Beruflich Dokumente
Kultur Dokumente
The present overview of validation and verification procedures in clinical chemistry focuses on the use of harmonized
concepts and nomenclature, fitness-for-purpose evaluations and procedures for minimizing overall measurement
and diagnostic uncertainty. The need for mutually accepted validation procedures in all fields of bioanalysis becomes
obvious when they implement international accreditation and certification standards or their equivalents. The guide
on bioanalytical method validation published by the US FDA in 2001 represents a sensible compromise between
thoroughness and cost–effectiveness. Lacking comprehensive international agreements in the field, this document
has also been successfully adapted in other fields of bioanalysis. European and international efforts aiming for
consensus in the entire field of bioanalysis are currently being made. Manufacturers of highly automated in vitro
diagnostic methods provide the majority of measurement methods used in unmodified in clinical chemistry. Validated
by the manufacturers for their intended use and fitness-for-purpose, they need to be verified in the circumstances
of the end-users. As yet, there is unfortunately no general agreement on the extent of the verification
procedures needed.
Validation and verification of measurement The pharmaceutical industry has been and is Elvar Theodorsson
methods are procedures that aim to establish still a driving force in the development of valida- Clinical Chemistry, Department of
realistic expectations with the analyst and con- tion practices given the regulatory environment Clinical & Experimental Medicine,
fidence with the end-user that the methods are they have been subject to early on. Clinical labo- Faculty of Health Sciences,
Linköping University, County
fit for their intended purposes. Different fields ratories are increasingly being accredited or certi- Council of Östergötland, Linköping,
of bioanalysis have historically lacked a common fied according to ISO 17025, ISO 15189 or other Sweden
Author for correspondence:
theoretical and practical ground due not only similar quality systems. These laboratories are Tel.: +46 1328 6720
to differences in the tasks at hand, but also to therefore in need of generally accepted and cost- Fax: +46 1010 33240
differences in terminology and in calibration, effective protocols for validation. Theoretically, E-mail: elvar.theodorsson@liu.se
validation and quality control practices. Recent there are no limits to the extent of validation and
harmonization efforts in these areas [1,101,102] verification procedures. However, in practice,
confirm that all fields of bioanalysis can share there are time and economic constraints. It is
the same principles and nomenclature cater- therefore crucial that validation and verification
ing for extensive harmonization of guidelines, efforts are optimized in order to maximize the
standards and practices. value gained for the resources spent.
In the early 1990s, the US FDA initiated This brief overview of validation and veri-
and supported conferences and harmonization fication methodologies in clinical chemistry
work on bioanalytical method validation [2,3] attempts to adhere to the currently accepted
that, in 2001, resulted in the ‘FDA Guidance guidelines in terminology and bioanalytical
for Industry – Bioanalytical Method Validation’ validation methodologies. The probable over-
guidelines [4,101] . They have been widely used, emphasis on certain aspects, for example, on
being suitable not only for the needs of the verification procedures and fitness-for-purpose
pharmaceutical industry but also for bioana- investigations, may be explained by the authors’
lytical methods in general [4] . In fact, lacking background in laboratory medicine and basic
similar international guidelines, this FDA docu- research. The current already extensive and
ment is widely used as standard reference for increasing use of commercially available meas-
validation of bioanalytical measurement meth- urement instruments and methods underscores
ods. European efforts in the field of validation the need for agreement on reasonable, but suf-
(European Medicines Agency’s Guidelines on ficient, methods for end-user verification of the
Validation of Bioanalytical Methods) [5] are manufacturer’s performance claims.
currently in progress.
10.4155/BIO.11.311 © Elvar Theodorsson Bioanalysis (2012) 4(3), 305–320 ISSN 1757-6180 305
R eview | Theodorsson
Figure 1. Qualitative and quantitative expressions of the type of error and their
combination. Since bias should be eliminated whenever possible, a dotted line is drawn between it
and measurement uncertainty.
Adapted with permission from [7] © Springer Science+Business Media.
