Basically I have 40 subject and for each I collect

- coronary cannulation before TAVR as : selective, non selective e unsuccessfull.

Then i collected the same data after TAVR with the same 3 level of outcome.

This is a case of repeated measure with multilevel outcome. In addition my contingency table is not "square" due to the fact that there aren't "non selective" outcome in the group "before TAVR".

Here my dataset

AfterTAVR

BeforeTAVR 0 1 2

0 1 0 0

1 1 16 22

McNemar, Stuart-Maxwell’sTest and Cochran’sQTest due to the Not Binary Outcome and Non Square matrix (3x2) of the my dataset.

Can someone have some suggestions??? I will really appreciate it