You are on page 1of 9

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/226357758

An approach for more precise statements of


metrological traceability on reference material
certificates

Article in Accreditation and Quality Assurance · April 2010


DOI: 10.1007/s00769-010-0644-2

CITATIONS READS

16 153

3 authors:

Robert Koeber Thomas P J Linsinger


European Commission European Commission
16 PUBLICATIONS 377 CITATIONS 121 PUBLICATIONS 2,403 CITATIONS

SEE PROFILE SEE PROFILE

Hendrik Emons
European Commission
270 PUBLICATIONS 3,621 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

EC Recommendation for a definition of the term "nanomaterial" View project

Nanolyse View project

All content following this page was uploaded by Thomas P J Linsinger on 21 May 2014.

The user has requested enhancement of the downloaded file.


Accred Qual Assur (2010) 15:255–262
DOI 10.1007/s00769-010-0644-2

DISCUSSION FORUM

An approach for more precise statements of metrological


traceability on reference material certificates
R. Koeber • T. P. J. Linsinger • H. Emons

Received: 22 September 2009 / Accepted: 25 January 2010 / Published online: 17 February 2010
 Springer-Verlag 2010

Abstract Metrological traceability of measurement results for CRMs into a description of the identity of the measurand
is a key requirement for the comparability of such data in and the quantity value and provides a solution to overcome
time and space and with legal or product specifications. the conflict as to whether a measurement result should be
Similarly, traceability of the certified values of a certified declared as traceable to the SI or traceable to a method.
reference material (CRM) is a prerequisite for their useful-
ness. Consequently, a statement on metrological traceability Keywords Metrological traceability  Certified
of the certified value of the specified property is a require- reference materials  Quantity value  Identity 
ment of the ISO Guide series on Reference Materials. The Traceability statement  Measurand
establishment of the traceability of a value assigned to a
property of a reference material to a stated reference is piv-
otal to the use of the material. Metrological traceability is Introduction
related to a combination of four attributes that are assigned to
a clearly identified sample: description of the property, Metrological traceability and measurement uncertainty are
description of the quantity/kind of quantity, a number with its the two key concepts of metrology. The two concepts are
corresponding uncertainty and the measurement unit. The interlinked, as demonstrated by the definition of ‘‘metro-
first two items define the identity of the measurand, whereas logical traceability’’ as ‘‘property of a measurement result
the latter two describe the value that is assigned to this whereby the result can be related to a reference through a
measurand. Therefore, the concept of traceability rests on documented unbroken chain of calibrations, each contrib-
two anchor points, namely identity and quantity value. The uting to the measurement uncertainty’’ [1]. This definition
identity of a measurand can be defined by its structure alone clearly shows that only results with an appropriate uncer-
or by an operational/procedural definition, the quantity value tainty statement can be traceable.
of the measurand can refer to the SI, or to an artefact, if The concept of calibration directly leads to calibrants,
standards of arbitrary values have been used for the cali- which must fulfil all requirements of certified reference
bration of equipment (e.g. WHO standards; international materials (CRMs), namely to have property values charac-
units). The presented approach splits traceability statements terised by a metrologically valid procedure [2] and such
calibrants can then be used to characterise secondary cali-
brants, leading to a calibration hierarchy. Whenever
Presented at BERM-12, July 2009, Oxford, UK. possible, the property values of the CRMs shall be traceable
Papers published in this section do not necessarily reflect the opinion
to an external reference. Although it is actually possible to
of the Editors, the Editorial Board and the Publisher. have ‘‘primary standards’’ defining a unit, this approach
causes problems whenever a batch representing this primary
R. Koeber (&)  T. P. J. Linsinger  H. Emons standard is exhausted and needs to be replaced or whenever
Institute for Reference Materials and Measurements (IRMM),
stability of the current batch needs to be confirmed.
Joint Research Centre, European Commission,
Retieseweg 111, 2440 Geel, Belgium Usually, traceability is discussed in the context of
e-mail: robert.koeber@ec.europa.eu measurement results, with relevant documentary standards

