Agriculture Reference
In-Depth Information
Table 3.2. Method Validation Parameters and Criteria
Applicability to
MRMs
Parameter
How to Address
Criterion
Accuracy
Determine mean recovery
from spikes
70-120%
Quantitative
only
Linearity
Construct calibration curve
Residuals
<±
20%
Quantitative
only
LoD
The lowest concentration
where 95% confidence of
detection of analyte(s) is
achieved
Less than or equal to
default MRL
(0.01mg/kg)
Qualitative
LoQ
The lowest concentration at
which criteria for accu-
racy and precision are met
Quantitative
only
Matrix effect
Comparison between detec-
tor response for standards
made up in solvent and
that for standards made up
in sample matrix
No criteria. Matrix
effects may vary
from analyte to
analyte as well as
between samples
Qualitative and
quantitative
Precision
(RSD
r
)
a
Determine repeatability
from replicate spikes
analyzed in same batch
of samples
20%
Quantitative
only
Precision
(RSD
R
)
Determine reproducibility
from replicate spikes
analyzed on different days
20%
Quantitative
only
Qualitative
b
and
quantitative
Selectivity
Response should be attrib-
utable to the analyte
<
30% LoD
Check for any response in
reagent and sample matrix
blanks
<
30% LoQ
Robustness
a
How often the method fails
to meet the criteria that are
applicable above
Qualitative and
quantitative
Source
: Ref. [16].
a
It is not essential to address these parameters during initial method validation as they can be derived from
ongoing QC data generated as the method starts to be used for routine analyses.
b
No requirement has been set since any detect is supposed to be followed up by an additional confirmatory
analysis. However, selectivity should be such that the number of false detects is low enough for efficient use
in routine practice.
AQC requirements relating to recovery or linearity, only selectivity. As it is unlikely
that any method can detect all possible analyte/commodity combinations, it has
become widely accepted [16] that a 95% con
dence level (i.e., analyte detected in 19
out of 20 samples) is suf
cient. This means that the SDL of a method is the lowest
concentration for which it has been demonstrated that a particular analyte can be
detected (without necessarily meeting unambiguous identi
cation criteria) in at least
