Survival Analysis


I'm trying to use SPSS to perform a survival analysis. I am comparing two groups (Primary surgery versus 2-stage surgery) for recurrence after their surgery. There are some confounders (e.g. smoking / age etc). Do I just perform a cox-regression analysis and see what factors are significant? How can I adjust the Kaplain meier curve and the regression analysis to adjust for significant confounders?

Many thanks in advance.
Since there are confounders to adjust for, you need to use semi-parametric methods of survival analysis, like Cox proportional hazards model, accelerated failure time model, etc. Whether each parameter equals 0 can be determined using the corresponding Wald test or likelihood ratio test,
Thank you for reply. If I get a Wald statistic that is statistically significant (which we are defining as p<0.05), then how do I include it in the survival analysis to adjust the curves in SPSS?

Thanks again.


Not a robit
K-M curves are typically like the initial bivariate analyses before logistic regression. As noted you would use Proportional Hazard Regression (e.g., Cox). The interpretation of the model is comparable to that of logistic regression, effect of exposure while holding covariates constant.

Just looked at your attachment. You have sample size issues for sure. Too many terms given that you have only 20 events. You could try to collapse the covariates into a single propensity score, but you still will likely have too sparse of data without ability for conclusive results.