SQL Server 2008 - datediff month - sql-server

Question

I have the following query below --
SELECT *
FROM
dbo.patient AS p
INNER JOIN
dbo.study AS s ON s.patient_fk = p.pk
/***where p.pk = s.patient_fk***/
WHERE
s.study_custom1 LIKE'%hosp%'
AND s.study_datetime >= DATEADD(month, datediff(month, 0, getdate()) -1, 0)
AND s.study_datetime < DATEADD(month, datediff(month, -1, getdate()) -1, 1)
I want to find all dates for the most recent prior month. In this case it is the month of July, but the query is picking up '8/1/2014' along with July dates.
How do I need to modify this to exclude August and only find July?
Thanks

Answer 1

Try this
select *
from dbo.patient AS p
INNER JOIN dbo.study AS s
on s.patient_fk = p.pk
/***where p.pk = s.patient_fk***/
where s.study_custom1 like '%hosp%'
and DATEPART(year,s.study_datetime) = DATEPART(year,DATEADD(month,-1,GETDATE()))
and DATEPART(month,s.study_datetime) = DATEPART(month,DATEADD(month,-1,GETDATE()))
EDIT: Note: You could speed execution by setting variables instead of computing the month and year of today for every row, but as you had your original in a query without variables I provided a less computationally efficient solution for the sake of making sure you could use it in your application.
Also note that the arithmetic for getting the year and month of "last month" must be done inside the DATEPART so the query will function correctly in January.
EDIT 2: Explaining optimization
My first edit was written before I saw your comment, so I'm not sure if you're asking for an explanation of my original query, or asking for an explanation of my suggested optimization, but I'll guess the optimization, I'll be happy to further explain the query if that's what you meant.
So the weakness of my first solution is it makes 8 function calls per row. Not a big deal, but it could add up for a large database. If you instead define variables that you populate outside the query, you can reduce it to a pair of index scans per row, which is significantly faster. Significantly being maybe 1/10 of a second if you only have thousands of patients :-)
But regardless, here is a more efficient solution.
DECLARE #StartM DATE
DECLARE #EndM DATE
SET #StartM = DATEADD(month,-1,CONVERT(DATE,SUBSTRING(CONVERT(VARCHAR(50),GETDATE(),20),1,8)+'01'))
SET #EndM = DATEADD(month,1,#StartM)
select *
from dbo.patient AS p
INNER JOIN dbo.study AS s
on s.patient_fk = p.pk
/***where p.pk = s.patient_fk***/
where s.study_custom1 like '%hosp%'
and s.study_datetime >=#StartM
and s.study_datetime < #EndM
Note that #EndM will be the midnight of the first day of this month, so even if your s.study_datetime is 11:59 PM on the last day of last month it still gets included. If you build an index on s.study_datetime, this will be very fast.