[Top][All Lists]

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

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

From: Jim Busser
Subject: [Gnumed-devel] Re: Kirk's penicillin allergy
Date: Sun, 11 Jul 2010 10:32:59 -0700

On 2010-07-11, at 1:59 AM, Eric Maeker wrote:

> Please remember that Drugs can :
> - have a specific ATC (that is different to the ATC molecules)
> - have multiple molecules with different ATC
> That's important for the explanation.
> Actually, FreeDiams calculates allergies throught :
> - DrugUID -> one uid == one allergy
> - Drug ATC -> one ATC == N allergies (all drugs with the same ATC code)
> - Drug INN -> one INN == N allergies (all drugs with the specified INN)
> I have to add:
> - Molecule ATC -> one ATC == N allergies (all drugs that contains the
> molecule ATC - which is different to the drug's ATC)
> I've just finished code for (see screenshot):
> - ATC class -> one ATC class (less than 7 letters) == N allergies -->
> all drug's with ATC begining with the class, all drug's that contains
> a molecule ATC begining with the class.

These are good points.  FD is panning to populate a MOLECULE_2_ATC table. 
Therefore, when a drugs-xx.db contains drugs ATCs, more may be possible. It 
should give cause to rethink
- how in the EMR, to store allergy information
- how much to leave for the prescribing program

 Presently, GNUmed tries to store 3-level information

- the drug (and I believe that, with this, GNUmed can store the drug ATC)
- level of certainty (~ is definite?)
- can tolerate others in class (can only really be known to be true for 
individual drugs in a class when subsequently tried and tolerated).

At the time of deciding how widely a particular allergy should be "known as", 
for example in the case of penicillin G, whose ATC is J01CE02, it might be 
handy to supply the user with an ATC tree informing the 5-character level like

and so the EMR would store the equivalent of

        allergic to penicillin G
                (treat as more-widely allergic to J01CE)

and the infile to FD could include

        InnAllergies value=penicillin G
        ATCAllergies value="J01CE02; J01CE"
        DrugsUidAllergies value="uid of the brand of penicillin G"/>

However a few more scenarios remain to be worked out...

- when the adverse reaction occurred while taking more than one drug at a time
        --> here the clinician must decide on each drug, its likelihood i.e. 
whether to catch it in the "web of suspicion"
        --> here in case of blaming the innocent and missing the guilty
                it would be helpful to later in the EMR be able to review:
                "which drugs was this person taking on this date
                (or between dates X and Y)??"

- when the patient had a reaction while taking a multi-molecule drug
        (say, ramipril with hydrochlorothiazide)
        -->here the drug will have an ATC that represents a combination C09BA05
        --> let us suppose the user did no more work...

        FD could still know

        from the combined "drug" UID (if captured in the EMR)
                (no matter that the pharmacy gave a different brand)
                --> the FD tables can still link to each of the molecules

        from the combined entity ATC (but it is not clean):
                        C09BA05 identifies *which* ACE = ramipril
                        C09BA05 does *not* identify *which* diuretic 

        so, without a UID, the clinician/EMR may need to capture
        suspected allergy to each component

- btw the ATC system sometimes mixes, within any one level of ATC
        a drug class
        the same drug class in combination with some other class
        a different drug class

        thus flagging may be desirable at multiple (but not all) at an ATC 


- where any one molecule has more than one ATC (say, amoxicillin as part of an 
H Pylorii eradication regimen)
        my brain is tired, I'll think on this one

-- Jim

reply via email to

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