How to get all orders with maximum update_count value only? - sql-server

Question

I have laboratory orders system and these orders can be update more than one time for example
order_id test_id update_count
10 1 1
10 1 2
10 1 3
11 2 1
11 5 1
12 3 1
12 3 2
I want to select all orders without duplicates and select orders with maximum update count
I tried a lot and i checked WITH clients as and self join but always select statement returned all rows with all update counts not only maximum update count for each order_id
This is my Select statement:
SELECT LAB_RESULTS.ORDER_ID as 'Order Number'
,LAB_RESULTS.PATIENT_NO as 'Patient No'
,Patients.Patient_Name as 'Patient Name'
,Patients.Age as 'Patient Age'
,LabTests.TestName as 'Test Name'
,LAB_RESULTS.RESULT_NUMBER as 'Result'
,LAB_RESULTS.RESULT_REPORT as 'Text Result'
,LAB_RESULTS.APPROVED_DATE as 'Approved_Date'
,LAB_RESULTS.REQ_FORM_NO as 'Request Form Number'
,LAB_RESULTS.CUSTID as 'Customer Id'
,Machines.Machine_name as 'Machine Name'
,LAB_RESULTS.SAMPLE_ID as 'Sample Id'
,LAB_RESULTS.packageid as 'package id'
,LAB_RESULTS.GROUPID as 'group id'
,LAB_RESULTS.EXAMINED_BY as 'Examined By'
,LAB_RESULTS.EXAMINED_DATE as 'Examined Date'
,LAB_RESULTS.APPROVED_BY as 'Approved By'
,LAB_RESULTS.update_count
FROM LAB_RESULTS
inner join patients on LAB_RESULTS.patient_no = Patients.Patient_No
inner join labtests on LabTests.TestId = LAB_RESULTS.TESTID
inner join Machines on Machines.Machine_id = LAB_RESULTS.machine_id
where LAB_RESULTS.APPROVED_BY is not null
and LAB_RESULTS.SAMPLE_STATUS = 6
and LAB_RESULTS.update_count in (select max(update_count) from LAB_RESULTS where LAB_RESULTS.SAMPLE_STATUS = 6 and LAB_RESULTS.deptid = 2 )
and LAB_RESULTS.deptid = 2
I expect to get the following result :
order_id test_id update_count
10 1 3
11 2 1
11 5 1
12 3 2
then i added this condition but this return only the maximum update count only not group by each order_id only the maximum for all orders.
and LAB_RESULTS.update_count in (select max(update_count) from LAB_RESULTS where LAB_RESULTS.SAMPLE_STATUS = 6 and LAB_RESULTS.deptid = 2 )
How can I do this ?

Answer 1

then i added this condition but this return only the maximum update
count only not group by each order_id only the maximum for all orders.
The only reason your update_count in() didn't work is because you didn't correlate the subquery.
Instead of this:
and LAB_RESULTS.update_count in (
select max(update_count)
from LAB_RESULTS
where LAB_RESULTS.SAMPLE_STATUS = 6
and LAB_RESULTS.deptid = 2
)
You need this:
and LAB_RESULTS.update_count in (
select max(update_count)
from LAB_RESULTS l2
where l2.SAMPLE_STATUS = 6
and l2.deptid = 2
AND l2.order_id=LAB_RESULTS.order_id --correlate to outer query
)
Although I recommend also aliasing the table in the main query and using both aliases in the subquery.

Answer 2

Add this into your select statement:
,DENSE_RANK() OVER
(PARTITION BY LAB_RESULTS.ORDER_ID ORDER BY LAB_RESULTS.update_count DESC) AS rank
Then you will wrap that result in a WITH clause and select and filter doing something like this:
WITH base_query as (
the current query you have with the extra column I suggested before) Select order_id, test_id, update_count where rank=1

Answer 3

CREATE TABLE t1
(order_id INT, test_id INT, update_count INT)
INSERT INTO t1 VALUES
(10,1,1),
(10,1,2),
(10,1,3),
(11,2,1),
(11,5,1),
(12,3,1),
(12,3,2)
SELECT * FROM dbo.t1
SELECT order_id, test_id, MAX(update_count)
FROM dbo.t1
GROUP BY order_id, test_id
ORDER BY order_id, test_id
DROP TABLE dbo.t1