Reaction cell
Sample
Calibration
Reagent Matrix effects
Measurement uncertainty
Laboratory Operator
components can be measured and/or estimated or variance component analysis of the data. The
and added according to general rules for error top-down approach can identify components
calculation [105] . The uncertainty concept pub- of variation caused by different laboratories in
lished in 1993 as the ‘Guide to the expression of the same organisation, different measurement
uncertainty in measurement’ (GUM) [105] was instruments, different chemical principles in
the result of many years of dispute and delibera- measurement, different reagent lots, and so forth.
tions under the auspices of several international
organizations and replaces the former ‘error’ Validation or verification?
concept. According to International Vocabulary of
The components of variation are preferably Metrology 3 (VIM3), verification is ‘provision
depicted as fishbone/cause and effect/Ischikawa of objective evidence that a given item fulfils
diagrams (Figure 2) . Adding the contributions of specified requirements’ [102] and validation is
individual components of variation are appropri- ‘verification, where the specified requirements
ate when the components are well known and are adequate for the intended use’ [102] . If the
when their individual contribution to the over- method (reagents, procedure and the measure-
all measurement uncertainty can be measured ment instrument) is manufactured by a company
(bottom-up approach). or other reliable source that has performed proper
In laboratory medicine, measurement instru- method validation and who is providing you with
ments and reagents are commonly designed by the detailed results, a single laboratory method
the manufacturers as ‘black boxes’ precluding validation is not needed. Method verification as
the end-user controlling or measuring individ- described below is then appropriate.
ual components influencing the measurement
process. In these cases the end-user cannot Validation of measurement methods
measure the individual components of measure- Method validation is a specific kind of validation
ment uncertainty independently, and is unable “the process of defining an analytical requirement,
to add them using the bottom-up approach. A and confirming that the method under consid-
top-down approach for estimation of measure- eration has performance capabilities consistent
ment uncertainty can be used in this case and with what the application requires” [106] . Method
is also endorsed by GUM [105] . It measures the validation includes procedures that both establish
combination of the individual components of the performance characteristics and limitations
variation using a stable control material and of a measurement method (e.g., trueness, preci-
Key Term
can elucidate its individual components using sion, recovery, linearity, robustness) and establish
classical principles of experimental design and whether the performance characteristics of the Method verification:
Procedures to test to what
variance component analysis. The same batch of measurement method being investigated are fit extent the performance data
a stable control material is measured daily or sev- for the intended purpose [8,9,106] . obtained by manufacturers
eral times daily over long periods of time, com- Method validation is performed to a varying during method validation can be
monly in the order of 2 years, and measurement extent depending on its intended use. Single reproduced in the environments
of end-users.
uncertainty estimated using analysis of variance laboratory method validation [10] is appropriate
Key Term where the method is used for a specific purpose n The method should be fully developed and
Full method validation:
in a specific laboratory by personnel with the optimized;
Consists of both a technical part appropriate training. Full method validation
n A written standard operating procedure for
and fit-for-purpose studies. includes, in addition to the procedures employed
the method should be available;
in single laboratory validation, an interlabora-
tory study (collaborative study/collaborative n The measurement instruments to be used
trial), which usually includes in the order of ten should be regularly technically controlled and
laboratories with many measurement instru- well maintained;
ments, several operators and so forth. The per-
n The persons performing the measurements
formance characteristics of the measurement
should have sufficient training and experience
method over extended periods of time are also
for the task;
studied in full method validation, including the
effects of lot-to-lot variation. n Appropriate calibrators should be available
Several national and international organi- and a supply (for at least 1 year) of suitable
sations have published general guidelines on stable materials (for at least two concentration
method validation [3,4,6,11–37,101,106–114] and spe- levels) for internal quality control purposes;
cific literature are available in analytical chem-
n The needs of the end user regarding fit-for-
istry [18,19,21,38–40] , toxicology [41] , clinical chem-
purpose of the method should be known.
istry/pathology [42–50,115,116] , the food industry
[10,50,51] and the pharmaceutical industry [52–57] .
