Hi,

I need some reference to the following question:

I need some rules of a thump for a retrospective multivariate survival analysis. We are looking at cases that got cancer of urinary bladder, prostate and kidney. All together there are 320 cases of urinary bladder, 1500 of prostate and 352 of kidney.

We have around 40 variables and want to do a multivariate survival analysis (approximately 300 deaths happened in all groups together) and a multivariate analysis of how those variables are related to the incidence of cancer (variables such as number of symptoms at their visit of GP,..).

For the multivariate analysis there is role of a thump to have 15-20 variables for each independent variable.

Does something like this exist for multivariate survival analysis for retrospective data (for example we have 320 cases of cancer in one group, how much deaths must occur to be able to calculate the multivariate survival analysis)?

Thank you very much for your help.

Špela