123
256 Accred Qual Assur (2010) 15:255–262

[3] and guidelines [4] providing on how to establish materials for in vitro diagnostic devices [7] uses a hierar-
traceability for them. These documents emphasise the role chical concept, where a value is assigned to a primary
of calibration standards to achieve comparability of mea- calibrator using a primary reference procedure. Subsequent
surements. The fact that about half of the interlaboratory secondary calibrators are then produced, whose values are
comparisons amongst metrology institutes (Key Compari- assigned by comparison with the primary calibrator.
sons under the CIPM Mutual Recognition Arrangement) do
not give consistent data [5] shows how challenging this Pure materials
task is in practice. While in theory one single, traceable
measurement result would be sufficient to establish a ‘‘Primary reference materials’’ are often misleadingly
property value for a CRM, virtually all CRM producers feel understood to be substances with a very high purity.
that the significantly higher impact of CRMs when com- Assignment of traceable purity values is conceptually
pared with single measurement results warrants additional simple: all impurities need to be quantified and the purity is
efforts. This means that CRM producers try to find evi- obtained as 100% minus the sum of impurities. The
dence for the absence of bias rather than being complacent applicability of this approach has been demonstrated with
with the absence of evidence for a bias. Several approaches considerable efforts for some pure metals, where the
for material characterisation with consequences for the number of impurities is limited. For organic substances, the
traceability of the assigned values have been described for potential number of impurities is virtually infinite. The use
various types of CRMs. of various methods and knowledge of the synthetic path-
way has been proposed to avoid overlooking major
Materials characterised by reference methods impurities [8]. Still, the problem of different analytical
responses of the often unknown impurities exists. The ICH
This approach is conceptually simple. A reference method guidelines on impurities in drug substances, therefore,
exists that is internationally agreed and delivers results of require identification, synthesis and calibration of all
accepted quality. Measurements are performed using this impurities present above 0.1–1% (peak area percentage,
method and the traceability of the assigned value is the depending on the daily dose) [9]. Apart from these exam-
same as the one of the measurement results. In principle, ples, there might be cases where even traces of impurities
two situations exist: would have to be assessed as they could show adverse
Measurands of rational methods [4] for which the ref- effects (e.g. inhibition).
erence method simply delivers results of a higher quality
but is not the only method available: this situation is typ- Materials characterised by multiple methods
ically encountered in the case of metrologically high-order
(or sometimes ‘‘primary’’) methods of measurement. As The certification of matrix materials for measurands of
other methods usually have higher measurement uncer- rational methods (determining the total concentration of the
tainties, nothing is gained by combining these results with analyte in the test sample) poses a particular problem.
those from so-called primary methods (however, examples Strictly speaking, when a sample preparation step with a
have been reported where primary methods of measure- phase transfer is involved, this sample preparation step
ment give results with higher uncertainties than of routine breaks the chain of direct comparisons/calibrations.
instrumental methods [6]). Nevertheless, IRMM requires Because of the number of potentially unexpected effects
that also results from ‘‘primary methods’’ are confirmed by (rationalised as ‘‘matrix effects’’ or ‘‘interferences’’), it has
independent methods to rule out the possibility of gross in fact become common practice to require that certified
errors. Conceptually, this approach using ‘‘reference values are determined from a combination of at least two
methods’’ also applies to materials prepared by gravimetric independent methods [10, 11]. Although it is recognised
weighing. that this concept does not provide traceability in a rigorous
However, the concept of reference methods is of an even way, it has, for the time being, been put forward as state of
higher importance for method-defined measurands. In this the art [12].
case, the respective community (e.g. the International
Federation of Clinical Chemistry and Laboratory Medicine Statement of the reference
(IFCC), or AOAC International) has agreed on a harmo-
nised method for a certain measurand (e.g. enzymatic Apart from the problems on how to establish traceability,
activity). This measurement procedure defines the measu- discussions arose on how to phrase the traceability state-
rand and is hence at the top of the traceability chain. ment correctly. No disagreements existed for measurands
Consequently, the ISO standard on metrological trace- of rational methods, i.e. measurands which can be descri-
ability of results assigned to calibrators and control bed without reference to a specific method. Certified values

