Thanks for clarifying that this is about cancer staging schemes and for the attached cross(?)-tabulations.
I find it confusing that both the tables in your initial post and the new ones have titles suggesting cross-tabulations (var1*year etc.), but have no row headers. This makes it difficult to compare different tables appropriately, because it is not clear which categories (rows) in different tables should be compared.
Each of the tables in the attachment has a total row at the bottom, which can be guessed only from the numbers.
Also, you use variable name VarComb (which I used for the CATX-concatenated variable) for a variable derived using the COALESCE function (an approach which I advised against).
The attached tables present a different picture than those in your initial post: Table "var3*year" in your initial post suggested that for the years 2004-2006 VAR3 has only very few non-missing values. The attached table with the same title, however, shows large numbers in all (displayed) categories. At the same time, table "Var2*year" has substantial numbers in only two of the eight categories shown. Of course, this must have to do with the significant change in the assignment of years to variables which has occurred between your initial and current post.
I tend to believe that, at the current stage, this is not a programming question. You should rather talk to a medical expert who is familiar with those different cancer staging schemes and who can advise you if and how different schemes can be unified for further analysis. As soon as you can describe the rule clearly, many forum members will be able to help you implementing it in SAS.
... View more
