LETTERS TO ThE EDITOR 131

ceeding the reference range of the population. This can be the case for serum ferritin considering its large reference range even though the within-individual biological variation is lower than other indexes. Another index could be more diagnostically useful even though it presents a higher within-individualbiological van-ance if a smaller relative divergence from the homeostatic point is needed before exceeding the population reference range. Un-fortunately, the authors did not estimate (or at least did not report) the between-subject biological variance so that we can compare it with the within-subject component. As the authors mentioned in their introduction, the decision regarding which indicator to use in determining iron status is influence by a number of factors, including variability, scnsitiv-ity, and specificity of these measurements. This decision cannot solely be based on the biological variability. An indicator with a very small biological variability is of no use if it does not reflect adequately the iron status. However, considering two valid in-dicators of iron status, data on biological variation can help in determining their relative utility in different clinical contexts. When the objective is to serially follow the iron status of a pa-tient, an index with a small biological (and analytical) variability is more useful. For diagnostic purposes, however, the decision depends on the ratio of total within-individual variance (biologi-cal and analytical) to between-subject variance (called the index of individuality) (1). A low ratio indicates that the test is of little diagnostic value if conventional population-based reference val-ucs are used