Question about a strange type of variable in Cox regression table

#1
I have a question about this Cox regression analyse table;

View attachment 1337

If you take a look at the valuation of variables in this table, you will see that some are coded binair (0 or 1).
But for some reason the author of this table used for 'tooth type' a valuation from 1 to 6.

So my questions are;
Is that even possible in a cox regression analyse?
And how do we interpreted the data then?

Thank you for any comment!
 
Last edited:

WeeG

TS Contributor
#2
I think the right thing was to build 5 dummy variables out of this 6 level variable. try to read the article or research to see if he didn't do it. I don't think you can put a variable like that in this way, unless your software knows to handle it when you tell it that it's categorical
 
#3
I think the right thing was to build 5 dummy variables out of this 6 level variable. try to read the article or research to see if he didn't do it. I don't think you can put a variable like that in this way, unless your software knows to handle it when you tell it that it's categorical
Thank you for your quick respons!

Unfortunately, the article doens't describe how the variable is build up. But there is mentioned that they used the SPSS statical software system (SSPS Inc., Chicago, IL). Can this program handle such a categorial variable?

On the other hand; before the cox regression analyse is made, the researcher of the article makes a bivariate analyse of the data with both a logrank-test and a breslow (generalised wilcoxon) test.

View attachment 1338

Could these result's relate to building up the strange type of variable in table 5?
 
Last edited: