> This forum is about open systems. Is there a successful
> argument that open systems will deliver, say a prescribing
> system, that follows a different (and better?) purpose then
> the other 20 or so commerical prescribing systems?
Well, the argument I would put forward, is as follows:
First of all, its worth saying at this point that most arguments relating to
open source have grown up in relation to discussions about how to develop
software devices. To fulfil our hopes for intelligent medical systems - like
prescribing support - its not primarily software devices that we're missing,
but sufficiently large, complex and interlinked knowledge bases. Resources
like SNOMED and UMLS are just the tip of the iceberg in this department.
UMLS may be vast, but you couldn't build a prescribing system using it and
nothing else.
The level of detail that must be captured about drugs and diseases, in order
to support a prescribing system that does not infuriate clinicians, is very
complex. The knowledge base alone to cover essentially all drugs, the
diseases they treat, interact with or cause, their mechanisms of action,
chemical substructure etc, is some 400,000 facts.
Additionally, whatever ontology/terminology you use to represent that
complexity has to be able to interface with whatever terminology is used in
the medical record you intend to link to. You have to be able to get from:
atenolol - CONTRAINDICATION - asthma
...to whatever code or codes represent asthma in your local record.
This makes the whole endeavour both very large, expensive to quality assure
and vulnerable to changes in the external schemes that must be referenced
(e.g. central mandate that an entirely new coding scheme must be used in the
EPR). A single centralised approach, especially a commercial one, is
unlikely to take the economic gamble. Additionally, medical end users have a
tendency to resent 'paying for words', and also believe that things which
appear to be public good should be free or at cost. Thus the market is only
willing to pay a modest price, and major investment can be hard to recoup.
But even assuming that some rich benefactor or state body did fund such a
project, one of the key open source arguments is that any centralised
development - however funded - develops cultural blind spots. The result
performs adequately, but never becomes the best it could be. The more
complex the project, the greater the risk that this happens.
So, the argument summarised is: better clinical systems require (much)
better clinical knowledge bases. These are expensive to build, hard to sell
for profit and so complex that a closed organisation or editorial process
will naturally settle for a compromise that is significantly less complex
than ideal, but can at least be managed.
Open sourcing such KBs spreads the cost, eliminates the question of profit
and removes the centralised accountability for possible failure in a very
complex project. This removes those barriers that otherwise frequently act
to prevent something becoming all that it could be.
Jeremy
