How to select data where more than one column is populated

Question

I have a sample table, called [SAMPLES] which I've added multiple columns to the end for different types of lab analysis results, however for the majority of samples only one analysis has been completed. I want to be able select only the samples where more than one analysis has been completed on the same sample.

The table structure is something like:

For the above, the query should select only samples 'ABC122' and 'ABC123'.

Is there an easy way to select this data out?

• SQL 2012 database.

Answer 1

You can use cross apply and values to easily unpivot your columns back into the rows they should be, then count and filter accordingly

with x as (
    select distinct sampleId, SampleType, Count(*) over(partition by sampleid,sampletype) qty
    from samples s
    cross apply(values (analysis1),(analysis2),(analysis3),(analysis4),(analysis5),(analysis6))v(analysis)
    where analysis is not null
)
select s.*
from x join samples s on s.sampleid=x.sampleid and s.sampletype=x.sampletype
where x.qty>1

Answer 2

A better database design would be:

samples table
-------------
sampleID
sampleType


analyses table
--------------
sampleID
analysisType
analysisResult

Then you can query your desired result like this:

select sampleID
from analyses
group by sampleID
having count(analysisResult) >= 2

With proper indexes this will be super fast.

Answer 3

You can go for UNPIVOT to get the sampleids with analysis > 1. then use those sampleids for filtering.

--
DECLARE @sample table(sampleid char(30), sampletype char(10), analysis1 decimal(3,2),analysis2 decimal(3,2), analysis3 decimal(3,2), analysis4 decimal(3,2))

insert into @sample
values ('ABC121','ROCK',0.23,null,null,null),('ABC122','ROCK',0.27,null,null,0.23),('ABC123','ROCK',0.22,null,null,0.28),
('ABC124','ROCK',null,null,0.25,null)

;WITH CTE_Moresamples as
(
SELECT sampleid FROM @sample 
unpivot
(
val for keyv in ([analysis1],[analysis2],[analysis3],[analysis4])
) as upvt
group by sampleid
having count(val) > 1
)
SELECT * FROM @sample as s
inner join CTE_Moresamples as c
on c.sampleid = s.sampleid 

sampleid	sampletype	analysis1	analysis2	analysis3	analysis4
ABC122	ROCK	0.27	NULL	NULL	0.23
ABC123	ROCK	0.22	NULL	NULL	0.28