Results of measurements of biomarkers regu- Measurement precision
lated by ISO 17025, ISO 15189 or by CLIA [117] Measurement precision is the “closeness between
can and are extensively used in the development indications or measured quantity values obtained
of pharmaceuticals. by replicate measurements on the same or similar
objects under the specified conditions of meas-
Single laboratory method validation urement” [102] . The quantitative expression of
Single laboratory method validation is appropri- precision is the SD or CV/CV%. The SD of the
ate when the method is to be used in a single estimate of the SD is:
laboratory for a well-defined group of custom-
ers who are using the particular laboratory and
SDs = 0.71 # SD
measurement method and no other. Single N
laboratory method validation is appropriate Equation 2
in the pharmaceutical industry where a single
laboratory and method is used for measuring all and is thus inversely proportional to the square
samples in a specified study, or in research labo- root of the number of replicates. Furthermore,
ratories where a single laboratory and method the computed SD (even when using N-1 in the
is used for the same study. In healthcare, the denominator – the Bessel’s correction) under-
samples from a patient are, over time, likely to estimates the true population SD, especially
be measured in different laboratories using dif- when the number of observations in the sam-
ferent methods as the patient visits primary care ple is low [58] . Therefore a ‘c4 correction’ [118]
and different levels of hospital care. In this case, is appropriate when estimating the population
full validation of all the methods a patient is SD from small samples. When N = 2 the SD is
likely to encounter is crucial, including efforts to too low by approximately 20%, when N = 10 it
minimize measurement uncertainty, in particu- is low by approximately 3%, and only low by
lar the effects of bias, which usually is by far the approximately 2% when N = 15.
major component contributing to measurement Measurement precision falls into one of the
uncertainty. If this not feasible, samples from following three following types: repeatabil-
the same patient should be measured using the ity, intermediate measurement precision and
same measurement method and measurement measurement reproducibility.
instrument over time using reference intervals
established for that particular method. Repeatability precision
Repeatability measurement precision is esti-
The basic prerequisites for method validation mated when ‘the same measurement procedure,
Before initiating method validation the following same operators, same measuring system, same
prerequisites should be fulfilled: operating conditions and same location, and
Adept method
in an outpatient department
Adept method
with individual patients
Figure 3. Mentor (split sample) technique used to monitor bias between a mentor
laboratory and its adept laboratories. Amongst the characteristics of the mentor laboratory are
highly motivated personnel and optimal calibration, traceability, internal quality control and external
proficiency testing. A perquisite for a practical system like this is appropriate information
technology [119] .
Figure 4. Broad overview of mentor (split sample) results from several measurands and measuring instruments in a large
laboratory organisation.
quality control. After measurement in any of the the Levy–Jennings plot by a linear regression
other laboratories in the conglomerate, the sam- and bias plots showing the relation between bias
ple is sent to the mentor laboratory (split sample and the measurement level (Figure 6) .
technique) for analysis in order to measure bias
(Figures 3 & 4) [119] and to estimate the relative Measuring interval
contribution of bias and random error of indi- The measuring interval/working interval/work-
vidual methods and measurement instruments to ing range/measuring range/measurement range
the measurement uncertainty (Figure 5) . is a “set of quantities of the same kind that can
In order to monitor the bias over time using be measured by a given measuring instrument
graphs it is an advantage to normalize the results or measuring system with specified instrumen-
in the following manner: tal uncertainty, under defined conditions” [102] .
