|
Statins
Oseltamivir
Diabetes
Packaging
Product names
Alcohol metabolism
Decision support
Modernisation
President's message
Fellowships
Retention fee
Christmas miscellany
Letters to the Editor
|
Decision support
Choosing between a rock and a hard place
From Mr R. Glasspoole, MRPharmS, and Ms H. Judd, MRPharmS
Following the recent Broad
Spectrum, “Crying wolf” (PJ,
22 November 2003, p708), and the ensuing correspondence (PJ, 6 December
2003, p773) from a number of readers I thought it would be valuable to
give the perspective of a supplier of computerised drug interactions.
With clinical decision support you are damned if you do and damned if
you don’t. Decision support can be considered analogous to pharmacy
practice in that no matter how many positive interventions you make you
tend to be judged only by the errors that have been missed. What has
not been mentioned is that computerised interaction alerts have provided
valuable warnings to health care professionals in identifying “overlooked” interactions
of sufficient clinical importance to require alternative treatment.
First DataBank (FDB), the suppliers of the Multilex Drug Data File used
by a number of pharmacy system suppliers, is acutely aware of the need
to cut down “background noise” to reduce information overload.
This is one of our biggest challenges because no two health care practitioners
agree on the need for an alert — if it is known to them, it is “noise”;
if not, the alert is required. Therefore this challenge affects not only
suppliers of health care applications but also individual health care
professionals.
Fundamentally, drug knowledge base suppliers provide the core knowledge
and rules. Computerised drug interactions cannot be considered in isolation
without understanding a patient’s physical condition and the clinicians’ knowledge.
An interaction considered of minor clinical importance can still have
major therapeutic consequences, particularly in patient populations (consider
the implications of one fit on a previously stable epileptic patient).
An important interaction can still be overlooked by an experienced clinician
as we were recently reminded by a director of pharmacy investigating
a case where a consultant had overlooked a warfarin-azole interaction.
In order to resolve the issues with computerised drug interactions a
user profile is required where the user can individualise the alerts
that they want to see. We consider that it is impossible to produce centrally
a set of rules that meets “one size fits all” criteria. The
core rules provided by the decision support suppliers need to be able
to be customised at various levels including by the individual users
of the application. However there needs to be a full audit trail to ensure
that it is recognised that clinical decision support information has
been purposely filtered. One of the key requirements for “user
friendly” decision support is greater collaboration between the
providers of clinical systems, purchasers of such systems and the health
care professional.
This year FDB has been involved with three projects led by the National
Patient Safety Agency looking at different issues with decision support.
One of the projects involved the NPSA commissioning an academic research
project to identify the drug-drug combinations that should generate “critical
prescribing hazard alerts”, which cannot be easily overridden by
GPs using electronic prescribing systems in the UK. FDB supported this
project by supplying information on nearly 200 drug interaction combinations
that we consider to be the most clinically important for the panel to
review. One important aspect of our collaboration is that the results
from the study will help to specify future requirements for drug interaction
functionality. This will enable the knowledge gained
from the study being routinely available in the clinical environment.
Last month’s announcement of the first contracts being awarded
for the National (NHS) Programme for Information Technology (PJ, 13 December
2003, p802) heralds the start of a new era where the everyday use of
IT will be unavoidable and will probably have far greater impact on the
practice of pharmacy than anything else.
Decision support based on an individual’s full health record will
become routine and a standard tool for improving patient safety and care.
When the majority of the routine patient and clinical checks are automated
a valuable opportunity is created for pharmacists to engage patients
and physicians more readily in pharmaceutical care.
We would be pleased to hear from any reader with comments about decision
support since these are invaluable in helping to develop functionality
that meets clinical requirements.
Robin Glasspoole
Product Manager
(e-mail robin_glasspoole@firstdatabank.co.uk)
Hillary Judd
Head of Knowledge Base Services,
FirstDataBank Europe
First DataBank www.firstdatabank.co.uk |
* PDF files on PJ Online require Acrobat
Reader 4 or later.