Re: [Gnumed-devel] Re: Kirk's penicillin allergy

[Jim Busser]

On 2010-07-11, at 2:50 PM, Rogerio Luz Coelho wrote:

> I agree Jim, but this should not be the priority, priority should be to make 
> FD a frasewheel to populate GNUmed´s medication list ... after this I assume 
> it is all downhill


You are maybe thinking

        EMR user sets focus into medication field
        get ready to supply a GNUmed phrasewheel
                establish connection to FD
                pull down all of the FD "drug selector" values

but this is IMO not so easy because even if such functions (now missing) were 
added to FD, I am not sure it would work... in GNUmed you would still need to 
decide

        is what I am typing "brand (commercial)" or molecule?
        does this phrasewheel combine
                brand +- strength +- form +- route ?
        or, if these are done individually, does that then mean
                --> multiple "calls" back and forth?
        
How does it presently work for the German db?

As far as competing priorities, I agree that "intelligent" allergy handling is 
a future-item in terms of *implementation* but IMO may be good to think about 
it already. My big-picture question when discussing medical "intelligence" is 
how much to build into the EMR and how much to outsource to clinical decision 
support (CDS)?

When a patient reports an allergic reaction, the EMR may do as little as report 
the exact substance that was taken. How widely applicable to other substances 
becomes a matter of clinical judgement and expertise.

It would be nice for an EMR to support ~ "one optional level of extra clinical 
intelligence" while still letting outside resources help when a user may be 
unaware or incorrectly recall or judge a situation. So in the example of the 
penicillin allergy it may be good that

- an EMR, at the patient level, allows a judgement to extend an allergy more 
broadly (without a subsystem of expert knowledge inside GNUmed because this 
would be heavy lifting)
- the interoperating system (FreeDIams), when fed "dumb" allergy information, 
could still prompt the user... maybe FD can eventually support computational + 
table based prompts for drug-allergy and drug-disease interactions, or maybe 
this would be maintained in some separate open source expert system that both 
GNUmed and FreeDiams can access