The measuring interval is therefore dependent
Adept - Mentor on the extent to which the measuring system
Normalized = c m # 100
Mentor can produce results that are fit for the intended
Equation 3 purpose.
expressing the results of the adepts as percent- The lower limit of the measuring interval
age deviation from the 100% measured by the coincides with the LOQ (the lowest concentra-
mentor. The advantage of this is the ability to tion of the measurand that can be measured with
monitor changes of the bias over time using an acceptable uncertainty), which is higher than
Levey–Jenning plot to detect trends and monitor the LOD.
deviation using standard process/quality-control
methodologies (Figure 4) . The disadvantage is Linearity
that the level on the measurement scale is lost. There is a linear relationship between the con-
This can be compensated for by complementing centration and the signal in most, but not all
B, MCHC
Mtd Inst ColD Mean CV total% CV treat% CV error% CV% n
M1 2454 PPI 336.3 2658 2086 1804 2325 7
M1 2455 PPI 335.1 3126 07115 3180 4963 13
M1 3111 PPI 350.8 4719 2319 4222 4214 20
M1 3311 PPI 332.8 3546 2992 1946 2042 24
Figure 5. Mentor (split sample) and multivariate analysis techniques used to estimate bias
(CV treat%) and random error (CV error%) components of measurement uncertainty for
several hematology measurement instruments measuring mean corpuscular hemoglobin
concentration in whole blood samples.
www.random.org/has been shown to be truly where a is the intercept on the x-axis and SDyx
random, and is a personal favorite), and deter- is SD of the residuals around the calibration
mine the concentration at which the percent curve [76] .
positive or negative becomes unreliable.
There are several alternative acceptable meth-
ods for determining the LOD which are pre- LOQ
ferred depending on the basic technology being The LOQ is not a concept defined in VIM3,
validated: but has been defined as “the lowest concen-
tration of measurand that can be determined
n Repeated measurement of blanks: for methods with an acceptable level of repeatability preci-
yielding fair signal when measuring blanks, the sion and trueness” [106] . It is commonly taken
following method can be used. Perform all steps to be 10 × SD where SD is the SD of the blank
of the measurement procedure on ten aliquots or of a sample containing very low concentra-
of a blank sample and report LOD = 5 × SD; tions of the measurand. Perform all steps of the
Figure 6. A mentor (split sample) technique used to estimate bias between a mentor
laboratory and one of its adept laboratories regarding measurement of the
concentration of hemoglobin in whole blood. (A) A traditional Levy–Jennings internal quality
control plot of the mentor results around the optimum of 100%. Each horizontal line represents
one SD. (B) A linear regression of the concentrations measured by the mentor in relation to the
corresponding concentration measured by the adept. (C) and (D) Bias plots of the absolute and
relative deviation of the results in relation to the mentor results.
measurement procedure on ten aliquots of one back-calculated values for the standards at the
or more sample(s) with concentrations near to appropriate concentrations from the first day of
the LOD of the method, each aliquot from each long-term stability testing” [101] .
sample on separate days. Report LOD = 10 × SD
[39] or use the method proposed by Thienpont Sampling techniques
et al. [15] . The diagnostic relevance of “the accept- Patient preparation and preanalytical influences
able level of repeatability precision and trueness” on measurands in biological matrices are beyond
in clinical chemistry is not generally agreed, and the scope of the present presentation. They have
establishing the clinical use of diagnostic mark- been comprehensively dealt with, for example, by
ers is far from being only a technical matter, but Guder et al. [39] . The presence of coagulation pro-
rather is the subject of extensive clinical studies teins may interfere with the measurement method,
in patients [77] . making serum preferable to plasma. Similarly, anti-
coagulant additives may interfere in the measure-
Robustness ment process and their influence must therefore be
Robustness/ruggedness is “the ability of the meas- tested. Various enzyme inhibitors may be needed
urement method to withstand small changes in to stabilize the measurand, bur their influence on
its operating conditions” [39] . Robustness regard- the measurement procedure itself must also be
ing crucial measurement parameters particular investigated. The wealth of specialized literature
for the method, for example temperature or pH, should be consulted regarding each measurand and
should be tested systematically during validation the matrix in which it is to be measured.
