ASTM D6512-00

NOTICE: This standard has either been superceded and replaced by a new version or discontinued.
Contact ASTM International (www.astm.org) for the latest information.
Designation: D 6512 – 00
Standard Practice for
Interlaboratory Quantitation Estimate
This standard is issued under the fixed designation D 6512; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (e) indicates an editorial change since the last revision or reapproval.
1. Scope in routine field measurements. Avoidable sources would in-
clude, but are not limited to: modifications to the sample;
1.1 This practice establishes a uniform standard for com-
modifications to the measurement procedure; modifications to
puting the interlaboratory quantitation estimate associated with
the measurement equipment of the validated method, and gross
Z % relative standard deviation (referred to herein as IQE ),
Z%
and easily discernible transcription errors, provided there was
and provides guidance concerning the appropriate use and
a way to detect and either correct or eliminate them.
application.
1.4 The IQE applies to measurement methods for which
1.2 IQE is computed to be the lowest concentration for
Z%
calibration error is minor relative to other sources, such as
which a single measurement from a laboratory selected from
when the dominant source of variation is one of the following:
the population of qualified laboratories represented in an
1.4.1 Sample Preparation, and calibration standards do not
interlaboratory study will have an estimated Z % relative
have to go through sample preparation.
standard deviation (Z % RSD, based on interlaboratory stan-
1.4.2 Differences in Analysts, and analysts have little oppor-
dard deviation), where Z is typically an integer multiple of 10,
tunity to affect calibration results (as is the case with automated
such as 10, 20, or 30, but Z can be less than 10. The IQE
10 %
calibration).
is consistent with the quantitation approaches of Currie (1)
1.4.3 Differences in Laboratories (for whatever reasons),
and Oppenheimer, et al (2).
perhaps difficult to identify and eliminate.
1.3 The fundamental assumption of the collaborative study
1.4.4 Differences in Instruments (measurement equipment),
is that the media tested, the concentrations tested, and the
such as differences in manufacturer, model, hardware, electron-
protocol followed in the study provide a representative and fair
ics, sampling rate, chemical processing rate, integration time,
evaluation of the scope and applicability of the test method as
software algorithms, internal signal processing and thresholds,
written. Properly applied, the IQE procedure ensures that the
effective sample volume, and contamination level.
IQE has the following properties:
1.5 Data Quality Objectives—Typically, one would com-
1.3.1 Routinely Achievable IQE Value—Most laboratories
pute the lowest % RSD possible for any given dataset for a
are able to attain the IQE quantitation performance in routine
particular method. Thus, if possible, IQE would be com-
10 %
analyses, using a standard measurement system, at reasonable
puted. If the data indicated that the method was too noisy, one
cost. This property is needed for a quantitation limit to be
might have to compute instead IQE , or possibly IQE .
20 % 30 %
feasible in practical situations. Representative laboratories
In any case, an IQE with a higher % RSD level (such as
must be included in the data to calculate the IQE.
IQE ) would not be considered, though an IQE with RSD
1.3.2 Accounting for Routine Sources of Error—The IQE 50 %
<10 % (such as IQE ) would be acceptable. The appropriate
1%
should realistically include sources of bias and variation that
level of % RSD may depend on the intended use of the IQE.
are common to the measurement process. These sources
include, but are not limited to: intrinsic instrument noise, some
2. Referenced Documents
“typical” amount of carryover error; plus differences in labo-
2.1 ASTM Standards:
ratories, analysts, sample preparation, and instruments.
D 2777 Practice for Determination of Precision and Bias of
1.3.3 Avoidable Sources of Error Excluded—The IQE
Applicable Test Methods of Committee D-19 on Water
should realistically exclude avoidable sources of bias and
D 6091 Practice for 99 %/95 % Interlaboratory Detection
variation; that is, those sources that can reasonably be avoided
Estimate (IDE) for Analytical Methods with Negligible
Calibration Error
This practice is under the jurisdiction of ASTM Committee D19 onWater and E 1763 Guide for Interpretation and Use of Results from
is the direct responsibility of Subcommittee D19.02 on General Specifications,
Technical Resources, and Statistical Methods.
Current edition approved Feb. 10, 2000. Published May 2000.
2 3
The boldface numbers in parentheses refer to the list of references at the end of Annual Book of ASTM Standards, Vol 11.01.
this standard.
