, or median 76 IQR 70-81 vs. 79 IQR 75-83, P=.009 with ANOVA).
So, I think the different age could be a confounding.
The 2 groups have different numbers (200 vs 50 pts). The smaller group has older mean age.
What do you think I should do?
Should I stratify the larger group for age, then perform another KM?
How can I do that?
Thank you
Use a Cox proportional Hazards model and include age as a covariable.
I'd identify age range in the smaller set and filter the larger set to the same range and run KM. I would also use all the data and fit either a parametric or a cox proportional hazards model with age as a continuous predictor to adjust for covariation with age. Two general complementary approaches are to 1) adjust through sampling or data handling, or 2) adjust using statistical analyses.
In line with jochen's suggestion you might find this paper of interest.. The R programs are available without variable selection in Jared Lander's R for everyone available in the z-library for download. Best wishes, David Booth
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA