Re: st: matching within groups

[Jeph Herrin <junk@spandrel.net>]

Svend,

Thanks for the suggestion. The chart reviewers wanted it
laid out as I had here, so I was trying to work within
that format. But I may try to persuade them that it's
not worth the trouble.

cheers,
Jeph

Svend Juul wrote:
> Jeph wrote:
>  
> Though I see a long way to solve this, I'm hoping
> someone out there sees a more efficient solution.
>  
> My data consist of different reports of hospital
> admissions - one from the medical record (mr) and
> one from the patient (pat).  I have an admission
> date (admdt) & hospital code (hospcd) for each
> admission. Here -id- identifies patients, and I've
> added some separators manually for clarity:
>  
> id     mr_admdt   mr_hospcd    pt_admdt   pt_hospcd
> ---------------------------------------------------
> 1       1 Jan 07     35         6 Sep 06    35
> 1       6 Sep 06     35         1 Feb 07    36
> 1              .      .        23 Jun 06    35
> ---------------------------------------------------
> 2      11 Oct 07     34        21 Dec 06    34
> 2      21 Dec 06     34                .     .
> ---------------------------------------------------
> 3       1 Jan 07     33         1 Jan 07    33
> ---------------------------------------------------
> ...
>  
> The problem is to determine discrepancies. For instance,
> patient [3] has no problems - the two sources identify
> the same admissions (same date and place). However patient
> [1] has one match (both report 6 sep 06 at hosp #35) and
> then three discrepancies.
>  
> Ideally I would end up with:
>  
> id   mr_admdt  mr_hospcd  mr_dis pt_admdt pt_hospcd pat_dis
> -----------------------------------------------------------
> 1    1 Jan 07     35         1    6 Sep 06    35      0
> 1    6 Sep 06     35         0    1 Feb 07    36      1
> 1                            .   23 Jun 06    35      1
> -----------------------------------------------------------
> 2   11 Oct 07     34         1   21 Dec 06    34      0
> 2   21 Dec 06     34         0                        .
> -----------------------------------------------------------
> 3    1 Jan 07     33         0    1 Jan 07    33      0
> -----------------------------------------------------------
> ...
>  
> My solution would be to create two datasets and merge them
> on (hospdt hospcd), using _merge to identify which
> records don't match. However, I it gets messy trying to
> get the results back into the original file.
>  
> Is there any command or routine out there that will let
> me match up two variables (I can always group the date
> & code variables into one) without disturbing the original
> data? I don't mind sorting the admissions with patid
> (the current order is the arbitrary order in which they
> are reported).
>
> =========================================================
>  
> I don't quite understand what you want to end up with. I think
> I would want something like:
>  
>      +-------------------------------------------------------+
>      | id   hospcd      hospdt   patdata   hospdata   _merge |
>      |-------------------------------------------------------|
>   1. |  1       35   23jun2006         1          .        1 |
>   2. |  1       35   06sep2006         1          1        3 |
>   3. |  1       35   01jan2007         .          1        2 |
>      |-------------------------------------------------------|
>   4. |  2       34   21dec2006         1          1        3 |
>   5. |  2       34   11oct2007         .          1        2 |
>      |-------------------------------------------------------|
>   6. |  3       33   01jan2007         1          1        3 |
>      +-------------------------------------------------------+
>  
> I obtained it with this do-file:
>  
>       -----------------------------
>       clear
>       input id d m y hospcd
>       1 1 1 2007 35
>       1 6 9 2006 35
>       2 11 10 2007 34
>       2 21 12 2006 34
>       3 1 1 2007 33
>       end
>       gen hospdt = mdy(m,d,y)
>       format hospdt %td
>       drop d m y
>       generate hospdata=1
>       sort id hospdt hospcd
>       save hospdata.dta, replace
>  
>       clear
>       input id d m y hospcd
>       1 6 9 2006 35
>       1 2 2007 36
>       1 23 6 2006 35
>       2 21 12 2006 34
>       3 1 1 2007 33
>       end
>       gen hospdt = mdy(m,d,y)
>       format hospdt %td
>       drop d m y
>       sort id hospdt hospcd
>       generate patdata=1
>       save patdata.dta, replace
>  
>       merge id hospdt hospcd using hospdata.dta
>       sort id hospdt hospcd
>       list, sepby(id)
>       -----------------------------
>  
> Hope this helps
> Svend
>
> ________________________________________________________ 
>  
> Svend Juul
> Institut for Folkesundhed, Afdeling for Epidemiologi
> (Institute of Public Health, Department of Epidemiology)
> Vennelyst Boulevard 6 
> DK-8000 Aarhus C,  Denmark 
> Phone, work:   +45 8942 6090 
> Phone, home:   +45 8693 7796 
> Fax:           +45 8613 1580 
> E-mail:        sj@soci.au.dk 
> _________________________________________________________ 
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/