Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: Handling pharmacy data with multiple entries per subject


From   "Doernberg, Sarah" <Sarah.Doernberg@ucsf.edu>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   RE: st: Handling pharmacy data with multiple entries per subject
Date   Fri, 10 Jun 2011 15:05:23 -0700

Hi all:

Thank you for your responses. I'm attempting to be more specific about my question below. My data is currently formatted as follows (with an example entry):

ID     drug_name     start_date      stop_date        outcome
1       ceftriaxone      5/15/2001     5/17/2001       5/31/2001
1       ceftriaxone      5/19/2001     5/20/2001       5/31/2001
1       ceftriaxone      5/20/2001    5/24/2001        5/31/2001
1       ceftriaxone      7/24/2001     7/27/2001       .
This one person had 3 prescriptions for ceftriaxone during a hospitalization in May, 2001, including one day where the person was not given this drug (5/15-5/17 and 5/19-5/24). There was another hospitalization in July, 2001, where another prescription was given. The patient did not experience the outcome during the second hospitalization. The dataset only contains information about two different drugs.

My ultimate goal is to figure out the number of days each person received each drug during a 30-day period from the first day of receipt or before the date of the outcome (if <30 days from the start of antibiotic) to allow for a logistic regression with exposure = antibiotic days. In addition, I may also do a survival analysis using start of antibiotic as the start date, development of the outcome as failure, and censoring at 30 days in those without the outcome. 

I hope that helps to clarify things. Thank you for your advice.

Best,

Sarah
________________________________________
From: owner-statalist@hsphsun2.harvard.edu [owner-statalist@hsphsun2.harvard.edu] On Behalf Of Phil Schumm [pschumm@uchicago.edu]
Sent: Friday, June 10, 2011 2:00 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: Handling pharmacy data with multiple entries per subject

On Jun 10, 2011, at 2:59 PM, Doernberg, Sarah wrote:
> I have a dataset from our pharmacy with prescriptions for
> antibiotics in hospitalized patients. Each time a patient was
> transferred (from the emergency department to the ward or the ward
> to the ICU, for instance), a new prescription (and thus, a new row)
> was generated. This is compounded by the fact that some people
> received intermittent dosing (each start date with it's own row).
>
> Because this is a very large set of data, I am trying to figure out
> how to have Stata combine the rows. Ideally, I would like to have
> one entry per person with consecutive courses of antibiotics
> represented by start and stop days (for example, someone who
> received an antibiotic from 6/1-6/3 and 6/7-6/9 would have
> start_date_1 = 6/1, stop_date_1=6/3 and start_date_2=6/7 and
> stop_date_2=6/9).
>
> I have tried doing this with the collapse command but the best I can
> do is to get total days on antibiotic in a given month. Converting
> from long to wide also is not ideal because consecutive courses are
> not combined due to the multiple prescriptions based on location.


You'll have to be a bit more specific here to get the help you're
asking for.  For example, why do you want

     start_date_1   stop_date_1   start_date_2   stop_date_2
     ------------   -----------   ------------   -----------
         6/1            6/3           6/7            6/9

I'm guessing (but I could be wrong) that your next step after this
will be to do some further calculations, which can probably be done
more easily with the data in the original, long format.  Also, if you
want help with the code to translate between what you have now and the
layout above, then you need to show the actual layout of the current
dataset.  Otherwise, people will just guess, and the whole exercise
becomes quite inefficient.

Don't be put off by this -- I do calculations like this all the time
in Stata, and it is very easy to do once you know how.  So chances
are, this is definitely worth persisting with.


-- Phil

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index