Re: [Gnumed-devel] EMR tree display of allergy

[James Busser]

On 25-Sep-08, at 3:07 AM, Karsten Hilbert wrote:

> I have put down an analysis here:
>
>    http://wiki.gnumed.de/bin/view/Gnumed/AllergyAndIntoleranceStatus
>
> Comments and improvements welcome.

I continue to think that the current focus on the schema misses the  
point that we still lack agreement on proper and/or desirable  
*clinical* usage of allergy information.

I understand

        allergy state = NULL (unasked)

        allergy state = 0 (asked and convincingly "none")

however any other state is simply "other than none"

Clin.allergy rows already provide for a level of uncertainty  
( "definite" or not) so I do not see the value of capturing and  
maintaining some separate level of uncertainty in "allergy  
state" (nor the value, in allergy.state, of a comment that then needs  
to be separately maintained).

Therefore I ask again, if a clinician would enter any comment in a  
clin_allergy row (whether this would be that the patient remembers  
they did have some allergy or intolerance -- but can't recall the  
details -- or whether they are being evasive, or otherwise do not  
wish to fully say) ... why does such an entry not suffice?

It seems to me that once the question of allergy has been asked, the  
state should be entirely determined by the presence or absence of  
items in clin.allergy...

- if one or more items have been marked definite, then the allergy  
state is definitely allergic to at least something. You could even  
have a patient who cannot recall what it was that they had received,  
but that they nearly died and were told they should never have it  
again. They would still be properly regarded "definitely allergic to  
something" but how would you even use this information?

- if one or more items have been entered in clin.allergy but *none*  
have been marked definite, then the allergic state is "possible"

- if after having asked the question, there is nothing -- not even  
hesitation -- to put into the clin.allergy table, then the allergy  
state is "none" or more correctly "none that we presently know and,  
indeed, none that we have any question about"

This is the reason why the only intrinsic value in allergy.state is  
to track whether the question had ever been asked, and the rest  
(possible or definite) should simply be derived from what exists  
among the clin.allergy entries.

I would go further. Richard already raised the question of what value  
it serves to have asked it before, given that it should be re-asked  
prior to every new or repeat exposure to which we would be subjecting  
a patient. I have to concede that despite the temptation to ask it  
"at least once", it does little good to have asked it *only* once.  
Even when an entry of "no allergy" would have been true at the time  
of entry, supposing a patient would have newly developed an allergy  
in response to care from inside *or* outside the praxis, does it  
serve any good to have asked this before when it is not asked the  
next time a doctor in the praxis prescribes something? This is why it  
would be much more to the point that a reminder of whatever is the  
current allergy information be contained in a prompt to the clinician  
who would prescribe something to which the patient may be allergic.  
Otherwise we are talking about a once-only entry of information for  
whatever value it has to be recorded in *advance* of when it will be  
needed, even though the information could in the meantime have easily  
changed.