Submitted to: Food Control
Publication Type: Peer reviewed journal
Publication Acceptance Date: 12/22/2010
Publication Date: 7/1/2011
Publication URL: naldc.nal.usda.gov/download/49290/PDF
Citation: Lee, K., Armstrong, P.R., Thomasson, A., Sui, R., Casada, M., Herrman, T.J. 2011. Application of binomial and multinomial probability statistics to the sampling design process of a global grain tracing and recall system. Food Control. 22(7):1085-1094. Interpretive Summary: Section 306 of the Bioterrorism Act of 2002 administered by the Food and Drug Administration requires grain to be traced one step forward and backward. For a commercial grain storage facility serving as the first collection point, tracing grain back to the farms of origin and forward to a terminal grain elevator or processor is an essential requirement. The EU General Food Traceability Regulation (EC/178/2002) requires labeling and traceability for food and feed, including biotech grains and grain products, and identifying immediate suppliers or customers of the product. The use of small, coded, pill-sized tracers embedded in grain are proposed as one method for grain traceability. There is a trade-off though in a tracer system between tracer concentration and the amount of sampling required to obtain confident identification. As such, a statistical sampling process for traceability was designed and tested using a science-based sampling approach with the goal of accurately identifying grain in mixed lots. Sampling tests were conducted at predefined sampling points in a commercial-scale grain facility during simulated grain transportation and storage. Five lots of grain with different coded tracers were used. Conclusions of this study showed statistical predictions of grain mixing from sampling and known conditions were similar. Insignificant segregation of tracers in bin and truck operations was also observed. The sampling process was proven to be effective and provides assurance for accurately identifying grain origin.
Technical Abstract: Small, coded, pill-sized tracers embedded in grain are proposed as a method for grain traceability. A sampling process for a grain traceability system was designed and investigated by applying probability statistics using a science-based sampling approach to collect an adequate number of tracers for grain lot identification. For verification of the designed sampling process, a probability sampling test was conducted at predefined sampling points in a commercial scale grain facility while simulating grain transportation and storage in a grain supply chain. In general, the statistical results and observations showed similar concentration and insignificant segregation of tracers in bin and truck operations. For binomial probability tests to identify single-source grain, most of the tracer concentrations found in grain containers fell within confidence intervals estimated by statistical methods. For multinomial probability sampling tests for identification of more than two sources of grain, truck sampling successfully identified all five grain sources, but all bin samplings were not successful for the same sample size. The sampling process based on probability statistics was empirically proven to be applicable and provides better scientific assurance of accurately identifying grain origin. This would reduce economic risks and costs due to inappropriate sampling in the proposed traceability system.