315 / 407
AOAC Final Action Method 2016.02 (Biotin)
Page 9 of 16
Table 2016.02C: Retention times and correlation coefficients
Lab #
Retention Time (Min)
Correlation Coefficient (r
2
)
Biocytin
Biotin
Biocytin
Biotin
1
4.6
16.1
0.9981
0.9995
2
3.8
15.9
0.9996
0.9999
3
4.9
16.2
0.9987
0.9993
4
5.9
18.1
0.9983
0.9997
5
4.4
15.5
0.9997
0.9993
6
4.9
16.5
0.9995
0.9995
7
3.3
14.5
0.9990
0.9990
8
4.1
13.8
0.9991
0.9973
9
6.1
22.5
0.9998
0.9995
11
4.9
15.1
0.9995
0.9988
12
4.1
12.9
0.9990
0.9993
MLT Results and Statistical Evaluation
The biotin results from practice and MLT samples (Day 1 and 2) are given in Tables 2016.02D, E and F
respectively. Fourteen different product types were analysed in duplicate by nine laboratories which
includes two practice samples and 12 samples for MLT over two separate days. The practice samples were
analysed in duplicates and the blind coded duplicates were analysed during MLT on the same day by the
participants. Laboratories 10, 11 and 12 submitted results for practice samples only.
The results are closely comparable between the duplicates and among all the participants and all the data
have been used for statistical evaluation to calculate repeatability, reproducibility and HorRat. The
statistical analysis of the data was carried out using
International Study Workbook Version 2.1 for Blind
(Unpaired) Replicates
from AOAC International. The data as presented in the report as
µg/100g
were used
as such for statistical calculation with a Factor for Units of Measurement of 1.00E-08. The summary of
relative standard deviation of repeatability (RSD
r
) and reproducibility (RSD
R
) along with HorRat
performance ratios is tabulated in Table 2016.02G.
Relative Standard Deviation of Repeatability (RSD
r
)
Repeatability expresses the precision under the same operating conditions (intra-assay) over a short
interval of time. The AOAC SPIFAN SMPR 2014.005 for the biotin analysis specifies a maximum
repeatability standard deviation of not more than 6% for biotin levels greater than 1 µg/100g. The mean
RSD
r
of the matrices analysed in the study is 4.6% and is well within the limit of the SMPR. However MLT
samples 5 and 9 recorded slightly higher RSD
r
of 6.79% and 7.03% respectively. The duplicate results from
Lab 8 are causing the outlier for MLT sample 5 and removing the data from the lab bring the RSD
r
down to
5.68%. Similarly, the removal of duplicate data from Lab 9 will bring the RSD
r
of MLT sample 9 down to
5.81%. The removal of these values still leaves statistically significant duplicate sets (8 sets) for a valid
RSD
r
calculation. However the values are kept in the report as the overall repeatability meets the SMPR
criteria and the SLV demonstrated the repeatability precision before the present study.
Relative Standard Deviation of Reproducibility (RSD
R
)
Reproducibility expresses the precision among the laboratories. The relative standard deviation of
reproducibility (RSD
R
) exceeds the SMPR criteria of not more than 12% for all the 14 matrices analysed in
the study confirming the precision of the analytical method. The maximum RSD
R
noticed was 9.5% in the
MLT sample 1, which is partially hydrolysed milk based infant formula matrix. The reproducibility precision
confirms the suitability of the SPIFAN method AOAC 2016.02 as a strong candidate as a reference method
for global dispute resolution hence the inter-laboratory variation is minimal and exceeds the expectation of
the SMPR limits.
2016.02 (BIOTIN) FINAL ACTION REVIEW
FOR ERP USE ONLY
DO NOT DISTRIBUTE
207