123
Accred Qual Assur (2010) 15:255–262 257

for these are either made traceable to the international is based on the traceability policy from ERM [15], but it is
system of units (SI), to other systems of units or to an explained from the perspective and with examples of
artefact. In the latter case, assigned values are only IRMM. When considering the range of approaches and
meaningful if they are themselves traceable to the (often opinions published in the literature so far, we are aware
arbitrarily chosen) quantity value assigned to this artefact. that there are different views from different angles on this
The discussion arose for method-defined properties, such issue. However, in our opinion, the approach presented
as impact toughness, dietary fibre or enzymatic activity. here can overcome some of the problems that are associ-
The results for these measurands are only meaningful in ated with traceability statements for CRMs.’’
connection with a specific measurement procedure, but the
results are often expressed in SI units (J, kg, mol, s, etc.).
Clearly, results are only comparable if the measurement Traceability of chemical measurement results
procedure is strictly followed and if all input factors are
properly calibrated. Some CRM producers, amongst them The establishment of metrological traceability is usually
IRMM, therefore, emphasised the measurement procedure straightforward for many physical measurements. For
by stating that the certified value is ‘‘traceable to method chemical measurements, the situation is usually more
XXX’’, whereas other CRM producers (e.g. BAM) complex and establishment of traceability statements is
emphasised the calibration of input factors by stating confronted with many obstacles. In the field of chemical
‘‘traceable to the SI’’. Arguments can be brought for both analysis, the nature of analytes can range from elements
approaches, not the least as the most recent edition of the over small organic molecules to highly complex biological
VIM explicitly foresees a measurement procedure as a macromolecules. Analytical methods are often challenged
valid reference [1]. This led to the deplorable fact that not to provide more information about specificity and have to
even the traceability statements for similar measurands are come up with information about the state in which the
always comparable. analyte is present, e.g. by speciation of inorganic analytes,
Despite the high importance which is now attributed to discrimination between different isomers of organic mol-
and the problems with the establishment of traceability, the ecules or isoforms of proteins.
issue has been touched only lightly in guidelines on CRM Typical matrices can range from aqueous samples at the
production, which were developed several years ago. The one end to highly complex environmental or biological
version of ISO Guide 34 from 2000 devoted only two samples, such as sediments, sewage sludge, whole blood or
paragraphs on traceability and included a more extended tissue samples. As a consequence, the number of potential
discussion on the subject in an informative annex. The interactions between the analyte and the matrix is huge.
situation improved with the new version of ISO Guide 34 This complex situation is not always evident to analysts
[13] giving a clear traceability requirements for the various when the metrological correct term measurand is used. The
studies (homogeneity, stability, characterisation) during situation becomes even more demanding when the type of
CRM production. ISO Guide 35 expands on the issue, but analytical method comes into play. In chemical measure-
does not establish harmonised, internationally agreed rules ments, various methods may exist that claim to cover the
and requirements for establishing traceability [2]. same type of matrices and analytes. However, it is often not
ILAC Guide 12 [14] is a little bit more specific by clear whether different analytical methods determine really
suggesting the use of other CRMs in the characterisation the same measurand, which can depend on the combination
exercise. This approach is different from the hierarchical of matrix, analyte, and method. For example, is the sample
traceability chain, as the other (matrix) CRMs generally preparation able to disrupt all interactions between the
only serve to demonstrate the traceability of measurement analyte molecule and the surrounding matrix or is only the
results, but is not used to calibrate the measurement results. easily removable analyte fraction extracted and a remain-
Obviously, despite the generally acknowledged impor- ing residue still strongly bound to the sample? Analytical
tance of the traceability of assigned values and the laboratories might not be too concerned by this issue and
difficulties encountered in demonstrating it, there are at comparability between laboratories could be achieved
present neither clear rules available on how to establish using the same (standard) method, but this point is actually
traceability in practice nor on how to report it. a crucial one for producers of reference materials.
In this paper, we propose an approach that can overcome CRM producers do not only provide a material, they
the conflict as to whether a measurement result should be also have to provide certificates that state measurement
declared as traceable to the SI or traceable to a method. In uncertainties and the traceability to a reference [16]. The
our opinion, this approach is able to give more precise target uncertainty and the traceability of the certified
traceability statements on certificates of CRMs, resulting in property value should be the subject of a proper planning of
an added value for the user of such a material. This concept a CRM project design [13]. In this context, it is not only in

