41 / 71
Pang et al.:
J
ournal of
AOAC I
nternational
V
ol.
98, N
o.
5, 2015
1437
Results and Discussion
Evaluation of Collaborative Study Results
In 2010, the multiresidue determination method for 653
pesticides in tea was selected to be one of the priority study
projects of AOAC. After 3 years of preparation, 30 laboratories
from 11 countries and regions participated in this collaborative
study from March 1 through June 30, 2013 to evaluate the
reproducibility of the method. A total of 560 samples were
analyzed including green tea and oolong tea fortified with 40
selected pesticides, aged oolong tea, and pesticide incurred
green tea. Participants had the option to use one or more of
the three available instrumental analytical techniques, GC/
MS, GC/MS/MS, and/or LC/MS/MS, for the analysis. Thirty
laboratories submitted results to the Study Director (
see
Table 2, and the details can be found in Annex 1.2 available on
the
J. AOAC Int.
website). The first three columns in Table 2
report the total number of pesticide residues (6868) detected by
the 30 laboratories. The last three columns report the number
of results (8915) submitted by the 29 participating laboratories
for the precollaborative study. For GC/MS, the results of target
pesticides and precollaborative study were 1977 and 2740,
respectively. For GC/MS/MS, the results of target pesticides and
precollaborative study were 1808 and 2440, respectively. For
LC/MS/MS, the results of target pesticides and precollaborative
study were 3083 and 3735, respectively. Note that Laboratory 20
did not submit any precollaborative study results to the Study
Director. Columns 4–6 report a total of 41238 ions monitored,
columns 7–9 report the total ion abundance data (23205), and
columns 10–12 report the total number of calibration data points
(2233) generated in the collaborative study.
A review of the 6868 test data and related information
submitted to the Study Director from the 30 laboratories
revealed that one laboratory (Laboratory 20) had not
submitted precollaborative test results to the Study Director
but had gone ahead and conducted the collaborative study.
Since this laboratory had failed to meet the prerequisite
proficiency criterion for entering into the collaborative study,
the data submitted by this laboratory were not included in
the collaborative data package. A cursory analysis of an
example calibration data generated by Laboratory 20 shown
in Table 3 for oolong tea immediately revealed that there were
too many significant deviations ≥20% between the measured
and the expected concentrations, demonstrating that the
laboratory had not achieved the requisite proficiency level
required for the collaborative study and would not have been
qualified to participate in the analysis of the real samples.
Therefore, only 6638 results from the 29 laboratories that
had met all the precollaborative data requirements and had
been certified proficient to participate in the collaborative
study by the Study Director were considered for further
analysis and interpretation (
see
Annex 1.2 on the
J. AOAC
Int.
website).
(a)
Application of the Dixon and Grubbs outlier tests on
6638 data generated
.—The Dixon and Grubbs tests were used
to examine outliers in the 6638 data from the remaining 29
laboratories. For GC/MS, 65 outliers out of 1977 test data were
identified, accounting for 3.3%. For GC/MS/MS, 65 outliers
out of 1704 test data were identified, accounting for 3.8%.
For LC/MS/MS, 57 outliers out of 2957 LC/MS/MS test data
were identified, accounting for 1.9%. Therefore, a total of 187
outliers, making up 2.8% of the 6638 data were identified and
eliminated from the data package.
(b)
Method extraction efficiency and reproducibility for
fortified samples
.—The recovery efficiency and reproducibility
of the GC/MS, GC/MS/MS, and LC/MS/MS methods for the
analysis of the 20 selected pesticides in oolong and green tea
are presented in Tables 4–6. Further analysis of the method
efficiency parameters such as Rec. (recovery), RSD
R
, RSD
r
, and
HorRat values were summarized in Table 7.
(
1
)
By GC/MS.—
For green tea or oolong tea, the average
recoveries (% Avg Rec.) for the fortified samples by GC/MS
fall within the range of 75–100%, RSD
r
<8%, accounting for
100%. For green tea samples, the pesticides with RSD
R
<16%
accounted for 100%. For oolong tea samples, the pesticides
with RSD
R
<25% accounted for 100%. The HorRat values for
all the pesticides were less than 2.0.
(
2
)
By GC/MS/MS.—
For green tea or oolong tea, the average
recoveries (%Avg Rec.) for the fortified samples by GC/MS/MS
fall within the range of 75–100%, RSD
r
<8%, accounting for
100%. For green tea samples, the pesticides with RSD
R
<16%
accounted for 100%. For oolong tea samples, the pesticides
with RSD
R
<25% accounted for 90%. There were two pesticide
residues for which the RSD
R
exceeded 25%, accounting for
10%. The HorRat values for all the pesticides were less than 2.0.
(
3
)
By LC/MS/MS.—
For green tea or oolong tea, the average
recoveries (%Avg Rec.) for the fortified samples by LC/MS/MS
fall within the range of 75–100%, RSD
r
<15%, accounting for
100%. For green tea samples, the pesticides with RSD
R
<16%
accounted for 95%, and one pesticide had RSD
R
more than
25%, accounting for 5%. For oolong tea samples, the pesticides
with RSD
R
<25% accounted for 75%.
There were five pesticide residues for which the RSD
R
exceeded 25%, accounting for 25%. The HorRat values for all
the pesticides were less than 2.0.
According to
Evaluation of Collaborative Study Results
,
b
(
2
)
and
b
(
3
), there were seven pesticide residues (two pesticides
analyzed by GC/MS/MS and five pesticides analyzed by
LC/MS/MS) for which the RSD
R
exceeded 25%. The low
precision for the analysis of these seven pesticide residues
may be explained by the fact that the concentrations of the
pesticides added to oolong tea were determined to be too
close to the LOD of the older generation instruments used in
the analysis. It was with these older generation instruments
that the method generated significant variability in precision.
Additionally, the concentration of seven compounds added to
oolong tea were low and close to the lowest concentration point
on the calibration curve, thus contributing to the imprecision
at the low end of the calibration curve. For example, this
happened in the case of Laboratory 25, which determined that
Table 2014.09H. Recommended maximum permitted
tolerances for relative ion intensities using a range of mass
spectrometric techniques
Relative intensity
(% of base peak), %
GC/MS
(relative), %
GC/MS/MS, LC/MS/MS
(relative), %
>50
±10
±20
>20–50
±15
±25
>10–20
±20
±30
≤10
±50
±50