Thankfor for help in evaluating survival prognosis


I'am investigating into the survival between two cohorts of incident patients. One historical (1986-1996) and one contemporary (2002-2012). I was using the Kaplan Meyer survival curve. Using log rank to show a significant difference in the survival between the curves of the two population (our hypothesis is that the prognosis has improved). When using the log-rank I get a significant difference. This is the catch; I want to do this evaluation although only two years into the curve my contemporary cohort gets a lot of censoring due to the fact that the "fate" for these cases can only be evaluated for a totalt of two years. Other cases can in the best case be followd for 12 years (2002-2014). Does this uneven symmetry AND number of censoring between the cohorts make it impossible to use the Log rank? Or can the log rank be trusted since the censoring is strictly due to administration and should not be confounded by an bias?

Would be very thankfull to anyone who could share some thoughts on this!

Good night


Omega Contributor
If one group was followed for a substantial longer period that would seem to pose concerns.

Can you standardize the length of followup to only 1 year? If not, why?