123
258 Accred Qual Assur (2010) 15:255–262

the economic interest of the producer to develop a refer- Table 1 Example of certified values of a CRM (mass fraction of
ence material that is actually useful for a potential dietary fibre in dried apple; ERM-BC516) where the measurand and
customer. If the traceability of a CRM value is not the certified values depend on the method that was applied for the
determination
appropriate for a potential user, the material will not be
sold and used and hence the resources spent on such a Dietary fibre according to Certified Uncertainty
value (g/kg) (g/kg)
CRM would have been wasted and could have been better
used for other, more useful materials. There could be an AOAC 1990 985.29 [17] 164 4
even worse situation from the analyst’s point of view. If the Englyst (by GC) [18] 137 5
traceability statement of a CRM value would be incorrect, Uppsala 994.13 [19] 162 8
imprecise or misinterpreted, the use of the material could AOAC 1992 MES-TRIS 991.43 [20] 149 10
not be justified for a specific analytical problem and the Englyst (by colorimetry) [18] 134 5
conclusions that are drawn from the analysis of this CRM
could be wrong and lead to the laboratory astray. There-
standards are available whose units are chosen on an
fore, CRM producers should not only provide correct
arbitrary basis. Examples are the international units (IU) as
traceability statements, but formulate them also in such a
defined by the WHO for the measurement of vitamins,
way that the customer can judge if it fulfils the require-
hormones, vaccines and blood products. This concept is
ments for the purpose in question.
based on ‘biological equivalents’ where a conversion to the
Sometimes, there is a misconception that traceability
mass unit may sometimes be established.
can only be established to the SI. However, the ISO Guide
As already mentioned in the introduction, there can be
99 (VIM) explicitly defines two other possible anchor
cases for which there are different opinions on how to
points for linking traceability [1]. Apart from the preferred
express the traceability statement on CRM certificates.
link to the SI, traceability can be established using a
There is sometimes the perception that values obtained
measurement procedure (including the measurement unit
with ‘‘empirical’’ methods (i.e. for procedure-defined
for a non-ordinal quantity) or a measurement standard (i.e.
measurands) cannot be traceable to the SI. In the following
CRM) [1]. Although the use of the SI as the reference is
chapter, an approach is described that allows CRM pro-
commonly preferred, there are cases in chemical mea-
ducers to provide traceability statements on CRM
surements where this is not relevant.
certificates in a more concise manner.
For example, it might be metrologically ideal to express
the catalytic activity of an enzyme as a mass concentration
thus linking it directly to the SI unit ‘‘kilogram’’. In prac-
Splitting the traceability statement into two parts
tice, however, the mass of a protein has nothing to do with
its functional property, the biological activity. There is no
The certified (property) value is attributed to a quantity
proportionality between the mass of the protein and its
representing a property of the CRM [1]. Following the
primary, secondary and tertiary structure and the number of
terminology of ISO Guide 99, a ‘quantity’ is ‘‘a property of
relevant binding sites. The measurand in question needs a
a phenomenon/body/substance, to which a number can be
link to a specific method with specified parameters that
assigned with respect to a reference. This reference can be
serves as the reference here. This implies also that mea-
a measurement unit, a measurement procedure or a refer-
surement results which were obtained with different
ence material’’. Consequently, a quantity (e.g. amount-of-
methods can result in different quantity values for the
substance concentration of creatinium in a blood plasma
sample in question. If this is the case for certified values of
sample) would be the combination of the identification/
a CRM, the procedures that lead to the values in question
description of the property (creatinium) of a body/item
have to be specified. An example is given in Table 1.
(blood plasma) and the ‘base (or derived) quantity’
The CRM certificate specifies different certified values,
(amount-of-substance concentration). The IUPAC Provi-
all of them depending on the respective procedure that was
sional Recommendations on ‘‘Metrological Traceability of
used. The measurand ‘dietary fibre’, being the indigestible
Measurement Results in Chemistry’’ [22] describe this
portion of plant foods, depends on the definition by the
combination in the form of a sequence.
method. It is a sum parameter that comprises non-starch
polysaccharides, such as cellulose and many other plant System ) Component=analyte
ðblood plasmaÞ ðcreatiniumÞ
components, such as dextrins, inulin, lignin, waxes, chitins,
) Kind of quantity
pectins, beta-glucans and oligosaccharides [21]. The ðamount-of -substance concentrationÞ
measurand in this case covers a series of chemical structures.
The need to link traceability to an artefact is common in According to ISO Guide 99, the ‘measurement result’ is
fields such as clinical chemistry where measurement defined as ‘‘quantity value(s) being attributed to a

