Sql 删除数据透视表的子查询

我有一个使用子查询的查询，但我想让它更高效、更容易工作。我一直在研究PIVOT逻辑，但似乎无法使其发挥作用。下面是原始的子查询逻辑，是否有人可以帮助我或为我指出正确的方向

SELECT 
    Parent.id
    ,[sitecode]
    ,[started]
    ,[completed]
    ,cast(( completed - started) as time)               
    ,[profileId]
    ,(SELECT TOP 1 CAST(RTRIM(LTRIM(REPLACE(REPLACE(CAST(Child.message as NVarchar(4000)) , 'Initialized', ''),'demographic records',''))) AS NVarchar(4000)) FROM [backend].[dbo].[Transaction_StatusMessage] AS Child WHERE Child.transactionId =Parent.Id  AND child.message like '%Initialized%' AND child.message like '%demographic records%'  ORDER BY Child.created DESC) AS Dem
    ,(SELECT TOP 1 CAST(RTRIM(LTRIM(REPLACE(REPLACE(CAST(Child.message as NVarchar(4000)) , 'Initialized', ''),'clinical records',''))) AS NVarchar(4000)) FROM [backend].[dbo].[Transaction_StatusMessage] AS Child WHERE child.message like '%Initialized%' AND child.message like '%clinical records%' AND Child.transactionId =Parent.Id ORDER BY Child.created DESC) AS Clincial        
    ,(SELECT TOP 1 CAST(RTRIM(LTRIM(REPLACE(CAST(Child.message as NVarchar(4000)),'Total valid patient types:',''))) AS NVarchar(4000)) FROM [backend].[dbo].[Transaction_StatusMessage] AS Child WHERE child.message like 'Total valid patient types:%' AND Child.transactionId =Parent.Id ORDER BY Child.created DESC) AS ValidPatientTypes
    ,(SELECT TOP 1 CAST(RTRIM(LTRIM(REPLACE(CAST(Child.message as NVarchar(4000)),'Total invalid patient types:',''))) AS NVarchar(4000)) FROM [backend].[dbo].[Transaction_StatusMessage] AS Child WHERE child.message like 'Total invalid patient types:%' AND Child.transactionId =Parent.Id ORDER BY Child.created DESC) AS NotValidPatientTypes
  FROM [Backend].[dbo].[Transaction] AS Parent
  where completed > '2018-11-17 00:00:00.00' AND transactionType = 'BulkImport' AND profileId = 122 
  order by sitecode, completed desc

需要帮忙吗，谢谢。

---

即使Pivot不起作用，有人能给我指出一个更好的逻辑吗

您可以将这些子查询转换为

外部应用

SELECT 
    Parent.id
    ,[sitecode]
    ,[started]
    ,[completed]
    ,cast(( completed - started) as time)               
    ,[profileId]
    ,Child.Dem
    ,Child.Clincial        
    ,Child.ValidPatientTypes
    ,Child.NotValidPatientTypes
FROM    [Backend].[dbo].[Transaction] AS Parent
    OUTER APPLY
    (
        SELECT  TOP 1 
            MAX (CASE WHEN Child.message like '%Initialized%' AND Child.message like '%demographic records%'
                  THEN CAST(RTRIM(LTRIM(REPLACE(REPLACE(CAST(Child.message as NVarchar(4000)) , 'Initialized', ''),'demographic records',''))) AS NVarchar(4000))
                  END) AS Dem,
            MAX (CASE WHEN Child.message like '%Initialized%' AND child.message like '%clinical records%'
                  THEN CAST(RTRIM(LTRIM(REPLACE(REPLACE(CAST(Child.message as NVarchar(4000)) , 'Initialized', ''),'clinical records',''))) AS NVarchar(4000))
                  END) AS Clinical,
            MAX (CASE WHEN Child.message like 'Total valid patient types:%'
                  THEN CAST(RTRIM(LTRIM(REPLACE(CAST(Child.message as NVarchar(4000)),'Total valid patient types:',''))) AS NVarchar(4000)) 
                  END) AS ValidPatientTypes,
            MAX (CASE WHEN Child.message like 'Total invalid patient types:%'
                  THEN CAST(RTRIM(LTRIM(REPLACE(CAST(Child.message as NVarchar(4000)),'Total invalid patient types:',''))) AS NVarchar(4000))
                  END) AS NotValidPatientTypes
        FROM    [backend].[dbo].[Transaction_StatusMessage] AS Child 
        WHERE   Child.transactionId = Parent.Id  
        ORDER BY Child.created DESC
    ) Child
WHERE completed       > '2018-11-17 00:00:00.00' 
AND   transactionType = 'BulkImport' 
AND   profileId       = 122 
ORDER BY sitecode, completed desc

---

谢谢松鼠，这很有帮助。我遇到一个问题，子查询返回null，我想从主结果中删除该行。。再次感谢