gnumed-devel
[Top][All Lists]
Advanced

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

Re: [Gnumed-devel] Possible failure of GNUmed to void duplicate values o


From: Busser, Jim
Subject: Re: [Gnumed-devel] Possible failure of GNUmed to void duplicate values on a key
Date: Thu, 23 May 2013 16:15:36 +0000

On 2013-05-23, at 2:06 AM, Karsten Hilbert <address@hidden> wrote:

> On Thu, May 23, 2013 at 06:09:34AM +0000, Jim Busser wrote:
> 
>>>> Here is my idea … in the new patient creation window, before proceeding to 
>>>> commit the inputted data and create a new patient,
>>>> 
>>>> - can the middleware check whether the lastname, first initial, and date 
>>>> of birth already exist?
> 
> Why "first *initial*" ?
> 
> Liz had something to say on this quite some time back but I
> cannot find that off-hand.

If there exists a match on

        Lastname
        Date of birth

then already this is becoming suspicious, inside a praxis, whether it could be 
the same person. I do accept the point that twins will have the same lastname 
and date of birth, and that in some countries few family names (in China, 
apparently about 100) are shared by a high proportion of the population. 
Therefore to warn every time you create a patient with the same lastname and 
DOB could be excessive.

However if the warnings are supprssed until there is an exact match between

        Lastname
        Date of birth
        First names

then we miss to warn of already-existing patients when we have variations of 
inputs such as

        Smith, Jonathan Peter
        Smith, Jonathan P
        Smith, Jon
        Smith, J Peter

which is sometimes how a patient gets created in advance, based on information 
sent from another praxis, when the patient is not available to confirm their 
formal names. There exists also a case where the patient had omitted to offer 
their first name because they do not like it, or at least do not use it, and so 
in the example above it could be possible that the patient identifies themself 
as

        Smith, Peter

but I did not want to overreach by asking that the query be as complex as 
testing for (and showing the user) existing patients who match

        Lastname
        Date of birth
        First character of user input matches
                first character of firstnames
                or
                first chracter of any (space-separated) "word" in firstnames

or

        an already-assigned external id

(which logic can be mostly re-used when matching imported results to patients).

-- Jim
                
        


reply via email to

[Prev in Thread] Current Thread [Next in Thread]