preferably using one of the several available tools
for systematic experimental design [74] . Measurement uncertainty
Measurement uncertainty is a “non-negative
Storage stability of the measurand in the parameter characterizing the dispersion of the
sample matrix quantity values being attributed to a measur-
Stability of measurands in biological matrix and, based on the information used” [102] . It
depends on the properties of the measurand, includes both systematic and random compo-
the matrix and the storage conditions, includ- nents (Figure 1) . The initial estimate of meas-
ing the sample container, storage temperature urement uncertainty sufficient for single labo-
and exposure to light. A particular concern ratory method validation can be obtained from
regarding measurands in biological matrix is the method validation data [78,79] , but highly reliable
freeze–thaw cycles. estimates of measurement uncertainty are only
obtained after months and years of measur-
Freeze–thaw stability ing single batch of stabilized control materials,
The FDA recommends that “at least three aliq- including the contribution from all causes of
uots at each of the low and high concentrations variation (e.g., calibrations or changes of rea-
should be stored at the intended storage temper- gent lots) as part of full method validation. This
ature for 24 h and thawed unassisted at room was already emphasised by Walter Shewhart in
temperature” [101] . the 1930s [59] and needs to be rediscovered and
applied in practice by all engaged in measuring
Short-term temperature stability and monitoring variation.
The FDA validation document recommends that Since measurement methods are prone to
“three aliquots of each of the low and high con- matrix effects in stabilized control materials,
centrations should be thawed at room tempera- estimates of measurement reproducibility using
ture and kept at this temperature from 4–24 h these samples overestimate the uncertainty. Split-
(based on the expected duration that samples sample techniques using natural patient samples
will be maintained at room temperature in the as described in ‘Measurement reproducibility’
intended study) and analyzed” [101] . provide uncertainty estimates more close to the
real uncertainty experienced in healthcare [119] .
Long-term stability
The FDA document recommends “storing at least Full method validation
three aliquots of each of the low and high concen- Full method validation consists both of a techni-
trations under the same conditions as the study cal part and of fit-for-purpose studies – studies
samples. The concentrations of all the stability of the methods diagnostic properties in clinical
samples should be compared with the mean of chemistry.
laboratories are used. Bias between results of ongoing quality control using patient samples
individually calibrated measurement methods analyzed at the different locations within the
in healthcare systems is a major contributor ‘laboratory’ [83] .
to diagnostic uncertainty when diagnosing
diseases and when monitoring the effects of Verification of measurement methods
treatment. The second issue for making meas- Measurement methods and instruments used
urement results comparable is stating the meas- in laboratory medicine in general, and clinical
urement uncertainty as well as the diagnostic chemistry in particular, are overwhelmingly
uncertainty. manufactured by large international companies.
By tradition, a laboratory is taken to be a These companies are responsible for the valida-
coherent single physical entity within, for tion of these methods for specific diagnostic or
example, a hospital. Laboratories are increas- monitoring purposes [84,101,120] , but customers
ingly joined into larger laboratory organizations are responsible for verifying that the character-
encompassing several physical laboratories. istics shown during validation can be reproduced
This caters for important new opportunities in the practical clinical situation.
for redefining the concept of a ‘laboratory’ to Local verification practices have commonly
encompass all laboratories and measurement been established over time and are frequently
methods measuring the same measurand, for influenced by accreditation and certification
all levels of healthcare within a geographical authorities. Published practices for end-user
area caring for a defined population in the verification officially endorsed both by the end-
order of 0.5–1 million inhabitants. In order users and by the companies have appeared only
to make measurement results comparable a recently [115,116] .
measurement uncertainty has to be evaluated In vitro diagnostic (IVD) medical devices
for measurement results from this ‘laboratory’. are regulated in Europe by a third Medical
A major task in addition to all the routines Device Directive, the IVD Medical Device
already implemented would then be the task Directive 98/79/EC [120] , which has been
of minimizing the bias between results for mandatory since December 2003. Validation
patient samples transported between differ- of the performance of instruments and reagent
ent locations within the ‘laboratory’. In the kits are performed by the manufacturer. In the
clinical area, the bias can be monitored by an case of a measurement method established by a
Executive summary
Despite important developments during recent years, there is as yet no comprehensive agreement on validation and verification
procedures amongst the different fields of bioanalysis.
