![]()
28
| 2 | 2005
Klinisk Biokemi i Norden
The core laboratory of Helsinki University Central
Hospital, HUSLAB, has undergone a process of
extensive automation of the whole analytical
process in order to adapt to an ever-increasing
test volume. It became evident early in the auto-
mation process that the verification of laboratory
test results in the traditional way, by requiring each
result to be checked by a laboratory technician
before reporting it, would slow down the whole
process and would entail an increased risk of repor-
ting erroneous results.
Hence, we realised that an AV system would be
essential for us. Despite of the availability of com-
mercial AV systems we found that to satisfy our
requirements and to guarantee compatibility with
our LIS, the best alternative would be to develop our
own AV system in cooperation with the provider of
our LIS, Mylab Corporation, Finland.
The result of our joint project was the Modulab
AV system that is part of Multilab, that is our LIS
and also the most widely used LIS in Finland. The
Modulab AV system combines high-throughput and
reliability with the capability to use complex rules
freely defined by the user for each test and for each
analyzer. Since 2000 Modulab AV has allowed us to
routinely retain a high quality of test results while
achieving a shorter turnaround time (TAT).
Materials and Methods
In this context we define
manual verification
of
laboratory test results as verification done by a
person, usually a laboratory technician. In our labo-
ratory, manual verification involves reviewing the
results on a computer screen and comparing then
to experience-based limits and the previous results
of the same test for the same patient.
We define
autoverification
(also called autovali-
dation) as a process in which patient test results
are checked and reported by a computer program
without human intervention using a set of rules,
the current and other test results, internal quality
control results and analyzer flags.
The autoverification system Modulab AV is part
of our laboratory information system (Multilab)
and it has been developed as a cooperative project
between our laboratory and Mylab Corp. (Tampere,
Finland), the company that produces Multilab. The
system is highly customizable by the user, because
the rule sets that determine which results are
autoverified can be written by the user using a set
of simple modular rules most of which are param-
eterised and which can be combined in an almost
unlimited way to produce both simple and complex
rule sets. The autoverification system consists of
the autoverification program and of files where the
rule sets are defined, files where the intermediary
and final results are stored, and log files where the
entire autoverification process is documented.
When laboratory test results are handled by the
Modulab AV system, most results pass from the
analyzers to the clinicians who requested the tests
without any human intervention. The results that
are rejected by the autoverification system are
presented on screen to the laboratory technicians
who operate the analyzers and these results can
be handled by manual verification or rejection fol-
lowed by appropriate action, for example by check-
ing what is wrong with the process or the sample
and repeating the analysis when the problem has
been corrected.
In addition to the accepted and rejected result
groups some results can also be collected into sepa-
rate result groups that are handled in a special way,
for example results that need a written statement
of a human expert, or results of assays that were
done for non-clinical purposes such as development
projects.
In more detail, the autoverification process pro-
ceeds as follows: The interface program that con-
nects the analyzer to the laboratory information
system stores the analyzer’s results in an autoveri-
fication buffer file. The autoverification program
retrieves the results from the buffer and checks
them according to the rule set that has been defined
for the analyzer and for the assay.
The rule set consists of one or more chains of
individual rules. Table 1 shows examples of the
individual rules, or “rule modules”, currently avail-
able in the Modulab AV system and Table 2 shows
the rule sets we have defined for five common
clinical chemistry tests. The syntax of the rule
language is such that rule chains are enclosed in
square brackets and when a chain consists of more
than one rule, individual rules are separated by
semicolons. The arguments (parameters) of rules are
enclosed in parentheses and separated by commas.
(Fortsat fra side 26)
