Best way to do nested case statement logic in SQL Server

Question

I'm writing an SQL Query, where a few of the columns returned need to be calculated depending on quite a lot of conditions.

I'm currently using nested case statements, but its getting messy. Is there a better (more organised and/or readable) way?

(I am using Microsoft SQL Server, 2005)

---

A simplified example:

SELECT
    col1,
    col2,
    col3,
    CASE
        WHEN condition 
        THEN
            CASE
                WHEN condition1 
                THEN
                    CASE 
                        WHEN condition2
                        THEN calculation1
                        ELSE calculation2
                    END
                ELSE
                    CASE 
                        WHEN condition2
                        THEN calculation3
                        ELSE calculation4
                    END
            END
        ELSE 
            CASE 
                WHEN condition1 
                THEN 
                    CASE
                        WHEN condition2 
                        THEN calculation5
                        ELSE calculation6
                    END
                ELSE
                    CASE
                        WHEN condition2 
                        THEN calculation7
                        ELSE calculation8
                    END
            END            
    END AS 'calculatedcol1',
    col4,
    col5 -- etc
FROM table

Enriquetaenriquez

Hi, did you use a tool (like SQLinForm) to indent the nested queries so nicely? – Bey

Appreciate you teaching me a nicer formatting method for nested CASE WHEN – Homoeo

There are certainly times when this would be efficient and helpful, but in general, I like to keep logic away from my SQL queries. Just a tip for posterity. – Handcar

Answer 1

You could try some sort of COALESCE trick, eg:

SELECT COALESCE(
  CASE WHEN condition1 THEN calculation1 ELSE NULL END,
  CASE WHEN condition2 THEN calculation2 ELSE NULL END,
  etc...
)

Applied

Nice, I just tested to ensure that it short-circuits, and was suprised to find that it does. So if condition1 checked for a divide by zero, it appears that it's safe to do it in condition2. Not sure if this is guaranteed. – Abe

One catch is that if one of your cases legitimately wants to return a NULL, it's no longer possible. – Applied

This is a great trick, but keep in mind that it may not perform as well as using CASE statements alone. This is "documented" in the Community Additions section here - msdn.microsoft.com/en-us/library/ms190349.aspx. My DBA just put the hammer down when I implemented this solution... – Karolkarola

When is this better than just a case statement (@beach's answer)? – Medico

How to rename a column in this answer.. e.g if I want to rename 'calculation1' to 'CAL1'... How is it possible syntactically ? – Intellectuality

While it is a nice trick, since it does not appear to short-circuit (ref: https://mcmap.net/q/119624/-coalesce-guaranteed-to-short-circuit), I'd think this would be far less efficient, running numerous CASE statements that will not be used. – Goose

this would return null when condition1 is not true... at least it did for me – Bouffant

Answer 2

Wrap all those cases into one.

---

SELECT
    col1,
    col2,
    col3,
    CASE
        WHEN condition1 THEN calculation1 
        WHEN condition2 THEN calculation2
        WHEN condition3 THEN calculation3
        WHEN condition4 THEN calculation4
        WHEN condition5 THEN calculation5
        ELSE NULL         
    END AS 'calculatedcol1',
    col4,
    col5 -- etc
FROM table

Endostosis

Answer 3

You can combine multiple conditions to avoid the situation:

CASE WHEN condition1 = true AND condition2 = true THEN calculation1 
     WHEN condition1 = true AND condition2 = false THEN calculation2
     ELSE 'what so ever' END,

Bookworm

Answer 4

I personally do it this way, keeping the embedded CASE expressions confined. I'd also put comments in to explain what is going on. If it is too complex, break it out into function.

