Hello everyone. I am new to the forum Please could you help advise about this:

i am looking at role of routine blood tests in overall survival in a disease...i will keep it simple

I did a univariate analysis of each test (a, b, c, d) followed by a multivariate analysis (a, b remained independent predictors).

Then, I thought of combining the tests results (high, low results-categorical: high a&b, low a&b, high b&c, low b&c, high a&c, low a&c), and did a univariate analysis. Then, i did a separate multivariate analysis for the combined tests only without re- adding the single tests to the same analysis, and found high a &b, and high a&c were independent predictors.

Please, my question to you: is this a valid method of doing a multivariate analysis of single tests and doing a separate multivariate analysis of combined tests and presenting the results?

Or do you think it is only valid if i run a multivariate analysis including single tests and combined tests together? I am using SPSS.

Any help or advice is appreciated! Thank you. Prit