66 / 258
12
B
runt
et al
.:
J
ournal of
aoaC I
nternatIonal
V
ol
.
100, n
o
.
3, 2017
(
3
) Sample No. 10, Infant Formula Powder, Milk-Based, Lot
No. 4044755861.
(
4
) Sample No. 12, Child Formula Powder, Lot No.
00866RF00.
(
5
) Sample No. 14, Infant Formula Powder, FOS/GOS-
Based, Lot No. 50350017W1.
(
6
) Sample No. 19, Adult Nutritional RTF, High-Fat, Lot No.
0729RF00.
These six samples were used in the precision study in both
laboratories, with an additional two samples (Nestlé Ref 1
and 2) included at NRC. The repeatability and intermediate
reproducibility were assessed by analyzing all samples in
duplicate on 6 different days, with the exception of Nestlé
Ref 2, which was analyzed in duplicate on 24 different days
(Table 2).
The repeatability (RSD
r
) achieved in both laboratories
(Table 2) is well below the upper level of 6% defined in SMPR
2014.002 (5), with only one exception, sample No. 14, which
had an RSD
r
of 8.9% at CCC. The results for sample No.
14 may be expected to be more variable because this sample
contains the lowest fructan content, about 0.03 g/100 g,
which is close to the LOQ. The intermediate reproducibility
(RSD
iR
) achieved in both laboratories is, in general, below
8%, again with the exception of sample No. 14 analyzed at
CCC. Sample No. 14 was analyzed with an RSD
iR
of 14%,
which is almost certainly linked to the low concentration of
analyte in that sample. No limits were defined for RSD
iR
in
SMPR 2014.002 (5), but the data suggest that achieving an
RSD
R
of below 12% (as defined in SMPR 2014.002) for a
multilaboratory trial could be possible, although it may be
expected that samples containing fructan levels close to the
LOQ may have higher variability.
The analytical results and the established precision data of
NRC and CCC agree very well with each other. There is no
significant difference in the mean fructan content of the samples
[paired
t
-test (α = 0.05)], but the RSD
iR
values of CCC are slightly
higher than those of NRC (Table 2). This could be explained by
the fact that CCC was able to introduce more variability in the
execution of the intermediate reproducibility experiments than
NRC: data at CCC were collected over the course of 4 months,
whereas at NRC, the data were mostly collected over the course
of 1 month; and data at CCC were acquired by two different
people using three different instruments, whereas at NRC, most
of the precision data were generated by a single person using
two different instruments.
Accuracy/Trueness
Primary indications of the accuracy of the method were
obtained at CCC during the determination of LOD and LOQ,
because this was performed by spiking low levels of fructans
into a blank matrix (Table 1). The fructan addition rate was
0.046 g/100 g on both days; thus, the average recoveries were
95.6% on day 1 and 104% on day 2, which are well within the
defined target range of 90–110%.
Table 1. Determination of LOD and LOQ at CCC
Parameter
Day 1
a
Day 2
a
Fructan content,
g/100 g
Meets the
SMPRs
Fructan
content,
g/100 g
Meets the
SMPRs
Average
b
0.044
NA
c
0.048
NA
SD
b
0.002
NA
0.003
NA
LOD
d
0.005
Yes
0.009
Yes
LOQ
e
0.016
Yes
0.030
Yes
a
Days 1 and 2 were separated by 3 months.
b
Average and SD values were calculated on 10 replicates measured on
a single day.
c
NA = Not Applicable.
d
LOQ calculated as 10SD.
e
LOD calculated as 3SD.
Table 2. Summary of results from the precision study
Sample
No.
Sample description
n
a
Mean fructan content,
g/100 g
RSD
r
, %
Meets SMPR
2014.002 target
RSD
iR
, %
NRC CCC NRC CCC NRC CCC NRC CCC NRC CCC
1
Child Formula Powder
6 × 2
6 × 2
0.27
0.29
b
0.8
1.4
b
Yes
Yes
b
1.1
7.8
b
9
Toddler Formula Powder,
Milk-Based
6 × 2
6 × 2
0. 21
0.22
2.0
1.6
Yes
Yes
2.4
6.1
10
Infant Formula Powder,
Milk-Based
6 × 2
6 × 2
0.28
0.26
2.4
1.2
Yes
Yes
2.8
6.9
12
Child Formula Powder
6 × 2
6 × 2
0.28
0.28
1.2
1.1
Yes
Yes
1.1
6.1
14
Infant Formula Powder,
FOS/GOS-Based
6 × 2
6 × 2
0.03
0.04
0.9
8.9
c
Yes
No
c
2.5
14.2
c
19
Adult Nutritional RTF,
High-Fat
6 × 2
6 × 2
0.48
0.51
1.8
2.6
Yes
Yes
1.6
4.2
Nestlé
Ref 1
Infant formula
6 × 2
NA
d
2.9
NA
1.1
NA
Yes
NA 1.8
NA
Nestlé
Ref 2
Infant formula
24 × 2
NA
0.39
NA
3.3
NA
Yes
NA 4.3
NA
a
n
= Number of determinations, represented as days × replicates.
b
There was one (Grubbs) outlier in one of the duplicates; therefore, the results are based on a set of five duplicates instead of six.
c
The fructan content was close to the LOQ, resulting in somewhat increased RSD
r
and RSD
iR
values.
d
NA = Not Applicable.
66