different publications is often contradictory.
Therefore, for a certain drug-laboratory test
combination, we have collected available
publications and tried tomake a summary ofthese.
Our conclusion iswhatwecould call ameta-analy–
sis of the available literature. During theworkwe
have realised that it would never be cost-effective
to carry out the work for a single hospital.
Therefore, the initial aim of theworkwas to carry
out the coding using the highest possible
professional standard and to make the product
available toother laboratories andhospitals as weil.
TheDLI code has turned out to be a useful too!
to code theknowledge.
It
is language-independent
and thus easy to localize in other language areas.
For coding drugs, we have used the ATC coding
scheme, which is the official international stan–
dard issuedby theWHO. Since theDLI knowledge
base has become aproduct of its own, it can easily
be included in any laboratory information system
or electronic patient record application in which
drugs and laboratory tests are stored in a structured
way. The DLI knowledge base is marketed by
Multimedica Ltd and a fully functioning demo
version can be downloaded from the company's
web site:
.
TECHNICALSOLUTION
To provide automatic warning to clinicians, we
have to !inkpatientmedication and laboratory test
data and analyze the information using the DLI
code.When the program detects that adrugwith a
clinically significant effect on laboratory test is
used, it displays a warning together with the
laboratory test. From the data, the format of the
warning can be automatically built, e.g., "Note:
The patient uses haloperidol that increases serum
prolactin level''.
Since the need for alarms maybe very different
for university hospital specialists and for general
practitioners, we soon realised that we had to add
the option of switching certain alarms off. Since
this is not itself a feature of the drug effect on a
laboratory test, we did notwant to include it in the
DLI code. The decisionwhen to produce an alarm
has to be based on the local needs of the hospital
or laboratory.
96
PROBLEMS INTHE PILOTVERSION
The experiences from our pilot version have been
mainlypositive.Clinicians regard the information
that will soon be automatically provided forthem
very useful.We havemet some problems that need
to be solved.
If the same laboratory test is repeated several
times in a short period of time, the same warning
is repeated each time. This may seem ridiculous
and make the laboratory form difficult to read.
Technically, it would be easy to show the warning
only the first time or once a week. However, in
this case the clinician is easily mislead to believe
that the drug that constituted the problem some
daysago has beendiscontinued and the latter result
is therefore reliable. In our opinion, the warning
has to be given every time even if the laboratory
test is repeated daily.
SOLUTIONS INTHE FUTURE
The results of our pilot system have encouraged
us to continue the work and extend the DLI
knowledge base to new drugs and laboratory tests.
TheDLI knowledge base is gaining popularity in
laboratory systems in Finland and abroad. Now
when the knowledge itself is codedin a structured
format, it is technically quite easy to include this
knowledge in new systems. The DLI knowledge
base is suitable for many departmental systems
outside clinical laboratories as weil, such as
intensive care unit systems, dialysis departments
etc.
Ideally, the warning should be provided before
the laboratory test is performed.When a laboratory
test request is written on a laboratory information
system, thecomputercould imrnediately check the
medication of the patient and warn if a drug may
affect the test. Then we would be able to reduce
unnecessary tests and costs.
The expansion of telecomrnunications and the
Internet inparticularwill make it possible to prov–
ide on-line services to clinical laboratories. We
could sendupdatedversions of theDLI knowledge
base to end-users regularly. Maintaining the last
available knowledge in the system will become
quite easy.
Klinisk Kemi
i
Norden3, 1998
