[COAS-List] Various Issues (ii) - Context

["Richard Dixon" <R.M.Dixon@comhealth.hull.ac.uk>]

(ii) Reports, Documents, Contexts, ...

There have been various discussions on this recently, many focused around
(if Ive got this correct), the Report as a form of observation, possibly an
XML blob, among others.

One of the things which, IMHO, the COAS model currently does not address,
but needs to is that of the context in which the Observations appear.  I
know this has cropped up in discussions previously, but there seems to be a
number of issues being raised that bring it back into the front.

One of the most fundamental notions from the GEHR work (which has been taken
on board in the EHCR-SupA work and is currently being considered for the CEN
model) is that of the Transaction.  (OK, maybe the term isnt brilliant,
but that doesnt really matter).

The Transaction is seen as of importance for a number of reasons, including:
It is a boundary for the information entered into the record at one time
under a clinicians authority
It is the minimum contextual unit of information that it is believed
sensible and safe to transfer

So, for example, a transaction might be a consultation, or a composed
report.

In GEHR, there were 7 transaction subclasses.  The EHCR-SupA model has
(IMHO) a better set of sub-classes for transaction.  Essentially, a
Transaction may be New Administrative data, New Clinical Data or a Reference
Transaction.  A Reference Transaction is one composed of references to
entries that already exist in the record (it is under debate as to whether
there may be any new entries in there as well, but my personal feeling at
the moment is that there shouldnt be, other than comments).

So a Report Transaction (a type of Reference Transaction) may be formed by
some clinician electing to create a summary report, which includes
references to existing entries.  This Report Transaction also becomes part
of the record.  A similar subtype of Transaction (currently known as Nota
Bene) represents references to those entries, chosen by a clinician, that
they believe are of vital importance (to be seen when first opening the
patients record  cf. The front cover of the physical notes).

Now, if I were using a COAS server to access data from a record that
included Transactions (or some similar context grouping), I would want the
reference to it to be maintained.

For example, if I wanted to get a list of High Blood Pressures via COAS, I
would want the information returned to include some reference to the
Transaction or context group in which it had been recorded, so that I was
aware that I may well be taking this out of context and that if I wanted to
go back and extract the whole transaction to get the fuller picture, I had a
means of so doing.

If COAS models an entity similar to the EHCR-SupA notion of Transaction, I
would see that Reports could well be a type (subclass or not) of these.  I
think the Nota Bene Transaction (or equivalent) would serve nicely to deal
with some of the questions raised regarding the grouping of Alerts (in the
sense of important information to note) and/or Allergies.

(If anyone wants more info on EHCR-SupA, I can point you at the public web
site (if its not live at the moment, it should be very shortly))

Now I realise that COAS is not just modelling entries from a record, but
also those data in existence about a patient which have not been explicitly
entered into a record.  In this case the context envelope may be just the
single observation, or perhaps a whole batch of observations issued by  a
machine  either way it is sensible that some responsible clinician (or team
or authority) makes the decision as to what is a safe context.  Just a
thought.