Analysis of industry-generated data. Part 1: a baseline for the development of a tool to assist the milk industry in designing sampling plans for controlling aflatoxin M1 in milk.

The presence of aflatoxin M1 (AFM1) in milk was assessed in Italy in the framework of designing a monitoring plan actuated by the milk industry in the period 2005-10. Overall, 21,969 samples were taken from tankers collecting milk from 690 dairy farms. The milk samples were representative of the consignments of co-mingled milk received from multiple (two to six) farms. Systematic, biweekly sampling of consignments involved each of the 121 districts (70 in the North, 17 in the Central and 34 in the South regions of Italy). AFM1 concentration was measured using an enzyme-linked immunoassay method (validated within the range of 5-100 ng kg(-1)) whereas an HPLC method was used for the quantification of levels in the samples that had concentrations higher than 100 ng kg(-1). Process control charts using data collected in three processing plants illustrate, as an example, the seasonal variation of the contamination. The mean concentration of AFM1 was in the range between 11 and 19 ng kg(-1). The 90th and 99th percentile values were 19-34 and 41-91 ng kg(-1), respectively, and values as high as 280 ng kg(-1) were reached in 2008. The number of non-compliant consignments (those with an AFM1 concentration above the statutory limit of 50 ng kg(-1)) varied between 0.3% and 3.1% per year, with peaks in September, after the maize harvest season. The variability between different regions was not significant. The results show that controlling the aflatoxins in feed at farm level was inadequate, consequently screening of raw milk prior to processing was needed. The evaluation of the AFM1 contamination level observed during a long-term period can provide useful data for defining the frequency of sampling.