Is there a way to calculate adjusted hazard curves for specific levels of a single variable which are adjusted for all other covariates in the model? Basically, I would like to plot hazard curves for my outcome for three different race/ethnicity groups, but there are a number of other covariates in the full model, all of which are categorical, that I would like these curves to be adjusted for. The SAS documentation for PROC PHREG only describes how to calculate hazard curves when values are specified for all of the covarites, and if no values are specified, then reference values are used by default. However, I don't want to show these curves for specific subgroups of my population; I'd like the curves to apply to the entire population, adjusted for these other variables. Is this possible?
If you recode your categorical covariates into 0/1 numeric variables, then PHREG will set them to their mean in setting up the adjusted curves. I think that is what you are saying when you use the term 'entire population.'