123
Accred Qual Assur (2010) 15:255–262 259

measurand’’ (with ‘measurand’ as the ‘‘quantity intended to no further specification of the measurand is needed. If the
be measured’’). A certified value on a CRM certificate identity of the measurand depends on the method, e.g. in
belongs to a specified quantity and is the combination of a case of the determination of lead in aqua regia extracts of
number (with its uncertainty) and the measurement unit. sediments according to ISO 11446 [23] where the sample is
What has to be kept in mind is that traceability chains consist not entirely digested, the identity of the measurand has to
of values, but how is it then possible to specify that a be specified by citing the method. Otherwise, the mea-
particular value is based on a method-dependent procedure? surement results could not be repeated or a certified value
We can reduce all the components that define a measurement of a reference material could be misleading if other
result to the identity of the measurand and the respective methods were used. Other examples can be found, for
quantity value. For many chemical measurements, the instance, in bioanalysis where it can be crucial to determine
identity of the measurand is defined via the analyte, an analyte in a specific binding state in pharmacology/
typically a specific atom, ion, molecule or parts of the toxicology. The measured in vivo concentration of a free
latter. If its identity is additionally influenced by analyte– form of a substance and its potential conjugate or immo-
matrix interactions which are not overcome by the bilised form (e.g. through protein binding) could depend on
measurement procedure, a method-dependent measurand is the method applied and the often abundant equilibria are
the consequence and such operationally defined identities or likely to be altered by the measurement procedures. Hence,
measurands can also be the result of unspecific measurement it is important to specify if the identity of the measurand is
procedures where groups of atoms or molecules or other method dependent or not.
entities cause the quantifiable signal. Exceptions exist in the There can be different reasons why a measurand is
case of functional properties (see ‘‘How to transfer this defined by a specific method. For example, a procedure can
concept into practice’’). Note that only the ‘‘qualitative part cover a group of analytes based on the composition and the
of the measurand’’, i.e. its identity is addressed here. The external interactions that these (often chemically similar)
quantitative part is covered and defined by the quantity value substances share together. As a result, the measurand could
which is described by the kind of quantity, a number and a represent a sum parameter covering a range of different
corresponding measurement unit. chemical entities. A different case is encountered when
The solution to the described problem of how to express functional properties are addressed. Here, the measurand is
the traceability for CRM values is achieved by splitting the not defined by an explicit chemical structure, but depends
traceability statement into two parts, namely the identity of on the functional activity exhibited by the sample and such
the measurand and the quantity value. By this, it is possible measurands are, for instance, relevant for food or clinical
to accurately describe if the identity of the measurand of a chemistry. One example is the previously mentioned case
reference material is defined by its structure alone or if it is of dietary fibre (Table 1). The measurand in question is the
defined by a specific procedure. This is in the spirit of the pool of all chemical entities in the sample that show the
ISO Guide 99 (VIM), where in the definition of CRMs in same behaviour under the analytical conditions applied
note 3 of clause 5.14 it is stated that ‘‘traceability covers (extraction, gravimetry). In clinical chemistry, for instance,
both ‘metrological traceability of a quantity value’ and enzymatic activity is a functional property that can com-
‘traceability of a nominal property value [1]’’. Further- prise a mixture of different (although similar) molecules
more, the traceability of the quantity value can be that form part of a ‘‘family’’ and may even be totally or
expressed by linking it to the relevant reference in ques- partially unknown [24]. The quantitative result depends on
tion, i.e. to the SI or to another reference. Through this the functional assay that is used for the measurement. In
concept, it is possible to allow that the certified properties general, the quantity values depend on all these cases on
of CRMs are method defined and that their values are made the experimental conditions under which the measurements
traceable to the SI at the same time. were performed.
The identity of the measurand can be structurally The traceability of the quantity value can be based on
defined or procedurally defined (i.e. by a method or func- the link to the SI or to an artefact (or ‘measurement stan-
tionality). If it is structurally defined, all the respective dard’ as defined in [1]) as laid down in ISO Guide 99. In
ions/atoms/molecules in the sample can be considered as both cases, it is important to establish an appropriate cal-
being chemically in an equivalent state (which does not ibration hierarchy using well-characterised calibrators.
mean that they are identical) and their determination does When traceability is established to an artefact, it has to be
not depend on the method applied for such a measurand. ensured that the whole calibration hierarchy is also based
For example, the determination of lead in sediment could on the artefact in question and such cases are common in
be based on a sample preparation that digests the sample the health sector. An example for using arbitrary units is
completely followed by a subsequent detection of the lead. the system based on IU as defined by the WHO Expert
Given that no losses occur, the analysis is quantitative and Committee on Biological Standardization (ECBS). The

