Statistical analysis used in the nutritional assessment of novel food using the proof of safety

The safety assessment of Novel Food, including GM biotechnology-derived crops, starts with the comparison of the Novel Food with a traditional counterpart that is generally accepted as safe based on a history of human food use. Substantial equivalence is established if no meaningful difference from the conventional counterpart was found, leading to the conclusion that the Novel Food is as safe and nutritious as its traditional counterpart. In general, the non-significance of p value is used for the proof of safety. From a statistical perspective, the problems connected with such an approach are demonstrated, namely that quite different component-specific false negative error rates result. As an alternative, the proof of safety is discussed with the inherently related definition of safety thresholds. Moreover, parametric and non-parametric confidence intervals for the difference and the ratio to control (conventional line) are described in detail. Finally, the treatment of multiple components for a global proof of safety is explained.