The need for mutually accepted validation procedures becomes obvious when international accreditation and certification standards or
their equivalents are implemented.
A giant leap forward could be accomplished if all fields of bioanalysis fully accept and implement the agreements and fundamental
documents in metrology – the mother discipline of all branches of bioanalysis.
The principles of metrology can be, and are being, implemented in the field of clinical chemistry, a field of bioanalysis where common
validation procedures need to be extended by investigating the diagnostic properties of methods, and the added value they bring to the
clinical diagnosis of human disease conditions and to the monitoring of treatment effects.
Measurement methods and instruments used in clinical chemistry are overwhelmingly manufactured by large international companies.
Such companies are responsible for the validation of these methods for specific diagnostic or monitoring purposes, but the customers
are responsible for verifying that the characteristics shown during validation can be reproduced in the local clinical situation. Proposals
for end-user verification officially endorsed both by the end-users and by the companies have appeared only recently.
4 Viswanathan CT, Bansal S, Booth B et al. Critical investigation of major validation 29 Hibbert DB. Method validation of modern
Quantitative bioanalytical methods validation criteria and associated experimental protocols. analytical techniques. Accredit. Qual. Assur.
and implementation: best practices for J. Chromatogr. B 877(23), 2180–2190 (2009). 4(8), 352–356 (1999).
chromatographic and ligand binding assays. 16 Hibbert DB. Method Validation. Elsevier, 30 Hibbert RB. Method validation of modern
Pharm. Res. 24(10), 1962–1973 (2007). Oxford (2005). analytical techniques. Accredit. Qual. Assur. 4,
5 Blume H, Brendel E, Brudny-Kloppel M et al. 17 Rozet E, Marini RD, Ziemons E, Boulanger 352–356 (1999).
Workshop/conference report on European B, Hubert P. Advances in validation, risk and 31 Barwick VJ, Ellison SLR, Raffaerty MJQ,
Medicine Agency draft guideline on uncertainty assessment of bioanalytical Gill RS. The evaluation of measurement
validation of bioanalytical methods. Eur. methods. J. Pharmaceut. Biomed. 55(4), uncertainty from method validation studies.
J. Pharm. Sci. 42(3), 300–305 (2011). 848–858 (2011). Part 2: the practical application of laboratory
6 Haeckel R. Verification, validation and 18 Thompson M, Ellison LR, Wood R. protocol. Accredit. Qual. Assur. 5, 104–113
evaluation of analytical procedures in Harmonized guidelines for single-laboratory (2000).
laboratory medicine. Clin. Chem. Lab. Med. validation of methods of analysis. Pure Appl. 32 Knottnerus JA, Muris JW. Assessment of the
42(1), 111–112 (2004). Chem. 74(5), 835–855 (2002). accuracy of diagnostic tests: the cross-
7 Menditto A, Patriarca M, Magnusson B. 19 Hartmann C, Smeyers-Verbeke J, Massart sectional study. J. Clin. Epidemiol. 56(11),
Understanding the meaning of accuracy, DL, Mcdowall RD. Validation of 1118–1128 (2003).
trueness and precision. Accredit. Qual. Assur. bioanalytical chromatographic methods. 33 Bossuyt PM, Reitsma JB, Bruns DE et al.
12, 45–47 (2007). J. Pharmaceut. Biomed. 17(2), 193–218 Towards complete and accurate reporting of
(1998). studies of diagnostic accuracy: the STARD
nn The crucial concepts of accuracy, trueness
20 Wood R. How to validate analytical methods. initiative. BMJ 326(7379), 41–44 (2003).
and precision still cause controversy and
confusion. Menditto and co-workers have Trends Anal. Chem. 18(9–10), 624–632 34 Williams DR. Minimum criteria for
created a diagram that brilliantly explains (1999). validation of analytical methods. J. Assoc. Off.
these concepts and the relation between 21 Wieling J, Hendriks G, Tamminga WJ et al. Anal. Chem. 69, 432–436 (1986).
