Conclusion

Most aggregation methods currently used by laboratories use substitution or censoring methods for non-significant or negative values, and therefore generate biases that can be very significant.

The methodology proposed in this article overcomes these biases. It is based on the following two fundamental rules:

• the accumulation of measurement values must be established on the basis of all individual measurement values, including non-significant values, such as negative values. These values are expressed in units of the measurand, and their associated uncertainty takes into account all sources of uncertainty. Modifying these values, on the pretext that they are insignificant, can only introduce bias into the cumulative result, and distort the assessment of its uncertainty;

• in cumulative...