123
260 Accred Qual Assur (2010) 15:255–262

precise definition of one IU differs from substance to Table 3 Examples for possible combinations of splitting the identity
substance and is established by international agreement for of the measurand from the quantity value for stating traceability (for
each substance. It is a unit of measurement for the reactive details, see text)
amount of a substance and is based on a measured bio- Identity of the measurand
logical activity or effect. Typical examples for which this
structurally operationally
unit is used are vitamins, hormones, some medications,
defined defined
vaccines, blood products, and similar biologically active
SI total cadmium dietary fibre
substances. The ECBS arbitrarily sets the number of IU [kg/kg] [kg/kg]
Quantity
contained in a preparation, and specifies a procedure to value artefact vitamin A coagulation time
compare other preparations of that substance to the refer- (traceable to) [I.U.] [INR]

ence preparation. The goal in setting the standard is that


different preparations with the same biological effect will
be attributed to the same number of IU. For some sub- gravimetric approaches, the measurement result depends
stances, the equivalent mass of 1 IU is later established. If on the definition of the method via the applied method. In
that happens, the former IU definition for that substance is the case of vitamin A, a structurally well defined molecule,
officially abandoned, in favour of a newly established references are often made to the IU although in this par-
mass. However, the unit count often remains in use, ticular case a link to the SI is also possible via conversion
because it is convenient for the scientific community. factors. However, despite this fact the use of IUs is still
Table 2 depicts some of the commonly used conversion widespread in clinical chemistry even though it is not
factors for some selected substances that are often referred recommended from a strict metrological point of view. For
to the IU. the coagulation time, an important parameter in clinical
The traceability statement that is split into the identity of chemistry, the available methods often differ in their ana-
the measurand and the quantity value can be expressed by lytical set-up, such as their reagents and standards, and
four possible combinations. In Table 3, an example for significant differences between routine methods were
each scenario is given. The case of total cadmium can be reported [25]. As a measurement unit, the international
regarded as the best-case scenario in which the measurand normalised ratio (INR) is used and is not expressed in SI
is independent from a method and traceability of the units, but is based on the coagulation time of a patient
quantity value can be easily established to the SI with the sample in comparison with a ‘healthy control sample pool’
quantity expressed, e.g. as a mass fraction. The example of [26].
dietary fibre was discussed in Table 1; although the
quantity value can be linked to the SI via spectrometric and
How to transfer this concept into practice
Table 2 International units, an example for a measurement standard
used as a reference point for metrological traceability
For the characterisation of CRMs, some specific points
Substance Conversion have to be accounted. The selectivity of the methods used
factor IU to lg
is crucial, particularly when structurally defined measu-
Vitamin A retinol 0.300 rands are concerned. It is important to use the same method
Vitamin B2 (riboflavin) 0.025 for operationally (procedurally) defined measurands; i.e.
Vitamin C 50 for homogeneity, stability and characterisation studies.
Vitamin E (D-alpha tocopherol) 667 Appropriate method validation that confirms the fitness for
Androsterone 100 purpose is mandatory in any case. As stated in ISO Guide
Insulin 45.45 35 [2], reference materials can be certified by measure-
Chlortetracycline 1.000 ments in a single laboratory (using so-called primary or
Doxycycline 1.149 reference methods) or involving multiple laboratories.
Penicillium G sodium 0.598
Excepting cases where a method-specific approach
Vancomycin 0.993
obtaining method-specific assessed property values using a
network of laboratories was chosen, the use of at least two
Only substances are given that can also be converted into mass independent methods is recommended for the certification
Note that the respective conversion factors change for every of matrix CRMs to prove the absence of a bias caused by
substance
matrix effects and/or interferences. Although this concept
There are many other substances where no conversion is possible and
where the amount-of-substance concentration (‘potency’) expressed
is implemented in the assignment of property values, it
in IU changes from material to material, depending on the biological should also be taken into consideration for making trace-
activity ability statements. In the case of matrix CRMs, statements

