Simple Query Help Needed

Question

I must not be in the right mode today, because this is giving
me a hard time...
I have a SQL Server database with 250,000 records. Each
record represents a medical visit.
Each record has 25 fields that contain patient diagnosis. A
visit may show one diagnosis, or it may show as many as 25. Any
specific diagnosis (250.00 for example) may appear in a field
called Code1, or a field called Code2.... Code25.
There are a total of about 500 patients that make up the
250,000 visits.
What I need is a report that shows each DISTINCT diagnosis,
and a listing (and count) of people that have been assigned that
diagnosis.
My thought is to first query the table to arrive at the
unique list of diagnosis codes. Then, in a nested loop, query the
database to see who has been assigned that code, and how many
times.
This strikes me as an incredibly database intensive query,
however.
What is teh correct way to do this?

Answer 1

The correct way to do this is to normalize your database.
Rather than having 25 separate colums that essentially
contain the same data (different codes), break that out into a
separate table.
Patient
patientid
Visit
patientid (foreign key referencing Patient)
visitid
Diagnosis
visitid (foreign key referencing visitid in Visit)
diagnosiscode
order (if the 1-25 ordering of diagnosis codes is important)
Once you correct your datamodel, what you're trying to do
becomes simple.
-- get all diagnosis codes
select distinct(diagnosiscode) from diagnosis
-- get a breakdown of diagnosis codes, patients, and counts
for that diagnosis
select diagnosiscode, patient.name, count(diagnosiscode) from
patient left join visit on (patient.patientid =
visit.patientid)
left join diagnosis on (visit.visitid = diagnosis.visitid)
group by diagnosiscode, patient.name