Stratified log rank vs cox regression

#1
Hello everyone,

I am performing a survival analysis to compare survival with two treatment regimen (with SPSS).

With Log-rank p value is significant (p: 0,033)

When stratified with a binary variable significance is lost (p:0,062)

With Cox regression (backward LR method) only variable left in the model is treatment (HR: 2.4). The binary variable that is shown to effect survival outcome with stratified log rank, is not significant in cox model.

How should I interpret that? Is the treatment significantly improve survival? Anything else to do?
 
Last edited:
#3
Variables that I have added to the model are known prognostic factors of the disease, and used in many previous studies. So I think data meets the proportional hazards assumption.
I found this paper helpful; a quote "This method (stratified log rank) not generally regarded as a formal statistical model, but is of use where a very small number of covariates are to be considered, if only as an exploratory method of analysis."
http://www.ncbi.nlm.nih.gov/pubmed/12888808