2015.14 Vitamin B (July 2018)

2015.14 (JULY 2018) BVit-02 Total B_AOAC_2015.14_MLT_Final

FOR ERP USE ONLY DO NOT DISTRIBUTE

For all product matrices, reproducibility averaged 6.6% RSD with a range of 3.66-16.8% RSD. Thirteen of the 14 fortified products met the standard method performance requirements for reproducibility of ≤ 10% RSD for fortified products. Total Niacin For all product matrices, reproducibility averaged 7.9% RSD with a range of 4.78-12.4% RSD. Thirteen of the 14 fortified products met the standard method performance requirements for reproducibility of ≤ 10% RSD for fortified products. Statistical Analysis of MLT Data Using the AOAC International Interlaboratory Study Workbook for Blind (unpaired) Replicates, Version 2.1 to analyze B 1 , B 2 , B 3 , and B 6 MLT data, statistical outliers were identified. There was a total of 21 lab/sample/analyte pairs of data that were removed when flagged. 17 of these came from lab 2 as one of their days of data had many anomalies. I did reach out to see if they could evaluate the data after finding contact information but have not received a response as of the date of this report. Lab 1 had significant deviations in Riboflavin standards. These deviations carried through to the samples and 6 of their 15 sample/analyte pairs were flagged and removed. There were 3 samples where both lab 1 and lab 2 were removed, resulting in 8 labs of data. Outside of Riboflavin for lab 1, and sporadic data from lab 2, no other statistical flags resulted in removal of data. General Comments In general, B vitamins are stable, but Thiamine is slightly sensitive to temperature, oxygen, humidity, and light. This study has taken over two years and there does seem to be a downward trend of thiamine data during this time. This may help explain why repeatability did not fail in any of the edited sample sets, but reproducibility was above the allowed 10% for 5 of the 14 fortified samples. There were several deviations to the MLT protocol during the MLT study, but none of these deviations appeared to impact final results. One laboratory did not analyze the practice samples. One laboratory used a sample size outside of the recommended weight. There were at least 2 variations of enzyme cocktail that were used.