163 / 258
1708
Pacquette & Thompson:
J
ournal of
AOAC I
nternational
V
ol.
98, N
o.
6, 2015
intermediate precision was mostly in the range of 2–5% RSD,
with the highest at 6.5% RSD. These results are consistent with
the subsequent collaborative study, for which reproducibility was
about 2–3% higher than the intermediate precision. Repeatability
was not probed extensively in the SLV because the short-term
precision is very good and the method, after all, does require
the duplicate precision to be below the 5% RSD required by the
SMPR.
Accuracy was checked via overspike recoveries in the SLV.
Table 4 shows the mean recovery of triplicate overspikes on each
of 3 days. Spike levels were at 50–200% of the native levels. All
recoveries were within 90–110%, meeting the SMPR.
Table 5 shows the SRM 1849a results during the SLV. This
SRM was tested 16 times with excellent precision. The mean
results for Cr and Se were within the certified range, whereas Mo
was just barely low. The subsequent collaborative study means
(one analysis in duplicate from each of the eight laboratories
using five different ICP/MS instrument models) were almost
identical to those from the Abbott SLV, including the Mo value
obtained.
Data for linearity and LOQ were obtained during the SLV, but
these figures of merit have to be proven for a given instrument
setup/model in the same way that the participating laboratories
did the prework for the MLT.
Table 6. RSD
r
, RSD
R
, and HorRat values for 2011.19 collaborative study
a
Matrix
Cr, µg/kg
RTF Labs
RSD
r
,
%
RSD
R
,
% HorRat
Mo, µg/kg
RTF Labs
RSD
r,
%
RSD
R,
% HorRat
Se, µg/kg
RTF Labs
RSD
r
,
%
RSD
R
,
% HorRat
Infant powdered milk
partially hydrolyzed
<PLOQ 8 N/A N/A N/A
20
8 3.3 6.7 0.33
27
8 2.4 2.5 0.13
Adult powder low fat
48
8 4.7 7.1 0.39
63
8 1.6 3.1 0.18
30
8 5.9 7.2 0.37
Adult powdered milk
16
7 3.4 12.1 0.57
33
7 1.0 7.9 0.42
24
8 6.1 6.1 0.31
Child formula powdered
30
7 5.5 9.2 0.48
30
8 3.3 4.6 0.24
24
8 3.8 7.3 0.37
Infant elemental powdered 24
7 3.8 13.4 0.67
18
8 1.7 7.9 0.38
23
8 6.4 9.3 0.46
Adult RTF high protein
130 8 7.0 8.1 0.37
150
7 1.0 3.0 0.14
93
8 2.3 8.1 0.36
Adult RTF high fat
140 8 2.1 5.8 0.27
190
8 1.2 3.8 0.19 133
8 4.7 5.0 0.23
Average
4.4 9.3 0.46
1.9 5.3 0.27
4.5 6.5 0.32
a
RSDs are shown for means that are at or slightly below PLOQ of 20 µg/kg RTF for Cr and Mo because they are indicative of method performance.
No results are above the SMPR required 15% RSD
R
. For the five cases in which the number of participating laboratories is listed as
n =
7, the eighth
laboratory’s data could be included and still meet SMPR reproducibility requirements (
see
footnotes for Tables 7–9).
Table 7. Collaborative study individual results for Cr (mg/kg, as is)
a
Lab No.
Adult milk
powdered
Infant powdered
hydrolyzed milk
Adult powdered
low fat
Child powder
Infant elemental
powdered
Adult RTF high
protein
Adult RTF
high fat
1
a
0.025 0.018 <PLOQ <PLOQ 0.083 0.073 0.036 0.045 0.026 0.016 0.020 0.022 0.019 0.022
2
0.121
b
0.237
c
<PLOQ <PLOQ 0.424 0.359 0.371
c
0.254
c
0.220
d
0.150
d
0.143 0.126 0.112
e
0.108
e
3
0.130 0.145 <PLOQ <PLOQ 0.385 0.386 0.240 0.250 0.192 0.185 0.105 0.125 0.128 0.122
4
0.140 0.166
f
<PLOQ <PLOQ 0.427 0.430 0.273 0.274 0.205 0.200 0.117 0.119 0.135 0.131
5
0.154 0.156 <PLOQ <PLOQ 0.448 0.438 0.278 0.288 0.230
g
0.250 0.128 0.129 0.130 0.135
8
0.121 0.124 <PLOQ <PLOQ 0.438 0.474 0.293 0.272 0.218 0.197 0.125 0.125 0.126 0.130
9
0.172 0.168 <PLOQ <PLOQ 0.437 0.447 0.331 0.286 0.213 0.229 0.135 0.126 0.127 0.127
10
0.194
h
0.165 <PLOQ <PLOQ 0.441 0.470 0.299 0.326 0.266 0.273 0.128 0.108 0.127 0.127
11
0.137 0.138 <PLOQ <PLOQ 0.427 0.432 0.266 0.268 0.200 0.206 0.111 0.116 0.126 0.125
SLV (
n
= 6)
0.142
<PLOQ
0.428
0.272
0.215
0.130
0.141
a
Other rejected data are indicated by footnotes below. The value in each cell is the mean of the duplicate analyses required by the method. The paired
results are the blind duplicates tested during the study.
b
Rejected by Cochran’s test; if not excluded RSD
R
rises from 12.1 to 22.4%; if just the 0.237 value is rejected (so that eight laboratories have repre-
sentative data), RSD
R
is 12.8%.
c
Rejected by Cochran’s test; if not excluded RSD
R
rises from 9.2 to 11.7%; if just the 0.371 value is rejected (so that eight laboratories have represen-
tative data), RSD
R
is 9.3%.
d
Rejected by Cochran’s test; if not excluded RSD
R
rises from 13.4 to 15.0%; if just the 0.150 value is rejected (so that eight laboratories have repre-
sentative data), RSD
R
is 12.7%.
e
Although rejected by Single Grubbs’ test these data points were kept in the final statistical analysis, as the data appear extraordinarily tight; if ex-
cluded, the RSD
R
falls to 2.9% and the HorRat to 0.13 for this product.
f
Rejected, 23% duplicate RSD; no retest result supplied.
g
Rejected, 13% duplicate precision; no retest result supplied.
h
Rejected, 15% duplicate RSD; no retest result supplied.
163