There could be some helpful tips here
http://support.sas.com/resources/papers/proceedings12/332-2012.pdf
and
http://support.sas.com/resources/papers/proceedings15/SAS1919-2015.pdf
I have found that rescaling the response can be miraculous (sometimes anyway).
You may have collinearity problems with the predictor variables (both continuous and categorical).
You could start with a basic model that includes only a minimal number of predictors (i.e., cohort_num and quarter). If that converges, then you have something to work with. If it doesn't then that may point you towards a potential problem with patid.
... View more