them. Rational experimental design for bioanalytical 35 Permanent Liaison Group between EAL
methods validation – Illustration using an (European co-operation for Accreditation of
8 Lee JW, Devanarayan V, Barrett YC et al.
assay method for total captopril in plasma. Laboratories) and EUROLAB. Validation of
Fit-for-purpose method development and
J. Chromatogr. A 730(1–2), 381–394 (1996). test methods. General principles and
validation for successful biomarker
22 Rozet E, Ceccato A, Hubert C et al. Analysis concepts. Accredit. Qual. Assur. 3, 29–32
measurement. Pharm. Res. 23(2), 312–328
of recent pharmaceutical regulatory (1998).
(2006).
documents on analytical method validation. 36 Holmgren M. Validation of test methods.
9 Lee JW, Hall M. Method validation of
J. Chromatogr. A 1158(1–2), 111–125 (2007). General principles and concepts. Accredit.
protein biomarkers in support of drug
23 Massart DL, Dijkstra A, Kaufman L. Qual. Assur. 3, 29–32 (1998).
development or clinical diagnosis/prognosis.
J. Chromatogr. B Analyt. Technol. Biomed. Life Evaluation and Optimization of Laboratory 37 De Bièvre P, Günzler H. Validation in
Sci. 877(13), 1259–1271 (2009). Methods and Analytical Procedures. Elsevier Chemical Measurement. Springer, Berlin,
Scientific Publishing, Amsterdam, The Germany (2004).
10 AOAC guidelines for single laboratory
Netherlands (1978). 38 Taverniers I, De Loose M, Van Bocksraele E.
validation of chemical methods for dietary
supplements and botanicals (2002). 24 Clinical and Laboratory Standards Institute Trends in quality in the analytical laboratory.
(CLSI). User verification of performance for II. Analytical method validation and quality
11 Van Amsterdam P, Arnold M, Bansal S et al.
precision and trueness; approved guideline. assurance. Trends Anal. Chem. 23(8),
Building the global bioanalysis consortium
25(17), EP15-A2 (2006). 535–552 (2004).
– working towards a functional globally
78 Feinberg M, Boulanger B, Dewe W, Hubert P. 102 JCGM. International vocabulary of 112 Eurachem: the fitness for purpose of
New advances in method validation and metrology – basic and general concepts and analytical methods. A laboratory guide to
measurement uncertainty aimed at improving associated terms (VIM 3). method validation and related topics (1998).
the quality of chemical data. Anal. Bioanal. www.bipm.org/en/publications/guides/vim. www.eurachem.org/guides/pdf/valid.pdf
Chem. 380(3), 502–514 (2004). html 113 Harmonised guidelines for the use of
79 Patriarca M, Castelli M, Corsetti F, Menditto 103 Terminology in analytical recovery information in analytical
A. Estimate of uncertainty of measurement measurement – introduction to VIM 3 measurement (1998).
from a single-laboratory validation study: (2011). www.eurachem.org/guides/pdf/recovery.pdf
application to the determination of lead in www.eurachem.org/guides/pdf/ 114 Eurpean Medicines Agency. Guideline on
blood. Clin. Chem. 50(8), 1396–1405 (2004). TAM_2011_Final_web.pdf Bioanalytical Method Validation (2011).
80 Galen RS, Gambino SR. Beyond Normality: 104 An ontology on property for physical, www.ema.europa.eu/docs/en_GB/
The Predictive Value and Efficiency of Medical chemical and biological systems (2004). document_library/Scientific_
Diagnoses. John Wiley & Sons, NJ, USA http://ontology.iupac.org/ontology.pdf guideline/2011/08/WC500109686.pdf
(1975). 105 JCGM. Evaluation of measurement data – 115 Khatami Z, Hill R, Sturgeon C, Kearney E,
81 Pepe MS. The Statistical Evaluation of Medical Guide to the expression of uncertainty in Breadon P, Kallner A. Measurement
Tests for Classification and Prediction. Oxford measurement. verification in the clinical laboratory: a
University Press, Oxford, UK (2003). www.bipm.org/en/publications/guides/ guide to assessing analytical performance
82 Zhou X-H, Obuchowski NA, Mcclish DK. gum.html during the acceptance testing of methods
Statistical Methods in Diagnostic Medicine. n All measurements are frought with (quantitative examination procedures) and/
Wiley-Interscience, NY, USA (2002). or analysers. The Scientific Committee of
uncertainties. Understanding them and
the Association for Clinical Biochemistry
83 Dybkaer R. From total allowable error via handling them properly is crucial both in
(2010).