123
Accred Qual Assur (2010) 15:255–262 261

‘‘traceable to the SI’’ should only be made if at least two for the certified value stating the statistical parameters
independent methods were used for the characterisation (e.g. unweighted mean of…) and stating the metrological
[10, 11]. In the opinion of the IRMM, this principle is also traceability of these values, typically as ‘‘It is traceable to
applicable for major steps of an analytical method. Proto- the International System of Units (SI)’’. As for all certified
cols often contain separate parts such as sample preparation values of CRMs, respective uncertainties have to be
(extraction, clean-up), separation (e.g. by chromatography) provided that are explained in another footnote. For oper-
and detection (e.g. by a spectrometric technique). For each ationally/method-defined measurands no footnote is needed
of these steps, matrix interactions could interfere with the to specify the matrix or the kind of quantity as this is
process. If different laboratories or measurement proce- unambiguously stated in the heading. A footnote is required
dures in a characterisation study have common parts that stating the definition of the certified property, typically like
cannot be considered independent, the measurand cannot ‘‘as defined by the procedure according to ISO 12345’’ or:
be considered as entirely structurally defined; the respec- ‘‘as obtained by Soxhlet extraction and subsequent quanti-
tive common parts of the method must be considered as fication by GC–MS’’. The footnotes for the certified value
procedurally defined as only limited information on the and the uncertainty are the same as for the structurally
method independence is available. In Table 4, a certificate defined measurand mentioned above.
of a CRM is depicted to give an example on how this can In Table 4, an existing example of an ERM certificate
be translated into practice. from IRMM is given. In the characterisation of the feeding
The certified value as the mathematical product of a stuff material certified for the mass fraction of aflatoxins,
number and the measurement unit has to be derived via results from 7 to 8 different laboratories were accepted
properly calibrated measurement systems and it is this after technical scrutiny. As all methods employed used
calibration hierarchy that needs to be described. As this is reversed-phase liquid chromatography methods with post-
usually too much information to be inserted on the first page column bromination and fluorescence detection, method
of a CRM certificate (and possibly even on the certificate at independence cannot be guaranteed. As at least four dif-
all), this description of the traceability hierarchy may be ferent extraction solvents and techniques were used,
abbreviated, for example, in a footnote accompanying the independence to the extraction method is given. Therefore,
term ‘‘certified value’’ on page 1 of the certificate. Although the aflatoxin mass fractions as stated are defined by the
not mandatory for an ERM CRM, the certification report is employed reversed-phase liquid chromatography methods
the best place to deliver all information concerning cali- with post-column bromination and fluorescence detection.
bration hierarchies and metrological traceability. The This is reflected in footnote a in Table 4. The certified
following principles for the traceability statements on ERM values for the mass fractions of aflatoxins are traceable via
CRM certificates are applied: for ‘‘structurally defined’’ the common, certified calibrants used. The certified values
measurands quantified in total, footnotes are only required of the calibrants are traceable to the SI, as stated on the
respective certificate, due to the gravimetric preparation
Table 4 Example how to express metrological traceability applying
employed. Therefore, the mass fractions of aflatoxins in the
the proposed split into the identity of the measurand and the quantity CRM are traceable to the SI. This is stated in footnote b in
value Table 4. In any case, metrological traceability for all the
Mass fraction measurements to the stated reference must be established to
claim it for the certified value.
Certified valueb (lg/kg) Uncertaintyc (lg/kg)
What is now the important information for the user of
Compound feeding stuff such a CRM? The footnote of the certified value reveals the
Aflatoxin B1a 12.9 1.8 final reference of the traceability chain. The footnote of
Aflatoxin B2a 0.68 0.10 the certified property gives the information as to whether
Aflatoxin G1a 5.2 0.8 the CRM can be used regardless of the user’s method or if
a limitations apply. In the mentioned example, the customer
As obtained, after extraction, by reversed-phase chromatography
with post-column bromination and subsequent quantification by is informed that he is free to use the extraction method
fluorescence detection of choice. For the rest of the analysis protocol, high-
b
Unweighted mean value of the means of accepted sets of data and performance liquid chromatography in the reverse phase
each set being obtained in a different laboratory. The certified value mode with subsequent post-column derivatization by bro-
and its uncertainty are traceable to the international system of units mination and fluorescence detection must be used. If this
(SI)
c
condition is not fulfilled, the certified value (including the
The certified uncertainty is the expanded uncertainty estimated in
accordance with the guide to the expression of uncertainty in mea-
characteristics of traceability and measurement uncer-
surement (GUM) with a coverage factor k = 2, corresponding to a tainty) does not apply and the CRM cannot be taken for the
level of confidence of about 95% intended use.

