AOAC OMA# 2011.06 Final Action Review-OMB

B handari et al . : J ournal of AOAC I nternational V ol . 101, N o . 6, 2018  1889

Statistical Evaluation

All data were statistically analyzed using the AOAC harmonized guidelines for collaborative studies to establish overall mean, intralaboratory repeatability, repeatability relative standard deviation (RSD r ), interlaboratory reproducibility, reproducibility relative standard deviation (RSD R ), and Horwitz ratio (HorRat). Cochran ( P = 0.025, one-tail) and Grubbs (single and double, P = 0.025, two-tail) tests were used to determine outliers (19). Twelve laboratories participated in the study. One of the laboratories could not qualify for analysis of the MLT samples. Eleven laboratories could provide complete results of all the MLT samples in time to report to AOAC. The laboratories used the method as provided to them in the protocol without any deviation. The results of the analysis of all of the samples for total folate as folic acid (μg/100 g) were reported by each participating laboratory upon completion of the analysis. The results received from participants of all 11 laboratories are tabulated and summarized in Table 1. The submitted data were statistically analyzed using the AOAC protocols (19). Only two sets of outliers were identified and excluded as Cochran outliers. One of the outliers was the result for the partially hydrolyzed milk- based infant formula powder (Sample 2) from laboratory 7, and the other one was for the soy-based infant formula sample (Sample 5) from laboratory 8 (Table 1). No other outliers were identified, and all other results were used in the generation of averages and precision values. The statistical analysis of results submitted by participants was performed per AOAC protocol (19), and the results are provided in Tables 2 and 3. The method precision as repeatability ranged from 3.5 to 6.6% RSD r. The method reproducibility values ranged from 9.0 to 15.7% RSD R . The method repeatability (RSD r , %) as well as the reproducibility (RSD R , %) for all of the SPIFAN samples analyzed in the study were within the limits set in the folate SMPR 2011.006 (18). Acceptable reproducibility was also demonstrated with better than expected HorRat R values for the method, ranging from 0.61 to 1.06 (expected range 0.5–2.0; 19). The method accuracy has been verified earlier during the single lab validation of the method using NIST SRM 1849a and BCR 485 (Fol-22, AOAC First Action Official Method 2011.06 Single Laboratory validation report submitted to the SPIFAN meeting on March 16, 2016). The results obtained in the MLT study for the NIST SRM 1869 for folic acid are compared with the NIST consensus value to further support the accuracy of the method (Table 4). The method’s accuracy was evaluated based on the preliminary value for folic acid, which has been documented through an interlaboratory study but not yet published and finalized by the NIST for this sample. They were kind enough to share those values with us. The NIST has reported a consensus value for folic acid in this SRM based on results from an interlaboratory study with participation of 17 laboratories. The grand mean value obtained by the MLT study was close to the NIST consensus mean and within the NIST consensus range (Table 4). The method’s main objective is to analyze total folate based on the analysis of the relevant folate vitamers in infant formula Results and Discussion

nutritional ready to feed – high fat – Liq

Total folate as folic acid, mcg/100 g 1 179.1 171.9 142.8 140.5 201.9 196.7 274.0 274.1 253.2 199.2 176.5 179.2 369.1 356.4 55.5 51.5 17.8 18.5 392.8 372.7 249.7 296.3 61.7 65.0 2 165.7 164.1 134.9 140.9 182.6 188.3 259.2 255.3 255.5 179.6 184.2 184.8 338.1 347.3 52.7 50.1 16.1 16.8 352.3 351.2 256.8 251.2 59.8 59.2 3 166.9 161.8 151.3 149.1 213.0 181.5 275.3 231.9 216.4 183.8 191.1 165.2 360.1 335.2 42.8 38.0 15.7 16.0 316.5 311.2 251.9 245.2 50.5 46.6 4 130.2 131.8 113.1 110.9 134.6 128.4 241.8 270.7 243.6 132.4 129.4 157.8 266.1 270.5 44.6 44.7 18.1 20.1 366.4 380.4 277.3 278.6 51.1 47.3 5 130.6 146.7 104.7 116.7 157.1 163.5 237.4 248.6 247.2 146.5 164.0 168.9 286.0 316.4 50.5 47.8 16.0 15.8 337.7 375.4 273.3 266.3 57.3 57.1 6 152.2 161.6 124.4 118.7 192.5 167.9 246.9 259.5 259.5 169.2 169.5 164.6 331.5 313.7 47.9 50.4 17.0 15.4 333.3 323.4 246.3 254.6 58.7 58.7 7 130.4 160.8 177.1 a 105.8 a 156.0 158.3 218.7 210.3 237.7 164.7 161.1 154.4 337.9 285.1 43.9 43.3 12.0 12.4 301.2 308.8 205.0 215.4 50.9 52.2 8 164.7 169.5 152.5 151.5 202.4 212.6 178.9 212.1 187.0 187.5 a 203.5 a 132.1 a 346.2 346.1 44.1 42.7 10.1 12.2 324.4 296.2 194.5 215.5 50.5 54.5 9 164.6 170.3 145.8 140.4 207.7 178.8 256.1 276.3 270.8 198.3 191.3 192.5 342.3 340.5 52.7 53.9 16.5 17.4 383.2 371.9 283.7 283.0 60.8 60.5 10 146.4 152.6 117.8 136.7 134.5 136.0 264.9 244.8 225.1 173.8 189.2 171.4 281.2 265.9 53.2 53.6 13.4 13.9 394.7 357.6 324.3 284.8 61.9 56.7 11 142.6 141.4 117.3 120.6 170.7 156.2 251.9 256.0 254.2 164.2 161.3 183.6 341.1 340.8 57.9 54.9 15.3 15.6 363.7 373.6 266.5 268.1 66.2 67.5 a  Results removed as Cochran outlier prior to average and precision calculations.

Sample 11: Adult

Sample 10: Infant

elemental powder

formula powder – milk-based

Sample 9: Child

Sample 8:

Infant formula

ready to feed –

milk-based – Liq

nutritional ready to feed – high protein – Liq

Sample 7: Adult

Low-fat adult nutritional powder

Sample 6:

Sample 5: Soy-based infant formula

Nutritional

Sample 4: NIST SRM

1869 Infant Formula Adult

Infant formula powder, partially

Sample 3:

soy-based

hydrolyzed

Sample 2:

powder, partially hydrolyzed milk-based

Infant formula

Sample 1:

Milk-based

infant formula

Table 1. Raw data for replicate analysis of the Folate MLT SPIFAN Kit samples Lab No.