Diagnostic Workup Bias in the Evaluation of a Test

Two studies report markedly divergent results about the usefulness of serum ferritin in diagnosing iron overload in relatives of patients with hereditary hemochromatosis. One study found the sensitivity of elevated serum ferritin to be 0%; another study found a sensitivity of 100%. Although different genetic abnormalities in iron or ferritin metabolism may explain the different results, our examination of these studies suggests that diagnostic workup bias also may explain the difference. In the study reporting a sensitivity of 100%, relatives with normal serum tests may have been excluded from consideration for liver biopsy, thus preventing detection of iron overload. The controversy may provide an empirical illustration of diagnostic workup bias.