How to create a frequency report

  • Hi,

    I have two tables containing test results in categorical scale.

    Table one contains PCR results for some factors (P1, P2, ....). The assigned values are (0='Not tested', 1='Positive', 2='Negative')

    IF object_id('tempdb..#TestTb1') IS NOT NULL

    BEGIN

    DROP TABLE #TestTb1

    END

    CREATE TABLE #TestTb1 ([Sample_ID] VARCHAR(10), [P1] INT, [P2] INT, [P3] INT, [P4] INT)

    INSERT INTO #TestTb1

    SELECT 'Sample1', 1, 1, 0, 2 UNION ALL

    SELECT 'Sample2', 0, 0, 2, 1 UNION ALL

    SELECT 'Sample3', 1, 2, 0, 1 UNION ALL

    SELECT 'Sample4', 0, 1, 0, 1 UNION ALL

    SELECT 'Sample5', 1, 1, 1, 1 UNION ALL

    SELECT 'Sample6', 2, 1, 1, 2 UNION ALL

    SELECT 'Sample7', 1, 2, 1, 2 UNION ALL

    SELECT 'Sample8', 0, 2, 0, 1 UNION ALL

    SELECT 'Sample9', 1, 1, 0, 0 UNION ALL

    SELECT 'Sample10', 2, 2, 2, 1 UNION ALL

    SELECT 'Sample11', 1, 1, 0, 0 UNION ALL

    SELECT 'Sample12', 2, 1, 2, 0

    Table two includes antibiotic resistance test results for different antibiotics (AB1, AB2, ....). The values are (0='Not tested', 1='Sensitive', 2='Intermediate', 3='Resistant')

    IF object_id('tempdb..#TestTb2') IS NOT NULL

    BEGIN

    DROP TABLE #TestTb2

    END

    CREATE TABLE #TestTb2 ([Sample_ID] VARCHAR(10), [AB1] INT, [AB2] INT, [AB3] INT, [AB4] INT)

    INSERT INTO #TestTb2

    SELECT 'Sample1', 3, 1, 0, 2 UNION ALL

    SELECT 'Sample2', 2, 1, 3, 3 UNION ALL

    SELECT 'Sample3', 3, 3, 3, 3 UNION ALL

    SELECT 'Sample4', 0, 2, 0, 3 UNION ALL

    SELECT 'Sample5', 1, 3, 1, 2 UNION ALL

    SELECT 'Sample6', 3, 1, 2, 3 UNION ALL

    SELECT 'Sample7', 1, 1, 1, 2 UNION ALL

    SELECT 'Sample8', 3, 2, 0, 1 UNION ALL

    SELECT 'Sample9', 1, 1, 0, 3 UNION ALL

    SELECT 'Sample10', 2, 3, 2, 1 UNION ALL

    SELECT 'Sample11', 3, 1, 0, 2 UNION ALL

    SELECT 'Sample12', 2, 1, 3, 0

    Now, I need to build a report to show the frequency of positive (=1) PCR factors (P1,P2,...) for each antibiotic (AB1,AB2, ....) when the sample is resistant (=3) for the antibiotic.

    The result report supposes to be like this :

    Antibiotic Factor Freq.

    (=3) (Factor=1)

    ---------- ------- -------------

    AB1 P1 3

    AB1 P2 3

    AB1 P3 1

    AB1 P4 2

    AB2 P1 2

    AB2 P2 1

    AB2 P3 1

    AB2 P4 3

    AB3 P1 1

    AB3 P2 1

    AB3 P3 0

    AB3 P4 2

    AB4 P1 2

    AB4 P2 3

    AB4 P3 1

    AB4 P4 3

    As we have around 20 antibiotics, 20 PCR factors and thousands of samples, any suggestions will be a great help and will be appreciated.

    Thanks in advance.

  • Hello again,

    I managed to find a simple solution that seems working on sample tables.

    ;WITH Final_TB AS

    (

    SELECT X.SAMPLE_ID, X.AntiBio, #TestTb1.P1, #TestTb1.P2, #TestTb1.P3, #TestTb1.P4 FROM

    (

    SELECT [Sample_ID],'AB1' AS [AntiBio]

    FROM #TestTb2 WHERE [AB1] = '3'

    UNION ALL

    SELECT [Sample_ID],'AB2' AS [AntiBio]

    FROM #TestTb2 WHERE [AB2] = '3'

    UNION ALL

    SELECT [Sample_ID],'AB3' AS [AntiBio]

    FROM #TestTb2 WHERE [AB3] = '3'

    UNION ALL

    SELECT [Sample_ID],'AB4' AS [AntiBio]

    FROM #TestTb2 WHERE [AB4] = '3'

    ) X JOIN #TestTb1 ON X.[Sample_ID] = #TestTb1.[Sample_ID]

    )

    SELECT AntiBio, PCR, COUNT(1) AS [Freq.]

    FROM

    (

    SELECT [Sample_ID], [AntiBio], 'P1' AS [PCR]

    FROM Final_TB WHERE [P1] = '1'

    UNION ALL

    SELECT [Sample_ID], [AntiBio], 'P2' AS [PCR]

    FROM Final_TB WHERE [P2] = '1'

    UNION ALL

    SELECT [Sample_ID], [AntiBio], 'P3' AS [PCR]

    FROM Final_TB WHERE [P3] = '1'

    UNION ALL

    SELECT [Sample_ID], [AntiBio], 'P4' AS [PCR]

    FROM Final_TB WHERE [P4] = '1'

    ) Y

    GROUP BY Antibio, PCR ORDER BY Antibio, PCR

    It returns the expected table. However, as I mentioned with too many variables (the variable numbers are dynamic), I search for a more flexible solution.

    Any suggestions ?

    Thanks in advance.

  • Wrong design.

    CREATE TABLE Test(

    PatientID int,

    TestID int,

    Reading decimal(10,4)

    TestDate DateTime

    );

    Then the rest is trivial.

  • This design:

    CREATE TABLE #TestTb1 ([Sample_ID] VARCHAR(10), [P1] INT, [P2] INT, [P3] INT, [P4] INT)

    is going to bite you. I would strongly recommend not doing that. (I worked on even small databases in Ack!cess that were like this, and performance was abominable. If you create a union query, you may lose the benefits your indexing provides, so you'll get table scans every time you query anything. Performance killer.

    If you changed your structure to something like

    CREATE TABLE #TestTb1 ([Sample_ID] VARCHAR(10)

    , [TestID], [P1] INT)

Viewing 4 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic. Login to reply