Beast mode count distinct based on subset of data

Hi - I am trying to count distinct patient ids based on a subset of data.  My beast calculation is shown below.  To explain further, I am in need of the unique patients who are first in the initial group (have at least 1 record with one of the indicated procedure codes), and then, of those patients, how many also have a record with this procedure code 0521F.  I can easily get a count of the unique patients for each independently.  I just can't get it to work with the dependency.  The calculation below indicates it's valid, but gives me a processing error.  I've tried adjusting without success.

Is there a way to do this without having to use ETL?  It is a huge dataset, so, I'm hoping there is an alternative way.   Thanks so much!

( Case when left(`PROCEDURECODE`,5) in ('77427', '99201', '99202','99203', '99204', '99205', '99211', '99212', '99213', '99214', '99215','99241', '99242', '99243', '99244', '99245') then

                              COUNT(DISTINCT

                              case when left(`PROCEDURECODE`,5) = '0521F' then `PATIENTID`

                              end       

               )

end)