Gordon Challand pinpointed the problem with QC rules in modern analyzers,
acting "stat" instead of "batch".
The first problem, i.e. that tightening the limits cause a number of false
rejections, is classical and independant on the type of analyzer or if the
method is performed manually.
The solution of this problem is, as I mentioned previously, the use of double
limits: One absolute limit based entirely on the clinical demand and one
warning
limit based on Westgards rules. The use of absolute limits is an incentive for
improving the method. For e good method those limits are practiaclly never
exceeded, but a bad method will cause a lot of trouble with re-runs etc.
If not working satisfactory (i.e. due to drift) the warning limit will be
exceeded also for "the good" method, but then you can do a planned service
in due time before the method detoriates beyond the absolute limit.
The second problem is analyzers working "stat", i.e. samples are inserted
as soon as they arrive. Those analyzers are often placed near the patient
("bedside"), and they may be used by personnell not trained in laboratory
work. Furthermore the manufacturers often overestimate their performance
stating that "no laboratory experience is needed". This is obviosly not true.
There is a considerable risk that a "simple" analyzer operated by a non-
trained person will produce an unacceptable result without being trapped by
the QC system.
It is possible to introduce QC runs in such instruments, but then you have
to compromize with the economy: Running one or two controls for each sample
will almost certainly trap errors, but obviously this is an expensive way!
Maybe controls should be run automatically with fixed intervals, e.g. each
forth hour and, as an action by the operator, before "important" runs.
In any case, when evaluating a new instrument, its characteristic pattern of
possible errors (fliers, drift etc) should be monitored frequently during
its first use in the laboratory. In order to trap seasonal effects, this
period should cover at least 6 months and at least 300 control runs. Plotting
the values in a Shewhard plot will reveal the pattern of errors of the
instrument.
After the period of evaluation QC rules for the instrument can be set according
to the characteristics of the instrument. As for all methods, these rules
should include absolute as well as warning limits.
Also note:
Beside the QC rules for the instruments, there are also a need for QC rules
covering pre-analytical and post-analytical errors. Also a perfect instrument
will yield erroneous results if the sample is not handled appropriately
before analysis!
Best wishes
Mr Sten Öhman, PhD
-----------------------------------------------------
From: Mr Sten Öhman, PhD
Postal address: p.o. Box 133, S-590 70 Ljungsbro, Sweden
E-Mail address: [log in to unmask]
Phone: int: +46 13 219020, nat: 013-219020
Fax: int: +46 13 219021, nat: 013-219021
Personal home page: http://hem1.passagen.se/stoh7971/
Company home page: http://www.elfinilab.se/
------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|