![]()
26
| 2 | 2005
Klinisk Biokemi i Norden
Modulab AV - A User-Customizable, High-Throughput
Autoverification System for Clinical Laboratory Test Results
Janne Suvisaari, Solveig Linko, Aija Helin, Kari Pulkki, and Martti Syrjälä
HUSLAB, Helsinki University Central Hospital Laboratory, Helsinki, Finland
E-post: janne.suvisaari@hus.fi
Abstract
Manual verification of laborato-
ry test results is labour-intensive
and error-prone and some form of
computerized result autoverifica-
tion is generally considered neces-
sary to speed up the total process
and raise the quality of test results.
In this case study we describe our
autoverification software, Modulab AV developed
in cooperation with Mylab Corp., Finland, with an
emphasis on the user-customizable verification rules.
We also compare the times required to verify results
manually and by autoverification. Autoverification
has been used in the clinical chemistry, haematology
and coagulation analyzers attached to the automation
line of our core laboratory since 2000. Eighty-six per
cent of the results of these analyzers are currently
released without human intervention. The autoverified
results are reported faster than the manually verified
ones. The median verification times for autoverified
and manually verified results were 0 and 8 minutes,
respectively. As a consequence of the highly skewed
distribution of manual verification times, the diffe-
rence is much more pronounced for higher percentiles.
The 95th percentile for autoverified results was only 1
minute while it was 30 minutes for manually verified
results. Use of autoverification has contributed to the
high throughput we have attained, but this is due less
to faster verification times and more to the fact that
autoverification frees laboratory technicians to other
laboratory activities. A high degree of user-customiza-
bility is a very desirable quality for an autoverification
system, because it makes it easy to design and develop
verification rules by a stepwise empirical process wit-
hout intervention of the system provider.
Introduction
Considerable development in laboratory automation
and total quality management thinking has chal-
lenged clinical laboratories during the last decades.
Regardless of the extent of laboratory automation,
error sources in laboratory work are many and may
occur in any phase of the laboratory process (1).
Human errors are arguably the most common, but
all errors may be critical (2, 3). Today, much target-
oriented attention is continuously put on patient
safety and error prevention in different hospital set-
tings (4, 5, 6). As total patient safety is concerned,
these issues are unavoidably also related to labora-
tory organizations as the sources of laboratory test
results. For this reason, quality management tools
should be extensively used, and not only those rela-
ted to technical aspects. This approach was introdu-
ced in the expanded model of the laboratory process
(7) broadened from the brain-to-brain-loop of
Lundberg (8) in the context of autoverification (AV)
software. The primary objective of AV is to prevent
results with laboratory errors, or at least those with
gross blunders, from passing to the clinician.
The number of available AV tools varies for dif-
ferent laboratory disciplines and different analytical
phases. Reference values together with the LIS (labo-
ratory information system) have traditionally been
used to check the plausibility of a patient’s result.
However, relatively few commercially available AV
software systems have been available during the
last decades. In the eighties, the French ER-EMS
Company launched VALAB (9), a rule-based AV soft-
ware system with more than 20000 rules that runs
on a personal computer. The Dutch LabRespond AV
system is designed to function on total quality man-
agement verification levels, such as administrative,
technical, sample, patient, and clinical verification
(10). The Australian LabWizard expert developed by
Pacific Knowledge Systems generates result inter-
pretation together with a report to facilitate decision
making of clinical pathologists (11).
(Fortsætter side 28)
