AOAC Final Action Methods in 2017

838 J OSEPH ET AL . : J OURNAL OF AOAC I NTERNATIONAL V OL . 101, N O . 3, 2018

Linearity and Range

Table 3. Retention times and correlation coefficients

r 2

Retention time, min

The retention time of biocytin ranged from 3.3 to 6.1 min and that of biotin ranged from 13.8 to 22.5 min. The retention times reported from Laboratory 9 stand out from the other laboratories, although the chromatographic conditions were followed as documented. The variation is likely due to a difference in the hardware configuration of the liquid chromatographs used for the analysis; however, this information was not available at the time of data processing. All the participants used more or less the same calibration range of 1 – 20 µg/100 mL for biocytin and biotin, as specified in the method. The range showed excellent correlation coefficients among the participants of not less than 0.999, confirming the linearity of the method over the calibration range (Table 3). The biotin results from practice samples and MLT samples (days 1 and 2) are given in Tables 4 – 6, respectively. Fourteen different product types were analyzed in duplicate by nine laboratories, including two practice samples and 12 samples for MLT over 2 separate days. The practice samples were each analyzed in duplicate, and the blind coded duplicates were also analyzed during MLT on the same day by the participating laboratories. Laboratories 10, 11, and 12 submitted results for practice samples only. The results are closely comparable between the duplicates and between the participating laboratories. All the data have been used for statistical evaluation to calculate repeatability, reproducibility, and HorRat values. The statistical analysis of the data was carried out using the Interlaboratory Study Workbook Version 2.1 – Blind (Unpaired) Replicates from AOAC (15). The data are presented in the report as micrograms per 100 g powder or product as received from AOAC for the study. The RSD r and RSD R , and HorRat performance values are summarized in Table 7. of not more than 6% for biotin levels >1 µg/100 g. The mean RSD r value of the matrixes analyzed in the study is 4.6% and well within the limit of the SMPR. However, MLT samples 5 and 9 recorded slightly higher RSD r values of 6.79 and 7.03%, respectively. The duplicate results from Laboratory 8 resulted in an outlier for MLT sample 5, and removing the data from the laboratory brings the RSD r down to 5.68%. Similarly, the removal of duplicate data from Laboratory 9 brings the RSD r of MLT sample 9 down to 5.81%. The removal of these values still leaves statistically significant duplicate sets (eight sets) for a valid RSD r calculation. However, the values are kept in the report because the overall repeatability meets the SMPR criteria and the SLV demonstrated the repeatability precision before the MLT study. MLT Results and Statistical Evaluation RSD r Repeatability expresses the precision under the same operating conditions (intra-assay) over a short interval of time. SPIFAN SMPR SM 2014.005 for the biotin analysis specifies a maximum s r

Lab No.

Biocytin

Biotin

Biocytin

Biotin

1 2 3 4 5 6 7 8 9

4.6 3.8 4.9 5.9 4.4 4.9 3.3 4.1 6.1 4.9 4.1

16.1 15.9 16.2 18.1 15.5 16.5 14.5 13.8 22.5 15.1 12.9

0.9981 0.9996 0.9987 0.9983 0.9997 0.9995 0.9990 0.9991 0.9998 0.9995 0.9990

0.9995 0.9999 0.9993 0.9997 0.9993 0.9995 0.9990 0.9973 0.9995 0.9988 0.9993

Downloaded from https://academic.oup.com/jaoac/article-abstract/101/3/831/5653969 by guest on 05 May 2020

11 12

more than 12% for all 14 matrixes analyzed in the study, confirming the precision of the analytical method. The maximum RSD R value recorded was 9.5% in MLT sample 1, which is a partially hydrolyzed milk-based infant formula matrix. The reproducibility precision confirms the suitability of AOAC Method 2016.02 as a strong candidate to become a reference method for global dispute resolution because the interlaboratory variation is minimal and exceeds the expectation of the SMPR limits. The HorRat is a normalized performance parameter indicating the acceptability of methods of analysis with respect to among- laboratory precision (reproducibility). It is the ratio of the observed RSD among laboratories calculated from the actual performance data, RSD R , to the corresponding predicted RSD (PRSD R ) calculated from the Horwitz equation. The formula for the HorRat, as presented in the Interlaboratory Study Workbook of AOAC, is applicable only when the concentration is in the unit/ unit form (e.g., µg/µg, g/g, etc.). When the analyte concentration is a mass fraction amount such as µg/100 g, the appropriate factor for unit conversion should be selected to generate correct HorRat values. The factor selected for the calculation in this study is 1.00E − 08 to convert the results to g/g. Under reproducibility conditions, the acceptable HorRat value range is 0.5 – 2 as per AOAC. The HorRat values obtained in this study were within 0.3 – 0.5 for all SPIFAN matrixes, except NIST SRM 1849a. The performance requirements of SPIFANmethods are generally tighter than routine analytical methods and, therefore, the PRSD R used in the formula may be slightly higher, pushing the HorRat values down. Nevertheless, the lower HorRat values confirm the enhanced performance of the method and the homogeneity of the matrixes used for the study. NIST SRM 1849a recorded the lowest HorRat value of 0.21 and, undoubtedly, is the most homogeneous certified Standard Reference Material (SRM) used in the MLT. HorRat

Accuracy (NIST SRM 1849a)

RSD R

The accuracy of an analytical procedure expresses the closeness of agreement between the value found and the

Reproducibility expresses the precision among the laboratories. The RSD R exceeds the SMPR criteria of not