Screening for fetal Down's syndrome has the peculiarity of combining the biochemical assay of 2 or 3 serum markers with the risk associated with maternal age. If the accuracy of measurement of each parameter is known by the biologist, the uncertainty of the ultimate risk to the patient is not. Indeed, the means of risk calculation involve numerous multi-parameter equations which are not practical for daily use. Defining a re-test limit on thresholds of 1/50 and 1/1,000 is therefore impossible. Since the use of an arbitrarily defined threshold is not being satisfactory, we propose, by default, a methodology based on the exploitation of patient files in the laboratory with risks close to the two decision thresholds. Modulations of the concentrations of all the markers according to their uncertainty allow new risks to be obtained, which can be averaged and framed by an interval of several standard deviations. Choosing the level of uncertainty, the number of files to include, the number of standard deviations framing the average risk, as well as the calculation software, are all choices available to the biologist. The proposed methodology is therefore highly empirical but open, and adaptable, to the specific environment and performance capabilities of each and every laboratory involved.
Keywords: Down's syndrome; maternal serum markers; uncertainty of measurement.