Is there any way to adjust other covariates when doing a Kaplan-Meier and log-rank analysis?

[Tom Hsiung]

Hi,

It seems that the Kaplan-Meier and log-rank analysis is only able to focus on one variable each time.

Tom

[German Rodriguez]

Yes, use a Cox proportional hazards model -help stcox-, which generalizes the log-rank test the same way linear regression generalizes the t-test.

[Steve Samuels]

sts test, sts graph, and sts list will adjust for covariates; they do it by adding covariates to a Cox model in the background. However, I agree with German that you should also run stox yourself. If you do the automated adjusment, you are assuming: that the Cox model is correct; i.e. that 1) that hazards are proportional for all variables. and primary predictors and 2) that a linear term is correct for each of them; and 3) that there are no interactions.

[Marcos Almeida]

You may also use - stcurve - for that, I mean, after a Cox regression.

[Tom Hsiung]

You may also use - stcurve - for that, I mean, after a Cox regression.

I will try it tomorrow. Much appreciated!

Tom