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.
NOTICE: This standard has either been superceded and replaced by a new version or discontinued.
Contact ASTM International (www.astm.org) for the latest information.
D6512–00
Interlaboratory Testing of Chemical Analysis Methods straight-line, and hybrid (proposed by Rocke and Lorenzato
(3)). Evaluation includes statistical significance and residual
3. Terminology analysis.
4.3 The chosen model is used to predict the standard
3.1 Z % Interlaboratory Quantitation Estimate (IQE ),
Z%
deviation of interlaboratory measurements at any true concen-
also denoted “LQ,” for “Limit of Quantitation” in accordance
tration within the study concentration range. If interlaboratory
with Currie (1)—The lowest concentration for which a single
standard deviations change systematically with respect to the
measurement from a laboratory selected from the population of
true concentration (that is, they are NOT constant), the predic-
qualified laboratories represented in an interlaboratory study
tions are used to generate weights for fitting the mean-recovery
will have an estimated Z % relative standard deviation (Z %
relationship (the assumed straight-line relationship between
RSD, based on interlaboratory standard deviation).
measured concentration and true concentration), using
3.2 Definitions of Terms Specific to This Standard:
weighted least squares. (Otherwise, ordinary least squares is
3.2.1 Censored Measurement—A measurement that is not
used.) The mean-recovery curve is evaluated for statistical
reported numerically nor is reported missing, but is stated as a
significance, for lack of fit, and for residual patterns. The ILSD
nondetect or a less-than (for example, “less than 0.1 ppb”).
model is also used to estimate the interlaboratory standard
There are two reasons why the measurement may not be
deviation at concentrations within the concentration range.
reported numerically. Either the measurement was considered
Either a direct or interactive algorithm (depending on the
insufficiently precise or accurate (these kinds of data should not
model) is used to compute IQE , the lowest concentration
10 %
be censored), or the identification of the analyte was suspect
with estimated RSD = 10 % (Z = 10). If there is no such
(these kinds of data should be censored). See §6.2.3.1. A
concentration, then IQE is computed instead, or IQE ,if
20 % 30 %
reported “less than” may have the same meaning as a non-
necessary. If supported by the data quality objectives (DQOs),
reported measurement, but a reported “less than” also implies
IQE may be computed for someZ<10.
Z%
(perhaps erroneously) that any concentration greater than or
equal to the accompanying value (for example, 0.1 ppb) can be
5. Significance and Use
measured, and will be reported numerically.
5.1 Appropriate application of this practice should result in
3.2.2 Quantitation Limit (QL) or Limit of Quantitation
an IQE achievable by most laboratories properly using the test
(LQ)—A numerical value, expressed in physical units or
method studied. That is, most laboratories should be capable of
proportion, intended to represent the lowest level of reliable
measuring concentrations greater than IQE with RSD = Z %
Z%
quantitation. The IQE is an example of a QL.
or less. The IQE provides the basis for any prospective use of
the test method by qualified laboratories for reliable quantita-
4. Summary of Practice
tion of low-level concentrations of the same analyte as the one
4.1 Every ASTM Committee D-19 test method is evaluated
studied in this practice, and same media (matrix).
to determine precision and bias by conducting a collaborative
5.2 The IQE values may be used to compare the quantitation
study, in accordance with Practice D 2777. That study, or a
capability of different methods for analysis of the same analyte
similar collaborative study, can also be used to evaluate the
in the same matrix. The IQE is not an indicator of individual
lowest concentration level of reliable quantitation for a test
laboratory performance.
method, referred to herein as the interlaboratory quantitation
5.3 The IQE procedure should be used to establish the
estimate (IQE). Such a study must include concentrations
interlaboratory quantitation capability for any application of a
suitable for modeling the uncertainty of mean recovery of
method where interlaboratory quantitation is important to data
interlaboratory measurement, preferably without extrapolation.
use. The intent of the IQE is not to set reporting limits.
