Dear Prof Ekins
Thank you for again raising these problems. 
Maybe I should regret that I responded to the initial remark that the 
CPA requested measurement uncertainty. But since this is a - or even 
the - accepted concept since 1993 I thought it might be worth while to 
point out some pertinent publications on the subject also to the UK 
colleages. Unfortunately, the Mailbase participants seem only 
moderately interested and ponders along with the total error concept 
as if it were new, when in fact it has been abandoned by most 
analytical chemists, clinical chemists and accreditation bodies using 
the international standard EN/ISO 15189.
I can assure you that the heterogenity of certain molecules has never 
been forgotten and I would give credit at this time to Dr Rudi Lequin 
who has been involved in writing many ISO standards and always 
reminded the working parties of the problem. I do not think it has 
been resolved in generic terms yet - or ever will be - but it is 
stated in all flowcharts and descriptions referred to in the field 
that the first thing to do is to identify the quantity. That may 
result in mentioning, in the name of the quantity, the calibrator, the 
antibody and the manufacturer and whatever specifications necessary. 
The IFCC/IUPAC nomenclature, as described by the IUPAC C-NPU 
(www.iupac.org), offers a complete description of the nomenclature and 
a list of some 25000 quantities that have been defined with unique 
descriptors. In reviewing this you should also consider the modern 
definition of bias, developed and changed since the previous edition 
of VIM (www.BIPM.org).
The imprecision, or uncertainty, profile is extremely important and 
should be part of the verification process. We usually advice a 
profile based on both standard uncertainty and relative uncertainty 
with the same arguments as providing difference plots (e.g. according 
to Bland-Altman) in both absolute and relative differences. To be 
feasible for the laboratories that perhaps do not afford, or want, to 
set up special experiments this prompts for user-friendly software 
that allows partitioning of results of measurements of patient samples 
and estimating the uncertainty in appropriate concentration intervals. 
The alternative, of course, is to make repeated measurements at 
intelligently chosen concentrations. I think the latter is preferred 
since the former may require assuming a homoscedactic uncertainty 
distribution in the particular partion.
With my best regarsds,
Anders Kallner       

------ACB discussion List Information--------
This is an open discussion list for the academic and clinical community working in clinical biochemistry.
Please note, archived messages are public and can be viewed via the internet. Views expressed are those of the individual and they are responsible for all message content.
ACB Web Site
http://www.acb.org.uk
Green Laboratories Work
http://www.laboratorymedicine.nhs.uk
List Archives
http://www.jiscmail.ac.uk/lists/ACB-CLIN-CHEM-GEN.html
List Instructions (How to leave etc.)
http://www.jiscmail.ac.uk/