>Hello list members,
>
>We are noticing increasing numbers of UK NEQAS results of "zero"
>(sometimes to three significant figures) being returned to us in
>various schemes.
COMMENT: Results depend on the analyte content of the samples being
distributed. Agreed that a competent scientist will not specify more
figures than are statistically significant.
>
>[1] Are labs reporting 'zero' on patients samples too?
COMMENT: why shouldn't they, if they obtain a zero result?
>
>[2] Do lab staff and end user clinicians understand that a result of
>'zero' is impossible and potentially misleading?
COMMENT: Interesting assertions! Why should a result of zero be uniquely
impossible? Moreover I would be in a state of panic if a value >0 were
reported on the concentrations of HIV or hepatitis B or C viral antigens in
my blood.
If you are suggesting that one should only report that the value falls
below the detection limit, then consistency requires that one should never
report any measured value (eg 5) as such; only that the value lies
between, say, 4 and 6.
It has always been my view that one should report the observed result
together with the SD (or some agreed multiple thereof). Moreover if the
detection limit is 1, there is absolutely nothing wrong in reporting a
value of zero or 0.5 (+/- 1) if one has obtained results of zero or 0.5.
Given this information, a reasonably educated clinician will conclude that
the analyte concentration MAY be zero or (slightly) greater.
One should NOT tamper with the result before reporting it.
>[3] Should analysers not 'trap' or 'flag' results well below
>detection limits as these could signal technical faults?
COMMENT: Another interesting proposition! Clearly you have never worked in
a blood bank!
If, on the other hand, one were suddenly to see an unusually large number
of zero (or low) values of serum T4, one would be justified in believing
the analyser was faulty. In other words one should be alerted by any
statistically significant departure from previous experience.
>I do not wish to debate assay sensitivity and detection limits again,
>but feel there is a training issue here in respect of the theory of
>measurement and what (clinical laboratory) measurements actually
>mean.
COMMENT: Agreed.
>I am no expert on this but someone out there is ...
>
>Regards to all
>
>Jonathan Middle
>
>
>Jonathan Middle PhD, Chairman UK NEQAS Executive,
>UK NEQAS, PO Box 3909, Birmingham B15 2UE, U.K.
>tel +44 (0)121 414 7300, fax +44 (0)121 414 1179
>[log in to unmask] http://www.ukneqas.org.uk
Regards
Roger Ekins
UCL
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|