123
262 Accred Qual Assur (2010) 15:255–262

Summary 8. King B (2000) Accred Qual Assur 5:429–436


9. Guideline Q3A (2006) ICH, Geneva
10. May W, Parris R, Beck C, Fassett J, Greenberg R, Guenther F,
The approach to split traceability statements of CRMs into Kramer G, Wise S, Gills T, Colbert J, Gettings R, MacDonald B
a description of the identity of the measurand and the (2000) NIST special publication 260-136. National Institute of
quantity value provides a solution to overcome the conflict Standards and Technology, Gaithersburg
as to whether a measurement result should be declared as 11. European Commission, DG Research (MT Program) (1994)
Guidelines for the production and certification of BCR reference
traceable to the SI or traceable to a method. It allows materials, 1st edn. Document BCR/48/93
providers of CRMs to better describe their reference 12. Pauwels J, Squirrell A (2000) Accred Qual Assur 5:2–67
materials and has thus an added value for (potential) users 13. ISO Guide 34 (2009) General requirements for the competence of
of these CRMs. reference material producers. ISO, Geneva
14. ILAC Guide 12 (2000) Guidelines for the requirements for the
competence of reference materials producers. ILAC: www.ilac.
Acknowledgments Ulrich Panne (BAM) and Steve Wood (LGC) org
and their co-workers are acknowledged for the fruitful discussions 15. Traceability policy of ERM. http://www.erm-crm.org
resulting in the new ERM traceability policy. 16. ISO Guide 31 (2000) Reference materials—contents of certifi-
cates and labels. ISO, Geneva
17. AOAC Official Methods of Analysis (1990) Total dietary fibre in
References foods. In: Enzymatic—gravimetric method, 15th edn. 985.29, pp
1105–1106
1. ISO/IEC Guide 99 (2007) International vocabulary of metrol- 18. Englyst HN, Quigley ME, Hudson GJ (1994) Analyst 119:1497–
ogy—basic and general concepts and associated terms (VIM). 1509
ISO, Geneva 19. Tyheander O, Åman P, Westerlund E, Andersson R, Pettersson D
2. ISO Guide 35 (2006) Reference materials—general and statistical (1995) J AOAC Int 78:1030–1044
principles for certification. ISO, Geneva 20. Lee SC, Prosky L, De Vries JW (1992) J AOAC Int 75:395–416
3. ISO 17025 (2005) General requirements for the competence of 21. AACC Report (2001) Cereal Foods World 46:112–126
testing and calibration laboratories. ISO, Geneva 22. http://www.old.iupac.org/reports/provisional/abstract09/debievre_
4. Eurachem, CITAC (2003) Traceability in chemical measurement prs.pdf
5. Cox M (2007) Metrologia 44:187–200 23. ISO 11446 (1995) Soil quality—extraction of trace elements
6. Wolff JC, Taylor PDP, De Bievre P (1996) Anal Chem 68:3231– soluble in aqua regia. ISO, Geneva
3237 24. Panteghini M (2007) Clin Biochem Rev 28:97–104
7. ISO 17511 (2003) In vitro diagnostic medical devices—mea- 25. Horsti J, Uppa H, Vilpo JA (2005) Clin Chem 51:553–560
surement of quantities in biological samples—metrological 26. International Committee for Standardisation in Haematology
traceability of values assigned to calibrators and control materi- (1985) ICSH/ICTH recommendations for reporting prothrombin
als. ISO, Geneva time in oral anticoagulant control. Thromb Haemost 53:155–156

123

View publication stats

You might also like