[Normal and pathological laboratory results: how to define "normal" serum iron levels? (author's transl)]

Apart from technical problems, the difficulty in defining "normal" biological laboratory values lies in the choice of reference population and in the way of expressing the results. This is illustrated by a study of 181 healthy, non-anaemic and non-microcytic adults of both sexes (108 men and 73 women) and of comparable age-range. All subjects had venous blood samples taken under the same conditions to determine serum iron levels, total iron binding capacity and percent transferrin saturation (PTS). In 12 of the women these laboratory tests were done before and after one month of supplmentary iron administration. The values obtained in women were significantly lower than in men. The differences, however, were reduced when iron stores, as measured by percent transferrin saturation, were quasi-normal (i.e. PTS greater than 0,16 or 0,19). They disappeared in women who had received supplementary iron. Normal laboratory results are usually expressed as mean values +/- two standard deviations, the range being supposed to include 95% of normal subjects. But this only applies to the values obtained when dispersion is low. When dispersion is high, the percentile method should be preferred, as it expresses the biological findings without reference to an assumed Gaussian distribution.