SELECT
    col1,
    col2,
    col3,
    CASE WHEN condition THEN
      CASE WHEN condition1 THEN
        CASE WHEN condition2 THEN calculation1
        ELSE calculation2 END
      ELSE
        CASE WHEN condition2 THEN calculation3
        ELSE calculation4 END
      END
    ELSE CASE WHEN condition1 THEN 
      CASE WHEN condition2 THEN calculation5
      ELSE calculation6 END
    ELSE CASE WHEN condition2 THEN calculation7
         ELSE calculation8 END
    END AS 'calculatedcol1',
    col4,
    col5 -- etc
FROM table

Anesthetic

Answer 5

Here's a simple solution to the nested "Complex" case statment: --Nested Case Complex Expression

select  datediff(dd,Invdate,'2009/01/31')+1 as DaysOld, 
    case when datediff(dd,Invdate,'2009/01/31')+1 >150 then 6 else
        case when datediff(dd,Invdate,'2009/01/31')+1 >120 then 5 else 
            case when datediff(dd,Invdate,'2009/01/31')+1 >90 then 4 else 
                case when datediff(dd,Invdate,'2009/01/31')+1 >60 then 3 else 
                    case when datediff(dd,Invdate,'2009/01/31')+1 >30 then 2 else 
                        case when datediff(dd,Invdate,'2009/01/31')+1 >30 then 1 end 
                    end
                end
            end
        end
    end as Bucket
from rm20090131atb

Just make sure you have an end statement for every case statement

Astraddle

You may want to format your answer if you want it to be read by others. – Arum

how is this an improvement on the method used in the original question? – Plank

Answer 6

We can combine multiple conditions together to reduce the performance overhead.

Let there are three variables a b c on which we want to perform cases. We can do this as below:

CASE WHEN a = 1 AND b = 1 AND c = 1 THEN '1'
     WHEN a = 0 AND b = 0 AND c = 1 THEN '0'
ELSE '0' END,

Misfortune

Answer 7

a user-defined function may server better, at least to hide the logic - esp. if you need to do this in more than one query

Myriammyriameter

Answer 8

I went through this and found all the answers super cool, however wants to add to answer given by @deejers

    SELECT
    col1,
    col2,
    col3,
    CASE
        WHEN condition1 THEN calculation1 
        WHEN condition2 THEN calculation2
        WHEN condition3 THEN calculation3
        WHEN condition4 THEN calculation4
        WHEN condition5 THEN calculation5         
    END AS 'calculatedcol1',
    col4,
    col5 -- etc
FROM table

you can make ELSE optional as its not mandatory, it is very helpful in many scenarios.

Strohbehn

Answer 9

This example might help you, the picture shows how SQL case statement will look like when there are if and more than one inner if loops

Illogical

Please put the text of the code instead of screenshots because screenshots are hard to reuse or confirm. – Daradarach

Answer 10

Less complex format. Make sure keep the indentation as it is to not to be lost in ENDs.

SELECT 
CASE WHEN 1=1 THEN 
            CASE 
                WHEN 11=11 THEN CASE WHEN 111=111 THEN '1-11-111' END
                WHEN 12=12 THEN CASE WHEN 122=122 THEN '1-12-122' END
                WHEN 13=13 THEN CASE WHEN 133=133 THEN '1-13-133' END 
            END
        ELSE
            CASE WHEN 2=2 THEN 
                CASE 
                    WHEN 21=21 THEN CASE WHEN 211=211 THEN '2-21-211' END
                    WHEN 22=22 THEN CASE WHEN 222=222 THEN '2-22-222' END
                END
            ELSE
                CASE WHEN 3=3 THEN 
                    CASE 
                        WHEN 31=31 THEN CASE WHEN 311=311 THEN '3-31-311' END
                        WHEN 32=32 THEN CASE WHEN 322=322 THEN '3-32-322' END
                    END
                ELSE
                    CASE WHEN 4=4 THEN 
                        CASE 
                            WHEN 41=41 THEN CASE WHEN 411=411 THEN '4-41-411' END
                            WHEN 42=42 THEN CASE WHEN 412=412 THEN '4-42-412' END
                        END
                    ELSE
                        CASE 
                            WHEN 51=51 THEN CASE WHEN 511=511 THEN '51-511' END
                            WHEN 52=52 THEN CASE WHEN 512=512 THEN '52-512' END
                            WHEN 53=53 THEN CASE WHEN 513=513 THEN '53-513' END
                            WHEN 54=54 THEN CASE WHEN 514=514 THEN '54-514' END
                        END
                    END
                END
            END
        END
        AS Sku
FROM YourTable

Plage