Question

我的查询是跟踪“患者日”或患者（csn_id）进入并离开同一天。这种情况以两种方式之一发生

以事件类型1
事件类型6的自动当天放电

在我添加Group By语句之前，一切看起来都很好。我已经阅读了收到相同“Msg 8120”的其他问题帖子，但我不确定我如何解决这个问题并对任何/所有建议持开放态度。

SELECT
 '1,*'+char(13)+char(10)
+'80,1006057'+char(13)+char(10)
+'100,10'+char(13)+char(10)
+'2405,'+cast(count(distinct adt.PAT_ENC_CSN_ID) as varchar(18))+char(13)+char(10) --Census events --as varchar(10)
+'2420,'+format(adt.EFFECTIVE_TIME,'M/d/yyyy') --as efftime
FROM CLARITY_ADT adt
inner join PAT_ENC_HSP hsp on adt.PAT_ENC_CSN_ID=hsp.PAT_ENC_CSN_ID
WHERE adt.EVENT_SUBTYPE_C <>2 AND --no canelled events
       datediff(day,adt.EFFECTIVE_TIME,cast(CURRENT_TIMESTAMP as date)) between 1 and 7 AND

  ((adt.EVENT_TYPE_C ='6' AND
    adt.DEPARTMENT_ID in ('101000020', '101000021', '101000013', '101000014', '101000002',
                         '101000018', '101000003', '101000114', '101000005', '101000017',
                         '101000006', '101000118', '101000007', '101000016', '101000004',
                         '101000009', '101000010', '101000011', '101000022', '101000023',
                         '101000012')) OR
  (cast(hsp.HOSP_ADMSN_TIME as date) = cast(hsp.HOSP_DISCH_TIME as date) AND
   adt.EVENT_TYPE_C ='1' AND
   adt.DEPARTMENT_ID in ('101000020', '101000021', '101000013', '101000014', '101000002',
                         '101000018', '101000003', '101000114', '101000005', '101000017',
                         '101000006', '101000118', '101000007', '101000016', '101000004',
                         '101000009', '101000010', '101000011', '101000022', '101000023',
                         '101000012')))

group by cast(adt.EFFECTIVE_TIME as date)
order by cast(adt.EFFECTIVE_TIME as date)

Answer 1

当您执行group by时，您唯一允许在查询结果中使用的是组本身或max或sum等聚合。< / p>

您按cast(adt.EFFECTIVE_TIME as date)进行分组，但您的查询包含format(adt.EFFECTIVE_TIME,'M/d/yyyy') - 虽然这些可能在逻辑上是等效的，但它们并不相同，因此SQL Server会告诉您你觉得它不好。

您可以将format(adt.EFFECTIVE_TIME,'M/d/yyyy')放入group by，或者将日期放入查询可能会更好，然后使用外部查询对其进行换行以格式化您的结果：

select
    '1,*'+char(13)+char(10)
    +'80,1006057'+char(13)+char(10)
    +'100,10'+char(13)+char(10)
    +'2405,'+cast(num_records as varchar(18))+char(13)+char(10) --Census events --as varchar(10)
    +'2420,'+format(the_date,'M/d/yyyy') --as efftime
from
(
    select
        cast(adt.EFFECTIVE_TIME as date) [the_date]
        ,count(distinct adt.PAT_ENC_CSN_ID) [num_records]
    FROM CLARITY_ADT adt
    WHERE ... /* where and join clauses */
    group by cast(adt.EFFECTIVE_TIME as date)
) x
order by x.the_date;

MS-SQL中的列无效错误

1 个答案: