But the only thing that would get a snomed is the diagnosis on the
summary line for me (id say Im representative of the general masses).
Does not every histology report have _some_ summary????   Its kind of
like not answering the question without a summary???? "unfinished business"

(any more than snomeding a summary of my time and im retiring now) -
would you want to snomed each patient consult for 5 variables even if it
was off a pick list?

The rest of your questions with margins being clear etc means you are
wading anyway even with a snomed code to start and will still require 2
neurones to fire to interpret context.  I can't see a future at all
where free text pathology reports routinely come with snomeded margin
status ...

If only you had those 2000 reports in electronic format.... ? Get your
pathology provider to just send everything you ever requested
electronically retrospectively ;)

I am _confident_ you could code a solution in about 20 mins that would
get around most of the freetext issues.  (eg restrict kw search to
summary line, check for not|no|etc issues, extract SCC|BCC|Solar K and
then add in a little rules based analysis of the rest of the message
based on the summary diagnosis - if SCC|BCC then find margin, parse line
for usual combination of margins comments  "clear of" "extends up to"
"involves", make a best guess based on that and present iffy ones for
validation by yourself)

I dont know how you solve the problem of site specific terminology
either eg nose|alar|columella|cheek|nasal bridge could all simply =
"nose"....Im a simple man...to me thats "skin, not otherwise specified"
T01000.

Then for audits you are scrambling to match clinical description of site
with whats actually on the jar and then even whats snomed'ed in the
summary.  All too often there are infact left nose, right nose and right
upper nose all in the one group.  Plus its all too common to see "L
nose" in the notes, "l alar" on the jar.  I refuse to start poring over
25 codes describing various names for nose bits and trying to marry
those up consistently.  (I figure in life you have a finite upper limit
on the number of decisions your brain can make in a day before it goes
to jelly, and I would rather save it for the diagnostic loop than
agonising over snomed codes which <0.5% of any population group looking
at the reports ever uses anyway)

now lets not even think about opening the can of worms discussing
synoptic reports (the ultimate picklist/proscribed format) completely
replacing free text...

I think you definitely lose the nuance of interpretation (which is the
not quite black and white art of histology reporting) with a synoptic
report (unless you pollute its purity with free text mods).

JD

> 
> Emphasis on the *not* - such text are common, and if there is no dedicated 
> "diagnosis" section in the text that's parseable, free text analysis hunting 
> just for key words out of context is problematic.
> 
> I'd love to have coded histopath - I am currently in the process of wading 
> manually through more than 2000 reports to review my own performance 
> regarding pre-op diagnostic accuracy, free margins, and wound infection rate 
> depending on site and size (well, actually my poor daughter does most of the 
> wading in order to resuscitate her ailing bank account)
> 
> Horst
> _______________________________________________
> Gpcg_talk mailing list
> [email protected]
> http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
> 
> 


-- 
=================================================
dr john dooley mbbs frcpa

aka "ron"

_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk

Reply via email to