Question

I have this query

SELECT
patientid, 
practiceid,
visitcount
  FROM
    (       
        SELECT 
          patientid, 
          practiceid ,
          visitcount,
          RANK() OVER (PARTITION BY patientid ORDER BY visitcount DESC) as Rank
        FROM
            aco.patients_practices
    WHERE practiceid in (select id from aco.practices where parentaco = 30982) and isprimary = 0
    ) AS A
WHERE
    Rank = 1

Here are some results

patientid     practiceid       visitcount
157053           30976            6
158463           30974            2
187772           30973            15
187797           30971            1
187797           30975            1

Notice the last 2 patientid’s are the same and have the same visitcount hence the same rank. How can I omit these records with equal ranks completely from the output?

Thanks!

Answer 1

You can eliminate them by counting them and including them in the where clause. The following query counts them using logic similar to the rank — the number of times the patient has the same visitcount:

SELECT patientid, practiceid, visitcount
FROM (SELECT patientid, practiceid, visitcount,
             RANK() OVER (PARTITION BY patientid ORDER BY visitcount DESC) as Rank,
             COUNT(*) over (PARTITION by patientid, visitcount) as RankCount
      FROM aco.patients_practices
      WHERE practiceid in (select id from aco.practices where parentaco = 30982) and isprimary = 0
     ) A
WHERE Rank = 1 and RankCount = 1

I do notice that the practiceid is different in the last two records. It seems that you still want to eliminate both, though.