E-drug: Electronic Essential Drugs Resources (cont'd)
-------------------------------------------------------------------------
I am the chief investigator of an Australian university research
group working closely with Therapeutic Guidelines Limited (TGL).
We are a health informatics research group, consisting of
individuals from wide backgrounds including public health,
medicine, pharmacy and computer science. The research has
government funding, with links to academic, health and industry
organizations.
A research project to optimise the production process and enable
decision support based on the guidelines was recently funded and
commenced in 2001. The motivation for the research is the
promotion of quality use of medicines.
Initial research efforts have included development of a data model
of the information in the guidelines, as well as the encoding of
conditions and drugs. A guiding principle for the research is that the
information being dealt with will be transferred to external,
third-party, systems. Thus, we are endeavouring to establish a
format that is as near-universal as is practical.
The model of the information we have will be unique (though it will
contain common concepts). The goal is to create a guideline
information format that has 'index hooks' into it. These hooks will
indicate the drugs and conditions that each guideline deals with. A
whole raft of functionality based on integration can be envisaged
from this. These could range from simple presentation of guidelines
in clinical systems at contextually appropriate moments, through to
active decision support that interacts with patient data, as well as
reference data.
The diversity of coding schemes and the lack of a comprehensive
drug and condition listings has created difficulties throughout this
research. This is the major stumbling block to easy access to
guideline information. While these difficulties can be solved and the
in-house information can be managed, it will be difficult to integrate
the information with external sources without common formats.
The Australian situation is changing and complex, so I only report
on our experience in the past 6 months:
We have commenced encoding each of the conditions in a list of
2000 guideline topics, using ICD-10 (Australian Modification), as
that is the scheme in use in Australian hospitals and that has been
sanctioned for use in General Practice (in the short term). We have
discovered some problems in this process, including the inability to
encode large proportions of the topics. As the guidelines are written
for clinical use, particularly community practice, this indicates
problems with the use of ICD-10. (SNOMED has been suggested as
an alternative, but that is a proprietary scheme and expertise in
Australia is low). So, coding of conditions is hard, but we expected
that.
In addition, we had planned to assign EAN codes to each of the
generic drugs in the guidelines. An independent coding project to
assign generic EAN codes to generic drugs in Australia is current
and we had planned to use the results of this coding. However, the
format of the generic assignment is not aligned to our requirements,
and so we are considering the use of ATC codes, which are then in
turn linked to EAN codes may be a more viable option. (The generic
EAN codes link to specific brand EAN codes, which are a part of
manufacturing supply chain logistics).
Are ATC codes an appropriate means of classifying generic
pharmaceutical substances? If so, establishing a link between an
ATC to a generic EAN then to brand specific EANs may be a
workable solution. It seems drug coding is also hard, and I
personally am surprised by this.
Apart from the technical coding problems, which could certainly be
solved for drug coding, a major difficulty is the maintenance of any
schemes and the assignment of codes that is developed. Such a
listing is essential for the integration of systems. Without it, each
integration is an isolated exercise.
The data model of guideline information we have developed is
coding scheme independent, but will include coding assignments. I
anticipate that multiple assignments will be required, to account for
values on different 'dimensions'. One drug coding scheme could be
created to serve as a comprehensive listing of drug name-drug code
that ensures uniqueness and authority. An ontology that uses the
code could then serve to describe the properties of the drugs.
Trying to combine these two functions into one scheme may not be
possible.
Bryn Lewis
Research Fellow
Dept. Computer Science and Computer Engineering
La Trobe University, Australia.
e-mail: brynlewis@yahoo.com.au
--
Send mail for the `E-Drug' conference to `e-drug@usa.healthnet.org'.
Information and archive http://www.healthnet.org/programs/edrug.html
Mail administrative requests to `majordomo@usa.healthnet.org'.
For additional assistance, send mail to: `owner-e-drug@usa.healthnet.org'.