5. AOACSPDSMethods-2018AwardsV3

124

18  B rown & Y u : J ournal of AOAC I nternational V ol . 96, N o . 1, 2013

(c)  Injection .—Make single 10 µL injections of each test solution.

Single Grubbs’ test (g, α = 0.05 single replicate, 0.125 blind replicate):

g x X S i i = −  − /

Analyzing Raw Data

= suspected outlier (either highest or lowest result),

where x i

The amount (µg/g) of individual and total ginsenosides for each matrix is to be determined using the seven-point calibration curves. The AOAC INTERNATIONAL Statistical Package (AOAC INTERNATIONAL Interlaboratory Study Workbook v. 2.0 for Blind Replicates) is an electronic Excel spreadsheet provided by Mark Roman from Tampa BayAnalytical Research (Tampa Bay, FL). The spreadsheet is to be used to statistically analyze the reproducibility and HorRat R value for each sample matrix submitted as blind replicates. For the blind duplicates (whole root, powdered extract, and negative spiked material samples), data should be assessed and outliers removed using Cochran’s test (α = 0.025), and single and double Grubbs’ tests (α = 0.0125). In addition, the recovery will be calculated for the negative spiked matrix blank material to verify relative accuracy of each participating laboratory. For the samples that are submitted as single replicate samples (capsule and tablet materials), data will be assessed for outliers using the single Grubbs’ test (α = 0.05). Reproducibility is evaluated only by looking at the HorRat r value. The equations used for calculations are as follows: The concentration of ginsenoside in the samples: Concentration, µg/g = [(P o – b o )/m o ] × (V/W) × D where P o = peak area of the ginsenoside in sample chromatogram (mAu*s), b o = y-intercept of the calibration curve for the ginsenoside (mAu*s), m o = slope of the calibration curve for the ginsenoside [mAu*s/(µg/mL)], V = final volume of test solution (mL), W = sample weight (g), and D = dilution factor (final volume/aliquot volume: 1.2 mL). Average or mean values ( X ): Calculations

X = average or mean value, and S = SD. Double Grubbs’ test (G, α = 0.0125):

/SS o

or G = SS 1.2

/SS o

G = SS n-1,n

where SS o or SS 1.2

= sum of squared deviation from the mean and SS n-1, n = sum of squared deviations obtained after removal of

the two highest or the two lowest values, respectively. RSD R among laboratory data:

RSD = (S×100) X R − where S = SD among laboratory data and x = average or mean value among laboratory data. Predicted RSD (PRSD): PRSD R = C –0.15 where C = measured analyte concentration in decimal mass units (µg/g, %, ppm, etc.). HorRat R or HorRat r : HorRat R or r = RSD R or R /PRSD R Participant Comments Two laboratories were unable to provide data and were excluded from the interlaboratory study. Laboratory 2 withdrew prior to the start of the study because it did not have an available technician to complete the analysis before the specified deadline of the project. Laboratory 3 did not have proper extraction equipment, specifically a conical centrifuge rotor, to adequately separate the supernatant from the sample material. As a result of the exclusion of two laboratories as described above, 12 data sets were compiled and analyzed from the remaining participants. All of the remaining laboratories reported the individual and total ginsenoside content for each sample matrix. For data assessment, the AOAC INTERNATIONAL Interlaboratory Study Workbook (v. 2.0) was used to statistically evaluate the individual and total ginsenosides reported for each matrix. All data were tabulated and screened for outliers. After the outliers were removed from each data set, the results were evaluated for reproducibility and their accompanying HorRat. The recovery of the spiked negative controls ranged from 89 to 112%, indicating that the laboratories were relatively accurate with respect to their analysis. The whole root, powdered extract, and negative spiked matrix blank samples submitted as blind duplicate were evaluated using the RSD R and HorRat R values. For the tablets Results and Discussion Interlaboratory Study Results

X= x /n i _ ∑ ( )

= individual value and n = number of individual values.

where x i

Standard deviation (S):

0.5

   

∑    

S= (x -X ) /n i _ 2

where X=average or mean value, x i n =number of individual values. Cochran’s outlier test (C, α = 0.025):

= individual value, and

C =S / S j j 2 j 2 i=1 N ∑

where C j = SD of data series j, N = number of data series that remain in the data set ( N is decreased in steps of 1 upon each iteration of the C test), and S i = SD of data series i (1 < i < N). = Cochran’s C statistics for data series j, S j