metrological traceability to uncertainty of metrology and in all fields of bioanalysis.
www.acb.org.uk/An%20Ver/
measurement of the unbiased result. Accredit. 106 De Biévre P, Böttger D, Eastwood C et al. Measurement%20verification_
Qual. Assur. 4, 401–405 (1999). The fitness for purpose of analytical final_090608.pdf
84 Dati F. The new European directive on in vitro methods. a laboratory guide to method
116 Burnett D. Measurement verification in the
diagnostics. Clin. Chem. Lab. Med. 41(10), validation and related topics (1998).
clinical laboratory: A guide to assessing
1289–1298 (2003). www.eurachem.org/guides/pdf/valid.pdf
analytical performance during the
85 Stankovic AK, Romeo P. The role of in vitro 107 Measurement uncertainty arising from acceptance testing of methods (quantitative
diagnostic companies in reducing laboratory sampling: a guide to methods and examination procedures) and/or analysers.
error. Clin. Chem. Lab. Med. 45(6), 781–788 approaches (2007). (2010).
(2007). www.eurachem.org/index.php/publications/ www.acb.org.uk/An%20Ver/David%20
86 Passing H, Bablock W. Comparison of several
guides/musamp Burnett%20Editorial.pdf
regression procedures for method comparison 108 European Medicines Agency concept paper/ 117 Clinical laboratory improvement
studies and determination of sample sizes. recommendations on the need for a amendments (1988).
Application of linear regression procedures for guideline on the validation of bioanalytical www.fda.gov/MedicalDevices/
method comparison studies in clinical methods (2008). DeviceRegulationandGuidance/
chemistry. Part II. J. Clin. Chem. Clin. www.ema.europa.eu/pdfs/human/ IVDRegulatoryAssistance/ucm124105.htm
Biochem.22, 431–445 (1984). ewp/53130508en.pdf
118 NIST/Sematech: e-handbook of statistical
87 Deming WE. Statistical Adjustment of Data. 109 European Medicines Agency draft guideline methods (2008).
John Wiley & Sons, Inc, Oxford, UK (1943). on validation of bioanalytical methods www.itl.nist.gov/div898/handbook/
88 Bland JM, Altman DG. Statistical methods for
(2009).
119 Norheim S. Computer support simplifying
assessing agreement between two methods of www.ema.europa.eu/pdfs/human/
uncertainty estimation using patient
clinical measurement. Lancet 327, 307–310 ewp/19221709en.pdf
samples (2008).
(1986). 110 CDC. CLIA 88, Certification of laboratories http://liu.diva-portal.org/smash/
89 Altman DG, Bland JM. Measurement in
(1988). record.jsf?pid=diva2:417298
medicine: the analysis of method comparison wwwn.cdc.gov/clia/pdf/PHSA_353.pdf
120 EU Directive 98/79/EC of the European
studies. Statistician 32, 307–317 (1983). 111 Eurachem: traceability in chemical Parliament and of the Council of 27
measurement. A guide to achieving October 1998 on in vitro diagnostic medical
comparable results in chemical devices (1998).
Websites measurements (2003). www.europa.eu/legislation_summaries/
101 FDA Guidance for Industry. Bioanalytical www.eurachem.org/index.php/publications/ internal_market/single_market_for_goods/
Method Validation (2001). guides/trc technical_harmonisation/l21010c_en.htm
www.fda.gov/downloads/Drugs/
GuidanceComplianceRegulatoryInformation/
Guidances/ucm070107.pdf