The variation between reporting values for laboratories using egg test kits (Fig. 2A–D) were smaller than for the milk (Fig. 2E–L) kits. All kits underestimated the “target” concentration of egg in the samples; kit 1 was the least accurate of the egg kits with kit 5 being the most accurate at the 3 mg kg−1 level according to the ISO criteria, reflecting the fact this kit reported 97% of the incurred egg white protein (Table 4). Like the egg kits, there was also a relatively wide spread of data between laboratories for the individual milk (casein) kits SCH727965 price (Fig. 2E–H). Kit 1 was the most accurate of the “casein” kits and the only kit that returned an estimated milk protein content with an acceptable accuracy at the 6 and 15 mg kg−1

levels. “Casein” kit 3 was quite accurate but overestimated the casein concentrations compared with target values. Again this is reflected in the fact that “casein” kit 1 reported 103% and 101% of the incurred protein at the 6 and 15 mg kg−1 levels, over-reporting at the 3 mg kg−1 level and under-reporting at the 30 mg kg−1 level. However, results obtained using kits 1 and 3 were more variable compared to the other kits. Allergen levels reported from analyses undertaken with kits 2, 4 and 5 were less variable but also less accurate and led to underestimation of the concentration of milk in the samples, reporting between 51–62% (kits 2 and 5) and 77–90% (kit 4) of the incurred level of milk powder. Kit 3 consistently

over reported the doses, giving 120–134% of the incurred level of skimmed milk powder. None of the milk (“other”) assays based either on β-lactoglobulin (BLG) AG-014699 chemical structure or on “total” milk detection (Fig.

2I and J) were able to report the target value according to the ISO criteria although kit 6 was the most accurate. This reflects the fact that milk “other” kits reported between 17% and 131% of the incurred protein. All the other kits underestimated the milk concentration in the samples relative to the target value with recoveries of <40%). “Other” milk kit 6 returned the most anomalous data with three laboratories producing inconsistent results. However, the level of variation for this kit across the laboratories Fluorouracil in vivo that produced anomalous data was actually quite low, indicating the analytical basis of the kit is reproducible but errors in implementation are common. Over all the laboratories performed equally well, although two (14 and 19) returned relatively high numbers of outlying data, neither of which participated in the pre-ring trial. A pre-ring trial helps laboratories to become accustomed to working with a new matrix and unfamiliar assay kits, and their value to establish methodology is well accepted (Dumont et al., 2010; Abbott et al., 2010). Variation in the quality of calibration curve data across multiple laboratories using particular kits was observed, and demonstrates the importance of including calibration for each immunoplate used on the day of assay.