显示在SQL

Question

我想显示的30天之内

这里是我必须在录取表中的数据发生，患者全部入院内30天所有的再住院：

patient_id admid  adm_date     dis_date 
215   12   2011-05-17 14:43:00.000 2011-05-19 13:50:00.000 
421   13   2009-05-12 05:51:00.000 2009-05-13 18:35:00.000 
417   15   2014-02-12 16:44:00.000 2014-02-17 14:35:00.000 
417   17   2015-04-04 19:54:00.000 2015-04-05 15:00:00.000 
865   18   2009-09-28 05:40:00.000 2009-09-29 10:00:00.000 
992   19   2010-05-20 05:06:00.000 2010-05-25 17:51:00.000 
421   20   2009-05-29 07:15:00.000 2009-05-31 13:45:00.000 
421   17   2009-06-18 04:52:00.000 2009-06-19 10:55:00.000 
421   18   2009-06-30 13:39:00.000 2009-07-23 17:57:00.000 
326   21   2010-07-01 17:41:00.000 2010-07-01 17:41:00.000 
326   14   2010-06-07 11:35:00.000 2010-06-08 15:58:00.000 
215   16   2011-06-02 14:40:00.000 2011-06-03 13:25:00.000 

我试着此查询：

With adm as 
(select patient_id, admid, adm_date, dis_date,row_number() over(partition by patient_id order by adm_date) as r 
from admissions) 
select a1.patient_id, a1.admid, a2.admid, a1.adm_date, a1.dis_date 
from adm a1 inner join adm a2 on a1.patient_id=a2.patient_id 
where a1.admid <> a2.admid and a1.r+1=a2.r 
and a2.adm_date between a1.dis_date and dateadd(d,30, a1.dis_date) 
order by a1.patient_id 

我得到的结果为：

patient_id admid  adm_date     dis_date 
215   12   2011-05-17 14:43:00.000  2011-05-19 13:50:00.000 
421   13   2009-05-12 05:51:00.000  2009-05-13 18:35:00.000 
421   20   2009-05-29 07:15:00.000  2009-05-31 13:45:00.000 
421   17   2009-06-18 04:52:00.000  2009-06-19 10:55:00.000 
326   14   2010-06-07 11:35:00.000  2010-06-08 15:58:00.000 

但这不是我想要的结果，因为对于患者215有两个入院的时间间隔小于30，我想显示这个患者的两个入院，但是我的结果只有一个入院。

我想下面，而不是显示为上述结果：

patient_id admid  adm_date     dis_date 
215   12   2011-05-17 14:43:00.000  2011-05-19 13:50:00.000 
215   16   2011-06-02 14:40:00.000  2011-06-03 13:25:00.000 
421   13   2009-05-12 05:51:00.000  2009-05-13 18:35:00.000 
421   20   2009-05-29 07:15:00.000  2009-05-31 13:45:00.000 
421   17   2009-06-18 04:52:00.000  2009-06-19 10:55:00.000 
421   18   2009-06-30 13:39:00.000  2009-07-23 17:57:00.000 
326   14   2010-06-07 11:35:00.000  2010-06-08 15:58:00.000 
326   21   2010-07-01 17:41:00.000  2010-07-01 17:41:00.000 

请帮助！

Answer 1

我当然希望别人有比这更好的解决方案。但这是我的第一个传球。 CTE会在30天内找到所有首次入场的“下一次”入场。我使用row_number（）来确保“下一个”准入是下一个row_number。因此，任何具有“下一个”的事物，我们知道row_number将是当前行+1。联合CTE确保row_number值不同，并使用它们从原始CTE中进行选择。 就像我说的 - 我很想看到更好的答案。

;with cte 
as (
    select patient_id 
     , adm_date 
     , dis_date 
     , admid 
     , row_number() over (order by patient_id, adm_date) 'r' 
     from admissions 
    ) 
, cte2 
as (
     select adm1.r 
      , adm1.r+1 'nextR' 
     from cte adm1 
    left join cte adm2 
      on adm2.patient_id = adm1.patient_id 
      and adm2.r = adm1.r+1 
     where datediff(day,adm1.dis_date,adm2.adm_date) < 30 
    ) 
, CompleteSet 
as (
    select r 
     from cte2 
    union 
    select nextR 
     from cte2 
    ) 
select cte.Patient_id 
    , cte.admid 
    , cte.adm_date 
    , cte.dis_date 
    from CompleteSet 
    join cte 
    on cte.r = CompleteSet.r