Question about Kaplan-Meier analysis and log rank. I am currently working on a project using the National Cancer Institute's SEER database. A lot of the papers published using this database perform Kaplan-Meier analysis of the data and then run log-rank tests to determine if differences between the curves are significant. SEER survival search output is in the form of a life table, with survival modeled by Kaplan-Meier analysis (and presented in the form of %survival per time). Is there any way to perform log rank analysis on these curves using ready-made data such as this? It seems that every program I try requires that the data be input from scratch (time, event/censor), so that the program can then generate a life table, the curves, and perform log-rank analysis. Also, is this a valid way to test these differences? Based on what I've read, the KM model is not ideal in population-based data given that it deals with cause-specific survival rather than relative survival. Am I interpreting this correctly? Is Cox PH analysis a better way to deal with population-based survival such as in this case, or are they not interchangeable? Thanks in advance.