The Mig paper is worth the read but....
I agree that the potential for progress for Version 3 exists but I am less
than convinced that the potential can yet be realised. The development work
and the live evaluations of Read V3, such as they are, have concentrated on
the ability or otherwise of capturing nuance of meaning in a coded form.
'Does a code exist for the following disease/finding/test/result etc?' is a
typical question. Technically Version 3 was designed to provide as a
vehicle to give this flexibility.
Read V3 allows a user to make a statement, say - 'Primary replacement left
knee joint' in several different ways. To provide a meaningful count of
'Primary replacement left knee joint' done, each of the possible ways
should be included in the analysis. To do this, the meaning held in the
various possible codes has to be available to the analysis engine - yet the
data to support this computation of equivalence is not properly available
and certainly not tested. This may sound complicated, but it is absolute
simplicity compared to a hierarchically based query, or the complexity
required to carry out hierarchical analyses using a different hierarchy
from the one supplied with the codes. Yet the supplied hierarchy is
supposedly primarily to assist in only finding a term/code.
No doubt the potential is there, but field trials that demonstrate Read V3
(and the software to go with it) is usable for information recall, have
still to be made. Who wants to feed the machine with codes if there is no
meaningful output?
I have a further point about the current 4 byte and Version 2 Read Codes. I
would guess that many of those contributing to this list (or most that
don't) have yet to obtain any real value from the Read Codes in their
systems. Projects like the CHDGP Project, Department of General Practice,
Queens Medical Centre, University of Nottingham are the currently the
critical need. This project attempts to educate and standardise each GPs
use of the existing code set. Meaningful analysis becomes possible. We
have a series of existing tools available, but they are not being used to
anything like their true potential.
Lets keep the baby in the bath until V3 is ready and not chase potential
that cannot yet deliver. As the MIG paper points out, probably the biggest
hurdle to the adoption of Read to date has been the licensing structure -
which brings us back round to probity, and the NAO report, and what does a
GP or a supplier get for the license charge?
Martin Strange
I confess that I am not a GP
-----Original Message-----
From: Ewan Davis [SMTP:[log in to unmask]]
Sent: 15 March 1998 09:28
To: [log in to unmask]
Subject: RE: NAO Report on Read
The MIG paper referred to below is well worth reading.
It is certainly the clear view of Primary Care systems suppliers that READ
3 is essential to the progress of clinical computing.
Much of the publicity on Read focuses on the limited use of Read 3 (which
is true) but fails to recognise the widespread use of earlier version of
Read in Primary care and the ongoing support and development that NHS CCC
has provided and provides for these earlier versions. Suppliers have been
encouraged by recent developments around version 3 both by NHS CCC and
independent developers who have created a range of Read 3 tools and engines
which should make the effective implementation of Read 3 relatively
straightforward for system suppliers.
It is vital that unrelated issues of probity highlighted by the NAO don't
distract from the value and importance of Read 3.
------------------------------------------------------------------------
-------------------------
Ewan Davis
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|