The study must also be planned and conducted to allow the
known, routine sources of measurement variability to be
6. Procedure
observed at typical levels of influence. After the study is
6.1 The following procedure has stages described in the
conducted, outlying laboratories and individual measurements
following paragraphs: 6.2–IQE Study Plan, Design, and Pro-
should be eliminated, using an accepted, scientifically based
tocol; 6.3–Conduct the IQE Study, Screen the Data, and
procedure for outlier removal, such as found in Practice
Choose a Model; and 6.4–Compute the IQE. A flowchart of the
D 2777. The IQE computations must be based on retained data
procedure is shown in Fig. 1.
from at least six independent laboratories at each concentration
6.2 IQE Study Plan, Design, and Protocol:
level.
6.2.1 Choose Analyte, Matrix, and Method—At least one
4.2 Retained data are analyzed to identify and fit one of
analyte of interest is selected, typically one for which there is
three proposed interlaboratory standard deviation (ILSD) mod-
interest in trace or near-trace levels of concentration, such as
els. These models describe the relationship between the inter-
toxic materials that are controlled and regulated. For each
laboratory standard deviation of measurements and the true
analyte, an approximate maximum true concentration is se-
concentration, T. The identification process involves evaluating
lected, based on these considerations:
the models in order, from simplest to most complex: constant,
6.2.1.1 The anticipated IQE should be exceeded by a factor
of 2 or more,
6.2.1.2 A single model, (ideally a straight-line model in true
Annual Book of ASTM Standards , Vol 03.06. concentration, T) should describe mean recovery (that is, mean
NOTICE: This standard has either been superceded and replaced by a new version or discontinued.
Contact ASTM International (www.astm.org) for the latest information.
D6512–00
FIG. 1 Flowchart of IQE Procedure
measured concentration) for the entire range of concentrations, design. Three models are proposed herein for the relationship
from zero to the selected maximum concentration. between the interlaboratory standard deviation of measure-
ments and the true concentration: constant, straight-line (in-
NOTE 1—The IQE procedure uses the straight-line model for mean
creasing), and hybrid (increasing). See 6.3.3 for details. Chem-
recovery, thus implicitly assuming that a straight line is adequate. Thus,
istry, physics, empirical evidence, or informed judgment may
the IQE would not be appropriate for cases where this assumption is
make one model more plausible than others. However, it may
unreasonable. For example, it would not hold for cases where there was
systematic bias for most or all laboratories, such as a tendency to report
not be possible to anticipate the relationship between standard
values that are too high for some portion of the concentration range.
deviation and true concentration.
6.2.2.1 Select an IQE study design that has enough distinct
6.2.1.3 A single model in true concentration should describe
the standard deviation of interlaboratory measurements for the concentration levels to assess statistical lack of fit of the
models (see Draper and Smith (4)). Recommended designs are:
entire range of concentrations, from zero to the selected
maximum concentration. (a) the “semi-geometric” design at five or more true concen-
trations, {T , T , and so forth}, such as: {0, IQE /4, IQE /2,
6.2.1.4 The concentration range must be sufficient to enable
1 2 0 0
statistically significant coefficients to be estimated for the ILSD IQE ,2 3 IQE ,4 3 IQE ,8 3 IQE }, where IQE is an initial
0 0 0 0 0
estimate of the IQE (such as 10s8 where s8 is the interlaboratory
model and mean-recovery model. At least one matrix of
interest is also selected, and an accepted standard analytical measurement standard deviation at a trace-level, nonzero
concentration); (b) equi-spaced design: {0, IQE /2, IQE ,
method for those analytes is selected for study. If there is no
0 0
possibility of matrix interference, then it may only be neces- (3/2)3 IQE ,2 3 IQE , (5/2) 3 IQE ,3 3 IQE }; and (c) any
0 0 0 0
other design with at least five concentrations, provided that the
sary to determine a list of acceptable matrices that can be used,
design includes at least one concentration approximately equal
instead of selecting a specific matrix. For example, for a
particular analyte, concentration range, and method, it may be to 2 3 IQE , at least one nonzero concentration below IQE ,
0 0
and one blank, or unspiked sample. Preferably, the design will
supposed that reagent waters from different laboratories are
indistinguishable. However, that assumption may not hold for have at least seven concentrations, including a blank.
another analyte or another concentration range. 6.2.2.2 The study’s concentration levels must either be
6.2.2 Choose IQE Study Design—The design should be known (true concentration levels), or knowable, after the fact.
based (if possible) on an anticipated ILSD model. Section 7 of A concentration is considered known if reference standards can
Practice D 2777 can be followed for the study desig
...