Re: [Gnumed-devel] clinicians: coding use case survey - please respond !

[Karsten Hilbert]

On Wed, May 04, 2011 at 03:17:07PM -0300, Rogerio Luz wrote:

> > That's the sort of thing I asked for. Now, just to make
> > sure: This will force you to *again* pick codes each and
> > every single time you name an episode "Hypertension" -
> > regardless of whether you've picked the very same code
> > twenty-thousand times before ?
> >
> Yep ... you will pick that code 20 - 100 Thousand times, by the 4th time you
> will memorize it and it will be faster,

Got it. The second model (explicit linking) it shall be, apparently.

> but in NO WAY will the program be
> responsible for picking a code on your behalf.

Well, even in the implicit model GNUmed as such would never
guess anything. It is always the user - in this case the
user names an episode "Hypertension" - and that very fact
(the creation and naming of said episode) *also* means that
the user *implicitely* says "yes, this word 'hypertension'
is like all other words 'hypertension', you can show me the
same codes for it". Now, as soon as the user renames the
episode to "diabetes" this will immediately, automatically,
implicitely not be like any other "hypertension" anymore but
now rather like any other "diabetes" - simply by being
*named* "diabetes" - technically because "diabetes" matches
a term in a (user-modifiable) table which connects certain
terms (like "diabetes" or "hypertension") to certain codes.
No one needs to do anything. It simply happens by the user
using an episode description she's used before and for which
there are known codes.

> We can give you some hints,
> maybe, but it will be widely understood that it is the USER´S job to
> actually put the code together with the encounter.

There seems to be a strong reference for this behaviour so I
shall go with that model. Luckily I already implemented that
before I even asked. I only asked when it dawned on me that
- depending on user expectations - we might be able to
simplify the database schema.

BTW, I now also found a way to link codes to each of .RFE
and .AOE even though both are in the same clin.encounter
table :-)

Karsten
-- 
GPG key ID E4071346 @ gpg-keyserver.de
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346