I'm trying to understand a paper on kidney cancer research. The study uses a multivariate Cox regression. The outcome variable is death/no death from kidney cancer. There are several predictor variables, including pathological tumor stage (ordinal variable from 1 to 4, describing the anatomic extent of the cancer). From my understanding, you need to use dummy variables to include an ordinal variable with >2 possible categories in a multivariate Cox regression. Therefore, you should get a different p-value for each dummy variable in the final model. However, the paper I'm reading only gives one p-value for tumor stage in the final model. Can anyone tell me how you can get a single p-value for an ordinal predictor variable (with >2 possible categories) in a multivariable Cox regression? Any help is appreciated.